EPA - 3O3 -R-95-OOf-
           .               	j

Water Quality Functions of
Riparian Forest Buffer Systems in the
Chesapeake Bay Watershed
            Prepared by the
           Nutrient Subcommittee
               of the
          Chesapeake Bay Program

             EPA 903-R-95-004
             CBP/TRS 134/95
              AUGUST 1995


     Water Quality Functions of
   Riparian Forest Buffer Systems
                   in the
    Chesapeake Bay Watershed
                 A Report of the
              Nutrient Subcommittee
           of the Chesapeake Bay Program

                  August 1995
Printed by the U.S. Environmental Protection Agency for the Cftesapeake Bay Program
 Recycled/Recyclable • Printed with Vegetable Oil Based Inks on 100% Recycled Paper (50% Postconsumer) <


Principal Authors
                      - RICHARD LOWRANCE              ;
             USDA-Agricultural Research Service, Tifton, GA

                           LEES. ALTER
             USDA-Agricultural Research Service, Tifton, GA

                        J. DENIS NEWBOLD
              Stroud Water Research Center, Avondale, PA

                      RONALD R. SCHNABEL
         USDA-Agricultural Research Service,-University Park,  PA

                       PETER M.GROFFMAN
              Institute for Ecosystem Studies, Millbrook, NY   .

        ;                JUDITH M. DENVER
                   U.S. Geological Survey, Dover, DE

                        DAVID LCORRELL
       Smithsonian Environmental Research Center, Edgewater, MD

                       J. WENDELL GILLIAM
              North Carolina State University, Raleigh, NC

                       JAMES L ROBINSON
       USDA-National Resources Conservation Service, Ft. Worth; TX

                      RUSSELL B. BRINSFIELD
      University of Maryland, Wye Research Center, Queenstown, MD

                       KENNETH W.STAVER   ,
      University of Maryland, Wye Research Center, Queenstown, MD

               --• :       WILLIAM LUCAS
          Integrated Land Management Consulting, Malvern, PA

                         ALBERT H.TODD     '             :
      USDA Forest Service, Chesapeake Bay Program, Annapolis, MD


   This document is a research synthesis requested
by the, Forestry Work Group of the Nutrient Sub-
committee of the Chesapeake Bay Program. In devel-
oping the outline for the report, the authors agreed to
specifically focus on the existing Riparian Forest
Buffer System (RFBS) specification developed by
USDA and being used as a starting point for federal,
state, 'and local RFBS specifications. Although the
report contains a general review of riparian forest and
grass vegetated filter- strip literature, the goal was to
use this literature to help determine the applicability
of the forest buffer system recommended by USDA.
   The strategy for development ^of the document was
to bring together researchers in this field to: 1) dis-
cuss the current state  of knowledge of RFBS;
2) determine how that knowledge related to the
Chesapeake Bay Watershed; and 3) reach consensus
about the functions of RFBS in the Bay watershed
based on that current state of knowledge. The con-
sensus statements are very important but they do riot
ensure specific functions will result from RFBS, in a
given field setting.  Rather, they are Best Professional
 Judgements of the entire authors group and represent
 general agreement among the authors about the cur-
 rent validity of the statements. In addition to the
 authors, a large number of reviewers were asked to
. examine the report and  form their own judgements
 about the general conclusions. These reviewers,
 acknowledged below, generally agree'with the con-
 sensus statements contained in the report.
   As readers of this report will see, numerous scien-
 tific questions remain about the role of RFBS in all
 of the physiographic and land use settings of the Bay
 watershed.  Yet, incomplete scientific knowledge can
 not be used to avoid making informed management
 judgements, especially when society has. determined
 that a globally important natural resource such as
 Chesapeake Bay must be restored to ecological
 health.  The scientists involved with the preparation
 of this report have attempted to make the best judge-
,ments possible "to help guide the application of RFBS
 to improve  water quality in the Chesapeake Bay
 Watershed and ultimately in the Bay itself.


 Ackn owledg e m e n ts
   The authors are especially grateful to Mr. Spencer
 Waller, who provided logistical .support for.our meet-;
 ings and provided detailed minutes, of the discus- •
 sions. In particular, Spencer transcribed about 25
 hours of discussions which included our process of
 reaching consensus on the- applicability of RFBS in
 different physiographic settings.  Without these direct
 transcripts, we would not- have been able to ade-
 quately recapture these statements for the report.
   The authors also wish to thank the reviewers of the
 report. Most of them spent a large amount of time on
 their review and their .comments  were very helpful in
 completing the final draft. -The list of reviewers
 includes many scientists and natural resource man-
 agers active in the Chesapeake Bay Watershed. The .
 reviewers were Andrew Dolloff (USDA-FS), Richard
 Everett (U.S. Fish and Wildlife Service), Verna
 Harrison (Maryland Dept. of Natural Resources),.
 Thomas Jordan (Smithsonian Environmental
 Research Center), Larry Lubbers (Maryland Dept. of
Natural Resources), Kent Mountford (U.S. En-
vironmental'Protection Agency), Adel S^hirmo^
hammadi (University  of Maryland), George
 Simmons (Virginia Polytechnic Institute & State
University), Thomas  Simpson (Maryland Dept. of
Agriculture), Bernard Sweeny (Stroud Water
Research Center), and Donald Weller (Smithsonian'
Environmental Research Center).
   A number of people participated in at least one of
 our two meetings but are not-co-authors of the report.
 Among these people who provided useful input at the
 meetings were Ed_Corbett (USDA-Forest  Service),
 Rupert, Friday (Chesapeake Bay Foundation), Bob
 Merrill (Pennsylvania Bureau of Forestry), Kent
 Mountford (U.S. Environmental Proteetion  Agency),
 Ann Swanson (Chesapeake Bay Corfmiission), Bob
 Tjaden (Univ, of Maryland), and Dave Welsch
   As with any undertaking of this sort, much of the
 work was done by people who get little credit. Ms.
 Olive Sides, Office Automation Assistant with
 USDA-ARS, Tifton, GA helped with much of the
 correspondence and arrangements for th& two meet- .
 ings held to develop the report. Ms. Dalma Dickens,
 Secretary with USDA-ARS, Tifton, typed numerous
 versions of the report.  -Mr. H. L. Batten, USDA-
 ARS, Tifton, Ms. Wendy R. Pierce, USDA-NRCS,
 Ft."Worth, Texas, and Mr. Anthony J. Kimmit,
 USDA-NRCS, Ft. Worth, Texas prepared many of the
    Funding for this, report was provided by the
 Environmental Protection Agency, Chesapeake Bay
 Program,  and USDA Forest Service, Northeastern
Area, State & Private Forestry.      '


LIST OF TABLES AND FIGURES ... ........., :	!..........	 xl

EXECUTIVE SUMMARY	...:................. . .	 .  xiii

SECTION I. Water Quality Functions of Riparian Forest Buffer Systems	.......	...  1
      ;      A. Introduction .......... .-".'	  1
            B. Nonpoint Source Pollution Control Relative to Nutrient Load Reduction Strategies  	.  1
   :    .     C. Watershed Approaches to Nonpoint Pollution Estimation and Abatement	4
            D. Historical Overview of Scientific Interest in Riparian Ecosystems	:. . .  5
            E. Research Background for the Riparian Forest Buffer System Specification	5
            F. Current Understanding of RFBS Functions '....'.,...	•„.."	.".'...	   7
              1. Zone 1 - Control of the  Stream Environment,	'.	   7
               a. Temperature and Light ;	'.......	   8
               b. Habitat Diversity and  Channel Morphology	....'.	'.,.','.,*	   9
               c. Food Webs and Species Diversity	  10
             2. Zone 2—Removal of Nonpoint Source Pollutants ... .	 .......-,'., .  11
               a. Nitrate Removal	 1.,.,..	  11
  ,             b. Plant Uptake			;.'...;........' ...  12
               c. Microbial Processes	............;..........'.	'...'...-	 14
               d. Removal of Surface-borne Pollutants	......'............:.....  15
       •  , •   .3. Zone S^Sediment Removal and Spreading of Surface Runoff	  1,6
             4. Integrated Water Quality Functions of the Three-Zone Buffer System	17

SECTION II. Riparian Forest Buffer Systems in Physiographic Provinces  of the
            Chesapeake Bay Watersheds  .,.....:			    19
            A. Coastal Plain	  19
     j  • •       1. General Land Use and Hydrology	 ,19
                a. Inner Coastal Plain	,	  19
     .  ' - .       b. Well-Drained Upland	:'.	.........'	  20
                c, Poorly-Drained Upland ,	  20
         •   .   .. d'. Surficial Confined .  .'.  . . ..:	'.. . .	  20.
              2. Control of Nonpoint Source Pollutants	.'.	  22
           :    a. Nutrient Budgets for Riparian Forests	  23
               b. Removal of Nitrate from Groundwater	.".,	  24
               c. Nutrient Removal Processes	 ..... . ........  26
            ,   d. Removal of Sediments arid Nutrients from Surface Runoff	 .^ ...........  28
               3. Conclusions	  31

                        B. Piedmont	.:.....	   31'
                          1. General Land Use andHydrology	   31
                          2. Control of Nonppint Source Pollutants	   32
                             a. Removal of Nitrate from Groundwater	 . .*,  32
                             b. Removal of Sediment and Nutrients in Surface Runoff	 . .   33
                          3. Conclusions	   34
                        C. Valley and Ridge ..'	::.	•   35
                          1. General Land Use and Hydrology	   35
                          2. Control of Nonpoint Source Pollution	 .   35
                             a. Removal of Nitrate from Groundwater	•	   36
                             b. Streamflow Transport of Phosphorus	   37
                             c. Removal of Sediment and Nutrients in Surface Runoff	   38
                          3, RFBS in Forested Watersheds  ..		1	   38
                          4. Conclusions	   39

            SECTION III.  Applicability of the Three Zone Riparian Buffer System	   41
                          A. Control of the Stream Environment	   41
                          B. Control of Nonpoint Source Pollution	.....'.....	 .   41
                             1. CoastalPlain	.......;	 .., . .   42
                               a. Inner Coastal	   42
                               b. Outer Coastal Plain		,		• •   46
                                 1) Well Drained Upland .		   46
                                2) Poorly Drained Upland/Surficial Confined	   47
                               c. Tidally Influenced	   47
                             2. Piedmont	   49
                             3. Valley and Ridge	   49
                          C. Loading Rates and Nonpoint Source Pollution Control 	.......:	 .   52
                          D. Stream Order/Size	   52
                          E. Establishment and-Sustainability	   53

            SECTION IV.  Research Needs	   55

            LITERATURE CITED	   57

 Tables  and   Figures

 1. Land use in physiographic regions Of Chesapeake Bay Watershed	.2
 2. Total nitrogen, nitrate-nitrogen, and total phosphorus budgets for riparian forest      "
   ecosystems in the Coastal Plain	;	 23
 3. Above-ground woody vegetation uptake of N and Pin Coastal Plain riparian forests . .. :	26
 4. Sediment deposition in Coastal Plain riparian forests ..... ~	28
 5. Inputs, outputs, and % removals of sediment, total N, and total P from experimental
   Ky 31-Fescue vegetated filter strips in Maryland Coastal Plain .........		29
' 6. Effects, of different size buffer zones on reductions pf.sediment and nutrients from field surface runoff . 30
, 7. Removal of nitrate from groundwater—summary.	.......:.. -43
 8. Removal of phosphorus from.all sources—summary	......:	 ...	44
 9. Removal of sediment and sediment-borne pollutants—summary ....:.........	 .^.... 45


   1. Physiographic regions of the Chesapeake Bay Watershed ..'.....'	3
  2.- Schematic of the three zone Riparian Forest Buffer System ............:.............,........ 6
  3. Hydrogeomorphic regions of the Delmarva Peninsula	 21
  4. Nitrate concentrations in groundwater beneath riparian forests from five Coastal Plain sites .	 25
  5 , Conceptual model of below ground processes affecting groundwater nutrients in riparian forest  ..... 27
  6. Inner Coastal Plain	..'..;	 t.-......'. .......'-.:..-	 .......... 46
  7. Outer Coastal Plain - Well-Drained Upland	 ....	.... . .	 46
  8. Outer Coastal Plain - Poorly Drained Upland/Surficial Confined	 .... . 48
  9. Coastal Plain-Tidal Influenced flow system .	 s .'	48
 10. Piedmont (thin soils) ..../.	...,.'	.......:............	50
 11. Piedmont (schist/gneiss bedrock) ...."		 50
 12. Piedmont (marble bedrock)/Valley and Ridge (limestone bedrock)  . ..	,	 '50
 13. Valley and Ridge (sandstone/shale bedrock) 	,	'.	.-.,.'.... 51
 14. Valley and Ridge/Appalachian (low order streams)	:'... 1 .................. 51


 Executive  Summary
  Riparian (streamside) forests are known to reduce
delivery of nonpoint source pollution to streams and
lakes in many types of watersheds. In addition, ripar-
ian forests are known to be important in controlling
the physical and chemical environment of streams
and in providing detritus and woody debris for
streams and near-shore areas of water bodies.
Riparian forests were the original native vegetation in
most  streamside areas  of the .Chesapeake Bay
Watershed. This report assesses the state of scientific
knowledge concerning the Water Quality Functions
of riparian ecosystems.  This, assessment and specific
knowledge of riparian ecosys'tem function in physio-
graphic regions  of the Chesapeake Bay Watershed
were used to make consensus Best Professional
Judgements as to the potential water quality func-
tions of Riparian Forest Buffer Systems (RFBS) in
the Bay Watershed.       ,
  Research conducted in naturally occurring riparian
forests and experimental and on-farm grass filter
strips has been used by the U.S. Department of
Agriculture to develop a general "Riparian Forest
Buffer System specification" for controlling nonpoint
source pollution from agriculture and improving gen-
eral water quality. The specification calls for a three
zone buffer system, with  each zone having  specific
purposes but also having  interactions with the adja-
cent zones to provide  the overall RFBS function.
Zone 1 of the RFBS is an area of permanent forest
vegetation immediately adjacent to the stream chan-
nel and encompassing at least the entire stream chan-
nel system. Zone 2 is an area of managed forest, ups-
lope' from Zone 1. .Zone 2 is managed for control of
pollutants in subsurface flow and surface runoff
through biological and chemical transformations,
storage in woody vegetation,  infiltration, and sedi-,
ment deposition.  Zone 3  is a grass or other herba-
ceous filter strip upslope from Zone 2. Zone 3 is
managed to provide spreading of concentrated flow
into sheet flow and to remove sediment and sediment
associated pollutants.                    '...,'•.'
   The most general  function of Riparian Forest
Buffer Systems is, to provide control of the stream
environment. These functions include modifying
stream temperature, and controlling light quantity and
quality; enhancing habitat diversity; modifying chan-.
nel morphology; and enhancing food webs and
species richness. All of these factors are important to
the ecological health of a stream and are best pro-
vided by a RFBS which includes a Zone 1 that
approximates the original native vegetation. These
functions occur along  smaller streams, regardless of
physiographic region. These functions are  most
important on smaller streams,  although  they are
important for bank and near-shore habitat on larger
streams and the shoreline of tlie Bay. RFBS
contribute to bank stability and thus minimize sedi-
ment loading due to  instream bank erosion.
Depending on bank stability,-and soil .conditions in
Zone 1, management of Zone 2  for long-term rota-
tions may be necessary for sustainability of stream
environment functions  of Zone 1.
   The next most general function of RFBS is control
of sediment and sediment-borne pollutants carried in.
surface runoff. Properly managed RFBS should pro-
vide a high level of control of sediment and sediment
borne chemicals regardless of physiographic region.
Natural riparian forest studies indicate that forests
are particularly effective in filtering fine sediments
and promoting co-deposition of sediment  as water
infiltrates. The slope of the RFBS is the main factor
limiting the effectiveness of the sediment removal
function. In all physiographic settings it is important
to convert concentrated flow to sheet flow in "order to
optimize RFBS function. Conversion to sheet flow
arid deposition of coarse sediment which could dam-
age young vegetation  are the primary functions of
Zone 3-—the grass vegetated filter strip.

  The next most general function of RFBS is to con-
trol nitrate in shallow groundwater moving toward
streams. When groundwater moves in short, shallow
flow paths, such as in the Inner Coastal Plain (pri-
marily the western shore), 90% of the nitrate input
may be removed. In contrast, nitrate removal may be
minimal in areas where water moves to  regional
groundwater such as in Piedmont and Valley and
Ridge areas  with marble or limestone bedrock,
respectively. In these and some Outer Coastal Plain
regions,  high nitrate groundwater may emerge in
stream channels and bypass most of the RFBS.  In the
areas  where this occurs or where high nitrate  water
moves out in seepage faces, deeply rooted trees in'
Zone 1 or in seepage areas will be  essential. The
degree to which nitrate (or other groundwater pollu-
tants) will be removed in the RFBS depends on the
proportion of groundwater moving in or near the bio-
logically active root zone and on the  residence time
of the groundwater in these biologically'active areas.
  The least general function of RFBS appears to be
control of dissolved phosphorus in surface  runoff or
shallow groundwater. Control of sediment-borne P is
generally effective. In certain situations, dissolved P
can contribute a substantial amount of total P load.
Most of the soluble P is bioavailable, so the potential
impact of a unit of dissolved P on aquatic ecosystems
is greater. It appears that natural riparian forests have
very low net  dissolved P retention. In managing for
increased P retention, effective fine sediment control
should be coupled with use of vegetation which can
increase P uptake into plant tissue:
   Research on functions of natural, restored, and
enhanced RFBS  is needed in all portions of the
Chesapeake Bay Watershed. Research should be
directed into four general areas:  1) assessment of
existing riparian forests relative to the RFBS stan-
dard; 2) assessment of potential RFBS restoration for
NPS pollution control; 3) assessment of NFS pollu-
tion control in pilot restoration and enhancement
projects; 4) determine the effects of management fac-
tors on both pollution control and control of the
stream environment. The research,  because of the
need to do relatively large scale projects which last
for substantial periods of time, should be coordinated
with demonstration restoration/ enhancement pro-
jects. Some of the major research questions should
address the uncertainty associated with the functions
discussed above. Research should be directed toward
testing the hypotheses concerning which functions of
RFBS occur in specific physiographic settings and
the specific management conditions under which
these functions are likely to be enhanced. In'particu-
lar, research on the time to  recovery of RFBS  func-
tions and the processes which control the various
functions should  be integrated into demonstration

             Water Quality  Functions  of
             Riparian  Forest  Buffer Systems

  Riparian Forest Buffer  Systems (RFBS) are
streamside ecosystems, managed for the enhancement
of water quality through control of noripoirit source
pollution (NFS) and protection of the stream environ-
ment. The use of riparian management zones is rela-
tively well established as a Best Management Practice
(BMP) for water  quality improvement (in forestry
practices (Comerford et al., 1992), but has been much
less widely applied as a BMP in agricultural areas or
in urban or suburban settings. RFBS are  especially
important on small streams where intense interaction
between terrestrial and aquatic  ecosystems occurs.
First and second order streams comprise nearly three-
quarters of the total stream length in the United States
(Leopold et al., 1964). Fluvial activities influence the
composition of riparian plant communities along
these small streams (Gregory et al., 1991). Likewise,
terrestrial disturbances can have an immediate impact
on aquatic populations (Sweeney, 1993; Webster et
al., 1992). Small streams can be completely covered
by the canopies of streamside vegetation (Sweeney,
1992). Riparian vegetation has well-known beneficial
effects on the bank stability, biological diversity, and
water temperature's ;of streams (Karr  and  Schlosser,
1978). Riparian forests of mature trees (30 to 75 yrs.
old) are known to effectively reduce nonpoint pollu-
tion from agricultural fields (Lowrance et al., 1985b).
  Compared to other NFS pollution control mea-
sures, RFBS can lead to longer-term  changes in the
structure and function of agricultural  landscapes. To
produce- long-term improvements in  water quality,
RFBS must be designed with,an understanding of: 1)
the processes which remove or sequester  pollutants
entering the riparian buffer system; 2) the effects of ri-
parian management practices on pollutant, retention;.
3) the effects  of riparian forest buffers on aquatic
ecosystems; 4) the time to recovery after harvest of
trees, or reestablishment of riparian buffer systems;
and 5) the effects of underlying soil and geologic ma-
terials  on chemical,  hydrological, and biological
processes.      .  '
  . This report examines the scientific basis for apply-
ing the existing RFBS specification as an agricultural
Best Management Practice (BMP) in the. different
physiographic  provinces of the Chesapeake Bay
Watershed (CBW, -Table 1 and Figure 1). The report
briefly reviews NPS  pollution problems in the Bay
Watersheds and approaches to NPS pollution control
(Sections I. B & C); the scientific foundation for the
Riparian Forest Buffer System specification (Sections
I. D & E); and the water quality functions of each of
the three zones of the RFBS (Section I.  F). Included
is a review of the existing research on RFBS in dif-
ferent physiographic  provinces  that comprise the
CBW (Section II). Based on these results, the effec-
tiveness of RFBS for NPS pollution control is charac-
terized in different  parts of the Bay watershed
(Section III). Finally,  research needs are  discussed in
Section IV. RFBS are one of many factors that influ-
ence water quality and stream health. A complex suite
of interrelated functions and mechanisms contribute
to water quality and physical habitat parameters of the
aquatic ecosystem. Other important factors, outside
the scope of this report, that may affect the function-
ing of RFBS, and should be considered in their de-
sign, include: the type and intensity of land use in the
watershed;- the effectiveness of stormwater manage-
ment; streambank and streambed stability; and stream
uses (recreation, water supply, etc.).
  Nonpoint source pollution is the major cause of
surface water impairment in the United States (Baker,
1992; Long, 1991) and has been addressed as a na-
tional priority since passage of the Clean Water Act
(CWA), Section 319, which requires "that programs
for the control of nonpoint sources of pollution be de-

                                             TABLE 1
     Land use in physiographic regions of Chesapeake Bay Watershed (NCRI Chesapeake, 1982).

Valley & Ridge
Coastal Plain
% of TOTAL

, 659,700

. 1,607,900
.............ha—- ~ --

• 2,377,800


Rgure 1 shows generalized physiographic regions.
veloped and implemented." The effectiveness of the
RFBS is likely to be judged by their NFS pollution
control effectiveness.
  Although assessments are incomplete and do not
include all states, estimates are that about 30% of US
waters are impaired—i.e. they do not fully support
their designated uses (USEPA, 1990a). Of impaired
waters, about two-thirds of the problems are primarily
from NFS pollution  (USEPA, 1986). The nonpoint
sources of pollution vary, but agriculture is the major
contributor for rivers and lakes. Besides agriculture,
the other major contributors  of NFS pollution are
urban areas, mining, atmospheric deposition, and nat-
ural origins. Nutrients and sediments are still the prin-
cipal sources of surface water impairment (USEPA,
1986; USEPA, 1990a; USEPA, 1990b). Sediments are
the most important cause of impairment for rivers,
and nutrients are the most important cause of impair-
ment for estuaries. Pesticides, metals, and priority
pollutants are identified as problems in less than 20%
of the assessed waters. The extent of contamination,
especially for pesticides, may be underestimated.
  The earliest assessments of Chesapeake Bay water
quality in the 1980's identified non-point source pol-
lution as a major cause for water quality impairment
in the Bay (Correll, 1987; Chesapeake Bay Program,
1991). Reduction of NPS pollution has been a signif-
icant part of the strategy to  improve water quality in
Chesapeake Bay since that time. The main problems
were identified as nutrient enrichment, high levels of
toxic substances,  and excessive sediment  loads.
Effective control of all these types of pollution, espe-
cially nutrients and sediments, requires a watershed
based program for NPS pollution control.
   Improvement and maintenance of water quality is
the single most important component of the overall
protection and restoration plan established in the 1987
Chesapeake  Bay  Agreement  (Chesapeake Bay
Program, 1991). One of the most ambitious goals of
the 1987 and 1992 agreements is to reduce nutrient
loadings to the Bay by 40% by the year 2000 and to
retain this level as a permanent cap on nutrient levels.
Strategies for nutrient load reduction require control
of both point and nonpoint sources of  pollution.
Based on 1985 land uses and results from a Watershed
Model (Donigian etal., 1990), nonpoint sources dom-
inate both N (53% of total) and P (68% of total) loads
to the mainstem of the Bay (Chesapeake Bay Pro-
gram, 1991,  L.  Shu'yler, personal communication,
1995). The Watershed Model has been used to esti-
mate edge-of-stream  nitrogen and phosphorus load-
ings from various land uses  in the Bay watersheds.
Agriculture (including conventional cropland, conser-
vation cropland, pasture, and animal waste facilities)
accounted for 69% ,of total N and 79% of total P in
NPS pollution.
   Of the entire loadings of N and P to the mainstem
of the Bay (point plus NPS), 44% of the N and 50%
of the P came from agricultural nonpoint sources (L.
Shuyler,   personal  communication,  1995).  The


      Coastal Plain'

      Southern  Piedmont

      Northern .Piedmont

Susquehanna Basin (Pennsylvania, New York) and the
Eastern Shore (Delaware, Maryland, Virginia) con-
tribute the highest NFS loads of N and P. Loads from
these two regions were  dominated by agricultural
sources. In the Susquehanna, 74% of NPS  N loads
were from agriculture. Nonpoint source loads of N,
from the Eastern Shore were 81%  agriculturally re-
lated (Chesapeake Bay Program, 1991).
  Until 1990, approaches for NPS pollution control
in the Bay watersheds were largely focused on con-
trolling upland sediment  and sediment-borne pollu-
tants (Chesapeake Bay Program,  1990). These tradi-
tional  approaches were  a combination  of source
reduction (i.e. reduce erosion rates  in fields) and en-
gineered buffer systems or structural BMPs such as
ponds,  sediment detention  basins, terraces,  grass
water ways  and vegetated filter strips. In 1990, the
Chesapeake   Bay  Program's  Nonpoint  Source
Evaluation Panel recommended a systems approach
for nutrient  load reduction with regional and water-
shed management  strategies  based on  watershed
mass-balances (Chesapeake Bay  Program, 1990). A
systems approach for NPS pollution reduction will in-
clude structural BMPs and source load reductions, as
well as approaches which seek to integrate the man-
agement and restoration of landscape features which
retain pollutants through a combination of ecosystem
processes. Examples of these pollutant sinks include
natural wetlands, constructed wetlands, and riparian
forest  buffer  systems  (Fields, 1992). As pollutant
sinks increase in complexity from simple  physical
structures to diverse natural ecosystems, both the im-
portance and  difficulty of understanding processes
which sequester or remove pollutants also increase.


   Risk assessment and source reduction are new ap-
proaches for NPS pollution control (Baker, 1992). A
high percentage of total  pollutant  loadings  in some
watersheds comes from a relatively small portion of
the watershed area because of improper management
of sources, improper siting of facilities, problematic
environmental and site conditions,  or a combination
of  these  factors. Watershed scale risk assessment
seeks  to  identify  and reduce loadings from  areas
which contribute large amounts of NPS pollution.
  'Concurrent with identification of problem  areas
comes the opportunity for source reduction. Source
reduction has been responsible for  some of the more
impressive successes of NPS pollution reduction, in-
cluding the reduction of loadings of lead from auto-
mobile emissions and of prgand-chlorine pesticides
(Baker, 1992). Source reduction should be linked with
watershed-scale risk assessments because the poten-
tial for source reduction may be greatest (and proba-
bly most economical) in areas which are generating
highest unit area loadings. The linkage of risk assess-
ment and source reduction will depend on interacting
factors such as type of pollutant (e.g. purchased input
vs. by-product), reason for high risk (e.g. poor man-
agement, siting  of facilities, inherent regional risks),
and availability of alternative practices'and/or sites.
   Even when risk assessment and  source reduction
strategies lead to load reductions under average con-
ditions, a third aspect of watershed management  -
maintenance and restoration of buffer systems  be-
tween terrestrial and aquatic ecosystems - is neces-
sary to reduce the contributions of extreme events to
NPS pollutant loads.  Under the best of conditions,
source reduction will likely leave watersheds vulner-
able to extreme events, including both weather  ex-
tremes as well  as economically generated extremes
(e.g. intensification of pollutidn generating  produc-
tion practices). Watershed studies have  demonstrated.
the importance of extreme events to water and pollu-
tant transport. Extreme events within a year dominate
annual totals and wet years within multi-year cycles
dominate long-term loadings (Jaworski et al., 1992;
Lowrance and Leonard, 1988; Magnien et al., 1992).
Control of NPS pollution from extreme events will re-
quire integrating risk assessment and source reduction
approaches with buffer systems as landscape scale
"insurance policies."
   Buffer systems are also important components of
watershed NPS pollutant control efforts because of
•the limitations of other BMPs for NPS pollution con-
trol. For example, Hall (1992) monitored changes in
groundwater nitrate  (NOs—N) concentrations  be-
neath  heavily  fertilized and  manured fields in
Lancaster County, PA following the implementation
of "input management" techniques.  Fertilizer/manure
inputs were decreased from 39 to 67% (222 to 423 kg
ha'1) but groundwater nitrate, changed by -12 to 50%.
By the end of the. study, nitrate concentrations in
groundwater still exceeded federal drinking water
standards.  Shirmohammadi et  al.  (1991) used the
CREAMS simulation model to evaluate the effects of
seven different BMPs on groundwater nitrate concen-
trations beneath  cropping  systems on  the  eastern
shore of Maryland. Although CREAMS does not pro-

vide absolute predictions, none of the BMPs'were
predicted,to  reduce groundwater nitrate concentra-
tions to less than the federal drinking water standard.
Under appropriate conditions, described in this report,
RFBS are likely;to  be an  important component
of NPS pollution control when in-field BMPs are in-
adequate.             ,


   Most of the knowledge of riparian ecosystem ef-
fects on water quality comes from research conducted
since,1975. Two publications in 1978 galvanized sci-
entitle and management interest in riparian  ecosys-
tems. Karr  and, Schlosser (1978) concluded  that
stream environments are largely controlled by adja-
cent riparian ecosystems and provided ah overview of
relationships  between  water, resources and riparian
ecosystems (the  land-water interface). Johnson and
McCormick (1978) edited the proceedings of a sym-
posium which included 55 reports on various aspects
of riparian research^ management, and policy; While
the symposium proceedings contained  excellent dis-
cussions of the late 1970's state-of-knowledge con-
cerning riparian ecosystems and other types  of wet-
lands (Brown et al.,  1978; Wharton,  and Bririson,
1978) only one paper (Mitsch, 1978)  dealt  specifi-
cally and quantitatively with the water quality func-
tions of a riparian  ecosystem. The proceedings also
included a review of the general water quality func-
tions of wetlands (Kibby 1978) in which a number of
publications on nutrient cycling in riparian and other
wetlands were cited Only a few of the citations dealt
specifically with water quality effects of riparian
ecosystems (Kitchens et al., 1975; Lee et  al., 1975;
Kuenzler  et  al.,  1977; Richardson et al.,  1978).
Although the 1978 symposium contained numerous
claims about  the water quality functions of riparian
ecosystems, few data were presented.
  In the late 1970's a number of research projects
began to develop a more quantitative understanding of
the role played by riparian ecosystems in controlling
NPS pollution by sediment and nutrients in agricul-
tural  watersheds  (Jacobs and  .Gilliam,  1985b,
Lowrance efal., 1983; Peterjohn and Correll, 1984).
These studies were. primarily in .the Coastal Plain'
physiographic province of the Eastern U.S., where the,
typical land-use pattern is intensive row-crop agricul-
ture in upland areas with riparian forests along low-
order streams. These early studies shared at least two
otherlmportant characteristics: 1) a relatively shallow
aquiclude which forced most infiltrated water to move
laterally toward streams and pass through or near the
riparian forest root zone and 2) naturally regenerated
forests typical of the region rather, than'forests man-
aged specifically for water quality functions. These
studies focused on riparian processes related to nutri-
ents and sediment  with little or no attention to the
fates of other pollutants or to the effects of riparian
areas on the physical or trophic status of the stream.
   As interest in the nonpoint pollution control value
of riparian ecosystems increased, recognition of their
importance to the  physical  and  trophic status of
streams also developed.'  Karr and Schlosser (1978)
quantified the effects of riparian vegetation on sun-
light  penetration   and  temperature  of  streams.
Research in the 1980s confirmed the importance of
large woody debris and leaf litter inputs to the habitat
and trpphic status of most small  streams (Meyer and
O'Hop, 1983; Benke  et al., 1985; Harmon et al.,
1986). By 1987,  it was well established that woody
debris derived from riparian forests played an impor-
tant role in controlling channel morphology, the stor-
age and routing of organic matter and sediment, and
the amount and quality of fish habitat (Bisson et al.,


.   By the late 1980s, there was a clear need to syn-
thesize the existing knowledge into management rec-
ommendations for  the establishment, maintenance,
and management of riparian ecosystems for a broad
range of water quality functions (Lowrance, 1991). In
1991, the United States Department of Agriculture-
Forest  Service (USDA-FS)  with assistance from
USDA-Agricultural. Research" Service, USDA-Sqil
Conservation Service, Stroud Water Research Center,
PA, Pennsylvania Dept. of Environmental Resources,
Maryland Dept. of Natural Resources, and U.S. Dept.
of Interior Fish and Wildlife Service developed draft
guidelines for riparian  forest buffers.  This effort re-
sulted in a booklet entitled "Riparian Forest Buffers -
Function and Design for Protection and Enhancement
of Water Resources" (Welsch, 1991) which specified
a riparian buffer system consisting of three zones
(Figure 2).             '   -   ,                 :
   Zone 1 is permanent woody vegetation  immedi-
ately adjacent to the stream bank. Zone 2 is managed
forest occupying a strip^upslope from Zone 1. Zone 3

