EPA910-R-06-003
             United States
             Environmental Protection
             Agency
Region 10
1200 Sixth Avenue
Seattle WA 98101
Alaska
Idaho
Oregon
Washington
             Office of Water & Watersheds
              March 2006
             A Classification of Pacific
             Northwest Reservoirs with
             Respect to Nutrient Processing
                                Western Forested
                                Mountains

-------
                                                           EPA910-R-06-003
                                                                March 2006
   A Classification of Pacific Northwest Reservoirs with Respect to
                           Nutrient Processing
R.M. Vaga, US Environmental Protection  Agency,  Region  10, Office of Water and
           Watersheds, 1200 Sixth Ave. Seattle, WA 98101


AT. Herlihy,  Department of Fisheries and  Wldlife, Oregon State University, 104 Nash
           Hall, Corvallis, OR 97331

R. Miller, M.M. Sytsma, Environmental Sciences and Resources, Portland State University,
           Portland Oregon, 97202

-------
DISCLAIMER
The information in this document has been funded wholly or in part by the United States
Environmental Protection Agency.  It has  been subjected to the Agency's  peer and
administrative review and it has been approved for publication as an  EPA document.
Mention of trade names or commercial products does not constitute  endorsement or
recommendation for use.
(Cover Photos: Top - Lake Billy Chinook, a Xeric West reservoir in eastern Oregon;
bottom - Bumping Lake, a Western Forested Mountains reservoir in Washington).

-------
                        TABLE OF CONTENTS
                                                            Page
Abstract 	 iv
Introduction	  1
Methods	1
Results	5
Discussion	14
Conclusions 	 16
References 	18
Appendix I	 20
Appendix II	 21

-------
Abstract
To begin the process of developing nutrient criteria for Pacific Northwest reservoirs, 48
reservoirs were sampled in Oregon, Washington and Idaho during summer 2002.  The
National Inventory of Dams lists 2419 dams in the Pacific Northwest. Over 75% of them
represented reservoirs that were less than 30 hectares and were not considered in this
study.  After removing these small reservoirs and other non-target sites, there were 328
reservoirs that met our target criteria. The 48 sample reservoirs were selected from these
328  target reservoirs using a  systematic, randomized design with Omernik  Nutrient
Ecoregions (Xeric West or Western Forested Mountains) as a stratification variable.

Total phosphorus (TP) concentrations were higher and Secchi depths lower in the Xeric
West relative to the Western Forest Mountains.  Total nitrogen (TN) and chlorophyll-a
concentrations were not significantly different between the two Ecoregions.  All four nutrient
variables were strongly correlated with each other in both Ecoregions. However, reservoir
purpose class was not correlated with nutrient concentrations.

In the  Western Forested  Mountains, TP concentrations were very strongly related to
measures of ionic strength (e.g., base cations, alkalinity, conductivity); an indication that
there is a natural TP gradient related to watershed geology/soils. In the Xeric West TP was
less  strongly related to potassium and sodium and not related at all to the other indicators
of ionic strength.  The source of TP could be some specific geologic source that is  only
related to potassium or sodium or it could be associated with fertilizers that are high in
potassium.

Principal components analysis and cluster analysis were performed on reservoirs in the
Western Forested Mountains and  the  Xeric West.  Analysis did not reveal any clear
classification of reservoirs in the WFM.  In the Xeric West 5 variables were  (potassium,
alkalinity, absorbence at 326 nm, turbidity and chlorophyll) used to identify 4 clusters of
reservoirs. A reservoir typology resulting in 3 categories was developed from these  four
clusters.

Category 1 -  Low ionic strength, high  transparency, low absorbence, low turbidity, low
chlorophyll.

Category  2 - Moderate ionic strength, moderate transparency, moderate absorbence,
moderate turbidity due either to algae or particulate matter.

Category 3 - High ionic strength, very high absorbence, very low transparency due to algal
and/or nonalgal turbidity.

These three classes differed in distributions of total P concentration and water transparency.
Total N and chlorophyll values were not different among categories.

The  use of Nutrient Ecoregion  provides a  useful classification variable  in categorizing
reservoirs in the Pacific Northwest.  However, it appears that the Level III Ecoregion Eastern
Cascades should be  placed into the Xeric West Nutrient Ecoregion  when classifying
reservoirs.   The  reasons for  this are unclear but may have  to  do with  geology or
precipitation.

Future studies on reservoir nutrient dynamics in the Pacific Northwest should explicitly take
nonalgal turbidity into account to differentiate between the effects of nutrients and water
transparency on algal  growth.
                                        IV

-------
Introduction
According  to  the U.S.  Environmental
Protection Agency=s (EPA=s) 1996 report
to Congress, a large percentage of U.S.
lakes and  reservoirs are water quality
impaired, many due to excess nutrients.
In order to address this problem EPA has
devised a strategy for the development of
regional nutrient criteria (USEPA 1998).
Implicit in the strategy is  the recognition
that excess nutrients are a major cause of
water quality impairment in the United
States. The strategy also acknowledges
that because of diverse geology, climate,
and morphometry, single national nutrient
criteria for  lakes and reservoirs are not
appropriate. EPA's current approach for
developing  nutrient criteria for lakes and
reservoirs  involves  the  following  steps
(Kennedy 2000):