                  Runoff Control
   ZONE 2
Managed Forest
Undisturbed Forest
FIGURE 2. Schematic of the three zone Riparian Forest Buffer System^
is an herbaceous filter strip upslope from Zone 2. The
specification applies to areas where cropland, grass-
lands, and/or pasture are adjacent to riparian areas on
a) permanent or intermittent streams, b) margins of
lakes and ponds, c) margins of wetlands, or d) mar-
gins of groundwater recharge areas such as sinkholes.
Although referred to as a riparian forest buffer, inclu-
sion of the non-forested herbaceous strip as Zone 3
suggests that a more correct name would be "Riparian
Forest Buffer System".
    The primary purposes of Zone 3 of the RFBS are
to remove sediment from surface runoff and to con-
vert channelized flow to sheet flow.  The primary
function of Zone 2 is to block transport of sediment
and chemicals  from upland areas into the  adjacent
wetland or aquatic ecosystem. Vegetation and litter in
these zones forms a mechanical barrier to sediment
transport. Plant roots take up chemicals that become
sequestered in growing biomass. Vegetation also pro-
duces  organic matter that fosters chemical  and bio-
logical processes that immobilize or transform pollu-
tants. Although most Zone 2 functions also' occur in
Zone 1, the primary purpose of Zone 1 is to maintain
the integrity of the stream bank and a favorable habi-
tat for aquatic organisms.  Shade and litterfall  pro-
                vided by streamside vegetation has a direct influence
                on water temperature and dissolved chemicals.
                   The USDA-FS report  and specification  were
                based on a synthesis of literature existing  through
                1989 and on in-depth discussions with scientists and
                managers working on various riparian ecosystems
                (Welsch,  1991). Some of the generalizations which
                guided the design of the RFBS were based on studies
                of nutrient sequestering and nutrient transformations
                in agricultural watersheds (Correll, 1983; Lowrance
                et al., 1985; Yates  and Sheridan, 1983).-These water-
                shed-scale studies  indicated that riparian forests were
                important nutrient and sediment sinks in agricultural
                watersheds, but provided little or no guidance,on how
                to design an effective RFBS.  Process studies in these
                and other systems provided most of the original de-
                sign  guidance.  Several studies on nitrate  removal
                from shallow groundwater in riparian forest buffers
                found that  most reduction  in  nitrate concentration
                takes  place within the first 10 to 15  m of  forest
                (Lowrance et al., 1984a, Peterjohn and Correll, 1984,
                Jacobs and Gilliam,  1985b) and that the necessary-
                width for shallow  groundwater nitrate removal could
                be relatively  short. Although effective in  reducing
                sediment arid sediment borne chemical concentrations

 in sheet flow (Peterjohn and Correll, 1984), it was
 known that  channelized flow can  bypass riparian
 forests. To control channelized flow into a 'riparian
 forest, a herbaceous  strip in Zone 3 could be much,
 more easily reshaped and revegetated than a forest.
 Herbaceous buffers, especially grass filters, are effec-
 tive at removing coarse suspended  sediments and
 some sediment-borne pollutants but may require fre-
 quent maintenance and are not very effective at nutri-
 ent removal from shallow groundwater (Dillaha et al.,
 1989; Magette et al.,  1987; Magette et at 1989). -
   Long-term sequestering and removal of nutrients
 and'other contaminants in the RFBS is the main pur-
 pose of Zones 3 and 2. This can occur by 1) accumu-
 lating sediment and adsorbed contaminants; 2) micro-
 bial transformations  (for  N)  and biochemical
 degradation (for pesticides);  and 3) incorporation of
 nutrients and other chemicals into woody, biomass
 and soil organic matter. At least one study of Coastal
 Plain riparian forests showed substantial amounts of
 nutrient sequestering in Woody biomass (Fail et  al.,
 1986). The RFBS specification encourages produc-
 tion and harvest of woody biomass from Zone 2 to re-
 move nutrients and other contaminants. Once vegeta-
 tion has been removed  from the stream channel,
 recovery through plant succession may take long pe-
 riods of time  and revegetation may be dominated by
 undesirable species (Sweeney, 1993). Therefore, the
 need for permanent control of the stream physical and
 trophic environment requires directed succession to-
 ward desirable permanent vegetation in those portions
 of the RFBS which directly influence the; stream
 channel, in particular Zone 1.
   A number of practical concerns were  also con-
 sidered in the RFBS specification (Welsch,  1991).
 Most of the RFBS should be available, for manage-
 ment to provide an economic return without sacrific-
 ing water quality functions. Characteristics of soils,
 hydrology, and potential vegetation should guide de-
 sign and planning of effective RFBS. The  RFBS
 should be used  in .conjunction with sound upland
 management  practices  including nutrient  manage-
 ment and erosion control. In-stream woody debris're-
moval should be limited, but woody debris with po-
tential to form dams which cause inundation should ;
be removed. The dimensions of the RFBS-should de-
pend on the  existing and potential  NPS pollutant
loads and the  minimum size for sustained support of
the aquatic environment.
    Several of studies are underway to test the effec-
 tiveness ,of RFBS which correspond to or are similar
 to the USDA specification. Vellidis et al. (1993) and
 Sweeney (1993) describe RFBS restoration projects
 in the Georgia Coastal Plain and the Pennsylvania -
 Piedmont, respectively. Beare et al. (1994) describe
 preliminary results from management of an existing
 riparian forest which involves establishment of Zone
 3 adjacent to mature riparian forest and tree harvest
 treatments in Zone 2. Schultz et al. (1994) describe a
 multi-species three zone buffer system for use in agri-
 cultural areas.of Iowa and other parts of the Midwest.
 Much of the current understanding of RFBS has been
 incorporated into a Riparian Ecosystem Management
 Model which  simulates hydrologic and nutrient cy-
 cling processes in RFBS that conform to the .USDA
 specification (Altier  et al.,  1994; Sheridan et al.,
   It is important to note that bur current understand-
 ing of the functions, of the RFBS is based on studies
 that have been done in areas where riparian forests
 currently exist due to a combination of hydrology,
 soils, cultural practices, and economics.  Most of our
 current knowledge of the functions of the three zones
 of the RFBS specification is  derived from studies  in
 existing riparian forests and on experimental and real-
 world grass buffer  systems. Although  results can.be
 extrapolated from these existing forests to restored
 RFBS, most of the study sites are actually at some
 stage of restoration, following clearing within the last
 20-80 years.     .                \  '-^

 1, Zone 1—Control of the Stream
   Although reduction of NPS pollution is a -widely
 recognized function of RFBS,  they also contribute
 significantly to other  aspects of water  quality and
physical habitat (Allan  and Flecker, 1993;  Karr,
 1993). Habitat alterations, especially channel straight-
ening and removal of riparian vegetation, continue, to
impair the ecological health  of streams more often
and for longer time  periods than toxic chemicals
(Hughes et al.,  1990). Sweeney (1992)  considers loss
of riparian forests in eastern North America to be one
of the major causes of aquatic ecosystem 'degradation.
   Zone 1, the permanent woody vegetation at the
stream edge, enhances ecosystem stability and helps
control the" physical, chemical, and trophic status of

the stream. Healthy riparian vegetation in Zone 1 also
contributes to bank stability and minimizes instream
sediment loading due to bank erosion. Zone 1 also has
substantial ability to control NFS pollution through
denitrification (Ambus and Lowrance, 1991; Low-
ranee,  1992; Schnabel, 1986), sedimentation (Low-
rance et al.,  1986), or direct root uptake of pollutants.
   Zone 1 vegetation controls light quantity and qual-
ity, moderates temperature, stabilizes channel geome-
try, provides tree roots and woody debris for habitat,
and provides litter for detritivores (Barton et al., 1985;,
Beschta et al.,  1987; Hax and Golladay, 1993; Hill
and  Harvey,  1990;  Karr  and  Schlosser,  1978;
Sweeney, 1992, 1993). To maintain the biological in-
tegrity of the aquatic ecosystem,  an ideal managed '
buffer system should have patterns of vegetation,  lit-
terfall, and light penetration similar to those in a nat-
ural,  undisturbed riparian forest (Golladay. and
Webster, 1988; Karr, 1993; Sweeney, 1992, 1993).
However, for many locations, representative sites of
truly natural, undisturbed riparian ecosystems do  not
exist. In fact, after a long  history  of human distur-
bance in many  areas, the concept can be difficult to
define (Bren, 1993). Karr (1993) suggests that within
a homogeneous region, relatively pristine areas may
be identified as benchmarks for the evaluation of
other sites.
   Riparian  forest buffer functions  related to protec-
tion of the stream environment will not be reviewed
for different physiographic regions because  there is
general agreement among  literature  sources on  the
need for riparian forests in the Eastern U.S. for this
purpose. The major differences among physiographic
regions appear to be in the importance of stream temp-
erature control for cold-water vs. warm-water fisheries.

   a. Temperature and Light
   The diel and seasonal patterns of water tempera-
ture are critical habitat features that directly and indi-
rectly affect the ability of a given stream to maintain
viable populations of most aquatic species, both plant
and animal. Considerable indirect evidence suggests
that  the absence  of riparian  forests along many
streams and rivers in the Chesapeake drainage, partic-
ularly in agricultural areas, may have a profound ef-
fect on the  current geographic  distribution of many
species  of  macroinvertebrates  and fish. Sweeney
(1992) reviewed the effects  of temperature alterations
on the growth, development, and survival of stream
macroinvertebrates found in the Pennsylvania Pied-
mont. These studies showed that temperature changes
 of 2-6°C usually alter key life-history characteristics
 of most of the study species
    In the absence of shading by a forest canopy, direct
 sunlight can warm stream temperatures significantly,
 especially during summer periods of low flow. For ex-
 ample, maximum summer temperatures have been re-
 ported to .increase 6-15°C following deforestation
 (Beschta and Taylor,  1988, Lee and Samuel, 1976,
 Brown and Krygier, 1970). Streams flowing through
 forests will warm very rapidly as they enter  defor-
 ested areas, but excess heat dissipates quickly when
 streams reenter the forest. Burton and Likens (1973)
 demonstrated this alternate warming (by 4 to 5°C)
 and cooling as a stream passed through clear-cut and
 uncut  strips  in  the  Hubbard Brook  Experimental
, Forest, New Hampshire. In Pennsylvania (Valley and
 Ridge Province), average daily stream temperatures
 that increased 11.7°C through a clearcut area, were
 substantially moderated after flow through 500 m of
 forest below the clearcut. The temperature reduction
 was attributed primarily to inflows of cooler ground-
 water (Lynch et al.," 1980). The impact of deforesta-
" tion on stream temperature varies seasonally.  In the
 Pennsylvania Piedmont, Sweeney (1993) found that
 from April through October average daily tempera-
 tures  in a second-order meadow stream reach were
 higher than in a comparable wooded reach, but that
 the reverse was true from November through March.
    Riparian forest buffers have been shown to prevent
 the disruption of natural temperature patterns as well
' as to mitigate the increases in temperature following
 deforestation (Brown and Krygier, 1970; Brazier and
 Brown, 1973; Lee and Samuel,  1976). Brazier (and
 Brown (1973) found that buffer strips of 10'm width
 were  as effective as a complete forest canopy in re-
 ducing solar radiation reaching small streams in the
 Pacific Northwest. The exact width of Zone 1 needed
 for temperature control will vary from site-to-site de-
 pending on a variety of factors. Brown (1974) pointed
 out that streams oriented in a north-south direction are
 less easily shaded than streams flowing east or west,
 and that a buffer on the north side of a stream may
 have  little or no effect. Also, in larger streams and.
 rivers, the width of the channel prevents  a complete
 canopy cover, so that the effect of canopy shading
 may be reduced. In eastern North America, openings
 in the canopy immediately above streams occur when
 the channel width exceeds about 20 m in width (i.e.,
 about stream order 4  or  5).-  In a study  of five
 Minnesota Rivers,  Sinokrot and  Stefan (1993) in-
 ferred midsummer shading of 40-60% for rivers rang-

 ing from 15-50 m in width but effectively no shading
 along extremely wide rivers  (e.g., the 300 m wide
 Mississippi R.).  Stream orientation relative to solar
 angle may also affect the extent bf shading for larger
 streams. Although shading on larger rivers may have
 little or no effect on water temperature, shaded stream
'banks provide habitat microsites for fish and  other
 aquatic organisms.  ,   \ "            .
   The ability of a given width of streamside forest to
 maintain or restore the natural temperature character-
 istics of a stream segment  depends on how it. affects
 the factors that control the diel arid seasonal thermal
 regime of the stream. Such factors (other than shad-
 ing) include: flow, channel geometry,  solar radiation,
 evaporative heat  loss,  conductive  surface heat ex-
 change, and, in some ca,ses, conductive heat exchange
 with the streambed. Heat budget models can integrate
 local meteorological data with the above factors to
 predict stream and river temperatures with relatively
 high precision (e.g., Edinger et al., 1968; Brown,
 1969; Beschta, 1984; Theuer et al.,  1984; Sinokrot
 and Stefan,  1993; Edinger and Buchak, in press).
 These models indicate that solar radiation is the major
 factor influencing peak summer water temperatures
 and confirm that shading by the streamside forest is
 critical to the overall temperature regime of a stream
 or river. Stefan and Sinokrot (1993) estimated that re-
 moval of the  forest canopy along the  Straight R.,
 Minnesota, would Increase average  summer  water
 temperatures approximately 6 C.
   Hewlett and Fortson (1982) measured unexpect-
 edly  large  stream temperature fluctuations  in the
 Georgia Piedmdnt'on a clearcut site with a 5 to 8 m
 buffer strip left on each side of a  first-order stream.
 After logging and wind damage, about a 50% cover .
, canopy remained over the stream. Despite the partial
 buffer, as well as rapid regrowth  of  low vegetation
 over  the stream,  stream temperature fluctuations for
 four years following logging were much greater than
 in an uncut forest. Since the measured temperatures
 could not be accounted for by a stream temperature
 model, the authors suggested that in addition to the ef-
 fects of direct radiation on  stream temperature, efflu-
 ent groundwater temperatures may also have  been
 modified by the removal of vegetative cover.

   b. Habitat Diversity and Channel Morphology
    The biological diversity of streams depends on
 the diversity of habitats available.  Woody debris is
 one of the major factors in habitat diversity. Woody
 debris can benefit a stream in several ways: (1) by sta-
 bilizing the stream environment through attenuation
 of the erosive influence of stream flow;  (2) by in-
. creasing the  diversity and  amount  of habitat.for
.aquatic organisms;  (3)  by  providing  a  source of
 slowly decomposable nutrients; and (4) by forming
 debris dams, it enhances the availability of nutrients
 for aquatic organisms from mdre rapidly "decaying
 material. '   '             ;'-.•,..-.
   The quantity of woody debris in streams under
 forested canopies in the Eastern United  States has
 been reported to range from 0.4 to 23 kg m"2, averag-
 ing about 8 kg nr2 (Webster et al., in press). These
 figures are undoubtedly lower than would be encoun-
 tered  in streams flowing through undisturbed forest.
 because most eastern streams have been subjected to
 extensive removal of streamside vegetation and, in
 larger streams, clearing of woody debris for naviga-
 tional purposes (Webster et al., in press). Quantities of
 large woody debris (LWD) recommended for healthy
 streams in the George Washington National Forest in
 Virginia range from  34 pieces of LWD per km for
. warm water fisheries  to 136 pieces/km for cold water
 fisheries. Although the quantity of woody debris in
 streams without forested riparian zones would be ex-
 pected to be very low, there are few quantitative stud-
 ies. Sweeney (1992) found that the volume of woody
 debris  under  forested  canopies in  a Mid-Atlantic
 Piedmont stream was 20 times greater than the  vol-
 ume in a comparable .meadow reach. Following re-
 moval of a riparian forest, LWD present in the stream
 declines  through  gradual  decomposition,  flushing
 during storms, and lack of inputs. Smaller debris from
 second-growth stands promotes less stability of the
 aquatic habitat and tends to have a shorter residence
 time in the stream.
   Loss of streamside forest can lead to loss of habitat
 through stream widening where no permanent vegeta-
 tion  replaces  forest  or through stream  narrowing
• where forest is replaced by permanent sod. In the ab^
 sence of other perennial vegetation, bank erosion and
 channel .straightening can  occur  as unimpeded
 streamflow scours the streambed and banks (Hartman
 et al.; 1987; Oliver and Hinckley, 1987). The acceler-
 ated streamflow velocity allowed by straight channels
 promotes channel incision as  erosion from the stream
• bottom exceeds sediment entering the  stream, this
 process can eventually lead  to the development of
 wide,  shallow  streams that support an impoverished
 diversity of species (Shields et al., 1994). Bisson et al.
 (1987) point out that  stability of debris accumulation
 is important for aquatic habitat. Because of the greater

resistance to displacement by hydraulic forces,' LWD
is of greater benefit to stream stability. Longer mater-
ial is .relatively more important for the stability of
wider streams.
  In contrast, narrowing of stream channels has also
been reported following the replacement of stream-
side  forest with permanent  grassland or grass sod.
Zimmerman et al. (1967) found that the narrowing of
deforested stream channels was  evident for streams
up to drainage areas of 13 km2 (5 mi2) or about a third
or fourth-order stteam.  Sweeney (1992), quantified
the narrowing  phenomenon more  explicitly  in  a
Pennsylvania Piedmont basin, showing that: (1) first
and second-order wooded reaches averaged about 2
times wider than their meadow  counterparts of the
same order; and (2)  third and fourth-order forested
reaches were about 1.7 times wider than in deforested
areas.  The channel  narrows in  the  absence  of a
streamside forest because grassy vegetation, which is
normally shaded out', develops a sod  that  gradually
encroaches  on  the   channel  banks. For benthic
macroinvertebrates, microbes, and algae, which live
in and on the substratum, the loss in stream width
translates into a proportionate loss of habitat. The ef-
fects of channel narrowing on fish habitat are  more
complex and involve the influence of woody debris on
the pool and riffle structure (as discussed below)..
  Links between LWD in streams, the abundance of
fish habitat, and the  populations, growth, and diver-
sity of fishes have been documented (see reviews by
Dolloff,  1994; Harmon et al.,  1986;  Bisson et al.,
1987). Even when selective harvesting of trees has
been  allowed along streams, the removal of old
growth has caused a decline in aquatic habitat quality
due  to diminished inputs of LWD  (Bisson et al.,
1987). The surfaces of submerged logs and roots pro-
vide habitat that often support macroinvertebrate den-
sities far higher  than on the stream bottom  itself
(Rhodes and Hubert, 1991; Sweeney, 1992; Benke et
al., 1984).
   Woody debris, like boulders and bedrock protru-
sions, tends to form pools in streams either by directly
damming flow, by the scouring effects of-plunge
pools downstream of fallen logs, or by forming back-
water eddies where logs divert flow laterally (Dolloff,
1994a). In undisturbed forests, LWD can account for
the majority of pool  formation (Harmon et al.,  1986;
Hedman,  1992). As expected, removal of woody de-
bris by deforestation typically results in loss of pool
habitat (Bilby,  1984). Although pools are spatially
contiguous with riffles, there is little or no  overlap in
the species composition of the dominant macroinver-
tebrates occurring in the two habitats. The loss of
pools, therefore, translates directly into lower popula-
tions and diversity for this group. For fish, pools im-
prove habitat by providing space, cover, and a diver-
.sity of microenvironments. Greater depth and slower
velocity in pools afford protection  to fish during
storms, drought,  etc. (Dolloff,  1994a). The  habitat
provided by LWD may also offset the destruction of
stream habitat structures such as pools, riffles, and
cascades by catastrophic storm events (Dolloff et al.,
   Debris  dams of large woody material block the
transport of both sediment and smaller litter materials.
The impoundment and delayed  transport of organic
material downstream  enhances its  utilization by
aquatic organisms. By slowing transport rates, dams
on small order streams serve as buffers against the
sudden deposition of sediment downstream Bisson et
al. (1987). The capacity of a stream to retain debris,
therefore, is an important characteristic influencing
the aquatic habitat. (Bisson et al., 1987; Meehan et al.,
1977).                 '        -•"...
   Although it is often thought that LWD is less im-
portant on large rivers and openwater habitats, it has
been shown that woody debris derived from riparian
forests along tidal shorelines of the Bay provides an
important refuge habitat for numerous species  of fish
and crustaceans (Everett and Ruiz, 1993).  Shallow
water habitats with abundant LWD support greater
abundances of many- species of fish and crustaceans
than do areas with no woody debris bordered by nar-
row strips of march (Everett and Ruiz, 1993; Ruiz et
al., 1993). They hypothesize that the importance of
LWD along Bay shorelines has been increased due to
loss of habitat in submerged aquatic  vegetation and

   c. Food Webs and Species Diversity

   The two primary sources of  food energy input to
streams are litterfall (leaves, twigs, fruit seeds, etc.)
from streamside  vegetation  and  algal production
within the stream. Total annual food energy inputs
(litter plus algal production) are similar under shaded
and" open canopies, but the presence or absence of a
tree canopy has a major influence on  the balance be-,
tween litter input and primary production of algae in
the stream.
   Meehan et al.  (1987) noted that "streams flowing
through older, stratified forests receive the' greatest
variation  in quality of food for detritus-processing

 organisms." In the Piedmont, streams flowing through
 forested landscapes do  not subsidize  downstream
 channels that have been deforested (even contiguous
 reaches) because the large pieces of litter do not move
 very far (Sweeney,  1992). This means that a stream-
 side forest is needed along the  entire  length of a;
 stream in order to assure a proper balance of food in-
 puts appropriate to  the food chain of native species.
 Macroinvertebrate  populations  are  affected  by
 changes in litter inputs.-The activity of benthic organ-
, isms may  increase following:  strearhside plant  re-
 moval.  Woody material  decomposes more  quickly
 following riparian forest removal, thereby further re-
 ducing the stream's nutrient retention (Golladay and
 Webster, 1988).        ,
   The quantity and, quality of algal production in a
 stream is greatly affected by the quantity and quality
 of light striking its  surface. For-example, Bilby and
 Bisson (1992) showed that the algal community of a
 stream heavily  shaded by an old growth forest was
 dominated by diatoms all year, while a nearby stream
 in a deforested area contained mainly  filamentous
 green algae in the spring and diatoms at  other times.
 Other studies have  also shown that deforested sites
 tend to be dominated by filamentous algae while  di-
 atoms prevail under dense canopy cover {Lowe et al.,
 1986;  Feminella  et  al.,  1989): In  the eastern
 Piedmont, filamentous algae such as  Cladophorq can
 be dominant in deforested streams due primarily to
 the a combination of high nutrients, high light levels,
 and warm temperature. Although some macroinverte-
 brates such as crayfish (Feminella and Resh, 1989)':
 and waterboatmen, insects (Sweeney and Schnack,
 1976) readily consume this type of algae, most her-
•bivorous species of stream macroinvertebrates have
 evolved mouthparts  specialized for scraping diatoms
 from the surface of benthic  substrates (Merritt and
 Cummins, 1984) and cannot eat filamentous algae.
   The influence of-differences in the' quality of algal
 production  on the  aquatic, ecosystem is  complex.
 Algal grazing species generally benefit from an in-
 crease'.in algal growth (Wallace  and Gurtz,  1986;
 Perrin et al., 1987; Bilby and Bisson, 1992; Sweeney,
 1992). Because the growth efficiency of insects is
 often higher on algae than,on detritus, the opening of
 the canopy may increase the  production of macroin-;
 vertebrates in these reaches. For  example,  Behmer
 and Hawkins, et al. (1986) found both higher biomass
 and  densities, for most grazer species in deforested
 sites relative to forested sites. The pattern  is not clear,
 however, because Hawkins (1982) found  higher bio-
 mass but lower densities of grazers in deforested ver-
 sus forested sites. Newbold et al. (1980) observed in
 California^" streams that the  benthic  community in
 logged watersheds became dominated by a few algal
 feeding species.  The diversity of the macroinverte-
 brate community was significantly lower than in un-
 logged watersheds, except where the stream was pro-
 tected by a riparian buffer of 30 m or more. For buffer
 strips less than 30 m in width, the Shannon diversity
 was significantly correlated with buffer width.

 2. Zone 2—Removal of
   Nonpoint Source Pollutants

   The primary function of Zone 2 is to remove, se-
 quester, or transform nutrients, sediments, and other
 pollutants. Because of its proximity to Zone 1, Zone 2
 might also have direct impacts on the stream channel
 system and contribute to Zone 1 functions. The pollu-
 tant removal  function of a Riparian Forest Buffer
 System depends on two key factors; 1) the capability
 of a particular area to intercept surface and/or ground-
'water-borne pollutants and  2) the activity of specific
 pollutant removal processes.  Focusing on these two
 factors  as  regulators of buffer zone effectiveness is
 useful for evaluating the importance of a particular
 site  as  a buffer and for evaluating the three zone
 RFBS specification. In the sections below we review
 the major pollutant removal processes that operate in
 Zone 2 and discuss how these processes interact with
 pollutants 'in  either surface runoff or groundwater
 flow in the context of the three zone specification.,
                          ,'" '         •, ^
   a.    Nitrate Removal

   Nitrate  removal from shallow groundwater has
 been the focus of many completed and ongoing stud-
 ies. At least four,separate studies at different sites  in
 the  Gulf-Atlantic  Coastal  Plain  Physiographic
 Province have, shown that concentrations of nitrate  in
 shallow subsurface flow are markedly reduced after
 passage through, portions of .natural riparian forest
 analogous,to Zone 2 (Jacobs  and Gilliam,  I985a,b;
 Jordan et al.,  1993; Lowrance  et al.,  1983, 1984a;
 Peterjohn and Correll, 1984). Studies in other physio- .
graphic settings have also shown nitrate removal from
 shallow groundwater in areas analogous to Zone  2
 (Groffman et al., 1992; Simmons et al., 1992).  Most
studies with high levels of nitrate removal were in
areas  with high water  tables that  caused shallow
groundwater to flow through or near the root zone.
   The mechanisms for  removal of nitrate  in  these

study areas are thought to be a combination of deni-
trification and plant uptake. Linkages between plant
uptake and denitrification in surface soils have been
postulated as a means for maintaining high denitrifi-
cation rates in riparian ecosystems (Groffman et al.,
1992; Lowrance, 1992). In contrast, riparian systems
without substantial contact between the biologically
active soil layers and groundwater or with very rapid
groundvvater movement appear to allow passage of ni-
trate with only minor reductions in concentration and
load. Correll et al. (1994) reported both high nitrate
concentrations and high nitrate removal rates beneath
a riparian forest where very high nitrate flux and rapid .
groundwater movement through sandy aquifer mater-
ial  limited nitrate  removal efficiency. Staver and
Brinsfield (1990) showed that groundwater flow be-
neath the biologically active zone of a narrow riparian
buffer along a tidal embayment  in Maryland resulted
in little  removal of nitrate. It is also known that
groundwater  discharging through sediments of tidal
creeks may have up to 20 times the nitrate concentra-
tions found in the main stem of the creeks (Reay et al.,
  Phillips et al. (1993) indicated that groundwater ni-
trate might bypass narrow areas of riparian forest wet-
land and discharge into stream channels relatively un-
altered when  the forest is underlain by an oxygenated
aquifer. This pattern of groundwater flow was sup-
ported by modelling of a small Coastal Plain water-
shed in Maryland (Reilly et al., 1994). Isotopic analy-
sis of groundwater and surface water in this watershed
suggested that denitrification was not affecting the ni-
trate concentrations of discharging groundwater. In
these cases where nitrate enriched water surfaces in
the stream channel, a wide RFBS would have little ef-
fect on nitrate. Deeply rooted vegetation near the
stream might have some effect.
  Studies in New Zealand have shown that the ma-
jority of nitrate removal in a pasture watershed took-
place in organic riparian soils which received large
amounts of nitrate laden groundwater (Cooper, 1990).
The location of the high organic soils at the base of
hollows caused a high proportion of groundwater (37-
81%) to flow through the organic soils although they
occupied only  12% of the riparian zone. A related
study in New Zealand (Schipper et al., 1993) found
very high nitrate removal in the  organic riparian soils
but streamflow was still enriched with nitrate. The au-
thors speculated that water movement  through min-
eral soils was responsible for most of the nitrate trans-
port to streams. Puparian systems with  intermingling
of organic and mineral soils point out the need to un-
derstand where  groundwater is  moving and what
types of soils it will contact, especially in seepage
   b.   Plant Uptake

   Maintenance of active nutrient uptake by vegeta-
tion in Zone 2 should increase the potential for short-
term (non-woody biomass) or long-term (woody bio-
mass) sequestering of nutrients. Although plant water
uptake is chiefly a passive transpiration process, plant
nutrient uptake is mostly an active process, dependent
upon plant metabolic activity (Hoagland and Broyer,
1936). Most nutrients are transported into plants
against an electrochemical potential gradient (Bowl-
ing et al., 1966; Higinbotham et al., 1967). Obser-
vations of ion concentrations in plant xylem exceed-
ing external soil water concentrations by over  100
times indicate significant active uptake of P. (Russell
and Shorrucks, 1959). Transpiration tends to  influ-
ence the uptake of a nutrient when the external con-
centration of that nutrient is high. Transpiration in ri-
parian forests is  very high and can control, water
movement  to streams (Correll and Weller,  1989;
Bosch et al.,' 1993). Kramer and Kozlowski (1979)
pointed out that transpiration increases the mass flow'
of solutes toward root surfaces.
    Nutrient uptake by. flood-intolerant plants is
strongly influenced  by the  aeration status of the soil
(Hoagland  and  Broyer,   1936;   Hopkins,   1956;
Hopkins et al. 1950). As low oxygen supply .decreases
root metabolism, the uptake of most nutrients de-
creases. Flood-tolerant species, such as those found in
many riparian forests, may tolerate low-oxygen con-
ditions by means of adaptive metabolic responses
(Crawford, 1982). They may also avoid root anoxia
by morphological adaptations that facilitate the avail-
ability of oxygen. Under  flooded  conditions,' roots
may become thicker and increase in porosity, allow-
ing  an internal  downward diffusion of oxygen
(Armstrong, 1968; Courts  and Philipson, 1978). The
growth of adventitious roots may also allow water and
nutrient uptake from near-surface areas that are more
aerated  (Kozlowski, 1984; Sena  Gomes and  Koz-
lowski, 1980).
   Vegetation selection for restored or managed PJFBS
must consider the ability of different species to take
up and store nutrients under specific.conditions of the
site. Kozlowski  and Pallardy (1984), point out that
flooding can.enhance the nutrient uptake and growth

 of some  species.  Bottomland  hardwood seedlings
 grow faster under saturated conditions than under
 drained but Veil-watered conditions! Mote rapid in-
 creases in total.dry weight and N and P uptake were
 found in water tupelo (Nyssa aquatica L.) as well as
 .several other species under  saturated  conditions
 (Hosner and Leaf, 1962;,Hosner et al., 1965). Shoot
' weights of a majority of wetland and intermediate
 plant species  were either  unaffected or increased
 under  flooded conditions  (Justin and  Armstrong,
    Nutrient uptake and accretion in riparian forests
 will be affected by vegetation management. Nutrient
 demand by vegetation corresponds with growth rate
 (Cole, 1981; McDonald et al., 1991). Loblolly pine
 dominated  forests in the. Gulf  Coastal Plain attain
 maximum rates of growth of about 8t. ha"1  yr.1 during
 the first twenty years of age, for which 101 kg of N
 and 9 kg of P are required each year (Nelson et al.,
 1970; Switzer et al., 1979). Cole ,and Rapp  (1980)
 suggested a worldwide average annual N uptake rate
 of 70.5 kg ha'1 for deciduous tree species and 39 kg
 ha"1  for coniferous species. Temperate  deciduous .
 species produce 179 kg biom'ass kg-1 N uptake, and
 temperate coniferous species produce 103 kg biomass
 kg'1  N  uptake (Cole  and Rapp, -1980). However,
 Miller  (1984)  disputes  the  notion  that  coniferous.
 forests   require less nutrients  than broad-leaved
 forests. His review of nutrient uptake studies indicates
 that the ranges of measured uptake for coniferous and
 broad-leaved forests overlap.        •
    Compared to the "natural" riparian forests studied
 iii mosj existing research, managed riparian  forests .
have  the potential for increased accumulation of N
and P in biomass through both increased biomass pro-
duction and increased foliar nutrient contents. Trees
can respond to N subsidy by both increased growth
rates  and luxury N uptake. The growth rate of forests
is commonly N limited.  Cole (1986) suggested-that
high efficiency of N use by forests is an adaptation to
the N-deficient environments that they frequently in-
habit,      '".''•
   Often the potential N uptake rate is much  higher
than observed rates. Forest growth has been found to
respond readily to N applications (Miller and Tarrant,
1983; Schmidtling,  1973).  Mitchell  and Chandler
(1939) found large tree-growth responses to N.fertil-
izer applications up to 400 to 600 kg ha'1. Cole (1981)
found that after fertilizing with 400 kg N ha'1 .yr?  in
effluent'from a municipal sewage treatment plant for
three years, poplar  (Populus  riigra var.  italica
  Muench.) and Douglas  fir (Pseudotsuga menziesii
  (Mirb.) Franco) took up 213 and 78 kg N ha'1 yr1, re-
  spectively. This contrasted with an uptake of 16 kg N
  ha'1 yr1 by poplar and 23  kg N ha'1 yr! by Douglas fir
  in  unfertilized sites. Miller  and Cooper (1973)
  showed that trees can take up. "luxury" levels of N.
  Growth  responses  by 36-year-old Corsican pine
  (Pinus nigra var. maritima.(Ait.)  Melv.) to different
  levels of N fertilization showed that foliar N content
  reached a maximum of 26,400 mg kg-1 after applying
  the highest rate of 504 kg N ha'1 yr1 for three years.
  Maximum volume growth  corresponded  to a foliar
  content of about 20,000 mg kg'1, attained by applica-
  tions of 336 kg N ha"1 yr1 for three years.
     Conditions do exist where N is no longer the lim-
  iting nutrient for forest growth. Long-term inputs of
  nitrogen, such  as may occur from  atmospheric depo-
  sition in the northeastern  U.S., could result in N lev-
  els exceeding the total combined plant and microbial
 nutritional demands (Aber et al., 1989). Under these
 conditions, P might  become the limiting factor for
 tree.growth. Unlike upland forests,  P may often be the
 most limiting nutrient in wetland ecosystems (Taylor
 et al., 1990). Mitseh et al.  (1979) found the growth of
 bald cypress (Taxoditim  distichum^ (L). Rich.) in a
-southern Illinois swamp to correspond well with P in-
 puts from flooding. Foliar P content of loblolly pine
 on wet Coastal Plain sites in South Carolina has been
 observed to  correlate well with growth (Wells and
 Crutchfield, 1969). Analysis by Brinson (1977) of nu-
 trient ratios in decaying litter from tupelo gum trees in
 a North Carolina swamp forest suggested that P levels
 may limit decomposition rates. If P is the limiting nu-
 trient for tree growth,  it should make vegetation an ef-
 fective P sink.
  .  While several studies  have found plant uptake to
 be an important nutrient removal mechanism in areas
 analogous to Zone '2 of riparian forest buffers (Correll
 and Weller,  1989; Fail et al:,  1986; .Peterjohn and
 Correll 1984; Groffman et al. 1992), several factors
 may reduce the importance of plants as nutrient sinks.
 Pollutants in groundwater flowing into the riparian
 buffer will only be accessible to plants if the water
 table is. high in the soil profile {Ehrenfeld 1987) or if
 mass movement of water due  to transpiration de-
 mands moves water and solutes into the root zone.
 Coastal Plain riparian forests have been shown to con-
 trol localized downslope water transport by creating
 moisture gradients which  move water in unsaturated
 flow from both the adjacent stream and the upland
 field (Bosch et  al.., 1993).  Nutrients in surface runoff


and in water percolating rapidly through soil macrop-
ores as "gravitational water" may not be available to
plants. Large  rainfall events, that often transport a
high percentage of pollutants in the CBW (Jaworsjd et
al., 1992) often produce  concentrated surface  flow
and macropore-dominated percolation.
    Plant sequestering of nutrients is also limited by
seasonal factors. In the temperate deciduous ecosys-
tems that dominate riparian forest buffers in the CBW,
plant uptake will decline  or stop during the winter
season. A high percentage of surface and groundwater
flow occurs in the CBW during winter. There is also
concern that nutrients trapped in plant tissues can be
released back into the soil  solution following litterfall
and decomposition. However, nutrients released from
decomposing plant litter may be subject to microbial,
physical or chemical,attenuation mechanisms in the
root zone  of forest soils.  Storage of nutrients in
woody tissue is a relatively long-term attenuation, but
still does not result in removal of pollutants from the
ecosystem unless biomass is removed. A final con-
cern about plant uptake as a nutrient removal mecha-
nism arises from the possibility that the  ability of
trees in a buffer zone to sequester nutrients  in woody
biomass becomes less as  trees mature. The average
tree age in most riparian forest buffers in the CBW is
less than 100 years and should thus be accumulating
nutrients in woody biomass. Although net vegetation
accumulation of nutrients may reach zero, net ecosys-
tem accumulation may continue  as  nutrients  are
stored in soil  organic matter. Groffman et  al (1992)
describes a nitrate-enriched  riparian system  with
symptoms  of N saturation  (Aber et al.,  1989).
Nitrogen saturation is not likely to occur in RFBS be-
cause of high denitrification rates removing N from
the system.
   Little is known about the types  of vegetation
needed in new or reestablished RFBS. Crop tree man-
agement (the selection and release of desired trees by
removal of competing trees) will be possible in  many
natural successional riparian forests. Numerous native
tree species are recommended for water quality im-
provement in crop tree management  (Sykes et .al.,
 1994). The trees were selected based ori their ability
as nutrient filters although little data exist on individ-
ual riparian species.