1. Establish   regional   technical
   assistance groups.
2. Delineate nutrient ecoregions.
3. Classify or  group lakes/reservoirs
   using non-nutrient parameters.
4. Establish   an appropriate  nutrient
   related  database.
5. Establish   reference  conditions  for
   each group or classification.
6. Develop nutrient criteria for  each
   group or classification.
Although  the  same  basic  physical,
chemical  and  biological  processes
determine nutrient dynamics in lakes and
reservoirs,  the  age,   morphometry,
location  in  the  drainage  basin  and
hydrological  characteristics of reservoirs
make them a unique ecosystem (Cooke
and  Kennedy,  1989,  Thornton et  al.
1990).     As   such   developing   a
classification of reservoirs with respect to
nutrient processing must take the unique
nature of these systems into  account
(Straskraba, 1993). This report describes
the methods and results  of a study of
nutrient levels and  classification of  48
randomly  selected   reservoirs   in  the
Pacific  Northwest  states  of  Oregon,
Washington  and Idaho. In previous  work
we have demonstrated that it is possible
to  classify  lakes   using   Level  III
Ecoregions  (Vaga  and  Herlihy  2004,
2005).
Methods

Site selection and population definition

Reservoirs range greatly in size and type
across the Pacific Northwest from small
farm  ponds to giant hydropower/flood
control reservoirs on the Columbia River.
For this study,  the target population of
interest was restricted to specific reservoir
types and size range. The population of
interest only  includes reservoirs with  a
surface area between 30 and  10,000
hectares.     Target  types   include
hydropower, flood control, water supply,
irrigation, and recreation reservoirs. Mine
tailing, wastewater effluent, debris control,
run-of-the-river,  and  enhanced  lake
reservoirs are  not  part of the  target
population.  Determination of type was
based on the National Inventory of Dams
purpose  designations.  Enhanced lake
reservoirs were defined as lakes in which
the dam  height is less than 25%  of the
maximum depth of the reservoir.

Reservoir location was expected to be an
important factor in  explaining  nutrient
concentrations so ecoregion was used as
a stratification variable in site selection.
To aid in developing nutrient criteria at the
national level, EPA has  delineated the
conterminous   U.S.  into   nutrient
ecoregions according  to  geologic and
climactic characteristics (Omernik 1998).
Portions of three nutrient ecoregions lie
within  Region  10:   Western  Forested
Mountains (WFM), Xeric West (XW), and
Willamette   and   Central   Valleys.
Reservoirs within Region 10's portion of
the Wllamette and Central Valley nutrient
ecoregion are not common and for this
project  were  considered  part  of  the
Western   Forested   Mountains   region.
Thus there were two ecoregion strata for
site selection.

The sample frame or explicit list  of the
target population from which sites were
selected  was the  list of all reservoirs in
                                   Page 1 of 21

-------
the  National  Inventory  of  Dams  for
Oregon, Washington, and Idaho.  Based
on this frame, there are 2,417 dams in
Oregon, Washington, and Idaho (Table
1).  Reservoirs outside the type and size
cutoffs described above were  removed
from the frame.  Only 328 of the 2,417
reservoirs  in the Pacific Northwest met
our target criteria. The majority of dams
in  the database were non-target due to
small surface area (Table 1).  To make
inferences to all 328 reservoirs in this
target population, sample reservoirs were
selected from this set using a stratified,
randomized design.

The  desired  sample   size  was  48
reservoirs,  24 from  each  of  the  two
ecoregion  strata.  To ensure  a broad
geographic spread of reservoirs across
the  study  area, level  III ecoregions
(Omernik,  1987)  were used  to define
spatial clusters. The Northern Basin and
Range ecoregion was split into Oregon
section and Idaho sections because the
region is considerably larger than other
Level III ecoregions in Region  10.  The
328 reservoirs in the sample frame were
randomly sorted within level III ecoregion
and  then   level  III  ecoregion  blocks
randomly sorted within  each of the two
nutrient ecoregion strata. Sites were over
sampled so that there would be alternate
reservoirs. Therefore, 48 reservoirs were
selected from each strata, 24 primary and
24  secondary  (alternates).    After a
random  start, reservoirs were picked an
interval  of  N/48 from  the randomized
frame of all reservoirs in the strata (N is
the total number of reservoirs  in  the
strata).

Of  the  48 primary randomly  selected
reservoirs, 33 were successfully sampled
in the summer of 2002. Of the 15 lakes
not sampled,  6 were  due  to access
problems (private property, no launch site,
or remoteness), 2  were  dry,  2  were
enhanced, 1 had no dam, and 3 were less
than 30 ha when visited in the field.  In
addition, one reservoir wasn't  sampled
due  to  time constraints.   Thus 31%
(15/48) of the 328 reservoirs in the target
population were  not really target in  the
real world.  If a primary reservoir could not
be  sampled,  one  of  the  secondary
reservoirs  from   the  same  Level  III
ecoregion was sampled.  If that reservoir
could not be sampled another secondary
reservoir  from  the  same  Level   III
ecoregion was sampled.   If the pool of
secondary  reservoirs within a Level III
ecoregion was exhausted, the original
sample frame was revisited and a tertiary
reservoir was selected from  the same
Level III ecoregion.