   c. Microbial Processes

   In addition to plant uptake, there  are  microbial
processes that attenuate pollutants in RFBS. These
processes include immobilization of nutrients, deni-
 trification of nitrate and degradation of organic pollu-
 tants. Microbes take.up or "immobilize" dissolved nu-
 trients just as plants do. These immobilized nutrients
 can be re-released or "mineralized" following death
 and decomposition of microbial cells, just as nutrients
 sequestered by plants can be released following litter-
 fall. In ecosystems that are accumulating soil organic
 matter, there will be a net storage of immobilized nu-
 trients. Zone 2, if managed to foster soil organic mat-
 ter accumulation, may thus support significant long-
 term rates of nutrient storage by immobilization.
     Denitrification refers to the  anaerobic microbial
 conversion of nitrate to  N gases. •Denitrification is
- controlled by the availability  of oxygen (Oa), nitrate,
 and carbon (C). Although essentially an anaerobic
, process, denitrification can occur in well drained soils
 because of the presence of anaerobic microsites, often
 associated with decomposing organic matter  frag-
 ments which deplete available oxygen (Parkin, 1987).
 It is  likely that soil  moisture gradients,in riparian
 ecosystems cause  a  change in  controlling  factors
 within most three zone RFBS. In parts of the RFBS
 with better internal drainage and generally lower soil.
 moisture conditions, denitrification may be generally
 limited by their interacting factors of carbon avail-
 ability and aeration status. While many wetlands are
 often assumed to have high  levels of denitrification
 . because of high carbon soils and anaerobic condi-
 tions, denitrification in many wetlands will be N lim-
 ited (Groffman, 1994). In the more poorly drained or
 wetland portions of a RFBS, denitrification is more
 likely to be limited by nitrate availability.
    Wetland soils develop high levels of organic matter
 because of their slope position and hydrologic condi-
 tion.  Frequently inundated soils will have lower rates
 of litter decomposition because the flow of carbon
 from litter to microbial populations is reduced tinder
 anaerobic conditions  (Groffman, 1994). The  interac-
 tive nature of oxygen, nitrate, and carbon control of
 denitrification means that more denitrification gener-
  ally occurs in intermittently flooded sites than in per-
 manently  flooded  conditions (Reddy  and  Patrick,
     Denitrification, measured directly using the acety-
  lene  inhibition technique (Tiedje et al.,  1989), ac-
  counts for substantial nitrate loss from some  riparian
  ecosystems. Denitrification has been identified as the
  key nitrate removal mechanism in several riparian for-
  est buffer studies (Jacobs and Gilliam,' 1985b;  Pinay
  and  Decamps, 1988; Correll  and Weller,  1989;
  Groffman et  al.,  1992; Haycock  and  Pinay,  1993;

   Jordan et al., 1993). Estimates in the range of 30 to 40
   kg N ha'1 yr1 have been reported for natural riparian
   forests in the U.S. (Hendrickson, 1981; Hanson et al.,
   1994a, Lowrance et al., 1984b). In several studies of
   denitrification in riparian ecosystems,  denitrification
   has been concentrated in surface soil and rates are
x generally much lower below the top 12 to 15 cm of
   soil  (Hendrickson,  1981;  Grpffman  et  al.,  1992;
   Ambus and Lowrance,  1991;  Lowrance, 1992).
   Schipper et al. (1993) reported very high denitrifica-
   tion in the top 30 cm of an organic riparian zone soil
   in New Zealand. Denitrification rates  (measured on
   soil slurries made  anaerobic  with Argon gas) were
   over 11 kg N ha'1 d'1 in this site. This is likely an over-
   estimate of actual denitrification because the slurries
   were made anaerobic. The denitrification  rates mea-1
  ,sured were 1-3 orders of magnitude greater than most
   estimates in the literature. Measurements of denitrifi-
   cation in these organic soil zones showed that the den-
   itrification was greatest at the upslope edge of the ri-
   parian zone where nitrate-enriched water entered the
   organic riparian soil (Cooper, 1990). These studies in-
   dicated that most of the organic riparian soils in the
   watershed were  denitrifying at rates below their max-
   imum capacity  and could denitrify  more if nitrate
   loadings increased (Cooper, 1990; Schipper et al.,
.   1993). Denitrification is likely to be most important in
   wetland soils such as would be found in Zone 1 and
   some Zone 2 areas in the Chesapeake Bay watersheds
  (Lowrance et al. 1984b, Peterjohn and Correll 1984,
  Jacobs and Gilliam  1985b, Ambus and  Lowrance
   1991) but can also be  significant in drier forest soils
   subject to high nitrate loadings and in grass vegetated
  filter, strips (GVFS) (Ambus  and Lowrance,  1991;
  Groffman et al., 1991).         ,
     While the factors regulating denitrification in sur-
  face soils and aquifers are relatively well understood,
  the amounts of direct denitrification of groundwa-
  ter-rbprne nitrate are  much less  well established.
  Subsurface denitrification has been observed in sev-
  eral .studies (Truedell et al., 1986; Slater and Capone,
   1987; Smith and Duff, 1988;,Francis et  al.,:1989;
  Obenhuber  and  Lowrance,  1991), yet  other studies
  have found the potential for denitrification in the sub-
  surface  to  be  low  or  non-existent  (Parkin and
  Meisinger,  1989;  Ambus  and  Lowrance, 1991;
  Groffman et al., 1992; Bradley et al., 1992; Lowrance,
  1992; Yepmans et al., 1992; Starr and Gilham, 1993).
  Subsurface microbial activity is usually limited by
  carbon availability. In settings where the total and dis-
  solved carbon contents of aquifers are  low, they are
 poor quality substrates for microbial growth (Lind
 and Eiland, 1989; Hiscock~et al., 1991; Johnson and
 Wood, 1992; McCarty and Bremner, 1992) and anaer-
 obic conditions necessary for denitrification to pro-
 ceed are not generated.                       -
   Microbial attenuation of organic compounds arises
 from their ability to degrade these compounds as food
 sources or through non-energy yielding "cometabo-
 lism" reactions. There are many different microbial
 degradation mechanisms including aerobic, anaero-
 bic, chemoautotrophic and  heterotrophic pathways.
 The wide range of environments and high diversity of
 microbial metabolism in RFBS, should support many
 of these mechanisms. Further research into specific
 management strategies  to foster  a  wide range  of
 degradation strategies is  needed  (Paterson and
 Schnopr, 1992);          t
   In many cases, riparian zone'retention of ground-
 water-borne pollutants may depend on a complex in-
 teraction of hydrology, plant, soil and microbial fac-
 tors. The potential importance of these interactions is
 hypothesized based on studies where significant rates
 of nitrate removal from groundwater were measured,
 but the potential for denitrification in the subsurface
 was low.  Groffman et al.  (1992) and Hanson et al.
 (1994a,b) suggested that surface soil denitrification of
 groundwater derived nitrate is an important route of N
 removal  in  riparian forests. This route- depends on
 plant uptake of nitrate from groundwater, decomposi-
 tion and N release from plant litter, and nitrification
 and denitrification of this N in surface soil. In riparian
 forests where this route of N removal is important, the
 nitrate removal function may depend on complex in-
 teractions between hydrology, plant dynamics, and'
 soil microbial  processes. These  interactions vary
 within and between riparian forests and should be
 strongly influenced by soil drainage class, vegetation.
 and soil  type, climate,  and  groundwater, quality.
 Although soil denitrification should be sustainable in-
 definitely under proper conditions with a supply of ni-
 trate and available C, Hanson et al. (1994b) found that
 long term groundwater nitrate loading led to  symp-
 toms of "N saturation in the surface  soils of a riparian
 forest buffer.,

   d. Removal of Surface-borne Pollutants
   Fewer studies have  been published on NPS pollu-
tant removal from surface runoff in Zone 2 type
 forests. The primary function of Zone 2  relative to
 surface runoff is to remove sediment and sediment-
borne pollutants and tor infiltrate runoff. Daniels and

Gilliam (in press) found that mature riparian forests,
analogous to Zone 2 vegetation, were effective for
sediment load reduction with removal of 50 to 80% of
inputs from upland fields. Sediment trapping in ripar-
ian forest buffer zones is facilitated by physical inter-
ception of surface runoff that causes flow to slow and
sediment particles to be deposited. Effective sediment
trapping requires that runoff be primarily sheet flow.
Channelized flow is not conducive to sediment depo-
sition and can actually cause erosion of the RFBS.
Tvvo studies on long-term sediment deposition in ri-
parian  forests (Cooper et al., 1987, Lowrance et aL,
1986, Lowrance et al., 1988) indicated that long-term
deposition  is substantial. In both these studies, two
main actions occur: 1) the forest edge fostered large
amounts  of coarse sediment deposition within a few
meters of the field/forest boundary; 2) finer sediments
are deposited further into  the  forest  and near the
stream. Both Cooper et al. (1987) and Lowrance et al.
(1986) found much higher depths of sediment deposi-
tion at the forest edge than near the stream. A second
peak of sediment depth was often found in Zone 1,
possibly from upstream sediment sources deposited in
overbank flows (Lowrance  et al. 1986). The surface
runoff which passes through the forest edge environ-
ment is much reduced in sediment load because of
coarse sediment deposition but the fine sediment frac-
tion is enriched relative to total sediment load. These
fine sediments carry higher concentrations of labile
nutrients and adsorbed pollutants  (Peterjohn  and
Correll, 1984; Magette et al., 1989) which are carried
further into the  riparian forest and are  deposited
broadly across Zone 2.
    Movement of nutrients through Zone 2 in surface
runoff will be controlled by a combination of: 1) sed-
iment deposition and erosion processes; 2) infiltration
of runoff; 3) dilution by incoming rainfaiythroughfall;
and 4) adsorption/desorption reactions  with forest
floor soil and litter.  Studies that separate -the effects
of these various processes are not available. Peterjohn
and Correll (1984) found large reductions in concen-
trations of sediment, ammonium-N,  and ortho-P in
surface runoff which passed through about 50 m of a
mature riparian forest in the Maryland Coastal Plain,
analogous  to Zone 2. Although  the concentrations of
these pollutants were reduced by a factor of 3 or 4 in
most cases, the flow-length was about twice that rec-
ommended in the RFBS specification. Daniels and
Gilliam (in press) found that dissolved ortho-P  loads
in surface runoff were not reduced markedly in a
Zone 2-like area of riparian forest. The studies of sur-
face runoff through riparian forests agreed on the im-
portance of eliminating channelized flow through the
riparian forest and recommended spreading flow be-
fore it reached the forest buffer. Flow spreading is rec-
ognized as primarily a Zone 3 function in the RFBS
specification. In-field practices are also critical in pre-
venting channelized flow  from reaching the field

3.  Zone 3—Sediment Removal and
    Spreading of Surface Runoff

   The primary functions of Zone. 3 are to remove
sediment and sediment associated chemicals and to
spread surface runoff entering as concentrated flow.
Functions of grass vegetated filter  strips  (GVFS),.
analogous to Zone  3 of the RFBS, have been evalu-
ated in a number of replicated experiments. Most of •
the available research on GVFS is applicable to eval-
uating the potential for sediment deposition in Zone 3
of the RFBS.
   Several short-term experimental studies have found
that GVFS were effective for removal of sediment and
sediment-bound pollutants with trapping efficiencies
exceeding 50% if flow was shallow (< 5 cm depth)
(Young et al. 1980, Magette et al. 1987, Dillaha et al.
1989a).' Magette et al.  (1989) and Dillaha et al.
(1989a.) evaluated relatively narrow filter strips (4.6 m
and 9.2 m) for control of nutrients and sediment mov-
ing from row-crop plots. Magette  et al. concluded
that: 1) the performances of GVFS were highly vari-
able; 2) GVFS were more effective in removing sus-
pended solids than in removing nutrients; 3) GVFS
become less effective as more runoff events occur;
and 4)  the effectiveness of GVFS decreased as the
ratio of GVFS length  to source   area decreased.
Dillaha et al. (1989a) reached  similar conclusions.
They found that GVFS  were effective immediately
after establishment, removing up to 98% of the in-
coming sediment and that removal of incoming sedi-
ment bound total N and total P was nearly as effec-
tive. Soluble N (both NO3 and NH4) and soluble P
were not removed effectively. Both Magette et al.
(1989) and  Dillaha et al. (1989a) conclude that nar-
row GVFS would probably have relatively short use-
ful life spans. Dillaha et al. (1989a) reported that one
GVFS was nearly inundated with sediment during the
span of 6 rainfall simulator events. The sediment trap-
ping efficiency fell from 90% in run  1 to 5% in run 6.
GVFS were much less effective when flow was con-
centrated than  when  surface runoff was in shallow
 sheet flow (Dillaha et al.,1989a). Properly managed:

 Zone 3 areas are likely to perform similarly to GVFS
 in these experimental studies. Management of these
 areas will  likely include periodic removal of sedi-
 ment, reestablishment of vegetation, and removal of
 ephemeral channels:                    ,
    Trapping efficiencies for sediment decrease at high
 runoff rates  because of  increased depth of flow
 (Barfield et al.,  1979; Schwer and Clausen, 1989).
 Concentrations, of total N, total P, suspended solids,"
 and BOD were reduced up to 80% in feedlot runoff
 passed through  GVFS ranging from  92 to 262 m
 (Dickey and Vanderholm,  1981). The  need for rela-
 tively long filter strips was confirmed in other studies
 looking at  runoff from  chicken manure application "
 areas (Bingham  et al., 1980; Overcash et  al,  1981).
 They found that the ratio of buffer area to land appli-
 cation area in order to achieve complete removal of
 contaminants in  water leaving the GVFS was about
 1:1. Therefore a  filter area would need to be as large
 as the source area. This situation is often not possible
 due to inadequate land for filter areas or competition
 between land for GVFS and land for crop production. '
 A buffer source  area/length ratio of less than 1:1
 would be adequate for less than complete removal.
    Trapping efficiencies for sediment and nutrients
 also  decrease when runoff enters the GVFS in, con-
 centrated flow (Dillaha et al., 1986). When this is the
 case, very little of the filtration capacity of either the
 GVFS or the riparian forest is used. If field practices
 do not eliminate channelized flow,' it should be elimi-
'nated as near the upslope border of the RFBS as pos-
 sible. The RFBS specification suggests using level-
 lipped spreaders to convert concentrated flow to sheet
 flow before it reaches Zone 2 (Welsh, .1991). These
 spreaders, when needed, would be part of Zone 3 so
 they could be managed (cleaned out and, periodically
 restored) using farm equipment. Franklin et al. (1992)
 reported on the use of a level spreader to spread flow
 from agricultural fields before it entered a downslope
 forest filter zone (FFZ). Although they did not com-
 pare the FFZ with'and without spreaders for natural
.rainfall events, they did compare hydrologic response
 with and  without spreaders  for simulated  runoff
 events. Without a spreader, the time to reach peak
 flow at a flume below the FFZ was about 10 minutes •
 and the time to reach zero flow below the FFZ after
 the water supply was cutoff was only 20 min. In con-
 trast, with the level spreader in place, this artificial
 runoff took 45 minutes to reach a peak flow and 135
 minutes to stop  flowing  after  water was cut off
 (Franklin et al.,  1992). Although specific water qual-
 ity, data are not available from this study with and
 without'Spreaders, spreading the flow affected the
 timing of the event with a smaller effect on peak,and
 total flows.    •
    Used as part of the RFBS, GVFS should substan-
 tially reduce sediment and sediment-borne pollutant
 loads reaching the stream. Improperly installed GVFS
 may serve to accentuate channelization problems in
 the landscape, leading to erosiofi of the forested zones
 of the buffer. For;example,, in an analysis bf existing
 grass GVpS on 33 farms in Virginia (Dillaha et al.
 1986, 1989b) found that sediment trapping was quite
 poor in many cases. In hilly areas, sediment trapping
 was generally low because runoff usually crossed the
 GVFS as concentrated flow. Rapid (1-3 years) accu-
 mulation of sediment caused several GVFS to become
 vigorous  sediment producers. In cases where sedi-
 ment accumulation was  significant, runoff flowed
 parallel to the GVFS until a low point was reached
 where it crossed the GVFS as concentrated flow. Due
 to the uncertainties  in long-term performance of
 GVFS, overall buffer efficiency and  sustainability
 should be significantly increased by using a.combina-
 tion of grass strip and forest buffer as described here.

 4.  Integrated Water Quality Functions of
     the Three Zone Buffer System

   Although no published studies of  an integrated.
 three  zone buffer system are available, the" studies
 cited above provide useful irisights into the potential
 functions of each zone. Even with an integrated three
 zone system, it is possible that there will be conflict-
 ing objectives relative to the types  of water quality
 functions desired from the RFBS.
     Perhaps the .most basic potential conflict relative
 to NPS pollution control with RFBS is the need to si-
 multaneously control at least three  major  transport
 mechanisms of waterborne pollutants. It is likely that
 control of pollutants transported in the sediment ad-
 sorbedphase of surface runoff, the dissolved phase of
 surface .runoff,  and groundwater (dissolved phase
 only) may be optimal on different sorts of RFBS with
 differing soils, vegetation, and management. Riparian
 forest buffers must be effective in controlling multiple
 noripoint-sources of pollution.
     Phillips (1989) proposed a general model of ri-
 parian buffer effectiveness based on detention time of
 surface and subsurface runoff and comparison to a
 reference buffer of known'detention time and known
. effectiveness. Comparisons to a reference buffer with
 an.assumed  first order rate constant for nitrate ire-

moval were done. The detention-based model indi-
cated that relatively flat,  sandy, well-drained soils
with high infiltration capacities would be the most ef-
fective buffers for nitrate removal. This approach is
lacking relative to nitrate retention because it disre-
gards the effects of different soils on denitrification
and the unequal partitioning of nitrate between sur-
face runoff and subsurface transport paths. The deten-
tion model (Phillips, 1989) correctly concludes that
for surface-borne pollutants, increasing infiltration in
the RFBS will be an effective  measure for both dis-
solved and adsorbed pollutant control.  Conversely,
the sandy well-drained soils which have highest infil-
tration will likely have lowest denitrification rates and
may have rapid groundwater movement rates leading
to high rates of nitrate transport through the riparian
forest buffer.  This type  of situation is described by
Correll et al. (1994) for the entire riparian buffer and
by Cooper (1990) for the mineral soils in the riparian
  For nitrate removal via denitrification, a riparian
ecosystem where high nitrate water moves into high
organic matter soils or subsoils is the best way to pro-
mote denitrification and the best way to permanently
remove nitrate from the soil-water-plant system. This
is illustrated both by the New Zealand riparian studies
of organic riparian soils  (Cooper,  1990; Schipper et
al., 1993) and by the findings that denitrification is
highly stratified in mineral soils with most denitrifi-
cation occurring in the high organic carbon surface
soils  (Ambus and Lowrance, 1992;  Hanson et al,
1994a). Organic rich wetland soils can often respond
to increased nitrate loads with increased denitrifica-
tion (Groffmah, 1994). The same conditions which
are likely to promote denitrification are likely to  de-
crease the amount of retention of surface-borne pollu-
tants. Wetland  soils which are frequently inundated
will have little or no infiltration capacity or available
water storage capacity.

              Riparian Forest Buffer  Systems  in
              Physiographic Provinces  of the
              Chesapeake  Bay  Watershed

 1. General Land Use and Hydrology

   The Coastal Plain has higher proportions of both
 cropland (32%) and wetland (21%) than any other
 physiographic province of the Bay Watershed (Table
 1). The Coastal Plain portions of the CBW are com-
 prised of watersheds with low toppgraphic relief, rel-
 atively high moisture infiltration capacities, well-dis-
 tributed rainfall throughout the year, and unconfined
' surficial aquifers. Stre'amflow is mainly derived from
 groundwater discharge  from the  surficial aquifer.
 Direct .surface runoff in agricultural watersheds gen-
 erally accounts for about 5  to 15% of streamflow
 (Peterjohn and Correll, 1"984;  Staver ef al., 1988). The
 remainder of the precipitation either infiltrates and is
 available for either groundwater recharge or evapo-
 transpifation, or goes directly into surface water as
 stream or detention storage.' Although this general
 view of the Coastal Plain is useful, variations in soils,
 topography, subsurface  stratigraphy, and  land use
 within the Coastal Plain control the fate of NFS pol-
 lutants relative to RFBS.
  . The CBW Coastal Plain is  often divided into Inner
 and Outer Coastal Plains. The jhiner Coastal Plain is
 mostly the western shore of Chesapeake Bay and the'
 upper Eastern Shore. The Outer Coastal Plain is pri-
 marily the lower Eastern Shore/Delmarva Peninsula. ,
 Inner'Coastal Plain areas have relatively high topo-
 graphic  relief compared to Outer Coastal Plain sys-
 tems and generally have finer textured, nutrient-rich
 soils compared to the nutrient-deficient, sandy soils of
 the Outer Coastal Plain (Correll et al., 1992). A more
 detailed classification of the Coastal Plain was devel-
 oped by the U.S. Geological Survey for the Delmarva
 Peninsula (Phillips et al, 1993). This classification of
 hydrogeomorphic regions was  based.on qualitative
 analysis of geologic and geomorphic features, soils,'
 drainage patterns, and land cover (Figure 3). The up-
 land, non-tidal area of the Delmarva was divided into .
 Inner Coastal Plain which closely correlates with the
 Inner Coastal Plain of Correll et al. (1992) and three
 Outer Coastal Plain hydrogeomorphic regions: well-
 drained upland, poorly drained upland, and surficial
 confined region. Differences in the physical charac-
 teristics of these regions result in  variations in the
 functions of RFBS within them. The following dis-
 cussion presents the general hydrogeomorphic char-
 acteristics associated with each .region.

   a.  Inner Coastal Plain

   The Inner Coastal Plain (1C?) includes the portion
 of the Coastal Plain located on the  western shore of
 Chesapeake Bay and the area  immediately south of
 the Fall Line on the Delmarva Peninsula. Tidal sec-
 tions of rivers extend far into the ICP, near the Fall
 Line in some cases. Watersheds in .the ICP are char-,
 aeterized by well-drained soils on uplands with
 poorly drained soils limited to riparian zones. Land
 use is primarily agricultural on uplands and forested
 in riparian zones. Topography of this region is gently
 rolling with a high degree of stream incision.
   The, ICP is a hydrologically complex region be-
 cause sands' and gravels  that comprise the surficial
 aquifer are thin and overlie subcropping sands  or
 finer-textured confining beds of older Coastal Plain
 aquifers. Stream valleys  are commonly incised into
 the older units. As a result of this configuration, the
 surficial deposits-do not form an extensive aquifer as
 they do. in other parts of the.Coastal Plain. Shallow
 groundwater flow systems in the surficial sediments
 commonly  extend  from  topographic highs  into the
 deeper aquifer where they are close to the surface. In
 addition, older-water from deeper aquifers often dis-
 charges upward to streams.  If the  surficial aquifer
 overlies a shallow confining bed, groundwater flow is
 restricted to shallow depths where it comes into con-
tact with riparian zone, sediments  and soils near
 aquifer discharge areas.
   The Rhode R. Watershed along the western shore

of Maryland is representative of the hydrologic, con-
ditions common to much of the ICP. This 2286 ha wa-
tershed is 62% forest, 23% croplands,  12% pasture,
and 3% freshwater swamp (Jordan et al., 1986).  The
watershed is underlain by a relatively  impermeable
clay layer which forms an effective aquiclude. Most
groundwater flow to streams is in a shallow surficial
aquifer (Correll 1983). The 160 yr average rainfall is
108 cm. The long-term average precipitation by  sea-
son is 28 cm, 31.4 cm,  24.5 cm, and 24.6 cm for
December to  February, March  to May, June ,to
August, and September  to November, respectively
(Higman and  Correll,  1982 cited in Peterjohn  and
Correll, 1984). For the Rhode R. Watershed, slow
streamflow (baseflow or  groundwater discharge) av-
eraged 29.6 cm  of flow while  quickflow (mostly
stormflow  or  surface runoff from all contributing
areas) accounted for 4.97 cm (Correll, unpublished in
Peterjohn and Correll, 1984). Studies on Rhode R. in-
dicated that 86% of all watershed discharge  comes
from slow flow  or  groundwater discharge and  14%
from direct surface runoff. For  one year of study
March, 1981 to March, 1982, Peterjohn and Correll
(1984) estimated that about half of all quickflow  took
place in the Summer (June to August)  and that  over
half of slow flow (groundwater discharge) took place
in winter.

   b. Well-Drained Upland t

   Watersheds in the well  drained upland (WDU,
Figure 3) are  characterized by predominantly well-
drained soils on uplands and poorly drained soils on
fioodplains in  stream valleys. The topography is  rela-
tively flat to gently rolling and there is  a high degree
of stream incision (Phillips et al., 1993). Most of the
upland area is used for agricultural crop production
with wooded  areas generally confined to  narrow ri-
parian zones.  Sediments of the surficial aquifers are
primarily sand and gravel and range from about 6 to
12 m in the north to 24 to over 30m thick in the south
(Owens and Denny, 1979). The aquifer is unconfmed
and the depth  to water ranges from 3 to 10 m beneath
topographic highs,  to land surface in  surface-water
discharge areas.
   Groundwater flow paths range from  about 1 km to
several km in  length  in the  well-drained  upland
(Shedlock et al., 1993). The longest; oldest flow paths
originate at topographic  highs, extend  to the base of
the  aquifer,  and discharge  to  2nd and  3rd order
streams through the hyporheic  zone (beneath the
stream channel). The water contained in them is gen-
erally less than 50 years old near aquifer discharge
areas (Dunkle et al.,  1993).  Shorter,  younger flow
paths originate in near-stream recharge areas and are
the main source of baseflow to first-order streams.

   c. Poorly-Drained Upland

   Watersheds in the poorly drained uplands (PDU,
Figure 3) are characterized by interspersion of poorly
drained areas with forested land use, and moderately
well-drained and well-drained areas  with agricultural
use (Shedlock et al., 1993). In the northern part, the
"region has hurhmocky topography and low relief with .
many seasonally ponded wooded depressions. In the
southern part, topography is relatively flat with broad
poorly  drained  forested areas  that are  seasonally
flooded  (J.  M.  Denver, unpublished). Streams are
small and sluggish in the poorly drained upland and
flow through shallowly-incised valleys with low gra-
dients (Phillips et al.,  1993). Riparian zones are usu-
ally forested and often contain wetlands. Some parts
of the poorly drained upland have been ditched to pro-
mote drainage of agricultural fields.
   Sediments that make up the surficial aquifer in the
PDU are predominantly sands and gravels, similar to
those in the well-drained upland. The sediments range
in thickness from about 8 m in the north to more than
30 m in the south (Owens and Denny, 1979). The water
table is usually within 3 m of the land surface. This re-
gion is characterized as poorly drained because of the
combination of regionally high water table and small
degree of stream incision that results in groundwater
gradients too low to effectively drain the region,  rather
than a low permeability substrate (Phillips et al., 1993).
   Except for areas immediately adjacent to streams,
groundwater flow paths in the PDU  range from about
 100 m to about 1 km in the northern part of the region
where the aquifer is thin. In the southern part, where
the aquifer is thick, flow paths are up to several km in
length and  generally originate near the regional
drainage divide. Local flow patterns vary seasonally,
however, smaller localized flow paths associated with
the depressional wetlands and intermittent streams in
 the north and intermittent streams in the south are.ac-
 tive during wet seasons (generally winter and spring).
 A more regional flow system from topographic highs
 to perennial streams is active  throughout the PDU
 during the drier seasons (generally summer and fall).

    d.  Surficial Confined

    Watersheds in the surficial confined (SC, Figure 3)

                                         HYDROGEOMORPHIC  .
                     20 MILES
0   10
 Poorly-drained upland
 Well-drained upland
 Surficial confined
 Inner coastal  plain
 Poorly-drained lowland
 Fine—grained lowland
 Coastal wetland and
  beach region
   FIGURE 3. Hydrogeomorphic regions of the Delmarva Peninsula. (From Phillips et al, 1993).

region are geomorphologically similar to the southern
part of the poorly drained upland with low relief and
shallow incision of stream valleys, features that con-
tribute to the poor  drainage in  the  region.  Topo-
graphically, the  area is a flat sandy plain with low
ridges that rise a few meters above the surrounding
landscape. The plain is dominated by poorly drained
soils and the ridges  are  dominated by well-drained
soils. Throughout the region large tracts of forest are
interspersed with agricultural fields on the plains;
there are broad forested riparian zones and swamps
around the major drainageways. With the exception of
the sandy dune ridges, agricultural land is heavily
ditched to promote soil drainage and would probably
be  forested wetlands  in the absence of ditching
(Phillips et.al., 1993).
  The surficial aquifer is geologically heterogeneous
in the region, consisting of a major sand unit 25.to 30
m thick overlain by 0 to  13 m of complexly layered
clay, slit, and peat, which is itself overlain by 1  to 6 m
of wind-deposited sand with some peaty  sand, slit,
and clay lenses at the base (Owens and Denny,  1979).
The complex of fine-grained deposits acts as  a con-
fining unit between the sands of the upper and lower
units, except some areas where it is absent or entirely
composed of sand. The water table is generally less
than 3 m below land surface and occurs in the upper
sand unit. Local groundwater flow paths, in the upper
unit, are relatively shallow and generally less than 300
m long and extend from water-table  highs in inter-
fluves between ditches and streams into the ditches
and streams. Regional groundwater flow paths, in the
lower units, are up to ten kilometers long and  extend
from the uplands near the regional drainage divide to
major streams and rivers.  Local and regional flow
paths are separated  in most areas by the confining
layer, but local heads are higher than regional heads in
most places,  and shallow flow paths extend into the
lower  sand  where confining  beds  are   absent
(Shedlock et al., 1993). Residence time in the upper
sand is 15 years or less; in the deeper unit, groundwa-
ter residence tune is at least 40 to 50 years,  except
where there is hydraulic connection with the shallow
unit (Dunkle et al., 1993).