In three cases  this process  was  not
followed.

   1.  One Idaho Batholith reservoir: In
       the   Idaho  Batholith  Level   III
       ecoregion one primary reservoir
       was not sampled because of time
       constraints   (Fish   Creek
       Reservoir). A secondary reservoir
       with easier access  was sampled
       in its place (Brundage Reservoir).
   2.  Two Northern Rockies reservoirs:
       In the Northern Rockies Level III
       ecoregion the pool  of secondary
       reservoirs was exhausted and no
       suitable tertiary reservoirs were
       identified in the sample  pool. One
       reservoir located partially in  the
       Columbia  Plateau   and   the
       Northern Rockies ecoregions was
       selected  from  the  Columbia
       Plateau's   secondary  list.
       Laurence Lake was selected from
       the Cascade Level III ecoregion to
       replace  the  other   missing
       Northern  Rockies reservoir.
   3.  One Puget  Lowland  reservoir:
       The  primary and secondary  list
       was exhausted, and due to time
       constraints,  a reservoir close to
       Portland  was sampled  (Henry
       Hagg Lake,  tertiary, Wllamette
       Valley ecoregion)

Inspection  of the  data,    e.g.  cations
suggested  that East Cascade  reservoirs
were far more similar to reservoirs in  the
Xeric West that those in  the Western
Forested   Mountains.  Under  similar
considerations one reservoir (Thief Valley,
east  end   of  the   Blue  Mountains
Ecoregion)  which was originally in  the
WFM was  placed into the Xeric West.
Long  Lake (WFM) was   deleted from
analysis because it is  currently 303(d)
listed for nutrients  due to high nutrient
inputs  from point sources. In addition,
                                  Page 2 of 21

-------
after analysis of watershed versus dam
location  it was  determined  that  two
reservoirs  (Conconully,   Leader)   are
actually in the WFM  rather than the XW.
The resulting sample included 27 Xeric
west reservoirs and 20 Western Forested
Mountains (Fig. 1, Table 2).

Table  1.   Sample Frame  size  and target
population  size   for  Pacific   Northwest
Reservoirs. Non-target reservoirs were those
reservoirs excluded from the Frame  because
they did not meet criteria.  Of the  2419
reservoirs in NID  only 328  met the criteria.
About two-thirds of those were in the Western
Forested Mountains Ecoregion.
Table 2.  List of sampled reservoirs with location and
surface area.  Numbers refers to reservoir locations in
Figure  1. Most  sensitive  use:  HM:   Hydrologic
Management, RC: Recreation, WS: Water Supply.
Total Number in lnventory= 2419
A/on -Target Reservoirs
Too Small (< 30 ha)
Non-Target Types
Enhanced Lakes
Duplicate/Auxiliary Lakes
Inaccessible
Other
Total Non-Target
1828
122
71
53
7
10
2091
Target Reservoirs = 328
W. Forested Mt. Ecoregion
Xeric West Ecoregion
220
108

No
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
WESTERN FORESTED MOUNTAINS
Lake Name
Brundage Res
Bumping Lake
Clear Lake
Conconully Res
Diablo Lake
Fern Ridge Res
Galesville Res
Henry Hagg Res
Judy Res
Lake Simtustus
Laurence Lake
Leader Lake
Mayfield Res
N Fork Clackamas
Ochoco Res
Prineville Res
Sage Hen Res
Skookumchuck Res
Toketee Res
Willow Creek
Area
87.0
527.3
56.7
133.6
400.7
3788.0
257.0
468.2
56.7
218.5
64.8
74.9
910.6
141.6
493.7
1449.0
96.3
218.5
41.3
130.3
LAT
45.0
46.8
46.6
48.5
48.7
44.1
42.8
45.4
48.4
44.6
45.4
48.3
46.5
45.2
44.3
44.1
44.3
46.7
43.2
42.4
LONG
-116.13
-121.30
-121.27
-119.75
-121.13
-123.29
-123.18
-123.21
-122.19
-121.23
-121.66
-119.70
-122.59
-122.28
-120.73
-120.78
-116.19
-122.72
-122.42
-122.44
Use
HM
RC
RC
RC
RC
RC
WS
WS
WS
RC
RC
RC
RC
RC
RC
HM
WS
HM
RC
WS
XERIC WEST
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
Alexander Res
Barry Res
Ben Ross Res
Black Canyon Res
Blue Creek Res
Bray Lake
Bryant Mt. Res
Cedar Creek Res
Chickahominy
Daniels Res
Grasmere Res
JC Boyle Res
Kern Res
Little Blue Creek
Mann Creek Res
Mud Lake
Muddy Creek Res
Paddock Valley Res
Friday Res
Sand Creek Res
Scooteney Res
Stone Res
Thief Valley Res
Trail Storage Pond
Upper Rock Creek
Warmsprings Res
Weston Creek Res
494.9
60.7
142.9
445.2
76.1
82.6
48.6
424.9
214.1
151.8
31.6
170.0
43.7
53.8
127.5
68.0
78.9
542.3
85.8
31.6
617.2
100.0
373.5
104.8
155.4
1862.0
45.3
42.6
42.1
44.5
43.9
42.3
43.0
42.1
42.2
43.5
42.3
42.3
42.1
42.9
42.2
44.3
42.2
42.2
44.2
42.3
44.2
46.6
42.0
45.0
43.0
42.6
43.5
42.1
-111.70
-119.53
-116.44
-116.44
-116.18
-114.88
-121.33
-114.88
-119.61
-112.44
-115.90
-122.04
-118.78
-116.12
-116.89
-119.72
-120.52
-116.60
-119.90
-111.61
-119.03
-112.69
-117.78
-115.32
-119.31
-118.21
-112.13
HM
HM
WS
HM
WS
HM
HM
HM
HM
HM
WS
RC
HM
WS
RC
HM
HM
HM
HM
RC
HM
WS
RC
HM
HM
RC
WS
                                      Page 3 of 21

-------
Field and laboratory methods

Reservoirs   were  visited  during  the
summer growing season, May through
September,  2002.    Reservoirs  were
sampled by boat and those from different
nutrient ecoregions were  sampled on a
rotating schedule as much as Ipgistically
possible to  avoid seasonal bias.   The
deepest location in each  reservoir was
found by starting at the dam and using an
electronic depth finder. A water sample
and in situ measurements were collected
at the deepest spot in each reservoir.  In
situ profiles of temperature,  dissolved
oxygen, pH, Turbidity, redox potential,
light   extinction   coefficient,   and
conductivity  were collected  using  a
Hydrolab Sonde 4a meter. Secchi depth
was determined as the average of the
disappearance and reappearance depths
of a 20 cm Secchi disk. Two, 1  L water
samples were collected from 1 m depth
using a 2.5 L, Model 1010X Niskin bottle.
One  sample   was   preserved   by
acidification to pH < 2 for analysis of total
nitrogen and phosphorus.   The  other
sample was filtered with a Whatcom glass
fiber filter that had a nominal pore size of
0.7 urn. The filter paper was frozen, kept
in the dark and analyzed for chlorophyll-a.
The filtrate was kept and used for analysis
of anions, base  cations, DOC, alkalinity,
and light absorbence.  All water samples
were  kept  cold  and  in  the dark  until
analysis.

n the  lab, chlorophyll-a was determined
fluorometrically.    Base  cations  were
analyzed by flame atomic  absorption
spectrometry and sulfate and chloride by
ion  chromatography.   Alkalinity  was
determined by Gran titration, and DOC by
carbon analyzer. Total phosphorus and
nitrogen were  analyzed  by persulfate
digestion   and  colorimetry.     Light
absorbence  at  325 and  440 nm  was
measured with a spectrophotometer.

Field measurement of  reservoir physical
habitat and substrate  composition were
made using a condensed form of the U.S.
EPA EMAP  lake protocols (Baker et al.,
1997).  Sampling locations were set up
following  a  random start at  10 equal
interval  shoreline stations  around the
reservoir. Percentage category estimates
 Figure 1. Location of the 47 reservoirs used in
 this study.  Numbers refer to reservoir number
 in Table 2.
were made at each station for substrate
classes  (clay,  sand,   cobble/gravel,
boulder,  bedrock),  vegetation  cover
classes (barren, herbs,  shrubs, trees),
and  macrophyte   cover   classes
(submergent, emergent, floating) as well
as presence/absence of human influence
categories.  These data were analyzed by
averaging the 10 measurements together
to calculate shoreline cover estimates for
the entire reservoir.  In  addition, depth
below  maximum pool estimates  were
made by comparing lake level at the time
of sampling to shoreline evidence of
maximum reservoir conditions.

Data Analysis

Data   were   entered   into   Excel
spreadsheets and  converted  into SAS
databases for  analysis.  As is typically
observed for water chemistry data, the
data are not normally distributed.  Most
water chemistry variables  are   highly
skewed.   For  all  statistical  analyses,
chemical   variables  were   Iog10
transformed as were reservoir surface
area and maximum  depth.   Shoreline
percentage data and  Secchi depth were
not transformed. All correlation analyses
were   done   using   Spearman  rank
correlations in SAS.

A two-way analysis  of variance (ANOVA)
was used to test for  differences in nutrient
concentration due  to ecoregion (Xeric
West versus Western Forested Mountain)
and   reservoir  purpose  (Hydrologic
management,  Drinking  water,  or
Recreation).   Reservoir  purpose  was
taken from  the National  Inventory of
Dams database. Many reservoirs were
                                  Page 4 of 21

-------
coded  with  multiple  purposes.    The
hydrologic management class includes
hydropower,  irrigation,  or  navigation
categories. A single purpose class was
assigned to each reservoir as  a "most-
sensitive"  purpose rating sensitivity as
Water supply >  Recreation > Hydrologic
management.   In other words if  there
were any  one purpose listed as water
supply, the reservoir was coded as water
supply even though it may have been also
listed for other purposes. ANOVAs were
run in SAS using PROC GLM.