2.  Control  of Nonpoint Source Pollutants
   Although more studies have been done on Coastal
Plain riparian forests than in other physiographic re-
gions, a number of questions remain about the NPS
pollution control capacity  of naturally occurring ri-
parian forest buffers. Other questions remain about
 the NPS pollution control capacity of reestablished
 and managed RFBS. The following discussion will
 necessarily focus primarily on what is know about
 naturally occurring riparian forest buffers and ex-
 perimental  GVFS. Although discussion of reestab-
 lished RFBS will be limited, a number of useful con-
. elusions can be drawn from the existing Coastal Plain
    The studies on riparian forest buffer effects on NPS
 pollutants in the Coastal Plain have tended to concen-
 trate on the fate of nitrate in groundwater, with a sec-
 ondary emphasis on the fates of N, P, and sediment in
 surface  runoff.   Three areas  of the  Coastal Plain
 (Georgia, Maryland,  and North Carolina) have been
 studied  where gaged watersheds were used as the
 basis for nutrient budget estimates of riparian forest
 buffers.  The studies from Maryland (Rhode R.) have
 been used to develop nutrient budgets for watersheds
 and  riparian systems (Peterjohn and Correll, 1984;
 Jordan et al., 1986; Correll and Weller, 1989;  Correll
 et al.,  1992). The studies from Georgia (Little R.)
 have been used to develop both nutrient and sediment
 budgets (Lowrance et al., 1983,1984a,b, 1985; Fail et,
 ,al., 1986). The studies from North Carolina have been
 used to develop nitrate budgets  for riparian systems
 (Jacobs and Gilliam,  1985a,b). Hydrologic conditions
 for all of these studies were representative of ICP con-
    A second general type of study has been conducted
 on the fate and/or transport of potential NPS pollu-
 tants, primarily plant nutrients and sediment. These
 studies  have  also been  primarily in Maryland and
 Delaware (Correll et al., 1993;  Jordan et al., 1993;
 Whigham et  al, 1986), Georgia (Lowrance et al.,
 1988; Ambus, and Lowrance, 1991; Lowrance, 1992;
 Vellidis et al., 1993), and North Carolina (Cooper et
 al.,  1987; Cooper and Gilliam,  1987), In addition,.
 there are several studies of Coastal Plain hydrology or
 water quality which provide information on upland ri-
 parian interactions or provide limited data on NPS re-
 moval in riparian forest buffers. These are studies
 'which, in general, were not designed specifically to
• examine the removal of potential NPS pollutants in ri-
 parian forest buffers (Lowrance and Leonard,  1988;
 Weil et al.,  1990;  Staver and Brinsfield,  1990).
 Preliminary results  on  integrated  grass and forest
 buffers in the  Coastal  Plain have been  published
 (Parsons "et al.,  1991, 1994) and detailed studies of
 GVFS have been conducted in the Coastal Plain of
 Maryland (Magette et al., 1989).'.

   a. Nutrient Budgets for Riparian Forests
   The most direct means of determining the NFS pol-
 lution control function of a riparian forest is to de-
 velop  annual  or 'longer term  mass  .balances.
 Developing nutrient or sediment budgets  requires  a
' watershed from which hydrolpgic measurements  can
 be made which assure that all'watershed outputs are
 measured and sampled. If the riparian forest buffer is
 continuous  around  the  entire, stream  system  and
 groundwater discharging to streams moves through ri-
 parian soils and shallow sediments, the streamflow
 output can be treated as the output from the riparian
 forest system. The inputs to the riparian system must
 be estimated from sampling of groundwater and sur-
 face water inputs. The studies which have done  this
 for Coastal  Plain riparian forests  are summarized in
 Table 2. Total N and total P retention have been esti-
 mated in studies of Watershed-109  (WS-109) of the
                              Rhode R. in Maryland and the Heard Creek tributary
                             ' of Little* R. in Georgia. Both of these Coastal Plain
                              systems have effective aquicludes  at depths which
                              limit Techarge to deep groundwater and which cause
                              all or nearly all excess precipitation to move through
                              riparian systems and exit the watershed as streamflow,
                                Estimates of N retention were 89% of input (Rhode
                              R.), and 66% of input (Little R.). P retention in Rhode
                              R. was slightly less (80% of input) but much less in
                              Little R. (24% of input). Total N and P budgets for
                              Little R. (Table 2) did not include surface ^runoff in-
                              puts of N and P from the agricultural areas to the ri-
                              parian forest but did include all streamflow outputs of
                              N and P. Streamflow includes surface runoff, which
                              moved through the riparian forest and contributed to
                              stormflow. Therefore, the N and P retention (input-
                              output) estimates for the Little R. site, are underesti-
                              mates  of the actual retention. Peterjohn and Correll
                                              TABLE 2.
                   Total nitrogen, nitrate-nitrogen, and total phosphorus budgets for
                            riparian forest ecosystems in the Coastal Plain
Total N
Correll, 1984 -'
Rhode R.;MD
Input ,


Retention* Flux Notes*
74 NO3, NH4, Org-N in &
Lowrance et al., 1983  Little R., GA
Correll & "
Weller, 1989 '
Rhode R., Mb    45
Lowrance et al., 1983  Little R., GA      22

Cooper et al., 1986    Beaverdam Cr.,   35
     .-.-._   '    ,  „   NC

Peterjohn & Correll,    Rhode R., MD    3.6
1984                   "•'.• -;: -.':-.

Lowrance et al., 1983  Little R.. GA      5.1





 26          N03, NH4, Org-N In GW,
             P, SF.         ,
38.6         NO3linGW,SF
             (baseflow only).

19.9         NO3inGW,SF.

29.9         NO3 in GW, SRO, SF.
                                         2.9         Total P in SRO, GW, R
                                                     PSF, PQF.

                                         1.2         Total P in GW, P, SF.
+Retention = Input-Output        "

*SRO = surface runoff input; GW = groundwater input; P = precipitation input; SF = streamflow output; PSF = partitioned
 slowflow; PQF = partitioned quickflow

(1984) included  direct estimates of  both surface
runoff and groimdwater inputs and outputs for Rhode
R, Their budget estimates were based on these direct
measurements  rather  than streamflow.  outputs.
Streamflow outputs for Rhode R. were  different than
the riparian budget output for both total N and P. This
difference has only a negligible effect on the total N
budget, but has a large effect on the total P budget. If
streamflow outputs are considered the output from the
riparian forest for the Rhode R. site, .the total N reten-
tion is still 83% of inputs, but P retention is  zero.
   The Little R. and Rhode R. studies were both done
in systems which are likely to maximize retention by
natural riparian forests. Although the studies report
different ranges of percent retention for N and P, re-
tention of N was generally high. Both watersheds
have percentages of agricultural land typical  for the
more agricultural portions of the Coastal Plain and are
representative of potential inputs to riparian systems
in the absence of animal confinement facilities and
manure disposal systems. These natural riparian sys-
tems would appear to retain at least two-thirds of the
N inputs but perhaps as little as one-third of the P
input.                   •                '  •
   In both the Rhode R . and Little R. studies, nitrate-
in subsurface flow made up the majority of total in-
puts to the  riparian forest system.. The  input  in
groundvvater  for WS-109 of Rhode R. in the year re-
ported on in Peterjohn and Correll (1984) was 57 kg
NOa-N ha"1 yr1 based on the area of riparian forest.
This accounted for 69% of the total N input (Table 2).
Based on two more years  of  data for  WS-109 of
Rhode R., the input averaged 45 kg N03-N ha'1 yr1
(Correll  and Weller,  1989). Data from  Little  R.
showed that groundwater input was 22  kg NOs-N ha"
1 yr1, 56% of total N input. A  third study  of nitrate
budgets (Cooper et al., 1985) on a Coastal Plain wa-
tershed in North Carolina showed similar  results to
the MD  and GA studies. Nitrate retention rates of
85%, 86%, and 90% for the three studies (NC, MD,
GA, respectively) reflect removal of nitrate through
both  denitriflcation  and plant uptake. Plant uptake
(and perhaps microbial immobilization) contribute to
transformation of a predominately nitrate input to the
riparian zone into a predominately organic N output
in streamflow. Total N input to  the riparian forest on
Rhode R. was 69% nitrate. Streamflow was 51% or-
ganic N  (Correll et al,  1992, Correll,  1983).  On the
Little R., groundwater inputs to the riparian forest
were 74% nitrate. Streamflow outputs  were 18% ni-
trate and 80%  organic N. A later study of  the entire
Little R. watershed showed consistent trends of ni-
trate increase during stormflow, indicating that the ni-
trate  removal/transformation  capacity  of riparian
forests  is partially by-passed  when  water moves
through more quickly during high flows (Lowrance
and Leonard, 1988).

   b. Removal of Nitrate from Groundwater

   Although elemental, nutrient, chemical, and sedi-
ment budgets on a watershed scale are the most com-
plete way to evaluate the functions of riparian forest
buffers  and offer the  best  information on potential
load reductions, a number of studies have  examined
nitrate concentration changes in riparian forests/This
emphasis on nitrate is due to a number of factors in-
cluding the relatively high transport rate  of nitrate
from most agricultural systems, the availability of ni-
trate for algal uptake.as a stimulus for eutrophication,
and  possible impacts on  downstream  or  shallow
groundwater drinking water supplies. Studies  in at
least five separate Coastal Plain locations have exam-
ined the changes in nitrate concentrations as  shallow
groundwater moves from agricultural  fields  through
naturally occurring riparian forests (Figure 4). Studies
in four separate locations, have shown that average an-
nual edge-of-field nitrate levels of 7 to 14 mg NOs-N
L"1 decreased to 1 mg L'1 or less in shallow  ground-
water near streams. Some studies, have used chloride
concentrations and nitrate:chloride ratios to separate
the effects of dilution from the effects of biological
removal of nitrate. Decreases in chloride concentra-
tions were generally small compared to nitrate  de-
creases. Chloride concentrations usually increased at
some point in the shallow groundwater system proba-
bly due to exclusion of Chloride from the transpirar
tion stream (Peterjohn and Correll, 1986; Jordan et
al., 1993; Correll et al., 1993; Lowrance, 1992).
   Most studies of nitrate dynamics in riparian forests
have shown that removal of nitrate from groundwater
continued  year-round. Mechanisms to  explain  this
have not been elucidated, although it is likely that in
some of the  Southeastern  Coastal Plain areas,, rela-
tively warm soils and evergreen or tardily  deciduous
(broad-leaf trees that lose leaves in the spring) vege-
tation can provide biological removal of the nitrate.
Most Coastal Plain areas  of the CBW have  lower
groundwater and soil  temperatures in the. winter and
little or no evergreen vegetation. Weil et al. (1990) ob-
served  year-round reductions of groundwater nitrate
in streamside forests on tributaries of the Choptank R.
on the Eastern  Shore of Maryland.  Groundwater

     •'_, 26
       *  24
      2 20
     • z  • -
      2  16
                                               PETERJOHN  &  CORRELL,  1984 (MD)
                                               CORRELL ET  AL,  1993 (MD)
                                               LOWRANCE,  1992  (GA)
                                               JORDAN  ET AL,  1993 (MD)
                                               JACOBS  &. GILLIAM,  1985  (NC)
                                               STREAM  CHANNEL LOCATION
                                                 RELATIVE TO DISTANCE
                                                  SCALE (NOT N03 -N)
                                     30       40      50
                                  DISTANCE FROM FIELD  (m)
 FIGURE 4. Nitrate concentrations in groundwater beneath riparian forests from five Coastal Plain sites.
 under riparian forests always had less than 1 mg NO3-
 N L"1 while adjacent fields had concentrations of 15-
 40 mg NO3-N L'1. The decreases in chloride concen-
 trations  were much less than the nitrate  decreases.
 Year-round nitrate removal has been observed, but not
   At least one study has shown that in situations with
 relatively high nitrate concentrations entering from an
 adjacent field, substantial nitrate concentration reduc-
. tions can occur but still leave high  concentrations in
 shallow  groundwater at the stream (Correll et al,
 1993), (Figure 4).  This site, on a'tributary  of the
 Choptank R. on the Delmarva Peninsula is located in
 the Well Drained Uplands. Nitrate  concentration re-
 ductions were actually higher at this site than at two
 other Maryland  Goastal Plain sites (Peterjohn and
 Correll,  1984; Jordan et al,  1993)  but groundwater
 concentrations near the stream were 12 to 18  mg
    s-N L"1. Similar results were inferred from a study
  of nitrate in regional groundwater and nitrate levels in
  streamflow for the WDU hydrogeomorphic  region
  (Phillips et al., 1993). In .related work, Bohlke and,
  Denver (in press) concluded that, the riparian forest
  wetland next to  the stream in the Locust  Grove
  Watershed in Maryland  had little effect on  nitrate
  movement to the stream. Hydrologic data and ground-
  water flow modeling, show that groundwater  dis-
  charges upward directly  to  the streambed from the ;
  aquifer system, effectively bypassing the riparian
  zone (Reilly et al., 1994). Baseflow concentrations of
.  nitrate commonly exceeded 9 mg NOs-N L"1  in, the
  stream draining this watershed, and isotopic analysis
  indicated that denitrification was not significantly af-
  fecting nitrate concentrations (Bohlke, and Denver, in
    Nitrate transport  into tidal streams' is often domi-
  nated by direct recharge through sediments in inter-'
  tidal zones (Reay et al., 1992; Simmons et al., 1992;

Stayer and Brinsfield,  1994). Approximately 80 kg
ha'1 yr1 of NOs-N was discharged to a tidal creek in
Maryland with apparently most groundwater moving
at least 2 m below the ground surface in near-stream
areas (Staver and Brinsfield, 1994). These situations
may allow little chance for nitrate removal. The direct
NOa-N discharge to tidal streams make riparian
buffers desirable  (Simmons et al., 1992), if proper
management  could allow direct vegetation  uptake
from the groundwater.

  c.  Nutrient Removal Processes

  Removal processes were quantified in most of the
riparian forest research on nutrient budgets and nitrate
transport.  Studies in Maryland and Georgia have
made direct estimates of N and P uptake by vegetation
and storage of N and P in woody biomass. Estimates
from Watershed  109 of Rhode R.  (Peterjohn and
Correll, 1984; Correll andWeller, 1989) indicated that
total vegetation uptake of N and P was 77 and 10 kg
ha'1 yr1, respectively. N and P storage in woody bio-
mass was less than total uptake (Table 3).
  Extensive data on total N and P uptake and woody
storage were reported  by Fail et  al. (1986,  1987).
Values for P  uptake  and storage are similar  for the
Little R. and Rhode R. studies  (Table 3).  Major
differences between the two studies were found for N
                        woody storage and N uptake. Fail et al. (1986, 1987)
                        reported mean storage of N in wood as 52 kg N ha'1
                        yr1. The range was from about 35, to 98 kg N ha'1
                        yr1. The net primary productivity reported by Fail et
                        al. and Peterjohn and Correll are similar as are leaf N
                        concentrations and leaf and wood P  concentrations.
                        Wood N concentrations averaged 7900 ug g'1 in the
                        Little R. studies, compared to average sapwood val-
                        ues of about 900 ug g"1 in the Rhode R. study. Fail et
                        al. (1987) used branch wood samples to represent the
                        entire woody biomass of the tree and so overestimated
                        N accretion in wood. Based on a number of studies,
                        they pointed out that bole wood N content averaged
                        about 43% of branch wood N content. This correction
                        would bring the net wood accumulation of N down to
                        about 22 kg N ha"1 yr1. Total N uptake would be
                        about 84 kg N ha'1 yr1 if this correction is applied.
                          Denitrification has been  shown to be an important
                        N removal process in Coastal Plain riparian forests ei-
                        ther: 1) through indirect  measurement using the
                        acetylene inhibition technique; 2) through measure-
                        ment of environmental conditions which control den-
                        itrification (Eh,  water-filled  pore space, N and C
                        availability) and verifying that proper environmental
                        conditions exist; or 3) through measurement of deni-
                        trification  potential  (Ambus and Lowrance, 1991;
                        Lowrance  et al., 1984b; Hendrickson 1981, Jacobs
         Above-ground woody vegetation uptake of N and P in Coastal Plain riparian forests.
                                                            --kg  ha"1 yr1
Correll & Weller, 1989
Peterjohn & Correll, 1984
Failetal., 1986, 1987
Fail, 1986 (maximum)
Fail, 1986 (minimum)
Rhode R., MD
Rhode R., MD
Little R., GA
Little R., GA
Little R, GA
12 to 20
12 -~
3 to 5
 ' ND = not determined

 and Gilliam, 1985,b; Correll et al., 1994; Jordan et al,
 1993, Lowrance, 1992). The general conclusion of all
 these studies was that denitrification occurred in most
 riparian forest soils, especially in the root zone, or
 that  conditions were'favorable for denitrification.
 Recent work by-Bohlke and Denver (in press) indi-
 cated that denitrification can also occur in sediments
 beneath the influence of the riparian root zone.
    Denitrification was/measured in  riparian forests
 of Little R., GA in conjunction with water quality and
 hydrologic  measurements (Hendrickson  1981). A
 total of 1114  soil cores,(0 to 10 cm), were taken
 monthly for a year from 6 riparian forest sites on the
 Heard Cr. tributary of Little R. Summarized data from
 these ;studies were'used to estimate a denitrification
 rate of 31 kg N ha'1 yr1 for the top 5.0 cm of soil for
 the entire riparian zone of the watershed (Lowrance et
 al., 1984b). Denitrification rates under conditions of
 high N and C subsidy from a swine operation ranged
 up to 295 kg N ha -1 yr1 (Hendrickson, 1981). Lowest
 denitrification rates (1.4 kg N ha'1 yr1) were mea-
 sured in a riparian zone adjacent to an old field which
 received no fertilizer application. Hendrickson (1981)
 found that  the  active cores (those producing N2O
 above background levels) ranged from 1.1% to 6.6% of
 the cor.es taken, depending on the site. This study con-
 firmed the potential for denitrification in surface soils
 as well as the high variability to be expected in field
 measurements  of denitrification.  Soil  cores taken
 to 50 cm in 10 cm increments showed that, except
 near the stream channel, denitrification activity below
 20 cm depth was much lower than activity in the-top
 20cm.   -..'.....'"
     Later studies from Little Ri, GA have also shown
 that  denitrification potentials at the top of the water
 table are measurable, but very low (Lowrance, 1992).
 Nitrate which moves into upper soil layers is likely to
 be reduced by denitrification. Nitrate moving through
 a restored RFBS was reduced by high rates of denitri-
 fication averaging 68 kg N ha'1 yr1. These high rates
 were due to a relict forested wetland soil and move-
 ment of high nitrate water in the root zone (Lowrance
 et al.,  in press). In  addition, nitrate which moves
 through anoxia sediments in riparian zones is  .also
 likely to be reduced. In contrast, nitrate in groundwa-
 ter which moves through generally oxic aquifer mate-
'. rial or nitrate which.does not generally come in con-
 tact with the root zone soil layers is much less likely
 to be denitrified.
    The interaction of vegetation nitrogen uptake, or-
 ganic carbon production via. litterfall and  root senes-
 cence, and microbial denitrification appear to be dri-
 ving nitrate  removal in most Coastal Plain riparian
. forests. Correll  and Weller (1989) proposed a model
 of belpwground processes affecting nutrients (Figure,
 5) .which conceptualized the system' as being con-
 trolled  largely  by oxidation-reduction  conditions.
 Organic matter from decomposing litter and roots
   Cropland,  Forest
       Boundary _:
FIGURE 5. Conceptual model of below ground processes'affecting groundwater nutrients in riparian forest
(from Correll and Weller, 1989),

serves as an energy source and oxygen is consumed
through aerobic respiration, followed by nitrate re-
duction, followed by sulfate reduction when condi-
tions become sufficiently reduced. In the presence of
carbon rich sediments or relict organic matter hori-
zons,  these processes could potentially proceed with-
out forest vegetation. A similar conceptual model for
nitrate removal in Coastal Plain riparian forests was
proposed by Lowrance (1992). Stratified denitrifica-
tion potential in riparian forests of Little R. indicated
that denitrification coincided with the stratification of
N and C from litter and roots. These findings support
the hypothesis that nitrate removal in RFBS is depen-
dent on interactions in  the forest ecosystem rather
than just a poorly drained soil adjacent to a stream. It
is likely that nitrate removal in all Coastal Plain forest
sites  (where  substantial removal has been  demon-
strated) was due to these complex interactions of veg-
etation and the belowground environment. It should
be  noted, however, that hydrologic conditions in
which groundwater containing nitrate passes through
or near the root zone must be present for this mecha-
nism  to  operate effectively. Although most of the
Coastal Plain studies of nitrate removal were in areas
                        with relatively flat wetland soils near the stream, re-
                        moval often took place in areas immediately downs-
                        lope from the fields on better drained soils.
                            d. Removal of Sediments and Nutrients
                               from Surface Runoff
                           Removal  of nutrients and sediment from surface
                        runoff in the RFBS will be a function of both Zone 3
                        and Zone 2. Sediment and nutrient deposition from
                        surface runoff moving through a Coastal Plain ripar-
                        ian forest has been estimated from direct sampling of
                        surface runoff in the Rhode R. watershed (Peterjohn
                        andCorrell,  1984). Estimates  of sediment deposition
                        have been made based on soil  morphology and 137Cs
                        profiles in Little R., GA and in Cypress Creek, NC.
                        GVFS have been widely studied, with at least one de-
                        tailed study  of fescue buffers in the Coastal Plain of
                        Maryland (Magette et al., 1989).
                           The estimated range of sediment deposition rates
                        in riparian forests is large and apparently somewhat
                        dependent on estimation  technique (Table 4).  Al-
                        though the different methods  give widely  divergent
                        numbers, in all cases sediment deposition accounted
                                             TABLE 4.
                         Sediment deposition in Coastal Plain riparian forests.
Sediment Deposition   Notes
                                              Mg ha"1 yr1
Peterjohn & Correll, 1984    Rhode R. (MD)
Cooper etal., 1987

Cooper etal., 1987
 Lowrance, et al., 1987
Cypress Cr. (NC)      105-315*

Cypress Cr. (NC)    •   35-105*
Cooper et al., 1987         Cypress Cr. (NC)

Lowrance, et al., 1986       Little River (GA)
Little River (GA)


Annual measurements, first order ,
stream, runoff samples

137Cs measurements—forest edge

137Cs measurements^ephemeral &
Intermittent streams

137Cs measurements—floodplain

Watershed based, long term,
sediment delivery ratio, soil

Single  field/forest system 137Cs
 *Based on sediment depths in Cooper et al. (1987) and assumed bulk density of 1.4 g cm-3.

 for 80 to 90% of gross erosion  from the uplands.
 Relatively low overall deposition rates (4.2 Mg ha'1)
 reported from direct sampling were associated with
 90% reductions in sediment concentration in 19 m of
 flow through a riparian forest (Peterjohn and Correll,
 1984; Table 6). Sediment deposition estimates need to
 be compared to the gross erosion rates from'cropland
 with information on  the  contributing  area:riparian
. area ratio. With a field: forest ratio of approximately
 2:1,. the riparian forest would attenuate cropland ero-
 sion rates of about 2.1 Mgha^yr1 per year (Peterjohn
 and Correll, 1984). This is well below the tolerance
 value for the upland soils and many fields would con-
 tribute higher sediment loads from erosion. In con-
 trast, a sediment deposition rate of 35 Mg ha'1 yr!.at
 a 2:1 field to forest ratio would attenuate erosion from ;
 cropland contributing up to 17 Mg ha'1 yr1. Very" high
 sediment deposition rates (up to 315 Mg ha'1 yr1) re-
 ported from 137Cs' distribution studies (Table 4) were
 due to high deposition at field edge. This deposition
 was mostly coarse material and did not contain large
 amounts of adsorbed nutrients.
   Sediment removal in GVFS in Coastal Plain areas
 is very effective in relatively short distances (Table 5).
 The RFBS would generally include a grass strip of a
 little more than 4.6 m. If concentrated flow occurs
 across the GVFS, sediment removal is much less effi-
 cient. The  grass  strips also become less effective
 when multiple rainfall events take place in a few days
 or when sediment begins  to accumulate  and forms
 berms which can lead'to channelized flow (Magette et
                                                    al., 1989). Field evaluations of GVFS indicated that
                                                    they were more effective  in Coastal Plain areas of
                                                    Virginia than in steeper topography  (Dillaha et al.,
                                                    1989b). Slopes in Coastal Plain areas  were more uni-
                                                    form and field reconnaissance  indicated that signifi-
                                                    cant portions of stormwater runoff entered the GVFS
                                                    as shallow uniform flow. These GVFS needed regular
                                                    maintenance (sediment removal and possible revege-
                                                    tation every 1 to 3 years) because of the amounts of
                                                    sediment deposition (Dillaha et al, 1989b).
                                                       Nutrient removal from surface runoff has received
                                                    very limited study (Tables  5 and  6). The 4.6 m -filter
                                                    strips used by Magette et al, in the Maryland Coastal
                                                    Plain generally did not remove total N from surface
                                                    runoff and removed only 27% of the total P load. The
                                                    9.2 m filter strips had total N and P removals of nearly
                                                    50%. Peterjohn and Correll (1984) reported concen-
                                                    tration reductions of 74% for total N and 70% for total
                                                    P in flow through 19 m of mature riparian forest in
                                                    Watershed 109 of Rhode R. (Table 6). This width of
                                                    forest would be very similar to a Zone 2 which con-
                                                    formed to the RFBS specification.
                                                     .'   Data from Magette et  al.,  (1989) .and Peterjohn
                                                    and Correll (1984) have been combined to estimate
                                                    the /effects of combined Zones 3  and 2 on sediment
                                                    and nutrients in surface runoff (Table 6). The GVFS
                                                    of Magette et al. are analogous to Zone 3 and the 19
                                                    m of mature forest from Peterjohn and Correll is anal-
                                                    ogous to Zone 2. These widths, 4.6 m and 19 m, are
                                                    almost the exact widths specified in Welsch (1991) for
                                                   /Zones 3 and 2, respectively. Applying the 89.8% sed-
             Inputs, outputs, and % removals of sediment (total suspended solids), total N
            (particulate + dissolved), and total P (particulate + dissolved) from experimental
        Ky 31-Fescue vegetated filter strips in Maryland Coastal Plain. From Magette et al., 1989.
Strip  '   •"-•---
Width    Total Suspended Solids

        Input    Output   Removal*

m      —-Mgha'1-—      %

4.6     27.2       9.3       66

9.2   ,  27.2       4.9       82
                                            Total Nitrogen
                                        Input   Output   Removal

                                       -—kg ha-1	      %

                                         39.4     41.6      -5

                                         39.4     20.7   ,  47
    Total Phosphorus

Input    Output   Removal

-'—kg ha'1	    ~- %

32.3      23.6      27

32:3      17.4      46
 *Removal (%) = (Input-Output)/Input. Negative removal is percent increase in load after movement of runoff through filter strip.


                                              TABLE 6,
 Effects of different size buffer zones on reductions of sediment and nutrients from field surface runoff.
Buffer   Buffer
Width   Type


4.61     Grass

9.21     Grass

19.02-3   Forest

23.6s    Grass/

28.26    Grass/

Input. Output  Reduction4
Cone. Cone.

input Output Reduction4
Cone. Cone.

                                                        Input Output  Reduction4
                                                        Gone. Cone.
— mg L'1—

 7284  2841

 7284  1852

 6480   661

 7284   290

 7284   188





                             14.11 13.55

                             14.11 10.91

                             27.59  7.08

                             14.11  3.48



--mgL'1 —
i L
11.30 8.09
11.30 8.56
5.03 1.51
11.30 2.43
                             14.11   2.80     80.1        11.30  2.57
Calculated from masses of total suspended solids, total N, total P, runoff depth, and plot size (22 x 55 m) from Magette et
2lnput concentrations from Table 2, Peterjohn & Correll (1984). Nitrogen = Nitrate-N + exch. part, ammonium + diss. am-
 monium + part, organic N + diss. organic N. Phosphorus = part. P + diss P.
3Surface runoff concentrations at 19 m into forest reported by Peterjohn & Correll (1984). N and P constituents same as
 Input (footnote 2).
^Percent reduction = 100 * (lnput-Output)/lnput.
54.6 m grass buffer plus 19m of forest.
69.2 m grass buffer plus 19 m of forest.                                    ,
iment concentration reduction found in Peterjohn and
Correll (1984) to the output from a 4.6 m grass buffer
(2841 mg L"1) yields a sediment concentration of 290
mg L'1 from the 4.6 m grass and 19 m of forest (Table
6). This is an overall reduction of 96%. Applying the
same approach to total N and total P yields an output
concentrations of 3.48 and 2.43 mg L'1, respectively.
These  represent concentration  reductions of 75.3%
and 78.5% for total N  and total P,  respectively.
Increasing the width of the grass buffer to 9.2 m
would increase sediment retention by  1.4% of input,
N retention by 4.8%, but increase P  concentrations
    Although a number of experimental studies are
ongoing which link grass filters and riparian forests
for sediment  and  nutrient  removal  from surface
runoff, most have  only made preliminary reports.
Parsons et al. (1994) reports sediment load reductions
of 80 to 90% of field edge loads for both 4.2 and 8.5
                                m,Ky-31 fescue buffer strips at a lower Coastal Plain
                                site in North Carolina. Cutover riparian forests (per-
                                haps analogous to early natural regeneration in Zone
                                2)  showed somewhat  higher  sediment and total N
                                yields than the 8.5 m grass strips. In a study of Zone
                                2 management on a tributary of Little R., Georgia,,
                                sediment loads in surface runoff entering the stream
                                channel system were significantly higher from a clear
                                cut Zone  2 than  from a mature  or thinned Zone 2
                                (Lowrance et al., unpublished). Although these results
                                are preliminary they suggest the importance of the
                                GVFS in Zone 3 during the early regeneration of
                                Zone 2 after tree harvest. In a study of a reestablished
                                RFBS, Vellidis et al.(1993), reports consistent but rel-
                                atively minor reductions in PO4-P in surface runoff in
                                the first year after establishment of slash pine in a re-
                                stored riparian forest buffer system in Little R. water-
                                shed.          ,

    3.  Conclusions        ,

       For purposes of estimating riparian ecosystem
    functions in other physiographic regions^ results from
    Inner Coastal Plain RFBS, probably represent the
    upper limits for NFS pollution control in naturally oc-
    curring riparian forests equivalent to Zones 2 and 1.
    Other naturally  occurring  Coastal Plain  and  non-
    Coastal Plain systems are likely to be less effective
 , ,  than Inner Coastal Plain RFBS because of groundwa-
    ter flow paths that bypass the riparian zone. Although
,   numerous  questions  remain, the understanding of
    Coastal  Plain riparian systems  is much  advanced
   compared to other portions of the CBW..
       The ratio of source areas to RFBS which is re-
   quired for continued improvement in water quality
   can probably be greatest in ICP conditions. Under op-
   timum hydrologic conditions, such as the ICP, where
   groundwater moves in shallow pathways through nat-
:  urally occurring riparian forests, a ratio of 2:1 or 3:1
   (upland to riparian) is typical. These are the types of
  .systems where some of the first data linking riparian
   forests and water quality were  collected. However,
   data on nitrate concentration reductions suggest that
   much of the removal occurs within a relatively narrow
   ecotone at the field edge, implying that the ratio of
   field/forest can be increased; Management of upland
   source areas, to reduce NFS pollutants, andof RFBS,
   to increase effectiveness of removal of NFS  pollu-
   tants, should provide opportunities for  raising the
  , ratio of cropland to RFBS.
      Ongoing research on managed and experimental
   RFBS in the Coastal Plain suggests that restoration of
  the NFS pollution control function can be rapid, espe-
   cially when nitrate moves through relict wetland soils.
  These studies also confirm the need to control chan-
  nelized flow and to use an effective GVFS for sedi-
i  ment control when Zone 2 trees are harvested.
   1.  General Land Use and Hydrology
     The Piedmont Province is an upland region lying
   between the Coastal Plain and the Valley and Ridge
   Provinces at elevations ranging from 30 to 300 m. The
   Piedmont  accounts for 23% of the Chesapeake
   Drainage or 32,600 km2 (NCPJ Chesapeake, 1982).
   Of this area,'49% is in, woodland, 25%  is used as
   cropland, 4% is wetland, and 21% is in other uses.in-
   cluding pastures, and suburban and urban land uses
   (NCR!  Chesapeake, 1982). Of  the  total cropland
  within the Chesapeake drainage, 25% lies within the
  Piedmont                                    '.
     The Piedmont is underlain primarily by metamor-
  phic Precambrian and early Paleozoic rocks subject to
  several episodes of folding. The majority of Piedmont
  basementjmaterials are quartzites, gneisses, schists,
  and marbles. These rocks were metamorphosed from
  ancient sandstones, gabbros and granites, shales,  and
  limestones, respectively. During the Paleozoic, these
  basement rocks were interspersed with igneous peg-
  matite intrusions, and portions were covered by sedi-
 • mentary  deposits during  the  Triassic era.  In
  Pennsylvania and Maryland, the  marble belts form
  valleys; the gneiss,- schist, quartzite, and granites form
  uplands (Hunt, 1974). Pavich et al., 1989 described
  the upland residual mantle (regolith) of Fairfax Co.,
  VA  as  representative   of the  outer   Piedmont
  Crystalline Province  of Virginia and  Maryland
  (Thorhbury, 1965). The area has a high drainage den-
  sity with most of its perennial streams incised into  uh-
  weathered bedrock. .
   , Given the great age of the rocks, the high degree of
  weathering, and absence of quaternary glaciatipn,  the
  regolith (weathered rock, saprolite, subsoil, and soil)
  in< the Piedmont can be quite deep. The maximum
  thickness  of regolith is beneath flat upland hilltops.
  On schist, gneiss, and granite it is typically 15  to 30
  m deep. Rocks such as serpentine and quartzite which
  weather slowly have  thin-regolith (Pavich et  al,
  1989). Throughout, the outer Piedmont Crystalline
  Province,  unweathered bedrock crops out in streams
  and regolith is generally thin or absent in valleys of
.  perennial  streams (Pavich et al.,  1989). The contact
  between weathered and unweathered rock can be  es-
  timated on the side slopes of valleys by the location of
  heads of perennial   streams  at  minor  springs.
  Groundwater drains along the contact between weath-
  ered and unweathered  rock  and enters  surface flow
  through springs (Pavich et al., 1989).  Most of the
  groundwater storage in  the Piedmont is within the  re-
  golith above.the unweathered bedrock (Pavich et al.,
  1989). The saprolite acts as a relatively porous reser-
 voir for groundwater. To a large extent, the depth of
 the  regolith controls the hydrology of most Piedmont
 areas.         .                 •
   Based on hydrograph separations .in the Piedmont
 of Chester County, Pennsylvania, Sloto (1994) found
 that basefiow ranges from 57 to 66% of watershed
 discharge, similar to  estimates  for  the Virginia
 Piedmontof 60% (Pavich et al., 1989). The remainder
 of streamflow occurs during and following storms,

but the proportion that is surface runoff, as opposed to
enhanced subsurface flows (e.g., through near-stream
rise in groundwater,  drainage  from soil layers, or
rapid  lateral transport through macropores), is diffi-
cult to determine. In forested watersheds, very little
surface runoff occurs except from near-stream zones
of high soil moisture. However, cultivated fields in the
Piedmont generate greater surface runoff than fields
in lower gradient Coastal Plains.