To get at an initial indication of possible
reservoir clusters or types, a  principal
components   analysis  (PCA)   was
performed on the major reservoir typology
variables.  Axis or factor scores were then
clustered via cluster analysis to look for
groups  of reservoirs.   PCA was  done
using   SAS   PROC    FACTOR,
method=PRINCOMP.  Cluster analysis
were
done using PC-ORD using the Flexible
Beta method (beta=-0.5).

Results

Ecoregional Patterns
Of the total number of reservoirs included
in this analysis, 27 were in the Xeric West
and 20 were in the Western Forested
Mountains Ecoregion.  By most sensitive
purpose class, 11 were water supply, 17
were recreation, and 19 were hydrologic
management reservoirs.   Most of the
hydrologic management reservoirs were
in  the Xeric West  and most of the
recreation reservoirs were in the Western
Forested Mountains  (Table 2).
The  use of  Nutrient Ecoregion  as  a
classification  variable  for  reservoirs
proved  to  be useful  in characterizing
reservoirs (Table  3).  Reservoirs in the
WFM tended to be larger, deeper and of
greater volume that those  in the XW.
However, distributions of reservoir surface
areas within each of the two ecoregions
were virtually identical.  Similarly,  there
was little difference in surface area across
purpose class.   In a two-way ANOVA,
mean   surface   areas   across  both
ecoregion and purpose class were not
significantly  different  (p  >   0.25).
Maximum reservoir  depths,  however,
were significantly (F=7.0, p=0.01) higher
in the Western Forested Mountains than
the Xeric West (Tables 3, 4).   Depths
across   purpose   class  were  not
significantly different.

Water  transparency  as  measured  by
extinction coefficient  and Secchi depth
were  greater  in the  WFM  (Table  3).
Reservoirs in the XW were more turbid
that  in  the   WFM.    Surface  water
temperatures tended to be greater in the
XW.  Reservoirs in the WFM tended to be
more  dilute  that in the XW.   Median
conductivity, alkalinity and ANC were all
much lower in  the WFM than in the XW
(Table 3).   Chlorophyll concentrations
were  higher in the XW as was DOC.
Reservoirs in the XWwere higher in DOC
(Abs325) and  humics (Abs440)  than  in
the WFM (Table 3).

Cations (K, Na, Ca,  Mg) were all higher in
the XW  reservoirs as compared with
WFM reservoirs.  Anions (CL, SO4) were
also higher in the XW (Table 3).

The  primary nutrient  criteria variables,
total phosphorus (TP), total nitrogen (TN),
chlorophyll-a, and Secchi depth, and their
distributions   within   each   nutrient
ecoregion are  shown  in Figures 4 -  7.
Mean   TP   concentrations   were
significantly  higher in  the Xeric  West
Ecoregion and in Hydrologically managed
reservoirs (Tables  3,  4,  Fig. 4).  TN
concentrations  showed no  difference
across either ecoregion or purpose class
(Table  4,   Fig.   5).    Chlorophyll-a
concentrations   were  higher   in
hydrologically  managed reservoirs but
showed  no  ecoregion effect whereas
Secchi depth was higher in the Western
Forested  Mountains  but  showed  no
reservoir purpose effect (Table 4, Figs, 6,
7).  For all four variables, the ecoregion-
purpose interaction terms in the ANOVA
were not significant.

Turbidity, light  extinction coefficient, and
absorbence at 325 nm (DOC) and 440 nm
(fulvic acids) were  also analyzed. The
correlation  matrix  of  the  relationships
among these variables shows that in both
Ecoregions   the  variables  are  highly
correlated with each other (Table 5).   In
the Xeric West  there  were very high
                                  Page 5 of 21

-------
correlation coefficients  (r>0.8)  for  the
relationships  among  the transparency
variables (Secchi, extinction, absorbence)
and between transparency and turbidity.
The   relationship  between   extinction
coefficient and   chlorophyll-a     was
significant  but  much   lower   (r~0.5)
indicating that transparency in the Xeric
West  is  affected  by abiotic turbidity in
addition to algal production (Fig. 8).
  ^   BQOOj
      1000=
       100 =
        10-
        Westem Forested Mountains
                             Xerte West
                   Nutrient Ecoreglon
 Figure 2.  Box and whisker plot of reservoir
 surface area by nutrient ecoregion.  Boxes
 show  the  interquartile range and  median,
 whiskers  the  minimum/maximum values.
 I  40-
    30-
    20-
     10-
     o-
 8   Western Forested Mountains      Xeric West
                 Nutrient Ecoregion

Figure 3.  Box and whisker plot of maximum
depth by nutrient ecoregion. Reservoirs in the
Western Forested Mountains are significantly
deeper than those in the Xeric West.
On  the  other  hand,  in  the  Western
Forested  Mountains,  transparency   is
more  strongly related to  chlorophyll-a
(r~0.8), indicating  a more important role
for algal production in transparency. This
hypothesis  is  also  supported  by  the
scatterplot  of  Secchi   depth   versus
turbidity (Fig. 9). With one exception, the
Xeric West reservoirs have  almost a
Table 3.  Median (± 1 SE) values for physical
and  chemical  characteristics of reservoirs
sampled in the Western Forested Mountains
and Xeric West.  Volume is (m3 x 10s), depth
is maximum depth and ions are (uEq/L).