2. Control of Nonpoint Source Pollutants

    Direct studies of NPS pollution control by RFBS
have generally begun  since 1990 so most results are
preliminary at this point. The complex hydrogeology
of the Piedmont Province will make generalizations
from ongoing studies difficult even when final results
are available. Most of the discussion to follow will
focus on preliminary results from the North Carolina
Piedmont which are most applicable to the southern
portion of the  Piedmont in  the  Bay  watershed.
Discussions of the geohydrology of the Piedmont and
recent studies  of  the  sources of water  reaching,
streamflow will also be used to make inferences about
the roles of RFBS in this province.

   a.  Removal of Nitrate from Groundwater

   Groundwater in  the Outer Piedmont Crystalline
Province drains  along the contact between weathered
and unweathered rock and discharges through  springs
(Pavich et  al., 1989). There are thought to be three
pathways for groundwater discharge. In valleys un-
derlain by weathered saprolite (often near headwa-
ters), flow  through  the saprolite dominates baseflow.
Water in the flow system is often oxic and may dis-
charge nitrate directly to the stream channel.  In val-
leys where streams have cut through the regolith to
bedrock, springs begin in the  valley flanks.  Where
streams have eroded to bedrock, discharge from frac-
tures in  the  bedrock  also contribute to streamflow.
Stream discharge from the bedrock groundwater sys-
tem is bypassed by the shallower systems  if the re-
golith is not  entirely eroded away. Even where
bedrock contributes,  most of the water in  streams
originates in the regolith because the volume of water
in storage  is so much greater  than in the fractured
    Most groundwater recharge in the marble valleys
occurs rather rapidly  into fracture zones close to the
land surface. The regolith, although variable in thick-
ness, is usually thin and discharge to streams  is prob-
  ably from discrete fracture zones (in springs or di-
  rectly into stream channels). As a result, there is prob-
  ably little interaction of the groundwater with the root
 . zone of riparian systems in the marble valleys.
      In one study of nitrate transport in the Maryland
  Piedmont, McFarland (in press) found that streams
  contained nitrate concentrations of 5 to 10 mg NOs-N
  I/1. Most of the nitrate was attributed to discharge of
  water that was 0 to 5 years old from springs and from
  shallow flow systems in the regolith. Water in the
  bedrock was 20 to 30 years old with low or zero ni-
  trate  concentrations. Denitrification was suspected
  along older flow paths because of low dissolved oxy-
  gen and  high iron concentrations in the water. This
  study indicates  that riparian  systems  with deeply
 ' rooted vegetation may reduce nitrate.in streams by re-
  moving nitrate from spring flow and the shallow flow
  systems through the regolith.
     The only experimental study from the Piedmont
  that addresses the effectiveness of riparian buffers in
  mitigating subsurface flows of nonpoint pollutants is
  that of Daniels and Gilliam (in press), in which spa-
  tial and temporal patterns of groundwater nitrate at
  three sites in  North  Carolina  were  examined,.
  Cultivated fields were separated from ephemeral or
  intermittent stream channels by 3 to 20 m of grass and
  naturally forested riparian buffers. Nitrate concentra-
 ' tions in groundwater under the cultivated fields ex-
  ceeded  10 mg I/1, but declined to lower levels in
  downslope wells. At one site, concentrations declined
  by as much as 30 mg I/1 over a distance of 20 m from
  the field edge. The study did not include mass balance
  analyses of nitrogen losses, and interpretation is com-
  plicated by the fact that  streamflow nitrate concentra-
  tions exceeded those in near-stream wells. Thus, the
.  authors were unable to partition actual nitrogen re-
  moval within the'riparian zone from mixing (or dilu-
  tion) effects, although they speculated that both were
     The results of Daniels and Gilliam  (in press) are
  consistent with  findings from Coastal Plain  studies
  showing that high rates of nitrogen removal occur in
  areas with high water table conditions and shallow
  groundwater movement near the root zone. This sug-
  gests that the effectiveness of RFBS at particular sites
  throughout the Piedmont will depend strongly on the
  flowpaths of subsurface water in and near the riparian
  zone. Whatever the outcome of additional site-sper-
  cific studies, it seems likely that regional estimates of
  RFBS effectiveness will also require  data regarding
  hydrologic properties of near-stream zones.

    The Piedmont is topographically diverse. In areas
 of gentle slopes  and broad .alluvial floodplains, the
 depth to ground-water in near-stream areas is probably
 1  to  2 m 'as was the  case  in  the North Carolina
 Piedmont study. In such areas there may be ample in-
 tersection of the saturated zone with the» root system
 of riparian vegetation. This may also allow interaction
 between the saturated zone and soil  layers containing
 sufficient 'organic matter to induce  rapid denitrifica-
 tion.  However in upland areas, the'water table typi-
 cally lies 3 to 10m below the ground surface. In areas
 of steep terrain, it is common for smaller streams to
 be incised in relatively steep valleys. Near the  fall
 line,  larger streams also tend to have steep valley
 walls and minimal floodplains. Under these circum-
 stances, the area of riparian forest in which the soils
 and root zone intersect the water table may be quite
 small.                       '             .
    Perhaps equally; as important as water table eleva-
 tion is depth or thickness of the aquifer  in the near
 stream zone. While the bulk of subsurface water stor-
 age in the Piedmont occurs.in the regolith, which may
 vary in depth from less than one meter to approxi-
 mately  3 0 m, substantial  storage  occurs within a
 deeper  zone of unweathered  but fractured bedrock.
 The  depth of .the fractured  zone^  as  indicated by
 well-water yields, may range roughly from 60 to 200
 m, depending on  rock type (Sloto, 1994). The likeli-
 hood, that water  reaches streams via  shallow  path-
 ways, therefore, would depend both on the depth of
 the regolith in the vicinitylof the stream,.and on the
 proportionate contributions to streamflow from the re-
 golith and from  the fractured zone. Olmstead and
 Healy (1962)  concluded from analyses of temporal
 patterns in, baseflow and water table  elevations in the
 Brandywine Valley of Pennsylvania that most stream-
 flow originated from the regolith. Rose (1992, 1993)
 reached a similar conclusion from analyses of tritium
 variations in. streamwater  and groundwater in  the
 Georgia Piedmont. If the regolith is  beneath alluvial
 deposits, near, streams, much  of the water reaching
 streams may: pass through the riparian zone at sub-
 stantial depths.            '
   Another aspect of subsurface water movement that
may prove important to RFBS effectiveness is the po-
tential for lateral  flow through near-surface soil lay-
 ers. Lateral ;downslope water  flow  through  unsatu-
rated  or briefly saturated soils may occur through
macropores  (Bevin and Germann,  1982) or where
vertical  flowis impeded by a soil horizon of low per-
meability  (Gaskin et  al., 1989;  Schoeneberger and
 Amoozagar,  1990). There has been considerable, in-
 vestigation of shallow lateral drainage in .other re-
 gions (e.g*, Mosley, 1982; Mulholland et al.,  1990;
 McDonnell,  1990)  but only a few studies from the
 Piedmont Province.
    Hooper   et  al.  (1990)  used  end-member-
 .mixing analysis (EMMA) of water chemistry to dis-
 tinguish water sources in a Georgia Piedmont water-
 shed. Their model  used alkalinity, sulfate, sodium,
 magnesium, calcium and dissolved silica to identify
 three water sources: an organic soil horizon, hillslope
 drainage through subsoil and saprolite, and ground-
 water  in  bedrock.  They concluded  that hillslope
 drainage contributed a large portion of both baseflow
 and stormfiow drainage  during the  wet winter
 months. Groundwater dominated the baseflow during
 the dry summer months with significant contributions
 from the organic horizon during storms. Comparable
 results were obtained by Rose (1992,1993) in another
 study in the  Georgia  Piedmont. Rose inferred from
 analyses of tritium arid inorganic analyses, that while
 baseflow during dry summer months originated from
 groundwater with an average residence time of 15 to
 30 years, higher winter baseflows included a substan-
 tial component  of water with a much shorter resi^
 dence time (less than 10 yr) and lacking the chemical
 signature of groundwater.
    In the North Carolina Piedmont," Daniels and
 Gilliam (in press) noted that soil water in an alluvium
 overlying saprolite was chemically distinct and appar-
 ently isolated from thex deeper groundwater in  the
 saprolite. They attributed the isolation to low perme-
 ability of the Bt soil horizon (subsoil compacted by
 tillage), and inferred that water in the soil layers trav-
 eled laterally above the Bt horizon into the riparian
 zone. Kaplan and Newbold (1993) hypothesized ex-
 tended  periods  of  soil  water drainage following
 storms to explain patterns of dissolved.organic carbon
 concentrations in a Pennsylvania'Piedmont stream.
 The Bt horizon is well developed in Typic Hapludalfs
 and Typic Hapludults,  soil groups.which are common
 throughout the Piedmont, particularly in agricultural
 areas. In and near riparian zones, Aquic Fragiudults,
 are common. The fragipan associated with the latter
 soils probably also promotes lateral flow.

   b. Removal of  Sediment and Nutrients in
     Surface Runoff

   The ability of RFBS  to reduce nonpoint-source
pollutants in overland flow may be of greater signifi-
cance in the Piedmont than in the Coastal Plain be-

cause the steeper topography promotes greater veloc-
ities of overland flow. Daniels and Gilliam (in press)
studied sediment  and chemical reduction by GVFS
and riparian areas for two years at six sites in the
North Carolina Piedmont. They reported that the total
sediment load reduction by the vegetated buffers dur-
ing the study period ranged from 30 to 60 percent.
However much of the sediment (mostly sand) passed
through the vegetated buffers during one storm. When
the results of that one storm were omitted from the
calculations, the buffers removed  approximately
80 percent of the sediment. Removals of silt plus clay
averaged approximately 80% for the two-year  study
period. Total P removals in the filters ranged from 50
to 70%. Soluble orthophosphate removal was highly
variable and usually was 50% or less. Removal of var-
ious forms  of N was also  variable  and generally
ranged from 40 to 60%. Most of the reductions were
observed within 7 m of the field edge. The authors
noted that the slope of the GVFS was less than that of
the fields, so some of the sediment removal could be
attributed to the change in slope alone. They further
cautioned that the effectiveness of GVFS on steeper
slopes might be limited. Where runoff from fields was
directed as concentrated flow into riparian areas with-
out complete vegetative cover, little or no reductions
in either sediment or nutrients were observed.  From
these observations, Daniels and Gilliam (in press) rec- .
ommended upslope dispersal of drainage water di-
rected into forested areas.
   Parsons et al. (1994a, b) conducted plot-scale stud-
ies in the North Carolina Piedmont on sediment and
nutrient removals in grass  and forest vegetated filters.
They used 4 and 8 m grass buffers and 4 and 8m
forested filters to determine removal efficiencies. To
date, they have monitored 50 storms over a three year
period. They have observed that grass filters were
somewhat more effective  for sediment removal than
the forest filters because of greater tendency for chan-
nelization in the forested area. Comparison of the
grass  and forest buffers  is difficult because slopes
were 4 to 6% in the grass filters as compared to slopes
of 12 to 16% in the natural forested area. There was
generally more ground cover on the grassed plots than
in  the forest, especially after grasses were reestab-
lished in grass buffers.
   Preliminary data from these  studies are available
for a maximum of four storms in 1991 (Parsons et al.
 1994b). They found reduction of field edge sediment
loads was consistently over 90% for both 4.3 m and
8.5 m forest buffers.  Sediment loads were reduced
94% in the 4.3 m forest buffer (three storms) and 92%
in the 8.5 m buffer (two storms). Reductions of ni-
trate, total Kjeldahl N, ortho-P and total P were more
variable in these riparian forest  buffers in the four
1991 storms. Although data are not available for all
storms, it appears that the tendency to have channel-
ized flow through the riparian forest area caused the
high variation. For the available storm data lumped
together, nitrate was reduced 41% compared to edge
of field load for the 4.3 m forest buffer (four storms) ,
and reduced 63 %  in the 8.5 m buffer (two storms).
Total kjeldahl N was reduced 67% in the 4.3 m buffer
(three storms) but increased 14% in the one storm
monitored in the 8.5 m buffer.  Ortho-P, all of which
was dissolved, decreased 6% in the 4.3 m forest buffer
(three  storms)  but  increased  17%  in  one  storm
through the 8.5  m forest buffer. Total P (sediment-
bound + dissolved) decreased 50% in the 4.3 m buffer
(three storms) but only 17% in the 8.5 m buffer (one
storm).  Although these data are preliminary, they
show similar trends as some  of the Coastal Plain
runoff data with good control of sediment and sedi-
ment associated P' but variable control of dissolved
nutrients in  surface  runoff,  especially dissolved P.
More complete data from these  studies should help
guide design of RFBS for Piedmont landscapes.

3.  Conclusions
    Limited data from riparian forest studies in the
Piedmont makes, quantitative estimates of the NPS
pollution control functions difficult. Patterns similar
to ICP results seem to be present in studies from the
North Carolina Piedmont with good control of nitrate
in shallow flow paths and good control of sediment
and sediment-borne pollutants in surface runoff.
   Knowledge of the hydrology of certain parts of the,
Piedmont, such as the marble valleys of Pennsylvania
and Maryland indicate a minor role for RFBS in con-
trol  of groundwater borne  pollutants.  On  smaller
streams and in areas with thinner regolith, it  appears
that shallow groundwater movement through the sub-
soil and saprolite may be affected by RFBS. Buffer
systems in the headwaters of streams where springs
enter surface runoff may be effective, especially if the
RFBS promotes the eventual presence of high organic
matter  soils  in the areas  of springs and permanent
groundwater seeps.  RFBS  probably also intercept
water moving in relatively shallow flow paths above
texture discontinuities which promote lateral move-
ment in the  soil and subsoil. If extended periods  of
 soils drainage above these texture discontinuities does

 occur, these waters should be subject to nutrient re-
 moval rates in RFBS similar to those in the ICP situ-
 ations.                                 ,
    Although data are also limited on effects of RFBS
 on surface runoff, preliminary results indicate that the
 slope of RFBS  may limit effectiveness because, of
 channelization  through forests.  Relatively  steep
 RFBS will certainly benefit from the presence of a
 well managed Zone 3 at the field edge and may re-
 quire level-lipped spreaders,to control the tendency of
 surface runoff to create permanent channels. In rees-
 tablished areas on relatively steep slopes, such as  the
 12 to 16% slopes reported on from North Carolina, a
.high stocking density of trees in Zone 2 is warranted.
 This would have the effect of both increasing resis-
• tance to surface  flow by increased numbers of stems,
 as well as providing a high level of root biomass more
 quickly than lower stand densities.


 1.  General  Land Use and Hydrology

.'   Valley" and  Ridge physiographic province is  the
 area in which structures due to folding dominate  the
 topography. The Valley  and, Ridge and Appalachian
 Plateau make up about 60% of the CBW (Table  1).
 Geomorphologically, the Valley and Ridge province is
 one of folded mountains in which resistant strata form
 ridges and weaker rocks are-worn down to lowlands.
 Valleys within this province are underlain by lime-
 stone or shale and the ridges are capped by the moire
 resistant  rocks (well-cemented  siliceous sandstone
 and conglomerate).          .           .
    The physical characteristics of this province are in-
 timately connected with its-streams which are primar-
 ily causes of the present topography. Streams develop '
 mostly on belts of soft rock crossing hard rock ridges
 infrequently and usually at right angles. The .Bay wa-
 tershed encompasses the middle  section of the Valley
 and Ridge. Distinctive features of this section are con-
 spicuous trellised drainage patterns and a comparative
 absence of ridges on its southeastern one-quarter to
 one-third, the Great Valley (Fenneman, 1938).
     Heath (1984) placed the Valley and Ridge in the
' Central Nonglaciated groundwater region. The region
. is  characterized  by thin regolith underlain by frac-
 tured sedimentary bedrock. The principal water-bear-
 ing openings in the bedrock are fractures which de-
 velop both  along bedding planes and across them at
 steep angles. Openings developed, along the fractures
 are usually less than 1  mm wide. The principal ex-
»   ,     i                      /   -
 ception to this is in limestone,, where water moving '
 through the original fractures has enlarged them to
 form, at the extreme, extensive cavernous systems ca-
 pable of transmitting  large amounts of subsurface
 flow. Recharge of groundwater in this region gener-
 ally occurs in outcrop areas of the bedrock aquifers in
 the  uplands between  streams. Discharge  from the
 groundwater system,is by springs, seepage areas, and '
 direct inflow to the stream bed, and by evaporation
 and transpiration in the near-stream areas where the
 water table is near the land surface.     '     .'••.'
   . The aquifers, in  the Valley and Ridge are uncon-
 fined with little matrix permeability and low storage
 coefficients.  Groundwater circulation is limited at '
 depths greater than 100 m due to the decrease in frac-
 ture size and frequency. Even though the  entire Valley
 and Ridge  falls within  the Central Nonglaciated
 groundwater region^- there are substantial differences
 in flow  characteristics between  the limestone and
 shale valleys, arid among the limestone valleys. Flow
 characteristics are most complicated within the lime-
 stone aquifers and  connections between lower-order
 streams and regional groundwater are quite variable
 in time and space.                 •".;-.

 2. Control of Nonpoint Source  Pollution
    Despite the Valley and Ridge and  Appalachian
 Plateau comprising a large portion of the CBW, only
 a small number of research projects have been con-
 ducted to evaluate the amelioration of NFS  pollution
 in riparian buffers within this area. These projects ad-
 dressed within-stream water quality (e.g., cold water
 fisheries habitat, macroinvertebrates, and sedimenta-
 tion) and Bay-scale water quality (export of plant nu-
 trients, pesticides and suspended solids).
   The entire study of riparian ecosystems relative-to
 stream quality was begun within the  last 20 years.
 That little of it was conducted in the Valley and Ridge
 may be explained by the small amount of wetlands in
 this  province. The conditions  which promote im-
 provements in the  chemical composition of waters
 discharging through riparian ecosystems (small land
 surface slopes, high water tables and low  aeration sta- •
 tus)  are commonly associated with wetlands. Only
 1% of the Valley and Ridge located in  the CBW is
 classified as wetlands, which constitutes 7% of the  •'.
 wetlands in the CBW (Table 1). This contrasts with
 the 57% of CBW wetlands on the  Coastal Plain com-
 prising 21% of the Coastal Plain within the CBW.
.   The most intensive agricultural NFS pollution oc-
 curs in the limestone valleys of the Valley and Ridge.

Study of subsurface hydrology necessary to deter-
mine the extent of groundwater renovation in the ri-
parian zone is difficult and expensive in karstic lime-
stones. A paucity of studies in the Valley and Ridge
may be due to the major source  of NFS pollutants
being located over an aquifer with complex hydrology
for the scale of riparian zone studies.
   The processes which renovate surface and ground-
water within riparian ecosystems are the same in the
Valley  and  Ridge  as  in the other physiographic
provinces and much can be inferred from research
done in the Atlantic Coastal Plain. When watershed
mprphology and aquifer characteristics are compared
(Schnabel et al  , 1994), general  statements'can be
made about  the  likelihood of ground and  surface
water renovation in riparian zones of the Valley and
Ridge relative to the Coastal Plain. However, research
conducted within the Valley and Ridge must be eval-
uated to quantify the impact of riparian buffers on
stream hydrology, chemistry and biology.

   a. Removal of Nitrate from Groundwater
   The Mahantango Creek Watershed, a USDA-ARS
research  watershed,   is  located   within   the
Susquehanna R. Basin approximately 40 km north of
Harrisburg, Pennsylvania. Topography,  geology, and
land use of the Mahantango Creek Watershed are typ-
ical of upland watersheds in the unglaciated, intensely
folded and faulted Valley and Ridge Province. These
watersheds  generally  have relatively steep land-
surface slopes and minimal floodplain development
or alluvium. Most stream reaches  expose bedrock
over all or part of their length. Land use within the
watershed is approximately 57% cropland, 35% forest
and woodlots, and 8% permanent pasture. Elevation
ranges from 240 to 480 m msl. The northern  ridge is
covered with a mature deciduous forest, while agri-
cultural land use predominates .in the remainder of the
watershed. Climate is humid and temperate, and rain-
fall averages about 1150 mm yr1.
   Groundwater provides most of the  60 to  80% of
streamflow estimated to be subsurface return flow
(Gburek et al., 1986). Primary recharge occurs in the
late  fall, winter, and early spring  months, but minor
recharge can occur during the growing season follow-
ing  large single or grouped precipitation  events.
Because ridge-top soils are highly permeable, nearly
all rainfall infiltrates. In contrast, the finer-textured
poorly drained soils adjacent to the stream often func-
tion as groundwater discharge zones during the dor-
mant season.
   The Mahantango Creek Watershed is underlain by
two geologic  formations, Trimmers  Rock (Late
Devonian)  and  Catskill (Late  Devonian—Early
Mississippian). Previous analysis of well yields indi-
cated that rock fracture patterns are as important to
formation permeability as rock type, and based on
specific capacity data (Urban, 1977), the two forma-
tions are hydrologically similar. A shallow, approxi-
mately  10 to  15  m layer of  weathered fractured
bedrock overlays the entire  watershed and has hy-
draulic properties different from those of the deeper,
less-fractured layer (Gburek and Urban, 1990). The
two-layer aquifer, with  its upper, highly transmissive
layer, permits rapid horizontal groundwater, through-
flow while also leaking to recharge the deeper layer.
Differing land uses in the area recharging groundwa-
ter, the layered subsurface permeability distribution,
and the general pattern of groundwater flow are ex-
pected to result in a general pattern of higher nitrate
concentration in shallow groundwater and lower .con-
centration in the deep groundwater. In the experimen-
tal area, all aquifer waters, both shallow and deep, dis-
charge to the surface streams.
   Although it is a very small portion of the watershed
area, the near-stream zone exerts major controls on
stream flow chemistry  and hydrology. Because it is
hydrologically  dynamic, particularly  as related to
seep zone formation, the near-stream zone can control
the  amount and timing of surface  runoff and, thus
downstream  flooding.  The water table response to
storms strongly influences or controls subsurface dis-
charge, the nature and  extent of riparian, vegetation,
stream bank stability, "and the nature of the chemical
and biological systems to which chemicals in the dis-
" charge are exposed.
   Nitrogen and phosphorus species were measured in
surface runoff and seepage waters in a grassed buffer
between a first-order stream and a cropped field, dur-
ing  and immediately after storms to determine how
surface and  subsurface waters  interact to generate
streamflow during  storms  (Pionke et al., 1988).
Nitrate concentration in seepage and base flow were
similar and typically exceeded concentrations in sur-
face runoff, rainfall, and peak storm flow by 5 to 20
times. The median nitrate concentration observed in
seepage was similar to mean concentration observed
in stream base flow at the outlet to a 9.9-ha catchment
over a 2-yr period and similar to  those computed from
a hydrologic and nitrogen mass balance for agricul-
tural groundwaters of Mahantango Creek Watershed
(Pionke and Urban,  1984). They concluded that hy-

 drblogic conditions mainly determined nitrate con-
 centration and load delivered 'to the stream over the
 short-term, as hydrology affected  both the surface
 runoff: subsurface discharge ratio and the volume of
 subsurface  discharge. This  indicates  that  surface-
 groundwater  interactions  are more frequent and
 longer lasting on lower portions of the watersheds.
 Consequently, in this and similar watersheds the ni-
 trate content of groundwater is less likely to be altered
 in the riparian zone in  the,upper portions of head
 water streams.   !
   Another study on Mahantango Creek focused on ,
 the role of existing riparian zones. A strip of woods
 (20-60 m) was located between the stream and crop-
 land on both sides at the study site. There was a break,
 in slope on one side of the stream where the land sur-
 face flattens as it approached the stream. The woods
 were  removed from this flatter area bordering the
 stream 15 to  20; years before the study and it was
 seeded to grass. Thus the vegetation pattern moving
 up-gradient on one side of the stream was a relatively
 flat, well-maintained grass  strip, a .steeper strip  of
 woods and then cropland. On the other side a,steep
 strip of woods separated the  stream from cropland.
 Nitrate-N  concentrations in shallow groundwater
 under the grass strip were reduced by 25 to 50% be-
 tween 9 m and 6 m from the stream during the grow-
 . ing season. There were generally small differences in .
• nitrate-N concentrations in shallow groundwater 3 m
 from the stream and baseflow in the stream. The water
 table was frequently deeper  than 1 m, particularly on
 the wooded side of the stream.' The wooded side was
 much steeper and didn't develop seepage zones as fre-
 quently as on  the less steep grassed side of the stream
 (Schnabel, 1986). A pattern similar to the nitrate con-
 centrations measured in the grassed riparian zone was
 found in  deeper groundwater (3  m) beneath the
 wooded riparian zone; Gburek et al. (1986) estimated
 that nitrate reduction within'the riparian zone of the
 Mahantango Creek Watershed was equivalent to only
 4% of the mineral N exported from the watershed dur-
 ing the year. The limited impact of riparian processes
 on .total N export resulted from the small area near the
 stream thought to support denitrification at optimum
 rates,  combined, with the fact that the area generally
 expands after soil  temperatures begin to decrease,
 presumably limiting denitrification  and plant uptake
 rates.    ,   .
  . The chemical composition of the aquifer differs,
 with .depth. While recharge for the deeper part of the
 aquifer originates at the wooded ridge tops, the shal- .
 lower part of the aquifer is recharged in the agricul-i
 rural interior of the watershed. From simulation with
 a mixing nfbdel which, viewed baseflow as a mixture
 of discharge from the shallow fractured part of the
 aquifer  and deeper,  less fractured portion  of the
 aquifer,  Schnabel et al. (1993) concluded that the ri-
 parian zone was  not  the major control on temporal
 variation in nitrate concentration  at  the  outlet to
 Mahantango Creek Watershed.                     "
   A study  designed to examine groundwater nitrate
 dynamics was-conducted in the western portion of the
 Valley and Ridge Province. Altman  and Parizek
 (1994) conducted a study of nitrate movement from a
 field through the riparian zone of a tributary of Bald
 Eagle Creek at the western edge of the Valley and
 Ridge in Pennsylvania. They found that nitrate levels
 in groundwater decreased from 5 to 8 nig NOs-N L'1
 beneath  the field to Jess than 0.5 mg NOa-N I/1 in the
-riparian  zone. Based on flow-net analysis, they con- •
 eluded that water sampled in the riparian zone appar-
 ently did not originate from the crop area with ele-
 vated nitrate levels. The groundwater flow direction
 did not follow the surface topography but instead fol-
 lowed the local bedrock topography.  Groundwater
 was actually flowing  toward  the larger creek which
 the tributary stream was feeding. Their report did not
 address  the,fate of the nitrate enriched water  as it
 moved through the riparian,system associated  with
 Bald Eagle Creek, This study does point out the diffi-
 culty  of research on groundwater and associated
 solute movement in areas of complex hydrogeology
 such as the Valley and Ridge:

   b.  Streamflow Transport of Phosphorus

     Sediment  and water associated phosphorus ex-
 port  from the  Mahantango Creek Watershed and a
 9.9-ha subcatchment was examined for a 4-yr period
 to determine the mode of phosphorus transport from a
 typical Valley and Ridge upland watershed (Piorike
 and Kunishi, 1992). During storms, most of the labile
 P (sum of total soluble P and sediment P extracted by
 Cl. resin) was exported  from Mahantango  Creek
 Watershed in the dissolved rather than the particulate
 phase. The  dissolved  P dominated because the  dilu-
 tion of sediment by runoff (~3000:1) more than com-
 pensated for the greater  concentration of labile P
 compared to soluble  P..concentration (~1000':1). .In
 contrast, storm flow transport of algal-available (sed-
 iment P  extracted by 0.1 N NaOH) total P was mostly
 with-sediment,  largely because concentrations of both
 on sediment greatly exceeded labile sediment P con-

centrations. When combined with P concentrations in
base flow, which accounts for approximately 80% of
total flow, 50 and 28% of algal-available and total P,
respectively, were exported from Mahantango Creek
Watershed in  the dissolved phase. Thus, the most
readily available P components are transported in the
dissolved rather than particulate forms. This has im-
portant  implications for use of RFBS to  control P
losses from agricultural land. If RFBS are more ef-
fective at controlling particulate P than dissolved P,
higher proportions of P in the dissolved phase would
imply less effective  overall control of P transport.

   c.  Removal of  Sediment and Nutrients in
      Surface Runoff

   Studies by  Dillaha  et al.  (1988, 1989a,b) have
shown  the  potential efficacy  and limitations of
grassed filter  strips for controlling NPS pollution.
Near Blacksburg, Virginia, Dillaha et al. (1988) stud-
ied the use of orchardgrass (Dactylis glomeraia)
GVFS for controlling potential sediment and nutrient
losses from feedlots. Plots received 7,500 and 15,000
kg ha"1 of fresh dairy manure and had slopes of 11 and
16%.  In plots with shallow, uniform surface flow, 81
and 91% of the sediment and soluble solids were re-
moved by 4.6 and 9.2 m GVFS, respectively. In plots
where concentrated flow was allowed to occur,  re-
moval was much less. The GVFS were ineffective for
controlling dissolved nutrients and nutrients  associ"
ated with fine  sediment. Concentrations of soluble N
and P in effluents from GVFS were found to be high
enough  to cause eutrophication in receiving  waters.
Concentrations of soluble inorganic N were as high as
8.2 and  5.1 mg N I/1 from the 4.6 and 9.2 m GVFS,
   In  a  similar  study  of orchardgrass filter strips
below fertilized cropland, Dillaha et al. (1989b) ob-
tained comparable results to the feedlot experiment.
The sediment  was initially trapped at the top of the
GVFS. However, the GVFS became ineffective as it
gradually became inundated with sediment.
   In  surveys of farms that employed GVFS along
streams  in Virginia, Dillaha et al., 1989a,b) found that
in Valley and Ridge areas, the GVFS tended to be less
effective than in flatter Coastal Plain sites. Except for
localized  erosion control along  the  stream .bank,
GVFS did little to mitigate NPS pollution from.the
upland in the Valley  and Ridge because surface runoff
usually became concentrated within the fields in nat-
ural drainageways before entering the GVFS.  In gen-
eral, the GVFS were most effective below smaller
fields where water could enter the GVFS before it had
an opportunity to concentrate.
  Even where the GVFS had potential for sediment
trapping, in many cases inadequate maintenance had
rendered them ineffective (Dillaha et al., 1989a). Lack
of mowing sometimes allowed taller weeds to shade
put low ground cover, thereby reducing the capability
of the GVFS to trap  sediment. Erosion across the
GVFS had caused severe gully problems in some
cases. Heavy traffic had sometimes damaged the sod
and created ruts. Sediment buildup on some sites had
caused the upper margin  of the GVFS to be higher
than the adjacent  field. Or sometimes, ditches from
moldboard plowing were created parallel to the upper
edge of the GVFS. In either of these two latter situa-
tions, water would run parallel to the edge of the
GVFS until it could get across it in concentrated flow.

3.  RFBS in  Forested Watersheds

    Although the RFBS is designed for use adjacent
to agricultural areas,  a number of forestry experi-
ments in the Valley and Ridge Province-provide use-
ful general information on hydrology, sediment trans-
port, and sustainability of riparian forest buffers. A
series of experiments were begun in the late 60's and
early  70's  by Forest  Service  personnel to  design
BMPs for logging operations  in response  to  the
Federal Water Pollution Control Act Amendments of
1972 (P.L. 92-500). In many of these  experiments, a
strip  of  trees was left  standing along  perennial
streams to protect  the stream from NPS pollutants.
The  experimental sites  included  locations  in
Pennsylvania, West Virginia, and Tennessee.
  The Leading  Ridge  Experimental Watershed
Research .Unit is  located in the Ridge and Valley
Province of central Pennsylvania and consists of three
adjacent watersheds. BMPs  used on a commercially
clearcut watershed were designed to minimize stream
sedimentation  from silvicultural operations. These
practices included maintaining a 30 m buffer strip on
each side of perennial streams and restricting,slash
piles and log-landing sites'from the vicinity of stream
channels (Lynch and Corbett, 1990). A comparison of
suspended sediment concentrations on the Leading
Ridge Experimental Watersheds for the first two years
after clearcutting shows that the BMPs were effective.
Average suspended sediment concentrations of 1.7,
10.4 and 5.9 mg L'1 were reported for an uncut con-
trol watershed, a clearcut and herbicide-treated water-
shed without BMPs, and a commercially clearcut wa-
tershed with the riparian buffer strip, respectively, for

 the  first year after treatment  (Lynch et al., 1985; '
 Lynch and Corbett, 1990). Water, yield increased fol-
 lowing all clearcutting treatments. The  greatest  in-
 crease, equivalent to 32 cm over the area cut, occurred
 the first year after'clearcutting and herbicide'treat-
 ment to control regrowth. Annual yield was  sharply.
 lower the'second year and was not statistically differ- .
 ent from water yield on the control watershed at the
 end of the fourth year following, harvesting  (Lynch.
 and (Corbett,  1990).   Average  suspended sediment
 concentrations in  the  clearcut/herbicide .treatment.
jumped to  78.7 mg L'1 during the second year after
 treatment compared tp 5.1 and 9.3 mg LT' for the con-
 trol watershed and clearcut watersheds with riparian
 buffers, respectively (Lynch et al., 1985; Lynch and
 Corbett, 1990). The extremely high sediment concen-
 trations on the clearcut/herbicide treatment were at-
 tributed to channel cutting,  and  bank erosion  and .
 slumping  on the  lower portion of the channel.
 Increased sediment concentrations on the commer-
 cially clearcut watershed the second year (9.3 mg L'1 ,
 compared to 5.9 mg L"1) were attributed to tree blow-
 down along a 460 m length  of intermittent stream
 which did not have a buffer along it. The few remain-
 ing trees'in the section blew over and loosened the
 soil.  The increased 'water* yield resulting from the
 clearcut caused the intermittent stream to flow, per-
 mitting the transport of soil to the  stream- channel.
 This  illustrates  the need for  maintaining riparian
buffer strips along intermittent  streams (Lyrtch et al.,
 1985).   ,                  .   ---/':•,••••'
   Average annual' nitrate  concentration  in  the
 clearcut/herbicide watershed  was substantially  in-
 creased compared to the control watershed (2.54 vs.
 0.11 mg NOs-N L'1). Increased nitrate concentrations
 combined  with  increased water yields from  the
 treated watershed resulted in significantly greater N
 loading .than the control. However, rapid revegetation,
 which is almost impossible to  prevent in the humid,
 East, generally prevented any  major stream  enrich-
 ment problems. Where BMPs were used, nitrate con-
 centrations were  substantially less than the clearcut/
.herbicide treatment,  although  significantly  higher •
 than control (0.37 vs. 0.08 mg NO3-N L'1) for the first
two years after clearcut (Lynch et al., 1985).
   Mulholland et al. (1990) studied an area similar to
the Valley and Ridge region of the Chesapeake Bay.
They investigated the hydfogeochemical response of
the West pork of the Walker Branch Watershed in
eastern Tennessee to four large storms. The study area
was a 38.4-ha forested watershed with.deep, highly
weathered soils, a network of ephemeral stream chan-
nels, and a spring-fed perennial stream which flowed
over dolomite bedrock in.the lower portion of the wa-
tershed. The watershed has broad ridges which slope
steeply to harrow valleys. Surface soils have very high
hydraulic conductivities due to macroporosity associ-
ated with forest soil  formation processes. Reduced
hydraulic conductivities at depth in the soil are asso-
ciated with increasing clay content. The  weathered
zone ranges in depth from a meter near the stream to
about 30 m at the basin divides.
   In this watershed, water held above the shallow re-
strictive layer flowed through the rhizosphere and was
virtually depleted of nitrate. However, water passing
through the restrictive layer (apparently the layer was
quite  leaky)  had  higher  nitrate  concentrations.
Grburidwater transferred  between  catchments  or
leaked to deeper groundwater and discharged near the
watershed outlet bypassed the riparian zone closest to
the source  of NFS pollutants. Where these transfers
occurred, groundwater was less likely to be renovated
by riparian zone processes'.