Variable
Area (ha)
Volume
Depth(m)
Ext(rrr1)
Secchi (m)
Turb(NTU)
Temp (°C)
PH
Eh (mV)
Dosat (%)
TP (ug/L
TN (ug/L)
K
Na
Ca
Mg
SO4
Cl
Alk (mg/L)
Cond(uS)
ANC(uEq/L
Chla (ug/L)
DOC(mg/L)
Abs325
Abs440
WFM XW
Median
218.5
36.7
11.2
0.5
3.4
0.9
17.4
7.9
340.0
97.0
23.0
213.2
22.8
187.0
289.4
157.7
43.6
64.1
28.2
60.0
563.6
4.6
2.2
50.0
9.3
se
180.0
57.5
2.7
0.1
0.5
1.1
0.8
0.2
18.5
3.3
13.6
54.3
6.9
77.7
75.5
48.8
49.1
14.7
7.8
18.3
155.5
3.1
1.6
15.2
2.5
Median
104.8
6.3
6.0
2.1
0.7
14.7
22.0
8.4
318.0
96.9
109.0
430.9
88.2
357.6
524.0
244.1
77.2
101.8
57.6
113.4
1151.5
7.5
11.7
267.3
55.8
se
75.8
27.7
0.9
0.4
0.3
8.7
0.6
0.1
6.6
5.1
43.7
46.3
13.7
140.2
116.2
213.0
63.4
147.7
11.5
41.4
230.0
27.8
2.8
87.4
26.8
     Western Forested Mountains
                Nutrient Ecoregion
Figure 4.  Box and whisker plot  of total
phosphorus  by  nutrient  ecoregion.  Total
phosphorus is significantly higher in the Xeric
West reservoirs.
                                      Page 6 of 21

-------
 I
 z
 "ffl
 S
     Western Forested Mountains
              Nutrient Ecoregion
Figure 5.  Box and whisker plot of total
nitrogen by nutrient  ecoregion.   Total
nitrogen   concentrations  are  not
significantly different between ecoregions.
straight-line relationship between turbidity
and  Secchi depth (the one outlier is a
reservoir  with  chlorophyll-a=519  ug/L
which is responsible for the 0.15 m Secchi
depth).   In  the   Western  Forested
Mountains, the  plot isn't nearly as linear,
indicating non-turbidity factors influencing
Secchi depth.

Reservoirs with high TP tend to have high
TN (Fig. 10).   Reservoirs in the Xeric
West tend to have higher concentrations
of both TP and TN. However, the slope of
the  log(TN) -  log(TP)  regression  was
significantly (p < 0.05) higher in the WFM.
Chlorophyll-a concentrations  increased
with  TP in  both ecoregions (Fig. 11) but
again the slopes were significantly lower
in the XW.

Chlorophyll also increased with TN in both
nutrient ecoregions (Fig. 12).    Secchi
transparency decreased with increasing
TP  in   both   ecoregions   (Fig.  13).
However,  the   lowest  transparencies
observed in the Xeric West were not due
to chlorophyll but non-algal turbidity.

TP is often used  a measure of  trophic
status.  A commonly used scheme is to
use TP  concentrations of 10, 30 and 50
ug/L as  the cutoffs between oligotrophic-
mesotrophic, mesotrophic-eutrophic, and
eutrophic-hypereutrophic lakes.   Using
this scheme, none of the reservoirs in the
Xeric West are oligotrophic, and 62.5% of
the reservoirs  are hypereutrophic.   In
contrast,   in  the   Western  Forested
Mountains, one-third of the reservoirs are
oligotrophic and 25% are hypereutrophic.
As these data are from a random sample
of all Pacific Northwest reservoirs, these
percentages should  reflect conditions in
all 328 reservoirs that met our  target
criteria. Extrapolating the values in Table
3 to all reservoirs in the PNW using the
two Nutrient Ecoregion estimates and the
total  number  of reservoirs  in  each
ecoregion  strata (Table 1),  22.3%  are
oligotrophic, 25% are mesotrophic,  3%
are   eutrophic,   and  37.3%   are
hypereutrophic.
                                             §»  100:
 2.
 o
     Western Forested Mountains           Xeric West
               Nutrient Ecoregion

Figure 6.    Box  and  whisker  plot of
chlorophyll-a    by  nutrient  ecoregion.
Chlorophyll   concentrations   were  not
significantly different between ecoregions.

Watershed Chemistry and Nutrients

Base cation concentrations are  typically
controlled by watershed geology/soils and
weathering  rates.   There were  clear
differences  between ion  relationships
between  Nutrient Ecoregions.   In  the
WFM correlations among  most cations
and  anions  were high  and  significant
(Table 6). In the Xeric West, the divalent
base cations (calcium and magnesium)
are strongly related to each other and the
monovalent base cations  (sodium and
potassium) are strongly  related  to each
other but the two groups  are not strongly
related to each other.