4.  Conclusions
   Forested riparian buffers have proven effective in
controlling water  temperature and sediment delivery
to streams  in forest and agricultural settings within
the Valley and Ridge. Our knowledge of groundwater
renovation  in riparian ecosystems  is less certain.
Where regolith is thin and bedrock controls subsur-
face flow, seepage faces or springs produce relatively
small saturated areas with wetland characteristics.
Attenuation of nitrate concentrations, may occur if
RFBS are restored in these seepage areas. In contrast,
where regolith is  deep with flow-restrictive layers
near the land surface, shallow flow  systems develop
on the confining layers resulting in more extensive ri-
parian ecosystems where groundwater discharges to
the streams.  These conditions, more likely in the
glaciated Appalachian Plateau than in the Valley and
Ridge, are likely to have higher overall nitrate re-
moval rates.    .


             Applicability  of the Three  Zone
             Riparian  Buffer System
   The three-zone RFBS specification is based on
 studies of naturally. occurring riparian forests along
 low order (1st to 4th order) streams and experimental
 scale grass filter strips. Under natural conditions, ri-
 parian forest ecosystems formed a dynamic yet stable
 buffering system .along most shorelines,  rivers and-
 streams in the Bay watershed. Although few studies
 have documented the specific changes in water qual-
 ity functions during the establishment period of a ri-
 parian forest, established RFBS are expected to sus-
 tain water quality functions over the long term in a *
 manner similar to the natural system.
   The effect of upstream activities which modify hy-
 drology or pollutant loads, loading rates, or the
 change in functions due to management of the'RFBS,
, such as timber harvest, add uncertainty and risk to
•predicting changes in some water quality functions
 over time. However, existing research, knowledge of
 riparian ecology, and experience with related hydro-
 logic systems can form the basis for recommenda-
 tions  on the applicability of RFBS. The 12 member
 scientific panel that prepared this report utilized these
 resoufpes to produce the following  set of Best
 Professional Judgements (BPJ) of conditions and cri-
 teria for assessing the effectiveness of the  three-zone
 RFBS for use in the CBW.
   Control of the stream environment will occur in al-
 most all cases along smaller streams with Zone 1 veg-
 etation. The environments of tidal streams, tidal por-
 tions-of the bay, and larger, rivers maybe controlled by
 other  factors  more than the immediate riparian
 ecosystem. The consensus BPJ are:
    1) Gontrol of the stream-environment..for aquatic
       ecosystems is most likely to be achieved with
       vegetation approximating the original native
       vegetation along streams.
    2) Control of the stream environment will be af-
       fected less by physiographic regions than by
       size of stream. As the size of stream or water
       body increases, most effects of the riparian
       system, on the stream environment decrease.
       However, the habitat functions of large woody
       debris are important even on large river banks
       and on Bay shorelines.

    3) Just as Zone 1 may also play an important role
       in NPS pollution removal, Zone 2 may play an
       important supporting role in controlling  the
       stream environment.

   In many cases,  especially along  higher order
 streams, quality of the stream environment will reflect
 the  influence of both Zone  1  and Zone 2.  Sus-
 tainability of Zone 1 function may depend on proper
 management of Zone 2. Where windthrow of trees or
 stream bank stability is a problem, Zone 2 vegetation
 should be managed with long rotations, thinning cuts,
 or other practices which minimize the time and areal
 extent of a non-forest Zone 2. The general goal would
 be  to miniriiize both the  amount  of time and  the
 stream length for which Zone 1 would be the only ri-
 parian forest. Where Zone 1 will function alone, in-
 creased width and/or other adjustments may be re-
 quired to enhance sustainability.
    Unlike the processes involved :With control of the
 stream environment, the functions of riparian systems
,10 control NFS pollution are dependent on hydrologic
 Connection(s) of pollutant source(s) with the riparian
 for,est buffer system. Although generalizations will be
 made, the extent, timing, and spatial variability of the
 hydrologic  connections add uncertainty to BPJ, as-
 sessment of NPS pollution control. The hydrologic
 connection between source areas and riparian ecosys-

terns probably ranges from nearly 100% of the water
moving across the surface or in shallow groundwater
through the biologically active soil zones (e.g., ICP
Watersheds) to a very low percentage of flow moving
through riparian ecosystems. This lower limit is not
well defined, but a conservative estimate can be made
by hydrograph analysis to separate storm flow from
baseflow.  At  a minimum,  most stormflow  should
move in either surface runoff or shallow groundwater
and should be subject to processing in a RFBS.
  For either surface runoff or shallow groundwater,
removal of NFS  pollutants in RFBS is first deter-
mined by the hydrologic pathways and then modified
by interactions of hydrology, soils, geochemical envi-
ronments, management, loading, and vegetation. As
pointed out above in Sections I and II, some of these
factors are poorly understood and most are poorly
quantified, especially outside the ICP from which
much of the existing information is derived.
  As a means of conceptualizing the NFS pollution
control functions of riparian ecosystems in the CBW,
a series of flow diagrams for different physiographic
settings was developed (Fig. 6 through 14). These fig-
ures are generally representative of many of the dif-
ferent hydrologic  settings within the regions and pro-
vide  reference  points  for  discussions  (below)
concerning hydrologic controls on the NFS removal
function. It is important to note  that these diagrams
are generalized and that  more than one hydrologic
setting may be present in larger watersheds. The con-
sensus BPJ decisions  are summarized for nitrate re-
moval, sediment  and sediment-borne pollutant re-
moval, and phosphorus (from all sources) removal in
Tables 7 through 9.

1.  Coastal  Plain
   a.  Inner Coastal
   The best information on RFBS comes from Coastal
Plain systems represented by Figure 6. In these ICP
systems, most of''the excess  precipitation moves to
streams via subsurface runoff or shallow groundwater
movement. Most or all of this water moves in or near
the root zone or is subject to capillary transport due to
transpiration from the root zone. The ICP represents
one end of the spectrum of riparian ecosystems func-
tion for removal of NFS pollutants. In these systems,
riparian  ecosystems exert  substantial control. over
both the hydrologic and nutrient transport response of
agricultural watersheds.  ICP  areas,  represented in
Section II by Rhode R. in Maryland and areas in
Georgia and North Carolina, are typically areas with
a high density of stream channels,  well developed
"natural" riparian forests, and extensive connections
between agricultural fields and riparian forest ecosys-
tems. Most of the Western Shore  and the upper
Eastern Shore Coastal Plain in the .CBW is considered
ICP Because of the relatively large amount of scien-
tific data collected from ICP type systems, primarily
in MD, NC, and GA (see Section II), the scientific
panel was able to make the most comprehensive con-
sensus BPJ for these areas. Among these conclusions
are the following:
    1)  Based on mass balances, established RFBS re-
       move 20 to 39 kg NO3-N ha'1 yr1 from sub-
       surface flow.
    2)  Based on mass balances, total N retention in
       established systems ranges from 26 to 74 kg N
       ha'1 yr1.
    3)  For the RFBS to be applicable in systems with
       artificial drainage near streams, the drainage
       system will have to be modified to work in
       conjunction with the RFBS.
    4)  Newly established systems are likely to have a
       substantial  effect on subsurface nitrate loads
       in (at most) 5 to 10 years if anoxic sediments
       and high organic matter surface soils are al-
       ready in place. By  15 to 20 years,  reestab-
       lished RFBS should control groundwater ni-
       trate loads in most (if not all) ICP situations.
       Reestablishment of RFBS along all streams in
       the ICP is likely to lead to water quality im-
    5)' The nitrate concentration data from ICP sys-
       tems indicates  that higher  nitrate loadings
       could be removed in the RFBS if it was ex-
       posed to higher loadings than represented in
       the mass balance studies. This is most likely to
       be true in systems with highest denitrification
       rates or potentials.
    6) Based on the mass balances, net retention of
       phosphorus in established systems is 1.2 to 2.9
       kg P ha-1 yr1. Retention of phosphorus in sur-
       face runoff appears  to be mainly through re-
       tention of particulate phosphorus and infiltra-
       tion in  the RFBS. Retention of dissolved
       ortho-P appears to be considerably less effec-
       tive  for both surface runoff .and subsurface


Management Fat

Restoration/ Enhanct
Critical Constraints





Maintain vegetatjon/
denitrification link.
M J3
Important on all strearr
restoration of function
Bypass flow due to
subsurface drains.
5 c

to P •
E w 9
O) O O
if E 8

Inner Coastal Plain
«•: ...

Manage vegetation i
headwaters for deep
and N uptake.

re c
Concentrate on headw
areas. Zone 1 importa
nitrate removal.
Bypass flow due to
deeper aquifers. Long
flow paths surface in
stream channels.
w '
," f
~ p °"
m 0 5
E *g
°- 5 fe
_J C CO •

Owter Coastal Plain
-Well Drained upland.

3 2 §
| § «

Potential for higher r
with higher loadings
anoxic soils. Denitril
on larger streams.

re -j=
u. y
Restore in headwater a
Rapid restoration of fui
by denitrification.
Lower loadings. Lower
rates of removal in
head- water areas.
•._- . • -


Outer Coastal Plain
-Poorly Drained

i ^ - ,
i -
Enhance vegetation
link vegetation and
S ,' "
JC *
~ 0
> co
Limit practice to areas
marsh wetlands down
Depth to water-tables.
Bank erosion due to
unstable .soils.

 £ 5=
3 a 1


-thin soils/Tria'ssic
, -^ ••,-_'.
T5 ' '
Enhance vegetation
link vegetation and
Select deeply rooted
Vegetation. Restore in
areas. :
More flow: into regional
aquifers, bypassing
riparian zone.

3 . '
-Schist/Gneiss bedroc

LL. .
CC, <0
2 &
CO *
<0 -O
S >
C 0)
cpected level of f unctio
agricultural lands in th













2 -E
3 §
Q3 "43
(O p>
(Q rt
£ o •
o re -5
J= 0. CO
TJ g
re .E
- .«

*sr* /i
re o
P re
g m. «
to Jt £
s i i
re 2
.£ CD S

4%* 3 W
(/j oj Q)
p g
£ 2

> ^
o re
CO u_
CO *-
•6 2-d
2 8 .2
O 3 ~
O co .2


7/vier Coastal Plain
TJ i_
re 4^
. •—
CO —
CO .
re c
E o
o '•*
Q 4_f
<- ®
4-> •"•
£• c
= 0
CO 0)
re £3
o re ™
jE QL co
|_ •«*
° 8

g . i
co fc g
re o jj
« P CO
re 2 £
.E CD >2
CD re co
S t.E
+s 3 *i
co w .22
CC .£ CD
•E §
o_ 2

> fz
o re
CO .1
W *t
TJ g
2 8.2
4= re P
§! =
0 io .<£.


Outer Coastal Plain
-Well Drained


CO —-
CO »
re c
E o
0 *=
_O •*"*
— CD
.P O>
— CD
CD >
re P
Q- p
3 O
CO °+3
co ir
£ o •
o re -5
_ Q- co
Q fl) CO
""^ O .
E re P
j= p N
f* HI .-
g '.f
co !t g
ro o .._,
CD p to
re 2 £
C QJ ^^
CD re co
o t .E

co co .f2
DC .£ CD
P §
t 2
> P
o re
CO u_
CO it
S 1

I i!

Oirfer Coastal Plain
-Poorly Drained

TJ u.
P 0)
re g
CO — •
CO .
re P
E o
O -1g
_o ^^
p "*

'CD "*
f ~
3 o
CD 4=
re o
S re ••=
_ Q- co
m */\
re «2
3 Q)

C H™
« g
- s
Q- CD to
Q) 0 Q.
CD i'5
jC j£ co
g P CO
•^ LU co
£ •' 2
O H~ O)
^-* o
co g -u
CD S c
DC 2 re
P §
o_ 2
> P
o re
CO u,.
HT 3 .
O i- XJ
"o 8 S
^ ro £
Tidal-Coastal Plain


0 .
v- 0)
i i
•— CD

£ J
O j^
— O
re •
1 .
re co
.p 2
to °-
u. .. .
4-> O
CO c
DC 2



CO *t
"5 2 •
2 8
| re
O. co
-thin soils/Triassic


o .
•P 4-i
re P

£ •-
O ^
— O
ro -
re co
.£ —
'co 0-
U- ..
o !fc
4-> O
CO c
CD =;
CC. 2

CO uj
CO >t
"5 2
2 8
-re ,
0 3

o .
*» p
tp ^
p =5
CO ...

— O
is "°
re co
P J2
« Q-
O *t

CO (—
DC 2


TJ g
"5 2
o *
4= re
Piedmont/ Valley &


o .
'P1 4->
re P
|1 •
~ ""g
CO ...
re 2
g Hi
— O
8 •
re co
.£ -
CD °-

2 0
CO £
DC 2

CO uj
=5 g
"5 2
75 ® '
2 cj
•M re
Valley & Ridge
-Sandstone Shale


o .
•jg s

!E ^
09 CD
£ SP
O j^
— O
.ti ,

CO '
-; TJ
re co
.£ —
'CD D-
O !fc

CO =
DC 2

CO uj
=6 g
"6 2
p "t
Valley & Ridge/
-Low order







1 |

P «
O ,0
"O • O)
O 03
£ x:
. t • ,


Outer Coastal Pi
-Well Drained u,

£ ' :
Manage for sediment
in headwater streams.

2 c \
XI —
c £
— o
p: "
03 03
co re ...
> Jo 03
C X 03
uj os x:
•u. • .
*"* u.
.Q _£

3 ~
03 <°
0 03
re as
CO O 4-i
W s= W
o> >t >•
_J 03 OJ

E - •:
•s •§
Outer Coastal PI,
-Poorly drained

f§ '
Practices to increase
resistance to failure o1

« E
§ 8
^ CO
.£' S
03 0
o N
1 ° •
cc _T re
03 re w
«= .iC OT
03 « 03
0 CQ C
8 $ |
> € °> «



Tidal-Coastal Pla
C 4-1
0 C
Nl 03
H- E
Increase management
3 to control coarse se
and spread flow.
.2 S2
o **-
C c
13 "~"
u- co
OJ «
<" £
03 O
re N
_ c
re o
_ t->
C fT
03 uj-
f 8
5= D
O •£
Slope of

± .
-thin soils/Trias
C 4^
Nl 03

0 TJ
Increase management
3 to control coarse se
and spread flow.

^* ? o
5 UU r-
all areas
rests witl
.£ co£
03 •£ CO
S.o =
w D..J2
cS.i ol
« 1
re .2
03 H-
to £
^ CO
•§ ?>
CO ^ '
"o. w
O •« 05
c re 'w
C ~ _a3
O 03 ^
S E c
CO CO .t

3 ,
. -Schist/ Gneiss
C 4->
N 03
o ••&
Increase management
3 to control coarse se
and spread flow.
c ' ,

O 03 CO
u. u
'I iS N
all areas
rests witl
.£ o>£
<^ 1
re .2
03 H-
ro £
c ro
"o. 03
"O _
O •— s c/3
O tn 03
c re '55
C ~ _03
o os ™
S E c


03 i
- Marble/Limesi

C 4->
° §
Increase management
3 to control coarse se
and spread flow.
"re ?
-C .C Q
•| iS N
; Q) •'[
W D. .?>
03 C X
CE .h 03
Hh^ O
w CO
•— tx>
|| J.

4-1 0> -n
03 O o
CO ^ C

CQ •
Valley & Ridge
- Sandstone/Shal
C 4-
0 C
^ E
o .=5
Increase management
3 to control coarse se
and spread flow. (
(33 C- 03
• re £
^" ,C Q
> J£
{§ g |-
03 E >
«'1S £
.£ co £
S'| c
2 re 'g
w Q..J2
« c x
CC .£ 03
r- V
E. 03
o> _o
c •

-Si ^
1 5'i
4^ o5 -a
J-1 §
•— c *V
CO .£ c

E -'

Va//eK & Ridge
ypafachian .
-Low order strea
o> <*










                                                                                    « s

                                                                                    •S «
                                                                                    « •£
                                                                                    w _
                                                                                    55 .£
                                                                                    O 'C
                                                                                    J2 E
                                                                                    o ®

                                                                                    x .t;
                                                                                    UJ C

FIGURE 6. Inner Coastal Plain flow system.
FIGURE 7. Outer Costal Plain—Well-Drained Upland flow system (from Phillips et al., 1993)
    7) For a contributing area to RFBS area ratio of
       about 2:1, the range of sediment and sedi-
       ment-borne N and P reductions that could be
       expected under worst-case conditions is about
       96% for sediment, 75% for total N and 77%
       for total P. Most other cases— with a 2:1 area
       ratio and better upland conservation practices
       —would be expected to have lower concentra-
       tions leaving the RFBS. These numbers  are
       based on the assumption of non-channelized
       flow through the RFBS.

  Because of the lack of quantitative information on
RFBS functions in other hydrologic/physiographic/ •
transpiration settings, the more detailed information
from ICP settings will be used to  guide quantitative
estimates for the other settings. The consensus of the
scientific panel was that the ICP data represented an
upper limit on the functions of essentially unmanaged
RFBS. Numerous management options and manage-
ment factors discussed below could lead to increases
in the effectiveness and sustainability of nonpoint pol-
lution control functions. But in general practice, with-
out depending on the management improvements, the
effects of RFBS in the ICP would be representative of
other systems in the CBW where essentially 100% of
excess precipitation moves through  an unmanaged
RFBS. Where less than 100% of, excess precipitation
moves through the RFBS, the NPS pollution control
effects would be proportionally less.

  b.   Outer Coastal Plain

  1)  Well Drained Upland: Because much of the

  groundwater flow reaches the stream channel through
  the hyporheic zone, interactions with biologically ac-
 .tive soil, layers appear to be limited in the Well-
  Drained Upland (Figure 7). The consensus of the BPJ
  group was that for Coastal Plain systems, the WDU
  represented the other end.of the spectrum,from the
  ICP.  Processing  of groundwater-borne ]STPS  pollu-
  tants,  including nitrate, would be least in the WDU.
  Based on, the  present knowledge of. these systems,
  RFBS in the WDU would remove some nitrate from
  groundwater.  This removal function might  be en-,
  hahced by vegetation management, especially in the
  Zone 1 area where tree roots could access groundwa-
  ter discharge. Consensus decisions for the WDU are:
  .   1) Where  hydrologic  connections  between
        groundwater and biologically active soil lay-,
        ers are made, RFBS in the WDU should have
        about the same capacity for nitrate removal as
      •  in the ICP areas.
     2) The Zone 1 vegetation (adjacent to the stream
        channel) is very important because of poten-
        tial access to water 'and pollutants in the hy-
        porheic zone. Zone 1  vegetation should be
        managed for N  uptake and for formation of
        high organic matter surface soils. Provision of
        leaf litter  and other organic matter to the
        stream channels may increase denitrification
        in the channel and hyporheic zone.
     3) RFBS  in the WDU portion of the Coastal
        Plain would have about the same capacity to
        filter sediment and sediment-borne pollutants
        from surface runoff as RFBS, in the ICP
     4) RFBS in the WDU may have higher capacity
      .  for removing dissolved chemicals from sur-
        face runoff because of higher available storage
        ' for infiltrated surface .runoff. This function is
        directly related to lower water tables  in the,
     5) Reestablishment of RFBS in the WDU should
        focus on headwater- streams, many of which
        have been ditched. Enhancement of existing
        forests along both small and large streams
        should focus on control of surface runoff and
        surface-borne pollutants and on management
•}       of Zone 1 to intercept nitrate 'enriched ground-
        water.            '•

    2)  Poorly Drained  Upland/Surficial  Con-
 - fined: Functions of RFBS in the Outer Coastal Plain
 hydrologic systems designated Surficial Confined and
 Poorly Drained Upland are thought to be intermediate,
 between the"WDU and the ICP. These flow systems,
 are represented in Figure 8. Specifically, the consen- ,
 sus BPJ on these1 regions included the following: '
     1) Potential for nitrate removal is  intermediate
       between WDU and ICP. Generally lower re-
       gional groundwater concentrations of nitrate
       will lead to lower actual removal rates and to
       less important role for nitrate removal.
    2) Agriculture in these regions is commonly as-
       sociated with artificial, drainage which will
       need to be integrated with RFBS system.
    3) Potential for control of sediment and  sedi-
    ,   ment-borne chemicals should be  similar to
       RFBS in the ICP, but actual removal is proba-
       bly less because of lower loads of surface-
       borne pollutants.
    4) Potential for control of dissolved chemicals in
    '   surface runoff may be less than  in WDU be-
       cause of higher water tables and generally less
       available storage.

    c. Tidally Influenced
    Tidally influenced areas of the Coastal Plain pre-
 sent unique situations for a number of reasons. First,
 water and pollutants moving through the  terrestrial/
 aquatic interface move directly into the bay or tidal
 reaches of streams, providing a direct input of pollu-
 tants. Secondly, movement  of groundwater through
1 these tidal systems are affected by tidal movements of
 bay  water which serve to restrict discharge from
 freshwater aquifers. Thirdly, two main types of terres-
 trial/aquatic interfaces appear to exist, especially for
 groundwater fluxes. One case is  a tidal stream, em-
 bayment, or main stem location where a marsh system
 forms a buffer at  the terrestrial/aquatic  interface. In
 areas with marsh,  the nitrate removal function of the .
 RFBS is less significant due to groundwater discharge
 through the marsh being  stripped of nitrate in anaero-
 bic marsh sediments. The second case is when the in-
 terface does not include the marsh system and dis-
 charge takes place from a sand aquifer  directly into
 the bay or tidal stream. The second case is the one that
 is shown  diagrammatically in Figure 9.
    The nonpoint pollution control functions of RFBS
 in tidally influenced areas are dependent on two fac-
 tors:  depth to Water table and bank stability. The in-
 teraction  of water  table depth and nitrate removal via

denitrification has been discussed extensively in pre-
vious sections.  Bank stability is a major factor in
tidally influenced areas because of wave action, boat
wakes, storms, and rising sea level undermining trees
at the waters edge. It is likely that in tidal areas with
eroding shorelines, trees in a Z  one 1 position will
contribute to erosion and de-stabilization.
   The consensus BPJ on tidal areas of the Coastal
Plain include the following:
    1) In areas without a tidal marsh at the terres-
       trial/estuarine  interface,  nitrate  removal
       should be significant if the water table is
       within or near the root zone of trees in Zone 2.
       This  removal would be both through  direct
       vegetation uptake and through coupling  of
       vegetation uptake/denitrification  in  surface
       soil.  Where the water table is consistently
   below the root zone significant nitrate reduc-
   tion is unlikely to occur.
2) In areas where shoreline erosion is a problem
   or potential is high, Zone 1 trees at the water's
   edge are likely to contribute to shoreline ero-
   sion due to undermining of trees and tree fall
   into tidal waters. If established in these situa-
   tions, Zone 1 trees need to be put in a position
   that is not likely to contribute to,active ero-
   sion,  cliff destabilization,  or shading of
3) Functions of Zone 3 for sediment and some
   nutrient removal should be similar to function
   in ICP systems.
4) Restoration of RFBS in tidal areas should con-
   centrate on areas with shallow water tables, an
   absence  of tidal wetlands, limited shoreline
FIGURE 8. Outer Coastal Plain—Poorly Drained Upland/Surficial Confined flow system.
2-1 Om
FIGURE 9. Coastal Plain-Tidal Influenced flow system (based on Staver and Brinsfield, 1994).

     .  erosion problems and in areas with substantial
       surface runoff into tidal waters from adjacent
       land uses.

2.  Piedmont

   Although substantial work on RFBS has been done
in the North Carolina Piedmont and is underway in
Piedmont areas of the CBW (See Section II), less in-
formation is  available in the Piedmont than for the
Coastal Plain. The  consensus BPJ of the scientific
panel was that RFBS in the Piedmont represented a
range of conditions for NFS pollution control,  de-
pending on both the localized and watershed hydrol-
ogy and the proportion of excess precipitation which
moves through the. RFBS.  When  hydrblogic condi-
tions lead to surface runoff to streams and movement
of groundwater in or near the root zone of the RFBS,
the degree of NFS pollution control-should be similar -
to  conditions measured in the  North  Carolina
Piedmont and potentially as effective as the ICP con-
dition. When excess precipitation moves into deeper
groundwater and into larger streams through the hy-
porheic. zone, control of groundwater pollutants such
as nitrate may be minimal. As described above, base-
flow/stormflow separations  for Piedmont watersheds
should provide a conservative estimate of the quantity.
of water moving through RFBS.
   The first  hydrologic condition represented in" the
Piedmont is in areas with thin soils, direct flow paths
to streams^ and a large amount of water movement
through surface runoff and seepage faces (Figure 10).
These  conditions are  most likely in the Virginia
rPiedmont in the southern portions of the CBW. Under
these conditions, the consensus BPJ are:
    1)  Nitrate removal would be approximately as ef-
       fective as in ICP systems. Nitrate removaj
       may  be more  dependent  on  vegetation
       processes because of potential for deeper root-
       ing depth in  more aerated soils and the poten-
       tial  for longer  residence time- for water  in
       Piedmont RFBS.
    2). Control of sediment and sediment-borne pol-
       lutants in surface runoff should be as effective
       as ICP and North Carolina Piedmont systems.
       Control of sediment in surface runoff is likely
       to be limited by development of concentrated'
       flow channels,  especially  in  steeper RFBS
       areas  of the Piedmont. These areas may re-
       quire an expanded Zone 2.      .
     3)  Control of all sources of P should be repre-
     •  Dented well by ICP conditions and conditions
        from North Carolina studies. As in these sys-
        tems, control should be more effective for sed-
        iment-borne P than for dissolved P  in either
        surface runoff or groundwater.

   Piedmont areas with deeper soils and saprolite are
 likely to have longer flow paths and more water en-
 tering the stream channel directly from these longer
 flow paths  and the hyporheic zone. These  types of
 Piedmont systems are represented by areas with"pri-
 marily,  gneiss/schist bedrock and  primarily marble
 bedrock (Figures 11 and 12). Areas with  primarily
. schist bedrock should have substantial seepage which
 should be subject.to treatment in RFBS.
   For Piedmont areas represented in Figures 11 and
 12, the  Consensus BPJ include:
   .  1) Nitrate removal would  be  medium in the
        Piedmont areas  with Schist/Gneiss  bedrock
        and  should be used to control movement of
        water in both shallow water table conditions
      •  and  in seepage areas near streams. Nitrate re-
       moval should be least important in Piedmont
        areas underlain by marble because  of move-
       ment of groundwater and associated nitrate
      ,  into  regional  aquifer  systems  which  will
       recharge larger streams. This component of
       groundwater flow is likely to by-pass riparian
        systems. In both systems, nitrate removal will
       likely be enhanced by deeply rooted vegeta-
    2)  Control  of sediment and  sediment-borne
      .chemicals will depend  on  management of
       Zone 3 to reduce the effects of concentrated
       flow and to protect reestablished forests.
        Steeper slopes in riparian areas may limit both
     ;  the sediment filtering capacity and. the reten-
       tion  time of water. These conditions may re-
       quire expanded Zone 3 and/or .Zone 2.
    3) Control of all sources of phosphorus will be
       limited by ability to remove dissolved P in
       surface runoff. Areas  with high  sediment
       borne  surface  runoff P loads .should be re-
       stored on a priority basis because of potential
       for controlling these P types.

 3. Valley  and Ridge/Appalachian Plateau
   The Valley and Ridge is represented by larger order

FIGURE 10. Piedmont (thin soils) flow system.
FIGURE 11. Piedmont (schist/gneiss, bedrock) flow system.
FIGURE 12. Piedmont (marble bedrock)/Valley and Ridge (limestone bedrock) flow system.

 streams draining the main valleys with either lime-
 stone bedrock (Figure 12) or shale/sandstone bedrock
 (Figure 13) and by smaller order streams draining the
 ridges (Figure  14). The situation for^sediment and P
 sources is thought to be similar to the Piedmont hy-
 drologic settings. Nitrate removal will probably show
 the most variability among the different Valley types
 and with different valley configurations and flood-
 plain extent. Consensus BPJ for larger order streams
 in the Valley and Ridge for nitrate removal functions ,

    !)• Valley  and Ridge'areas with  limestone
       bedrock (Figure 12) will have the least poten-
       tial for nitrate removal due to most flow going
       into regional aquifers  which are intercepted
       primarily by  major rivers.  Seepage  areas,
       springs, and floodplains will  have'the most
       potential for nitrate removal1! Deep rooted veg-
   etation  should be used io control nitrate in
v  these areas.    "          •   -
 '.^-,   .y •       •   '•''.'     ~      •  •  •
2) Valley  and   Ridge  with  sandstone/shale  <
   bedrock (Figure 13) will have more potential
  ; for nitrate  removal due to less movement of
   groundwater and nitrate into regional aquifers
  ; and the  importance and prevalence of seepage
   areas, moving  nitrate into biologically active
   soil horizons.-The processing of nitrate is con-
   trolled by the presence and size of the flood-
   plain arid by the presence of seepage areas and
   springs. As in  other Piedmont and Valley and
   Ridge settings, deep-rooted vegetation should
   be used to  maximize the potential for N up-
 ,  take.       '     •'.;'.

3) Nitrate  removal from  low order streams in
   both  Valley  and  Ridge and  Appalachian
   Plateau  (Figure 14) settings will depend on
   residence time of water arid the presence of
FIGURE 13. Valley and Ridge (sandstone/shale bedrock) flow system.
FIGURE 14. Valley and Ridge/Appalaehian (low order streams) flow system (based on Mulhollandet al, 1990).

       seeps and floodplains. In these cases, as in
       other situations without extensive wetlands,
       the use of deeply rooted vegetation should en-
       hance nitrate uptake. Because of the limited
       extent of riparian systems in areas of high re-
       lief, Zone 1 will important for nitrate removal
       in these smaller streams.

  As a nonpoint pollution control practice, Riparian
Forest Buffer Systems represent a long-term invest-
ment which can change the structure of the agricul-
tural landscape. As a long-term management option, it
is quite likely that RFBS will be exposed to a wide
range long-term management option, it is quite likely
that RFBS will be exposed to a wide range of pollu-
tant loadings due to both  interannual variation, and
changes in management  practices in source  areas.
Information on how mature RFBS respond to chang-
ing pollutant loads is essential to understanding long
term sustainability of RFBS.
  As discussed above and in Section II, research on
some ICP systems indicates that higher rates of nitrate
removal would be possible under higher loadings of
nitrate. Published studies  indicate that this is most
likely to be true in areas where denitrification is the
primary means of nitrate removal. Given the range in
nutrient uptake observed both among different plant
species and within the same plant species, it is likely
that vegetation uptake will increase with increasing
loads,  if there  is significant hydrologic interaction
with vegetation.
  Increasing loads of P are likely to be less effec-
tively controlled than increasing loads of N, because
of the lack of biological processes to remove or  se-
quester P in the RFBS. If increasing P loads are to be
controlled, it will require both effective.management
of Zones 3 and 2 for sediment removal and manage-
ment of Zone 2 for infiltration. If dissolved or partic-
ulate P can be retained in the root zone, it will be
available for both biological and chemical removal
processes. If RFBS have some absolute removal po-
tential for P, reducing input loads should increase the
efficiency of removal.
  Management to control increasing loads of sedi-
ment and sediment-borne chemicals  will require spe-
cific management of Zones 3 and 2 for sediment re-
tention. As described above in Sections I and II, most
of the mass of sediment will be deposited in Zone 3
and most of the sediment-borne nutrients will be de-
posited in Zone 2. Increased sediment loadings to
Zone 3 will require increased management to elimi-
nate concentrated flows, remove accumulated sedi-
ment especially in berms, and restore the herbaceous
vegetation.  Increased sediment and sediment-borne
chemicals to Zone 2 should lead to higher amounts of
chemical deposition in surface litter. As with other
dissolved P in surface runoff, the ability of Zone 2 to
retain P may  be limited, especially under high load-
ings of dissolved P.
   Loading rate/buffer width relationships are only
poorly defined, especially for dissolved pollutants. In
published studies with water clearly in contact with
surface  litter or the  biologically active root  zone,
buffers of about 100  feet have been effective for at
least sediment and nitrate removal. One of the diffi-
culties in describing  these  relationships is that in-
creasing pollutant loads may also be accompanied by
increasing water volumes  in either'surface runoff,
groundwater,  or both. In the presence of increased
water movement^ denitrification for nitrate removal
should be enhanced and sedimentation and infiltration
may be decreased. Increased surface runoff and load-
ing-of sediment and sediment-borne chemicals can be
accommodated by management of Zones 3 and 2 to
increase  roughness  and control channelized flow.
Although mass balance approaches, used in Section II
may be extrapolated to higher loading rates, they pro-
vide only an estimate and may not predict real-world


   Regardless of the size of stream or the hydrologic
setting, water moving across the surface or through
the root zone of a RFBS should show reduction in ei-
ther nitrate (groundwater) or sediment and sediment-
borne chemical loads reaching the stream. As streams
increase  in size, the integrated effects of adjacent ri-
parian ecosystems should  decrease relative to the
overall water quality of the stream. On lower order
streams there is greatest potential for interactions be-
tween water  and riparian areas.  For NPS pollution
control, the change in impact of RFBS as stream order
increases can be estimated based on hydrologic con-
tributions from upstream  and  from' the  riparian
ecosystem. For first-order streams,  the potential im-
pact of the RFBS on chemical load or flow-weighted
concentration is directly proportional to the propor-
tion of the excess precipitation from the contributing
area which moves through or near the root zone or
surface of the RFBS.  For all streams above first order,

the contributing area is only one source of pollutants,
with upstream reaches providing the other source. For
second-order and above, the NFS pollution control
function of a given RFBS is based on both the pro-
portion of water  from the contributing/area which
moves through the riparian system and the  relative
sizes of the two potential pollutant loads - upstream
sources or adjacent land uses.. Clearly, the larger the
stream, the less" impact a RFBS along a particular.
stream reach can have on reduction in overall load
within that reach. If there are no RFBS upstream from
a particular stream reach, the water entering  the
stream reach is likely to be already contaminated.
   On a watershed basis, the higher the proportion of
total streamfiow  originating  from relatively short
flow-paths to small streams, the larger the potential
impact of RFBS. In comparing the potential effective-
ness of RFBS among watersheds,  drainage  density
(length of channel per unit area of watershed) should
provide a useful starting point. Higher drainage den-
sity implies greater potential importance for RFBS in
NFS pollution control.
   Control of the stream environment is most effective
when native vegetation forms a complete canopy over
the,stream. This is obviously only possible on rela-
tively ;Small streams. The effect of the RFBS on the
stream environment is not simply proportional to the
amount -of the channel  which is shaded As noted
above  in  Section I, besides  direct shading of the
stream channel, cooling of groundwater recharging
streams and provision of bank habitat will occur even
on larger  streams. Bank habitat, provision of coarse
woody debris and provision of leaf detritus remain
important functions, regardless of stream size.

   Some  aspects  of  establishment  are  discussed
above. RFBS should be used as part of an integrated
land management or conservation system which con-
sists of 1) watershed scale management, 2) NFS pol-
lution management, and 3) active management of the
RFBS. In this way, RFBS become part of conserva-
tion, stormwater, nutrient and farm management, tim-
ber harvest, and other land management planning ef-
   Watershed.management is essential to reduce over-
all pollutant loadings and integrate the riparian area as
part of a landscape influenced by upstream hydrology.
In a landscape context, RFBS which mimic  the  nat-
ural ecosystems of the area will increase the likeli-
 hood of long-term sustainability. Consideration, of ex-
 isting riparian forests and linkage of RFBS as contin-
 uous stream 'corridors is desirable. Source manage-
 ment and land conservation measures are important in
 conserving natural resources, reducing overall pollu-
 tion, and limiting stress on the RFBS.  These mea-
 sures, along with maintenance of buffer plantings, are
 especially important during the establishment phase
 and in  preventing excessive runoff or sediment and
 nutrient loading beyond the capacity of the  buffer.
 RFBS  management  such as periodic harvesting,
 runoff control maintenance, control of invasive plants,
 etc., is desirable to maximize performance and ensure
 long term effectiveness..Continued runoff control and
 protection of Zone 1 functions are essential to main-
 taining  optimum performance in RFBS.
   • Integration of,RFBS within  land management
 helps to prevent some of the primary reasons  for
 "acute" failure such as runoff inputs which exceed the
 design  of.the RFBS and cut gullies or  channels, or
 failure to address "chronic" problems such as a grad-
 ual decrease in phosphorus  retention.  Where gullies
 have formed into or through riparian forests, mea-
 sures other than  flow-spreading in Zone 3, will be"
 necessary to control channelized flow.  Because of the
 commitment of land required for RFBS  establish-
 ment, the approaches used for establishment and sub-
 sequent management should contribute to a RFBS
 which is sustainable for decades.
   At least one sustainability .question has been raised
 relative to each zone of the RFBS. The major sustain-
 ability  question for  Zone 3, discussed in .Section I,
 above is the need to remove accumulated sediment
 and reestablish herbaceous vegetation  periodically.
 Functions  of Zone  3 should be  sustainable  given
 proper  management of the sediment and vegetation.
 The other two sustainability questions are closely re-
 lated to Zones 2 and 1. In most cases, the sustainabil-
, ity of Zone 1 functions will depend on having a Zone
. 2 which is harvested infrequently. Biomass planta-
 tions which require frequent coppicing of trees or
 grassed Zone 2 areas are likely to expose Zone 1 veg-
 etation to catastrophic failure due to  blow down of
 .trees. Zone 2 functions,-if dependent on particular
 types of vegetation, such as deep-rooted  species or
 vegetation specific for high levels of nutrient uptake,
 will require some  management to control invasive


               Research  Needs
   Research needs are grouped into four general ob-
 jectives:  1)  assessment of existing riparian forest
 ecosystems relative to the minimum RFBS standards;
 2) assessment of the potential for RFBS restoration
 areas to control NFS pollution; 3) assessment of ef-
 fectiveness of NFS pollution removal in pilot restora-
 tion and enhancement areas; and 4) determination of
 the effects of management factors on the NFS pollu-
 tion  control functions  of restored and  enhanced
 RFBS, Ideally, objectives 1 and 2 would be completed
 as guidance  for pilot restoration and enhancement
 studies or large scale research projects which would  .
 be u'sed as the basis to achieve' objectives 3 and 4. If
 ongoing assessment work related to these first two ob-
 jectives is done in a timely manner, it will provide
 substantial guidance to achieve, objectives 3  and 4.
   The assessment and evaluation of existing riparian
 forest ecosystems will require the  use  of  remotely
 sensed data for delineation and classification of ripar-
 ian forests. Significant progress has been made in as-
 sessment of  the forest resources  of Maryland  in a
 "Comprehensive Forest Resources Inventory for the
 State of Maryland" undertaken by the Maryland Dept.
 of Natural Resources (Lade, 1994). The  objective of
 this project was to use Thematic Mapper data 'to cre-
 ate maps, statistical summaries and digital data sets to
 describe the location and extent of forest (especially
-streamside forests)  in the state of Maryland.  The
 study was designed for delineation of all forest re-
 sources with a minimum mapping unit of 1 acre and a
 minimum mapping unit of 100 feet for linear forest
 areas associated with streams. The data are then used
 in a Geographic Information  System to  characterize
 the extent and types or absence of forest in 30 feet arid
 100 m riparian buffers. This was done to explicitly as-
 sist in the identification of potential riparian forest
 buffer restoration sites for the entire state. The char-
 acterization of linear forest should be done at a finer
 resolution (10 to 20  m) in order to delineate riparian
 forest buffers of the width recommended in the RFBS
 specification. Data for these narrower linear forests
are needed for the entire CBWin order to character-
ize the riparian forest resources and forest cover in ri-
parian'areas.  '   ,      '
   One use of the forest inventory will be to overlay
other digital layers for further analysis of the relation-
ship of riparian forest buffers to other landscape char-
acteristics (Lade, 1994).  The classification scheme
developed here could be used as the starting point for
an assessment of the potential for existing, enhanced,
or restored RFBS to intercept surface runoff or sub-
surface borne NFS pollutants. Refinement of the clas-
sification scheme based on existing and new geohy-
drology data could be used to.produce basin .wide
maps of the relative potential for control of surface
and subsurface borne pollutants. These maps, overlain
with maps of riparian buffer vegetation and other data
layers such as wetland soils would make it possible to
quantify the riparian areas with different potential for
NFS  pollution  control which-Were  available for
restoration on a  subwatershed basis. Research sum-
marized in this report, as well as forthcoming research
results, could be used in conjunction with the mapped
information to make quantitative or comparative esti"
mates of the amount of NFS pollution reduction rela-
tive to the load reduction goals set for the Chesapeake
  .Concurrent with  development of a mapping ap-
proach is the need to make field assessments  of the
potential for hydrologic interaction between nonpoint
pollutant sources and the RFBS. Reliable indicators
of the degree of interaction between groundwater/sur-
face water and the RFBS will be necessary  when
making .field/farm/subdivision/ or watershed scale as-
sessments of nonpoint pollution  control potential.
Streamflow data  from USGS and other sources could
be used to assess  stormflow/baseflow proportions as a
screening technique. Watersheds with higher propor-
tions of stormflow could be targeted for more  inten-
sive reconnaissance investigations to .determine the
potential applicability of RFBS.
  The outputs of objectives 1 and 2 should be used to

guide the establishment of pilot restoration and en-
hancement projects and large scale research projects.
Only limited objectives can be accomplished in RFBS
restoration and management research  conducted at
the small scales normally associated with agricultural
plot research. A number of hypothetical examples can
be used to show the potential and the limitations of
small scale research in RFBS. For instance,  small
plots are being used to examine the effects of vegeta-
tion  management  on surface runoff spread evenly
through a restored RFBS. These same small plots can-
not be used for the study of effects of concentrated
flow on the filtering capacity of the RFBS. Similarly,
the effects of vegetation management on subsurface
flow cannot be studied on small plots. The minimum
size for plots to look at long term effects of clear-cut-
ting Zone 2 vegetation is constrained by the ability of
trees in adjacent reference areas to put  roots into the
clear-cut areas which will affect the hydrology of both
reference and clear-cut areas.
   The  ideal scale to .accomplish RFBS  research
should be based on the land uses contributing non-
point pollution and the hydrology of the system. It
may be necessary  to conduct work at the watershed
scale where accurate streamflow gaging data can be
used to assess the  effects of RFBS on watershed re-
sponses over time. At a minimum, the scale of re-
search is probably  that of the representative hillslope.
Ideally, integrated  research programs at a number of
spatial scales would be pursued simultaneously. For
instance, a number of hillslope studies with different
nonpoint pollution sources might be conducted in one
watershed. The hillslopes studies could be used to: 1)
examine   the  effects   of, differing  pollutant
loads/sources on similar RFBS; 2)  examine the effects
of differing RFBS management on similar pollutant
loads, or a combination of the two approaches; or 3)
examine the effects of differing hydrologic conditions
on RFBS functions. These research projects should
include sub-objectives to understand the processes re-
sponsible for removal of NFS pollution. At the same
time, restoration or enhancement of RFBS for signif-
icant portions of entire subwatersheds could provide
for a comparison with,other watersheds  without the
RFBS restoration/enhancement.
   The above discussion of objective 3 amounts to an
argument for the integration of research and demon-
stration projects on RFBS in the CBW and elsewhere.
The advantages to the research programs are in both
the ability to  conduct  research at  the  appropriate
scales and the ability to relate the research to "real-
world" restorations. The advantage to the demonstra-
tion or operational  restoration and/or enhancement
project is the potential for direct quantification of the
water quality  benefits  of  RFBS  in different land
use/hydrologic/buffer management settings.
   A long list of sub-objectives is possible for objec-
tives 3 and 4 of a general research program. Among
the potential research topics are: 1) effects of vegeta-
tion type and management on sustainabiliry of RFBS;
2) effects of vegetation type and management on NFS
pollution control by RFBS; 3) effects of chronic stres-
sors such as long-term N loading and N-saturation on
NFS pollution control; 4) effects  of acute stressors
such as large storms or extremes in temperature or
growing season rainfall.  For any given size and loca-
tion of RFBS, the actual degree of NFS pollution con-
trol may  be  dependent on  management  factors.
Although the existing research  provides little  guid-
ance in this area, management  factors are likely to
help control the effects of both chronic and-acute
stressors.                                     ,
   A viable approach to these research needs "would
be to continue funding for the assessment and map-
ping work under  objectives 1 and 2 while develop-
ing/enhancing  the coordination  between institutions
and individuals  involved with  pilot programs and
demonstration projects and  institutions and individu-
als interested  in pursuing  research associated with
these projects.

 Literature  Cited
Aber, J.D., KJ. Nadelhoffer, K Steudler and J.M. Melillo.
1989."Nitrogen saturation in northern forest ecosystems.
Bioscience 39:378-386.

Allan, J.D. and A.S. Flecker. 1993. Biodiversity conserva-
tion in running waters. BioScience 43:32-43.

Altaian, S. J., and R.R. Parizek. 1994. Evaluation of nitrate
removal from groundwater in the riparian zone. p. 277-290.
In Riparian ecosystems in the humid U.S.: functions, val-
ues and, management. Nat. .Assoc. Conserv. Districts,
Washington, D.C.

Altier, L.S., R. Lowrance, R.G. Williams, J.M. Sheridan,
D,D. Bosch, R. K. Hubbard, W.C. Mills, andD.L. Thomas. -
1994. An ecosystem model for the management of riparian
areas, p. 373-387. In Riparian ecosystems in the humid
U.S.;  functions/ values and management, Nat. Assoc.
Conserv. Districts, Washington, DC.

Ambus, P. and R. Lowrance. 1991. Denitrification in two
riparian soils: .Soil Sci. Soc; Am. J. 55:994-997.

Armstrong, W. 1968. Oxygen diffusion from the ropts of
woody species. Physiol. Plant. 20:920-926.

Baker, L.A. 1992. Introduction to nonpoint spurce pollu-
tion in the United States and prospects for wetland use.
Ecol.Eng. 1:1-26.

Barfield BJ. et al.  1979. Filtration of sediment by simu-
lated vegetation I.  Steady-state flow with homogeneous
sediment. Trans. Am. Soc. Agric. Eng. 22:540.

Barton,  D.R., W.D. Taylor,  and  R.M.  Biette.  1985.
Dimensions of riparian buffer  strips required to  maintain
trout habitat in southern Ontario streams. N. Am. J. of Fish.
Manage. 5:364-378.     '."  .-

Beare, M.H., R.R. Lowrance, and J.L. Meyer. 1994. Biotic
regulation of NOa- depletion in a Coastal Plain riparian for-
:est: experimental approach and preliminary results, p. 388-
397. In Riparian ecosystems in the humid U.S.: functions, -
values and management. Nat. Assoc. Conserv. Districts,
Washington, D.C.   "
 Behmer, D. J. and C. P. Hawkins. 1986. Effects of overhead
 canopy on macroinvertebrate production in a Utah stream.
 Freshwater Biol. 16:287-300.      --  • •  • .•  '

 Benke, A.C., R.L. Henry HI, D.M.  GJllespie, and R.H.
'Hunter. 1985. Importance of snag habitat for animal pro-
 duction in southeastern streams. Fisheries 10:8-13.

 Benke, A. C., T. C. Van Arsdall, Jr., D. M. Gillespie, and F.
 K. Parrish. 1984. Invertebrate productivity in a subtropical
 blackwater river; the importance of habitat and life history.
 Ecol. Monogr. 54:25-63.

 Beschta, R. L. 1984. TEMP84: A computer model for pre-
 dicting stream temperatures resulting from the manage-
 ment of streamside vegetation. Report WSDG-AD-00009,
 USDA Forest Service, Watershed Systems Development
 Group, Fort Collins, CO. 76 pp.

 Beschta, R.L., R.E. Bilby, G.W. Brown, L.B. Holtby, and
 T.D. Hofstra. 1987. Stream temperature and aquatic habi-
 tat: Fisheries and forestry interactions, p. 191-232, In E.O.
 Salo  and  T.W.  Cundy (Eds):  Streamside Management:
 Forestry and fishery interactions. Inst.  of Forest Resour.
 Univ. of Washington, Seattle.

 Beschta, R. L., and R. L. Taylor. 1988. Stream temperature
 increases  and land use in  a forested Oregon watershed.
 Water Resour. Bull. 24:19-26.

 Beven, K., and P. Germann. 1982. Macropore's and water
 flow in soils. Water Resour. Res. 18:1311-1325.

 Bilby, R.  E. 1984. Removal of woody debris may affect
 stream channel stability. J. For. 82:609-613.

 Bilby, R. E. and P. A. Bisson. 1992. Allochthonous versus
 autochthonous organic matter contributions to the trophic
 support of fish populations in clear-cut and old-growth
 forested streams. Can. J. Fish. Aquat. Sci. 49:540-551.
 Bingham,  S.C.,  P.W.  Westerman, and M.R. Overcash.
 1980. Effect of grass buffer zone length in reducing the pol-
 lution from land application areas. Trans. Am. Soc. Agric.
 Eng. 23:330-336.

Bisson, P.A., R.E. Bilby, M.D. Bryant, C.A. Dolloff, G.B.
Greite, R.A. House, MJ. Murphy, K V. Koski,  and J.R.
Scdell. 1987. Large woody debris in forested streams in the
Pacific Northwest: Past, present, and future, p. 143-190, In
E.O. Salo and T.W. Cundy (ed.) Streamside management:
forestry and fishery interactions. Inst. of Forest Resources,
Univ. of Washington, Seattle.

Bohlke, J.K. and J.M. Denver. In press. Combined use of
ground-water dating, chemical, and isotopic analyses to re-
solve the history and fate of nitrate contamination in two
agricultural watersheds. Atlantic Coastal Plan, Maryland.
Water Resour. Res.

Bormann, EH. and G.E. Likens. 1979. Pattern and Process
in a Forested Ecosystem. Springer-Verlag, New York.,

Bosch, D.D., R.K. Hubbard, L.T. West, arid R: Lowrance.
1993. Subsurface flow patterns in a riparian buffer system.
ASAE Intematl. Meeting, Chicago, IL. Paper No. 93-2551.

Bowling, D.J.F., A.E.S. Macklon, R.M. Spanswick. 1966.
Active and passive  transport  of the major nutrient ions
across the root ofRicinus communis. J. Exptl. Bot. 17:410-

Bradley, P.M., M. Fernandez Jr. and F.H. Chapelle. 1992.
Carbon limitation of denitrification in an anaerobic ground-
water system. Environ. Sci. &  Technol. 12:2377-2381.

Brazier, J. R. and G. W. Brown.  1973. Buffer strips for
stream temperature control. Research Pap. 15, Paper 865,
Forest  Res. Lab. School of Forestry, .Oregon State Univ.
Corvallis. 9 pp.

Bren, LJ. 1993. Riparian zone,  stream^ and floodplain is-
sues: a review. J. Hydrol. 150:277-299.

Brinson,  M.M. 1977.  Decomposition  and nutrient ex-
change of litter in an  alluvial swamp forest.  Ecology
58:601-609.  ,

Brown, G.W. 1969. Predicting temperatures  of small
streams. Water Resour. Res. 5:68-75.

Brown, G. W 1974. Forestry and water quality. O.S.U.
Book Stores, Inc., Corvallis, OR.

Brown, G.W. and J.T. Krygier. 1970. Effects of clear-cut-
ting on stream temperature. Water Resour. Res.  6:1133-

Brown, S., Brinson, M. M. and A. E. Lugo. 1978. Structure
and  function of riparian wetlands, pp.  17-31. In R. R.
Johnson and J. F. McCormick (ed.). Strategies for protec-
tion and management of floodplain wetlands and other
riparian ecosystems. Geh. Tech. Rep. WO-12, Forest Serv.,
U. S. Dept. Agric., Wash. D. C. 410 pp.

Burton, T. M., and G. E.  Likens. 1973. The  effect of
strip-cutting on stream temperatures in the Hubbard Brook
experimental  forest,  New  Hampshire.  BioScience

Chesapeake Bay Program. 1990. Report and recommenda-
tions of-the nonpoint source evaluation panel. CBP/TRS
56/91. Annapolis, MD.

Chesapeake Bay Program. 1991. Baywide nutrient reduc-
tion  strategy  - 1990 Progress repprt.  Report No. 2.
Annapolis, MD.

Cole, D.W. 1981. Nitrogen uptake and translocation by for-
est ecosystems, p. 219-232. In F.E. Clark and T. Rosswall
(ed.). Terrestrial nitrogen cycles. Ecol. Bull. 33,

Cole, D.W. 1986. Nutrient cycling in world forests, p. 103-
115.  In S.P. Gessel (ed.).   Forest site and  productivity.
Martinus Nijhoff, Dordrecht, Netherlands.

Cole, D.W. and M. Rapp. 1980. Elemental cycling in forest
ecosystems, p. 341-409. In  D.E. Reichle (ed.).  Dynamic
properties  of  forest  ecosystems. Cambridge University
Press, New York.                    .

Comerfqrd, N.B., D.G. Neary, and R.S. Mansell. 1992. The
effectiveness of buffer strips for ameliorating offsite trans-
port of sediment, nutrients,  and pesticides from silvicul-
tural operations. Nat. Coun.  Paper Ind. Air '&  Stream
Improvement Tech. Bull. 631. New York, NY.

Cooper, A.B. 1990. Nitrate depletion in the riparian zone
and stream  channel  of a  small  headwater catchment.
Hydrobiologia 202:13-26.

Cooper, J.R.y J.W.  Gillia'm, R.B. Daniels,  and  W.P.
Robarge. 1987. Riparian areas as filters for agricultural
sediment Soil Sci. Soc. Am. J. 51:416-420.

Correll, D. L. 1977. An overview of the Rhode River pro-
gram,  p.  105-124, In  D.  L. Correll  (ed.). Watershed
Research in Eastern North America: A workshop to  com-
pare results. Smithsonian Institution, Washington, DC.  \

Correll, D.L.. 1983. N and P in  soils  and runoff of three
Coastal Plain land uses. p. 207-224. In R. Lowrance et al.
(ed.) Nutrient cycling in Agricultural Ecosystems. Univ. of
Georgia Coll. of Agric. Special Publ. No. 23. Athens, GA.

Correll, D.L.. 1987. Nutrients  in Chesapeake Bay. p. 298-
320,  In S.K. Majumdar, L.W. Hall, Jr., and'H.M. Austin
(ed.) Contaminant  problems and management  of living

 Chesapeake Bay resources.  Pennsylvania Academy of
 Science, Philadelphia.

 Correll, D.L.,T.E. Jordan, and D.E. Weller. 1992. Nutrient
, flux in a landscape: Effects of coastal land use and terres-
 trial community mosaic on nutrient transport to coastal wa-
 ters. Estuaries  15:431-^442.           ;        '    .

 Correll, D.L., T.E. Jordan, and D.E. Weller. i994. Failure,,
 of agricultural riparian buffers  to protect surface waters
 from groundwater contamination. In Proc. internatl. conf.
 groundwater/surface water ecotones.  Cambridge  Univ.
 Press. -  .  ,

 Correll, D.L.  and  D.E.  Weller.  1989. Factors limiting
 processes  in freshwater wetlands: An agricultural primary •
 stream riparian forest, p.9-23. In R. Sharitz and J. Gibbons
 (ed.) Freshwater wetlands and  wildlife.  U.S:  Dept. of
 Energy, Oak Ridge, TN.       '      ,   "  '

 Courts,  M.P. and J.J. Philipson. 1978.  Tolerance of tree
 roots to waterlogging. New Phytol. 80:71-77.    •

 Crawford, R.M.M. 1982. Physiological responses to flood-
 ing, p. 453-477. In O.L. Lange, PS. Nobel, C.B. Osmond,
 and  H.  Ziegler  (ed.).  Physiological  plant  ecology.
 Encyclopedia of plant physiology (New Series, Vol.  12/B)
• 2nd ed.  Springer-Verlag, New York.

 Daniels, R.B., and J.W. Gilliam. In press. Sediment and
 chemical load  reduction by grass and riparian filters.  Soil
 Sci. Soc. Am. J.

 Dickey, E.G. and D.H. Vanderholm. 1981. Performance and
 design of vegetative filters for feedlot runoff treatment.
 Proc.  4th  International  Livestock Waste Symposium.
 American Society of Agricultural Engineers Pub., 2-81:357-
 360. St. Joseph, MLUSA.        .   '

 Dillaha, T.A.,  R.B. Reneau, S.  Mostaghimi and D.  Lee.
 1989a. Vegetative filter strips'for agricultural  nonpoint
 source pollution  control.  Trans. Am".  Soc. Agric.  Eng.

 Dillaha, T.A., J.H. Sherrard, and D. Lee. 1986. Long-term
 effectiveness and maintenance of vegetative filter strips.
 Virginia Water  Resour.  Res. Cntr. Bulletin, No.  153.
 Blacksburg, VA.                  ,

 Dillaha, T.A., J.H. Sherrard and D. Lee.  1989b. Long-term
 effectiveness of vegetative filter strips. Water Environ. Soc.
 1:419-42L                 ,    ,    '

 Dillaha, T.A., J.H. Sherrard, D. Lee, S. Mostaghimi, and
 V.O, Shanholtz. 1988. Evaluation of vegetative filter strips
as a best management practice for feedlots. Journal of the
Water Polution Control Federation, 60:1231-1238.   '

Dolloff, C. A. 1994a. Large woody debris—the common
denominator for integrated environmental management of
forest streams, p. '93-108. In J. Cairns, Jr., T.V. Crawford,
and H. Salwasser (ed.) Implementing integrated' environ-
mental  management.  Center  for  Environmental  and
Hazardous Material Studies, Virginia Polytechnic Institute
and State University, Blacksburg, VA.

Dolloff, C.A., P.A. Flebbe, and M.D. Owen. 1994b. Fish
habitat and  fish populations in a southern appalachian
watershed before and after Hurricane Hugo. Trans. Amer.
Fish. Soc. 123:668-678.

Dohigian; A. S., L. C. Linker, and C. H. .Chang. 1990.
Watershed model application to calculate bay nutrient load- •
ings: Preliminary Phase I findings and recommendations.
U.S. Environmental Protection Agency, Athens, GA.    '  •

Dunkle, S. A., L. N. Plummer, E. Busenberg, P. J, Phillips,
J.  M. Denver, P. A.  Hamilton, R. L. Michel, and T.  B.
Coplen., 1993.  Chlorofluorocarbons (CC12F2) as  dating
tools and hydrologic tracers in shallow groundwater of the
Delmarva Peninsula, Atlantic Coastal Plain, United States,
Water Resour. Res., 29:3837-3860.

Edinger, J. E., and E. M. Buchak. In press. Temperature,
modelling for instream flow studies. J. Am. Inst.  Hydrol.

Edinger, J. E., D. W. Duttweiler, and J. C. Geyer. 1968. The
response of water  temperature to mete'orolbgical  condi-
tions. Water Resour. Res. 4:1137-1145.

Ehrenfeld, J.G. 1987. The role of woody vegetation in pre-
venting  ground water pollution by nitrogen from septic
tank leachate. Water Resour. Res. 21:605-614.

Everett, R.A. and G.M. Ruiz. 1993. Coarse, woody debris
as a refuge from predation in aquatic communities: an ex-
perimental test. Oecologia 93:475-486.

Fail, J.L., Jr.,' B.L. Haines, and R.L, Todd. Riparian forest
communities and their role in nutrient  conservation in  an
agricultural watershed! Am.  J. Alternative Agric.  2:114-
121.."         '..       .  c           ;.

Fail, J.L., Jr., M.N. Hamzah, B.L. Haines, and R.L.,Todd.
1986. Above and below ground biomass, production, and
element accumulation in riparian' forests of an agricultural
watershed, p. 193-224, In D.L. Correll (ed.) Watershed re-
search perspectives. Smithsonian Institution, Washington, •.
D:C.                     .

Feminella, J.W.,  M.E. Power,  and -V.H," Res'h.  1989.

Periphyton responses to invertebrate grazing and riparian
canopy in three northern California  coastal streams.
Freshwater Biol. 22:445-457.

Feminella, J.W., and V.H. Resh. 1989. Submersed macro-
phytes and grazing crayfish: an experimental study of her-
bivory in a California freshwater marsh. Holarct. Ecol. 12:

Fcnneman, N. M.  1938. Physiography of eastern United
States. McGraw-Hill, New York.

Fields, S. 1992. Regulations and policies' relating to the use,
of wetlands  for nonpoint source pollution control. Ecol.
Eng. 1:135-142.

Francis,  A.J.,  J.M.  Slater  and  CJ.  Dodge!  1989.
Denitrification   in    deep   subsurface   sediments.
Geomicrobiology J. 7:103-116.

Franklin, E.  C., J. D. Gregory, and M. D. Smolen. 1992.
Enhancement of the effectiveness of forested filter zones
by dispersion of agricultural runoff. Rep. No. UNC-WRRI-
92-270. Water Resour. Res. Inst, Raleigh, NC.

Gaskin, J. W., J. F. Dowd, W. L. Nutter, and W. T. Swank.
1989. Vertical and lateral components of soil nutrient flux
in a hillsope. J. Environ. Qual. 18:403-410.

Gburek, W.J. and J.B. Urban. 1990. The shallow weathered
fracture layer  in the near-stream zone.  Ground Water

Gburek, W.J. and J.B.  Urban, and R.R.  Schnabel. 1986.
Nitrate contamination of ground water in an upland -water-
shed. p.352-380. In Proceedings of agricultural impacts on
ground water,  Omaha, NE, 11 to 13 August. Nat. Water
Well Assoc., Dublin, OH.

Golladay, S.W. and R.L. Sinsabaugh. 1991. Biofilm devel-
opment on  leaf and  wood surfaces in  a boreal  river.
Freshwater Biol. 25:437-450.

Golladay, S.W. and J.R. Webster. .1988. Effects of clear-cut
logging on  wood  breakdown in Appalachian  mountain
streams. Am. Mid. Nat. 119:143-155.

Gregory, S.V., FJ. Swanson, W.A. McKee, ,and  K.W.
Cummins.  1991. An  ecosystem perspective of riparian
zones. BioScience 41:540-551.

Groffman, P.M. 1994.  Denitrification in freshwater wet-
lands, p. 15-35, In  Current topics in wetland biogeochem-
istry. Wetland  Biogeochemistry Institute, Louisiana  St'.
Univ., Baton Rouge, LA.
Groffman, P.M., E.A. Axelrod, J.L. Lemunyon and W.M.
Sullivan. 1991.  Denitrification in vegetated filter strips. J.
Environ. Qual. 20:671-674.

Groffman,  P.M., A.J.  Gold, and  R.C.  Simmons.  1992.
Nitrate dynamics in riparian forests: microbial studies. J.
Environ. Qual. 21:666-671.

Hall, D.W. 1992. Effects of nutrient management on nitrate
levels in ground water near Ephrata, Pennsylvania. Ground
Water 30:720:730.

Hanson,  G.C.,  P.M. Groffman, and A.J. Gold.  1994a.
Denitrification in riparian wetlands receiving high and low
groundwater nitrate inputs. J. Environ. Qual. 23:917-922.

Hanson,  G.C.,  P.M. Grofrman and  A.J. Gold.  1994b.
Symptoms of nitrogen saturation in a riparian forest. Ecol.
Applications 4:750-756.

Harmon, M.E., J.F. Franklin, FJ. Swanson, P. Sollins, S.V.
Gregory, J.D- Lattin,  N.H. Anderson,  S.P.  Cline, N.G.
Aumen, J.R. Sedell, G.W. Lienkaemper, K.  Cromack, Jr.,
andK.W. Cummins. 1986. Ecology of coarse woody debris
in temperate ecosystems. Ad. Ecol. Res.  15:133-302.

Hartman, Q., J.C. Scrivener, L.B. Holtby, and L- Powell
•1987. Some effects  of different  streamside treatments on
physical conditions and fish population  processes  in
Carnation Creekj  a  coastal rain forest  stream in British
Columbia, p. 330-372. In E.O. Salo and T.W. Cundy (ed.)
Streamside management: forestry and fishery interactions.
Inst. of Forest Resources,, Univ. of Washington, Seattle.

Hawkins, C.P.,  M.L. Murphy, and N.H. Anderson. 1982.
Effects of canopy, substrate composition, and gradient on
the structure of  macroinvertebrate communities in Cascade
Range streams of Oregon. Ecology 63:1840-1856.

Hax, C.L. and S.W. Golladay. 1993. Macroinvertebrate col-
onization and biofilm development on leaves and wood in
a boreal river. Freshwater Biol. 29:79-87.

Haycock, N.E.  and  G. Pinay. 1993. Groundwater nitrate
dynamics in grass and poplar vegetated riparian buffers
during the winter. J. Environ. Qual. 22:273-278.

Heath, R. C. 1984. Ground water  regions of the United
States. U. S. Geological Survey Water Supply Paper 2242,
59pp.               '  -       -

Hedman, C- 1992.  Southern Appalachian riparian zones:
their  vegetative composition  and contributions  of large
woody  debris  to streams.  Ph.D.  Dissertation. Clemson
University, Clemson, SC.   ,

 Hendrickson, O.Q., Jr. 1981. Flux of nitrogen and carbon
 gases in bottomland soils of an  agricultural watershed.
 Ph.D. dissertation, Univ. Georgia, Athens, GA. 210  p.r
 (Diss. Abstr. 82-01544).

.Hewlett, J.D. and Fortson, J.C. 1982: Stream temperature
 under an inadequate buffer strip in the southeast Piedmont.
 Water Resour. Bull: 18:983-988.      '          :

 Higinbotham, N.,  B. Etherton,  and R.J, Foster.  1967.
 Mineral ion contents and "pell trans-membrane electro-po-
 tentials  of pea and. oat seedling tissue. Plant. Physiol.  ,
 42:37-46.             •                   '

 Higman, D. and D.L. Correll. 1982. Seasonal and yearly
 variation in meteorological parameters at the Chesapeake
 Bay Center for Environmental Studies,  p. 1-159, /n'D.L.
 Correll  (ed.) Environmental data summary for the Rhode
 River ecosystem.. Volume A.  part 1.  Chesapeake Bay
 Center for Environmental Studies, Edgewater, MD.

 Hill, W.R. and B.C.-Harvey. 1990. Periphyton responses to .
 higher trophic levels and light in a shaded stream. Can. J.
 Fish, and Aquat. Sci. 47:2307-2314.

 Hiscock, K.M,, J.W. Lloyd and D.N. Lerner. 1991. Review
 of  natural and  artificial denitrification of groundwater.
 'Water Res. 9:1099-1 111.              ,

 Hoagland,  D,R. and T.C. Broyer. 1936. General nature of
. the process of salt accumulation by roots with description
 of experimental methods. Plant Physiol. 11:471-507.
                               ' ,.
 Hooper, R.P., N. Christophersen,' and N.E., Peters. 1990.
 Modelling streamwater chemistry as a mixture of soilwater
 end-members—An application  to the  Panola Mountain
 Catchment, Georgia, USA. J. Hydrol. 116:321-343.

 Hopkins, H.T.  1956. Absorption of ionic species  of  or-
 mophosphate by barley roots: effects of 2,4-dinitrophenol
 and oxygen tension. Plant Physiol. 31:155-161.