These results suggest that in the Western
Forested  Mountains all of  the  base
cations  come from a similar source.  In
contrast, in the Xeric West there  appears
to be independent sources for these two
groups of base cations .
                                   Page 7 of 21

-------
    6-
    4-
    2
             T
    01

    Western Forested Mountains
g
                                 Xerfc West
                  Nutrient Ecoregion
Figure 7. Box and whisker plot of Secchi depth
by  nutrient  ecoregion.   Reservoirs  in  the
Western Forested  Mountains are  significantly
more transparent that those in the Xeric West.
                                                       • * * Western Forested Mountains
                                                          Xerlc West
                             Total Phosphorus (ug/L)
             Figure 10. Relationship between total nitrogen
             and  total  phosphorus  in  Pacific  Northwest
             reservoirs.  Total nitrogen increases  in  both
             ecoregions with total phosphorus.
                         • • ' Western Forested Mountains
                         " " - Xerlc West
                   •  -s  -.
                     Chlorophyll (ug/L)
  Figure 8. Relationship between Secchi depth
  and   chlorophyll   in   Pacific  Northwest
  reservoirs.    The  dependence  of  water
  transparency on chlorophyll is much stronger
  in the WFM as compared with the XW.
                                                       tt,0:
                        •Western Forested Mountains
                        " Xeric West
                                                                        H ."
                              Total Phosphorus (ug/L)
               Figure   11.      Relationship   between
               chlorophyll   and   total   phosphorus.
               Chlorophyll increases with  increasing total
               phosphorus in both ecoregions.
 E  6
                      * Vfestern Forested Mountains
                      3 Xerlc Wast
                  Turbidity (NTU)
Figure 9. Relationship between Secchi depth
and turbidity in Pacific Northwest reservoirs.
Highest turbidities in the Xeric west are due to
non-algal turbidity.
                                                  I   "
                                                  e
                                                  £
                                                  o
                       • Western Forested Mountains
                        Xerlc West
                                      200

                               Total Nitrogen (ug/L)
              Figure 12.  Scatterplot of chlorophyll versus
              total nitrogen.  Chlorophyll increases with total
              nitrogen in both ecoregions.
                                        Page 8 of 21

-------
Table 4. Two-way analysis of variance results
on  the  effects  of ecoregion  and  reservoir
purpose class on nutrient criteria  variables.
Bold values were significant at p < 0.05.
Variable
Total
Phospho
rus
Total
Nitrogen
Chlorop
hyll-a
Secchi
depth
Class
Ecoregion
Reservoir
Purpose
Ecox
Purpose
Interaction
Ecoregion
Reservoir
Purpose
Ecox
Purpose
Interaction
Ecoregion
Reservoir
Purpose
Eco x
Purpose
Interaction
Ecoregion
Reservoir
Purpose
Eco x
Purpose
Interaction
F
5.8
(0.021)
4.8
(0.013)
0.64
(0.53)
3.5
(0.071)
2.0
(0.20)
0.50
(0.61)
0.23
(0.63)
3.5
(0.039)
0.16
(0.85)
5.4
(0.026)
1.0
(0.37)
0.02
(0.98)
Means
Xeric > W.
For. Mt.
Mng > Rec
= Water
n.s.
n.s.
n.s.
n.s.
n.s.
Mng >
Water = Rec
n.s.
W. For. Mt >
Xeric
n.s.
n.s.
Nutrient Ecoregion: Xeric=Xeric West, W. For. Mt.
=  Western   Forested   Mountains.  Use:
Mng=Hydrologic  Management,  Rec=Recreation,
Water=Water Supply, n.s. = Not significant
In the WFM total P is highly correlated with
all cations but not correlated with Sulfate
and weakly correlated with Chloride (Table
6).  In the XW total P is weakly correlated
with Potassium and Sodium not correlated
with any  other ions.  Total N is weakly
correlated with  Sodium,  Calcium  and
Magnesium in the WFM but not correlated
with  any  ion  in   the  Xeric  West.
Furthermore,  total  P and total N are not
correlated  in the  WFM  but  loosely
correlated in the Xeric West.

The strong correlation among cations and
total P concentrations in  the WFM are an
indication  that there is a natural total  P
gradient  related   to   watershed
geology/soils.  In the XW, since total P is
related  to  the monovalent group but not
the divalent group, the total P source could
be a specific  geologic source that is only
related to Potassium or Sodium or it could
be associated with land use, e.g. fertilizers
that are high  in the monovalent cations.
This latter interpretation  is supported  by
the fact that  total P  and total  N were
correlated in the Xeric West but not in the
Western Forested Mountains,  i.e. due to
exogenous  inputs of  both nutrients  into
these watersheds.
                                               Table 5.  Spearman correlation  coefficient
                                               matrix  for nutrient related variables  in 24
                                               reservoirs in the Western Forested Mountain
                                               ecoregion and 24 reservoirs in the Xeric West
                                               ecoregion. Correlations significant at p<0.05
                                               are shown in bold, correlations significant at
                                               p<0.001 are  starred (*).
WESTERN FORESTED MOUNTAINS
Vari
TP
TN
Chl-
Sec
Turb
Ext.
Abs
Abs
rp
—
0.6
0.5
-0.5
0.0
0.5
0.2
0.2
FN

—
0.4
-0.5
0.3
0.6
0.6
0.6
;hl-a


—
-
1.
0.7
0.
0.
5ec



....
0.0
-
-0.5
-0.4
furb




—
0.7
0.5
0.6
Ext.