 -Hopkins, H.T., A.W. Specht, and S.B.  Hendricks. 1950.
 Growth and nutrient accumulation as controlled by oxygen
 supply to the plant roots. Plant Physiol. 25:193-209. '

 Hosner^ J.F. and A.L. JLeaf. 1962. The effect of soil  satura-
 tion upon the dry weight, ash content, and nutrient absorp-
 tion of various bottomland  tree seedlings. Soil Sci. Soc.
 Am. Proc. 26:401-404.

 Hosner, J.F., A.L. Leaf, R. Dickson,' and J.B. Hart, Jr. 1965.
 Effects of varying soil moisture upon the nutrient uptake of
  four bottomland tree species.  Soil  Sci. Soc.  Am. Proc.
  29:313-316.         •              .
 Hughes, R.M,, T.R, Whittier, C.M. Rohm, andD.P. Larsen.
 1990. A regional framework for establishing recovery cri-
 teria. Environ. Manage. 14:673-684.

 Hunt, C.B. 1974. Natural regions of the United States and
 Canada. W.H. Freeman, San Francisco. 725 p.

 Jacobs, T.C. and J.W. Gilliam. 1985a. Headwater losses of
 nitrate  from two Coastal Plain watersheds. J. Environ,
 Qual. 14:467-472.

 Jacobs, T.C. and J.W. Gilliam. 1985b. Riparian losses of ni-
 trate from agricultural drainage waters. 'J. Environ. Qual.
 14:472-478.   .                    '

 Jaworski, N.A., P.M.'Groffman, A. Keller, and A.C. Prager.
 1992. A watershed-scale analysis of nitrogen loading: The
 Upper Potomac River. Estuaries 15:83-95.

 Johnson, A.C. and M. Wood. 1992. Microbial potential of
 sandy   aquifer  material   in   the  London   basin.
 Geomicrobiology J.  10:1-13.                   •. •

 Johnson, R.R., and J.F. McCormick (ed.) 1978. Strategies
 for protection and management of floodplain wetlands and
 other riparian ecosystems. Gen. Tech.  Rep. WO-12, Forest
 Serv:, U. S. Dept. Agric., Wash. DC. 410 pp.

• Jordan, t.K, D.L. Correll, W.T. Peterjohn and D.E. Weller.
 1986. Nutrient flux in a landscape: the Rhode River water-
 shed and receiving waters. In D.L. Correll (ed.) Watershed
 Research  Perspectives. Smithsonian  Press, Washington,
 D.C.                 /        ,

 Jordan, T.E., D.L. Correll, and D.E. Weller.  1993. Nutrient
 interception by a riparian forest receiving Inputs from adja-
 cent cropland. J. Environ. Qual. 22:467-473.

 Justin, S.H.F.W., and W. Armstrong. 1987. The anatomical
 characteristics of roots and plant response to soil flooding.
 New Phytol. 106:465-495.        - '   .'•'..

 Kaplan, L.A., arid J.D- Newbold. 1993. BiogeOchemistry of
 dissolved organic carbon entering streams,  p. 139-166. In
 T. E. Ford (ed.) Aquatic microbiology, an ecological ap-
 proach. Blackwell, Boston.

 Karr, J.R. 1993. Measuring biological integrity: lessons
 from streams, p. 83-104. In S. Woodley, G.Francis, and J.
 Kay (ed;) Ecological integrity and  the management of
 ecosystems. St. Lucie Press, Delray Beach, FL.

 Karr, J^R. and I.J, Schlosser. 1978. Water resources and the
 land-water interface. Science 201:229-234.

 Kibby, H. V. 1978. Effects of wetlands on water quality, p.

 289-298. In R. R. Johnson and J.  F. McCormick (ed.)
 Strategies for protection and management of floodplain
 wetlands and other riparian ecosystems. Gen. Tech. Rep.

 Kitchens, W. H., J. M. Dean, L. H.  Stevenson,  and J. H.
 Cooper. 1975. The Santee Swamp as a nutrient sink in min-
 eral cycling. In F. G. Howell, J. B. Gentry and M. H. Smith
 (cd.) Southeastern Ecosystems. ERDA Symp. Series.Forest
 Serv., U. S. Dept. Agric., Wash. D. C. 410

 Kozlowski, T.T.  1984. Responses of woody plants to flood-
 ing, p. 129-163. In T.T. Kozlowski (ed.) Flooding and plant
 growth. Academic Press, New York, NY.

 Kozlowski, T.T. and S.G. Pallardy. 1984. Effect of flooding
 on water, carbohydrate, and mineral relations, p. 165-193.
 In T.T.  Kozlowski  (ed.) Flooding and  plant  growth.
 Academic Press, New York, NY. •

 Kramer, PJ. and  T.T.  Kozlowski.  1979.  Physiology of
 woody plants. Academic Press, New York, NY.

 Kuenzler, E.J.,  PJ.  Mulholland, L.A. Ruley, and R.P.
 Sniffen,  1977. Water quality  in North Carolina Coastal
 Plain streams  and   effects  of  channelization.  Water
 Resources Research Institute, UNC. Rpt. No. 127. Raleigh.
 160 pp.

 Lade, K.-P.  1994. Comprehensive forest resource inventory
 for the state of Maryland. Maryland  Dept.  of Natural

 Lee, G.F., E. Bentley and R. Amundson. 1975. Effects of
 marshes on water quality pp. 105-127. In A.D. Hasler (ed.)
 Coupling of land and water systems. New York: Springer-
 Verlag. pp. 105-127.

 Lee, R., and D.E. Samuel. 1975. Some thermal and biolog-
 ical effects of forest cutting in West Virginia. J. Environ.
 Qual. 5:362-366.

 Leopold, L.B., M.G. Wolman, and J.P. Miller. 1964. Fluvial
 processes in geomorphology. W.H. Freeman and Company,
 San Francisco.

 Lind, A.-M. and F. Eiland. 1989. Microbiological charac-
 terization and nitrate reduction in subsurface soils. Biol.
 Fertil. Soils  8:197-203.

 Long, C. 1991. National policy perspectives and issues re-
 garding the prevention and control of nonpoint source pol-
lution. USEPA-ORD Workshop on Nonpoint  Source
Pollution Control. USEPA, Wash., D.C.

Lowe,  R.L., S.  Golladay,  and J.R. Webster.  1986.
 Periphyton response to nutrient manipulation in streams
 draining blear-cut and forested watersheds. J. of the North
 Am. Benthol. Soc. 5:221-229.             •   .    ,   '

 Lowrance, R. 1991. Effects of Buffer systems on the move:
 ment of N and P from agriculture to streams, p. 87-96. In
 V. Holter (ed.) Proc.  of conf.  Nitrogen,  Phosphorus,
 Organic  Matter.   Danish .Ministry  of Environment,

 Lowrance, R. 1992. Groundwater nitrate and denitrification
 in a Coastal Plain  riparian soil. J. Environ, Qual. 21:401-

 Lowrance, R. andR.A. Leonard. 19.88. Streamflow nutrient
 dynamics in Coastal Plain watersheds. J. Environ. Qual.
 17:734-740.                    '      .  .

 Lowrance, R., R.A. Leonard,  LJE,, Asmussen, and R.L.
 Todd.  1985a. Nutrient budgets  for agricultural'watersheds
 in the southeastern coastal plains. Ecology 66:287-296.

 Lowrance, R., R. Leonard, and J. Sheridan. 1985b. Man-
 aging riparian ecosystems to control nonpoint pollution. J.
 Soil and Water Cons. 40:87-91.        .               -

 Lowrance, R., S., Mclntyre and J.C. Lance. 1988.  Erosion
 and  deposition in  a Coastal  Plain watershed measured
 using CS-137. J. Soil Water Cons. 43:195-198.

 Lowrance, R., J.K. Sharpe and J.M. Sheridan. 1986.
 Long-term sediment deposition in the  riparian zone of a
 Coastal Plain watershed. J. Soil Water Cons. 41:266-271.

 Lowarnce, R..R., R.L.  Todd, and L.E. Asmussen. 1983.
 Waterborne nutrient budgets for the riparian zone of. an
 agricultural watershed.  Agric. Ecocsys. Environ.  10:371-

 Lowrance, R., R.L. Todd,  and L.E. Asmussen. 1984a.
 Nutrient cycling in an agricultural watershed: I. Phreatic
 movement. J. Environ. Qual. 13:22-27.

 Lowrance. R.R., R.L. Todd, J. Fail., O. Hendrickson, Jr., R.
 Leonard, and L. Asmussen. 1984b. Riparian forests as nu-
 trient filters in agricultural watersheds. Bioscience 34:374-
 377.             •,

Lowrance, R., G. Vellidis, and R.K. Hubbard. In Press.
Denitrification in a restored riparian  forest  wetland.  J.
Environ. Qual.

Lynch, J.A. and "E.S. Corbett.  1990. Evaluation  of best
management practices for controlling nonpoint pollution
from silvicultural operations. Water Resour. Bull. 26:41-

 Lynch, J.A., E.S. Corbett, and K. Mussallem. 1985., Best
 management practices fpr controlling nonpoint-source pol-
 lution on forested watersheds. J. Soil Water Cons. 40:164-
 167.        "                          ,   '       ,

 Lynch,  J.A., E.S.  Corbett,  and W.E. Sopper.  1980.
 Evaluation of management practices on the biological and
 chemical characteristics of streamflow  from forested
 watersheds.  Technical  Completion Report  (A-041-PA).
 Institute for Research on Land and Water Resources, Penn.
" State Univ., State College, PA.            , ;

 Magette, W.L., R.B. Brinsfield, R.E, Palmer, J.D. Wood.
 1989. Nutrient and sediment removal by vegetated filter
' strips^ Trans. Am. Soc. Agric. Eng. 32:663-667.

 Magette, W.L., R.B. Brinsfield, R.E. Palmer, ID. Wood,
 T.A. Dillaha and R,B. Reneau. 1987. Vegetative filter strips,
 for  agricultural runoff treatment. U.S.E.P.A. Report No.
 CBP/TR.S 2/87, Washington, D.C.      '-

 Magnien, R.E., R.M. Summers, and K.G Sellner. 1992.
 External nutrient sources, internal tools, and phytoplankton
 production in Chesapeake Bay. Estuaries 15:497-516.

 McCarty, G'.W. and J.M. Bremner. 1992. Availability of or-
 ganic carbon for denitrification of nitrate in subsoils. Biol.
 Fertil. Soils 14:219-22.                 '           .,

 McDonald,  A.J.S., T. Ericsson, and T. Ingestad. 1991.
 Growth and nutrition of tree seedlings, p. 199-220. In A.S.
 Raghavendra (ed.) Physiology of trees. John Wiley & Sons',,
 New York, NY.   ,  .  '                -

 McDonnell, J.J. 1990. A rationale.for old water discharge
 through macropores in  a steep, humid catchment. Water
 Resour. Res. 26:2821-2832.                 :

 McFarland, E.R.  In press. Ground-water flow and geo-
 chemistry and effects of agricultural practices on nitrogen
 transport in two geological' settings, Patuxent River Basin,
 Maryland.  U.S., Geological  Survey Water Resources
 Investigations Report.                      -

 Meehan, W.R.,  F.J.  Swanson,  and J.R. Sedell. <1987.
 Influences of riparian vegetation on aquatic ecosystems
 with particular reference to salmonid fishes and their food •
 supply, p. .137-145. In  E.G. S'alo and T.W.  Cundy (ed.)
 Streamside management: forestry and fishery interactions. ;
 Inst. of Forest Resources, Univ. of Washington, Seattle.   /

 Merritt, R.W. and K.W. Cummins, editors. 1984. An intro-
 duction  to  the  aquatic  insects.   Second   edition.
 Kendall/Hunt Publishing Company, Dubuque,- LA.
 Meyer, J.L. and J. O'Hop. 1983. The effects of watershed
 disturbance on dissolved organic dynamics of a stream.
 Am. Midi. Nat. 109:175-183.

 Miller, H.G. 1984. Dynamics of nutrient cycling in plantar
 tion ecosystems, p. 53-78.  In  G.D. Bowen  and E.K.S.
 Nambiar (ed.) Nutrition of plantation  forests, Academic
 Press, London.

 Miller, H.G., and J.M. Cooper.  1973. Changes in amount
 and distribution of stem growth in pole-stage corsican pine
 following application  of nitrogen fertilizer.  Forestry

 Miller, R.E., and R.F. Tarrant. 1983. Long-term growth re-
 sponse of Douglas-fir to ammonium nitrate fertilizer. For.
 Sci. 29:127-137.     ,        .

 Mitchell, HX., and R.F. Chandler,  Jr. 1939. The nitrogen
 nutrition and growth of certain deciduous trees of north-
 eastern United States. Black Rock For. Bull. 11.

 Mitsch, WJ. 1978. Interactions between a riparian swamp
 and a river in southern Illinois, p. 63-72. In R. R. Johnson
 and J. F. McCormick (ed.).  Strategies for protection  and
 management of floodplain  wetlands and other  riparian
 ecosystems. Gen. Tech.. Rep. WO-12, Forest Serv., U. S.
 Dept. Agric., Wash.D. C. 410 pp.           '

 Mitsch, W.J., C.L.  Dorge, and J.R.  Wiemhoff.  1979.
 Ecosystem dynamics and a phosphorus  budget of an allu-
 vial cypress swamp in southern Illinois.  Ecology 60:1116-

'Mosley, M.P. 1982. Subsurface flow velocities through se-
 lected forest-soils. South Island,  New Zealand. J.  Hydrol.
 55:65-92.   •''    '                      ,         '•

 Mulholland,. P.J., G.V. Wilson,  and P.M.  Jafdine.  1990.
 Hydrogeochemical  response of  a  forested watershed to
 storms: Effects of preferential flow along shallow and deep
 pathways. Water Resour. Res'. 26:3021-3036.

 NCRI Chesapeake.  1982. Major land resource  areas in the
 Chesapeake Bay Watershed: aerial data  from the 1982
 USDA/SCS National Resource Inventory. National Center
 for Resource Innovations. Rosslyn, VA.

v Nelson, L.E., G.L. Switzer, and W.H.  Smith. 1970. Dry
 matter  and nutrient  accumulation in young loblolly pine
 (Pinus taeda L.). p. 261-273. In C.T. Youngberg and C.B.
 Davey:(ed.) Tree growth and forest soils. Oregon State
 University Press, Corvallis, OR.

 Newbold, J.D., D.C. Erman,  and K.B. Roby. 1980. Effects

of logging on macroinvertebrates in streams with and with-
out buffer strips. Can. J. Fish. Aquat. Sci. 37:1076-1085.

Obenhuber, D.C. and R. Lowrance.  1991.  Reduction of
nitrate in  aquifer microcosms by carbon additions. J.
Environ. Qual. 20:255-258.

Oliver, C.D. andT.M. Hinckley. 1987. Species, stand struc-
tures,  and  silvicultural manipulation patterns  for the
streamside zone. p. 259-276. In E.O. Salo and T.W. Cundy
(ed.) Streamside management: forestry and fishery interac-
tions. Inst. of Forest Resources,  Univ.  of Washington,

Olmsted, F.H., and A.G. Healy. 1962. Relation  between
ground water and  surface water in  Brandywine Creek
Basin, Pennsylvania. U.  S. Geological Survey Professional

Overcash,  M.R.,  S.C.  Bingham, and P.W. Westerman.
1981. Predicting runoff pollutant reduction in buffer zones
adjacent to land treatment sites. Trans. Am.  Soc. Agric.
Eng. 24:430-435.

Owens, J.P. and C.S. Denny. 1979. Upper  Cenozoic de-
posits of the Central Delmarva Peninsula, Maryland and
Delaware.  U.S. Geological Survey  Professional  Paper
1067-A. 28 p.

Parkin, T.B. 1987. Soil microsites as a source of denitrifi-
cation variability.  Soil Sci. Soc. Am. J. 51:1194-1199.

Parkin, T.B. and J.IMeisinger. 1989. Denitrification below
the crop rooting zone as influenced by surface tillage. J.
Environ. Qual. 18:12-16.

Parsons, J.E., R.B. Daniels, J.W. Gilliam, and T.A. Dillaha.
1991. The effect of vegetation filter strips on sediment and
nutrient removal from agricultural runoff. In A. Bottcher
(cd.) Proceedings of the environmentally sound agriculture
conference. IFAS, Univ.  of Florida, Gainesville, FL.

Parsons, J.E., R.B. Daniels, J.W: Gilliam, and T.A. Dillaha.
1994a. Reduction in sediment and chemical load in agri-
cultural  field  run-off by vegetated  filter strips. Water.
Resour. Res. Inst. of UNC. Report No. 286. No. Carolina
State Univ., Raleigh, NC. 75 p.
Parsons, I.E., J.W.  Gilliam,  R. Munoz-Carpena, R.B.
Daniels, and T.A. Dillaha. 1994b. Nutrient and sediment re-
moval  by grass and riparian buffers, p. 147-154. In A.
Bottcher  (ed.)   Environmentally   sound  agriculture,
Proceedings of the  second conference, IFAS,  Univ. of
Florida, Gainesville, FL.
Paterson, K.G. and J.L. Schnoor. 1992. Fate of alachlor and
atrazine in a riparian zone field site. Water Environ. Res.
64:274-283.               '

Payich, M.J.,  G.W.  Leo,  S.F.  Obermeier,  and  J.R.
Estabrook. 1989. Investigations of the characteristips, ori-
gin, and residence time of the upland residual mantle of the
Piedmont of Fairfax Co., VA. U.S. Geological Survey Prof.
Paper 1352. U.S. Govt. Printing Off, Washington, D.C. 58

Perrin,  C.J., M.L.  Bothwell,  and P.A. Slaney.   1987.
Experimental enrichment of a coastal stream  in British
Columbia: effects of organic and inorganic additions on au-
totrophic periphyton production. Can. J. Fish. Aquat. Sci.
44:1247.                                    •

Peterjohn, W.T. and D.L. Correll. 1984. Nutrient dynamics
in an agricultural watershed: Observations_.on the role of a
riparian forest. Ecology 65:1466-1475.^             ,  "

Peterjohn, W.T. and D.L. Correll. 1986. The effects of ri-
parian forest on the volume and chemical composition of
baseflow in an agricultural watershed, p. 244-262. In D.L.
Correll (ed.) Watershed research perspectives. Smithsonian
Press, Washington, D.C.

Petersen, R.C.  1992. The RCE-a riparian,  channel, and en-
vironmental inventory for small streams in the agricultural
landscape. Freshwater Biol. 27:295-306.

Phillips, J.D. 1989. Nonpoint source pollution control ef-
fectiveness of riparian forests along a Coastal Plain river. J.
Hydrol. 110:221-237.

Phillips, P.J., J.M. Denver, R.J7 Shedlock, P.A. Hamilton.
1993. Effect of forested wetlands on nitrate concentrations
in ground  water and  surface  water  on the  Delmarva
Peninsula. Wetlands 13':75-83.,

Pinay, G., and  H. DeCamps. 1988. The role of riparian
woods in regulating nitrogen fluxes between the alluvial
aquifer  and surface water: a conceptual model. Regul.
Rivers Res. & Manage. 2:507-516.

Pionke,  H.B., J.R. Hoover, R.R. Schnabel, W.J. Gburek,
J.B. Urban, and A.S.Rogowski. 1988. Chemical-hydrologic
interactions in the near-stream zone. Water Resour.  Res.

Pionke, H.B., and H.M.-Kunishi. 1992. Phosphorus status
and content of suspended sediment in a Pennsylvania wa-
tershed. Soil Sci. 153:452-462.

Pionke, H.B., and J.B. Urban. 1984. Agricultural land use
impact on nutrients  in  grouridwater and base flow. In

Proceedings  of the National Well  Water Association
Technical Division, Eastern  Regional Ground  Water
Conference, July 23-24, pp 377-393, Newton, MA.      '-

Reay, W.G., D.L. Gallagher, and G.M. Simmons, Jr.'1992.
Groundwater discharge and its impact on surface water
quality in a Chesapeake  Bay  inlet. Water Resour.  Bull.
28:1121-1134.    '              .    '

Reddy, K.R. and W.H. Patrick. 1984. Nitrogen transforma-
tions  and loss in flooded soils and sediment. CRC Crit.
Rev. Environ. Control. 13:273-309.        -     '

Reilly, T.E., L.N. Plummer, P.J. Phillips, and E. Busenberg.
1994. The use of simulation and multiple environmental
tracers to quantify ground water, flow in a shallow aquifer.
Water Resour. Res. 421-433.              ,'

Rhodes, H,A.,.and W.ArHubert. 1991. Submerged under-
cut banks as macroinvertebrate habitat in a  subalpine
meadow stream. Hydrobioldgia 213:149-153:

Rhodes,  J., C.M. Skau, D. Greenlee, and D.L. Brown.
1985. Quantification of nitrate uptake by riparian forests
and wetlands in an undisturbed headwaters watershed, p.
175-179. In R.R. Johnson (ed.) Riparian ecosystems and
their  management: Reconciling conflicting  uses, USDA
Forest Service Gen. Tech Rep. RM-120.

Richardson, C.J., D.L, Tilton, J.A. Kuclee, J.D. M. Chamie,
and A. W. Wentz. 1978, Nutrient dynamics of northern wet-
land ecosystems. In R. E. Good, D.  F. Whigham and R. L.
Simpson (ed.). Freshwater wetlands ecological processes
and management potential. Academic Press, New York,
NY.   • .   •'.        '             ,               :  .

Rose, S. 1992. Tritium in groundwater of the Georgia
Piedmont: implications for 'recharge  and -flow  paths.
Hydrol! Processes 6:67-78.

Rose, S.  1993. Environmental tritium systematics of base-
flow in Piedmont Province Watersheds, Georgia (USA). J.
Hydrol. 143:191-216..

Ruiz, G.M.,  A.H. Hines, and M.H. Posey. 1993. Shallow
water as  a refuge habitat for fish and crustaceans in non-
vegetated estuaries:  an example from  Chesapeake Bay.
Mar, Ecol. Prog. Ser. 99:1-16.

Russell, R.S. and V.M. Shorrpcks.  1959, The relationship
between transpiration and absorption .of inorganic, ions by
intact plants. J. Exptl.Bot. 10:301-316.

Schipper, L.A., A.B. Cooper, QGHarfoot, and WJ. Dyck.
1993. Regulators of denitrification in an organic riparian
soil. Soil Biol. Biochem. 25:925-933.   /  .;•
 Schmidtling, R.C. 1973. Intensive culture increases growth
 without affecting wood quality of young southern pines.
 Can. J. For. Res. 3:243-250.

.Schnabel,  R.R.  1986.  Nitrate concentration in  a  small
 stream as affected by chemical and hydrologic interactions
 in the riparian zone. p.  263-282. In D.L. Correll (ed.)
 Watershed  research perspectives.  Smithsonian Press,
 Washington, DC.                                  ,

 Schnabel,  R.R., WJ.  Gburek, and W.L.  Stout.  1994.
 Evaluating riparian zone control ,on nitrogen entry into
• Northeast streams, p. 432-445. In  Riparian ecosystems in
 the humid U.S.: functions, values and management. Nat.
 Assoc. Conserv. Districts, Washington, D.C.

 Schnabel,  R.R., J.B.  Urban, and  WJ.  Gburek.  1993.
 Hydrologic controls in nitrate, sulfate, and chloride con-
 centrations. J. Environ.  Qual. 22:589-596.

 Schoene.berger, P., and A. Amoozegar. 1990. Directional
 saturated hydraulic conductivity and macropore morphol-
 ogy of a 'soil-saprqlite sequence. Geoderma 46:31-49.

 Schultz, R.C., J. Colletti, W. Simpkins, M. Thompson, and
 C, Mize. 1994. Developing a multispecies riparian buffer
 strip agroforestry system, p. 203-225. /^Riparian ecosys-
 tems in the humid U.S.: functions, values and management.
 Nat. Assoc. Conserv. Districts, Washington, D.C.

 Schwer,  C.B.  and J.C. Clausen.  .1989.  Vegetative filter
 treatment of dairy milkhouse wastewater. J. Environ. Qual.
 18:446-451.                        >
 Sena Gomes, A.R. and T.T. Kozlowski. 1980. Growth re-
 sponses  and  adaptations  of  Fraxlnus pennsylvanica
 seedlings to flooding. Plant Physiol.  66:267-271.

 Shedlock,  R J., P.A. . Hamilton,  J.M. Denver,  and' P.J.
 Phillips.  1993.  Multiscale approach to regional  ground-
 water  quality assessment of the Delmarva Peninsula. In
 W.M.  Alley (ed.)  Regional Ground-Water Quality. Van
 Nostrand Reinhold, New York, NY.
                          . ^ ,                '•
 Sheridan, J.M., R.G. Williams, L.S. Altier, R.R. Lowrance,
 and W.C. Mills. 1993. Hydrologic modeling for riparian
 management. ASAE Technical Paper No. 93-2598. Am.
 Soc. of Agric. Eng., St. Joseph, MI.

 Shields,  F.D., Jr.,  S.S. Knight, and C.M. Cooper.  1994.
 Effects of channel incision on base flow stream habitats
 and fishes. Environ. Manage. 18:43-57.

 Shirmohammadi, A., W.L. Magette, and JLX. Shoemaker.
 1991.  Reduction 'of nitrate  loadings to  ground water.
 Ground Water Monit. Rev. Winter 1991.     .

Simmons, G.M., Jr., E. Miles, W. Reay, and D. Gallagher.
1992. Submarine groundwater discharge quality in relation
to land use patterns  in the southern Chesapeake Bay. p.
341-349.  In  J.A.  Stanford and JJ.   Simons  (ed.)
Proceedings of the first international conference on ground
water ecology. Am. Water Resour. Assoc. Publ. TPS-92-2.

Simmons, R.C., A.J. Gold, and  P.M. Groffinan.  1992.
Nitrate dynamics in riparian forests: groundwater studies.
J. Environ. Qual. 21:659-665.

Sinokrot,B.A. andH. G. Stefan. 1993. Stream temperature
dynamics: measurements  and modeling.  Water .Resour.
Res. 29:2299-2312.

Slater, J.M. and D.G. Capone. 1987.  Denitrification in
aquifer soil and nearshore marine sediments influenced by
groundwater   nitrate.   Appl.   Environ.  Microbiol.

Sloto, R. A. 1994. Geology, hydrology, and ground-water
quality of Chester County, Pennsylvania. Chester County
Water Resources Authority Water Resource Report 2, 129

Smith, R.L.  and J.H.  Duff. 1988.-Denitrification in a sand
and gravel aquifer. Appl. Environ. Micro. 54:1071-1978.

Starr, R.C. and R.W. Gillham. 1993. Denitrification and or-
ganic carbon availability in two aquifers.  Ground  Water

Staver, K.W. and R.B. Brinsfield. 1990. Groundwater dis-
charge patterns in Maryland Coastal Plain agricultural sys-
tems, p. 593-603. In  J.A. Mihursky and A. Chaney, (ed.)
New Perspectives in  the Chesapeake System: a Research
and  management partnership  Chesapeake  Research
Consortium,. Solomons, MD.

Staver, K.W. and R.B. Brinsfield. 1994. Groundwater/estu-
arine interactions in. a Coastal Plain riparian agroecpsys-
tem. p. 256-276, In Riparian ecosystems in the humid U.S.:
Functions, values and management. Nat. Assoc. Conserv.
Districts, Washington, D.C.

Staver, K. W, R.B. Brinsfield, and W.L. Magnetic. 1988.
Nitrogen export from Atlantic Coastal  Plain soils. ASAE
Paper No. 88-2040. Am. Soc.'of Agric. Eng., St. Joseph,

Stefan, H.G. and B.A. Sinokrot. 1993. Projected global cli-
mate change impact on water temperatures in 5 north cen-
tral United-States streams. Clim. Change 24:353-381.

Sweeney, B.W.  1992. Streamside forests and the physical,
chemical, and trophic characteristics of Piedmont streams
 in eastern North America. Water Sci. Technol. 26:2653-

 Sweeney, B.W. 1993. Effects of streamside vegetation on
 macroinvertebrate communities of White Clay Creek in
 eastern North America. Proc. Acad. Nat. Sci. Phila.  144:

 Sweeney, B.W. and J.A. Schnack. 1977. Egg development,
 growth,  and metabolism  of  Sigara alternata (Say)
 (Hemiptera: Corixidae). Ecology 58:265-277.

 Switzer,  G.L., L.E. Nelson, and L.E.  Hinesley. 1979.
 Effects of utilization on nutrient regimes and site  produc-
 tivity, p. 251-259. In S.P. Gessel, R.M. Kenady and W.A.
, Atkinson (ed.) Forest fertilization conference proceedings.
 Institute of Forpst Resources Contribution No. 40. Univ. of
 Washington,  Seattle.                  .

 Sykes, K.J., A.W. Perkey, and R.S. Palone. 1994. Crop tree
 management  hi riparian zones, p.  65-80.  In Riparian
 ecosystems in the humid U.S.: Functions, Values and man-
 agement.  Nat. Assoc. Conserv. Districts, Wash., D.C.

 Taylor, J.R.,  M.A.  Cardamone, and W.J. Mitsch. 1990.
 Bottomland hardwood forests: their functions and values.
 p. 12-86.  In J.G. Gosselink, L.C. Lee, and T.A..Muir (ed.)
 Ecological processes and cumulative impacts: illustrated
 by bottomland hardwood  wetland  ecosystems. Lewis
 Publishers, Chelsea, MI.

 Theuer,,F. D., K. A. Voos, and W J. Miller. 1984. Instrearn
 water temperature model. USDAFish and Wildlife Service,
 Cooperative Instrearn Flow Service Group, Instrearn Flow
 Information Paper no. 16, Fort Collins, Colorado. 221  pp.

 Thornbury, W.D. 1965. Regional  geomorphology of.the
 United States. John Wiley, New York.

 Tiedje, J.M., S. Simians,  and P.M. Groffinan. 1989..
 Perspectives On measurement of denitrification in the field
 including recommended protocols  for acetylene based
 methods. Plant Soil 115:261-284.

 Tradell, M.R., R.W. Gillham and J.A. Cherry.  1986. An
 in-situ study of the occurrence and rate of denitrification in
 a shallow unconfined sand aquifer. J. Hydrol. 83:251-268.

 Urban,  J.B.,  1977. The Mahantango Creek Watershed -
 evaluating the shallow groundwater regime, p. 251-275. In
 D.L.  Correll  (ed.) Watershed research in Eastern North
 America. Chesapeake  Bay  Center  for Environmental
 Studies, Smithsonian Institute, Edgewater, MD.

 USEPA. 1986. National Water Quality Inventory. Report to
 Congress. USEPA, Washington, D.C.

USEPA. 1990a. National Water Quality Inventory: Report
to Congress. USEPA, Washington, D.C.

USEPA: 1990b. Managing nonpoint source pollution: Final
report to Congress on section 319 of the Clean Water Act.'
USEPA, Washington, D.C.

Vellidis, G., R Lowrance, M.C. Smith, and R.K. Hubbard.
1993. Methods to assess the water quality impact of a re-
stored riparian wetland. J. Soil Water Cons;-48:223-230.

Vugts, H.F. 1974. Calculation of temperature variations of
small mountain  streams (thermal pollution).  J. Hydrol.
23:267-278.                      •':

Wallace, J. B.* and M. E. Gurtz.  1986. Response of Baetis
mayflies (Ephemeroptera) to catchment logging. Am. Midi.
Nat. 115:25-41.       ,                         '     '

Webster, J.R., S,W.  Golladay, E.F. Benfield,  J.L. Meyer,
W.T. Swank, and J.B. Wallace. 1992. Catchment  distur-
bance and stream response: an overview of stream research
at Coweeta Hydrologic Laboratory, p. 231-253. In EJ.
Boon, P. Calow, and G.E. Petts (ed.).River conservation and"
management. John Wiley & Sons, New York, N.Y.

Webster J.R., J.B. Wallace, and E.F. Benfield: (In press)
Organic processes in streams of the eastern United States.
In C.E. Gushing, G.W. Minshall, and K.W. Cummins (ed.)
Rivers of the .world. Elsevier.

Weil, R.RV, R.A. Weismiller, and R.S. Turner. 1990. Nitrate
contamination of groundwater under irrigated CoastaJ Plain
soils. J. Environ. Qual. 19:441-448.

Wells, C.G. andD.M. Crutchfield. 1969. Foliar.analysis for
predicting loblolly pine response to phosphorus fertiliza-
tion on wet sites. Forest Service Research Note SE-128.
S.E. For. Exp. St., USDA-Forest Service, Asheville, NC.

Welsch, D.J. 1991. Riparian Forest Buffers. USDA-FS Pub.
No. NA-PR-07-91. USDA-FS, Radnor, PA.

Wharton, C. H. and M. M. Brinson. 1978. Characteristics
of southeastern river systems, p. 32-40. In R. R. Johnson
and J.  F. McCormick  (ed.). Strategies for protection and
management of floodplain wetlands  and  other riparian
ecosystems. Gen. Tech. Rep. WO-12, Forest Serv., U. S.
Dept. Agric., Wash. D. C. 410 pp.

Whigham, D.F., C. Chitterling, B: Palmer, and J. O'Neill.
1986. Modification of runoff from upland watersheds—the
influence of a  diverse riparian ecosystem, p. 305-322, In
Watershed research perspectives.  Smithsonian  Institution,
Washington, D.C.   *

Yates, P. and J.M, Sheridan. 1983. Estimating the effective-
ness of vegetated floodplains/wetlands as nitrate-nitrite and
orthophosphorus filters. Agric. Ecosyst. & Environ. 9:303-
314.-    ' '   ."  '-''           :       "    c -,: •-  •   ;  .

Yeomans, J.C., J.M. Bremner and G.W. McCarty. 1992.
Denitrification  capacity and  denitrification potential of
subsurface soils. Comm. Soil Sci. Plant Anal. 23:919-927.

Young, R.A.,   T. Huntrods  and W.  Anderson.  1980.
Effectiveness of vegetated buffer strips in controlling pol-
lution from feedlot runoff. J. Environ: Qual. 9:483-487.

Zimmerman, R. C., J. C. Gpodlett, and G. H. Comer. 1967.
The influence  of vegetation  on'Channel  form of small
streams., p.  255-275. In Symposium on .river morphology.
International Assoc. of Scientific Hydrology, Publ. No. 75.