0.7
0.7
tos3






—
0.9
XERIC WEST
Vari
TP
TN
Chl-
Secc
Turb
Ext.
Abs3
Abs4
TP
....
0.68*
0.60
-0.86*
0.80*
0.84*
0.73*
0.68*
TN

—
0.55
-
0.49
0.56
0.45
0.35
;m-a


—
-0.55
0.48
0.50
0.23
0.16
Seech




-0.88*
-0.89*
-0.80*
-0.77*
Turb




—
0.96
0.83
0.87
Ext.






0.83
0.85
i\bs3






* —
* 0.96

)OC
—
0.9




0.8
0.7

DOC






....
....
TP = Total Phosphorus, TN = Total Nitrogen, Chl-a
= Chlorophyll-a, Secchi = Secchi Depth, Turb =
Turbidity, Ext. = Light Extinction Coefficient, Abs325
= Absorbence at 325 nm, Abs440 = Absorbence at
440 nm
                                      Page 9 of 21

-------
 f
 °
 1
 ffi
 ID
                   ••• Western Forested Mountains
                   """Xeric West
      1           10          BO         1000
              Total Phosphorus  (ug/L)

Figure 13.  Relationship between Secchi depth
and  total  phosphorus.   Water transparency
decreases with increasing total phosphorus.
                     * Western Forested Mountains
                      Xeric West
                 Potassium (ueq/L)
  Figure 14.  Scatter plot of total phosphorus
  versus potassium concentration in  Pacific
  Northwest reservoirs.  Total P has a higher
  correlation with Potassium in the WFM than
  the XW.
 Table   6.     Spearman  rank  correlation
 coefficients among base cations, anions, total
 phosphorus (TP) and total nitrogen (TN) in 24
 Xeric West and 24 Western Forested Mountain
 reservoirs.  Bold values are significant at p <
 0.0001.  Values with significance below 0.1 are
 shown as n.s.

K
Na
Ca
Mg
S04
CL
TP
Western Forested Mountains
K
A/a
Ca
Mg
S04
CL
TP
TN
-
0.90
0.72
0.77
0.56
0.55
0.84
ns

-
0.73
0.86
0.59
0.64
0.82
0.53


-
0.86
0.79
0.58
0.66
0.57



-
0.65
0.71
0.73
0.66




-
0.67
ns
ns






0.58
ns






-
ns
Xeric West
K
A/a
Ca
Mg
S04
CL
TP
TN
-
0.75
ns
ns
ns
0.80
0.61
ns

-
0.56
0.53
0.63
0.89
0.48
ns


-
0.91
0.76
ns
ns
ns




0.66
ns
ns
ns




-
0.65
ns
ns





-
ns
ns






-
0.63
 As  the  nutrient  variables  are  highly
 correlated  with  each  other,   it's  not
 necessary to consider how other variables
 relate to all of them. TP and chlorophyll-a
 were  selected   as   nutrient  indicator
 variables and analyzed to see how other
 reservoir variables related to them (Table
 6). In examining the data it  became
   -J IWtl:


   I
                                                                      • We Seen Forested Mountains
                                                                      Xeric West
                                                                  Potassium (ueq/L)
                                                   Figure  15.    Scatter   plot   of  total
                                                   phosphorus  versus   potassium
                                                   concentration  in   Pacific  Northwest
                                                   reservoirs.  Calcium and  Potassium are
                                                   more highly correlated in the WFM than in
                                                   the XW.
apparent that the two ecoregions behaved
differently  so the  results are  presented
separately for each ecoregion. In the Xeric
West, no variables were highly correlated
(p < 0.001) with either TP or chlorophyll-a.
TP was significantly related  (p  < .05) to
potassium, dissolved oxygen,  shoreline
disturbance,  and  maximum   reservoir
depth.     In   the  Western   Forested
Mountains,   many   variables  were
significantly  related  to   both  TP  and
chlorophyll-a.   Indicators of water ionic
strength  (base   cations,   alkalinity,
conductivity  were  highly  correlated with
both nutrient variables. Reservoir surface
area, depth below full pool, sulfate, redox
potential, and shoreline ground cover w
                                       Page 10 of 21

-------
were not related to either nutrient variable
in either ecoregion (Table 6).

Potassium was most strongly related to TP
in both ecoregions (Fig.  15). The log-log
plots   have  similar   slopes   in   both
ecoregions though the relationship is less
variable   in   the  Western   Forested
Mountains. On the other hand, calcium is
strongly  related to TP in the Western
Forested Mountains but is not related to
TP  in the Xeric West.   Plots of calcium
versus potassium show that the two base
cations are strongly related in the Western
Forested Mountains (r=0.74) but unrelated
\in the Xeric West (r=0.28;  Fig. 15).  In the
Western Forested Mountains, all four base
cations are correlated with  each other with
an r>0.75.

In the WFM DOC was strongly correlated
with absorbence at both 325 and 440 nm
(Table 5).    In contrast  there was no
correlation  among these variables  in the
Xeric West. Considering that median DOC
was five times higher in the XWthan in the
WFM,  the  absence  of  a  correlation
between  these  parameters in the XW is
problematic. This may have to do with the
high turbidities in the  XW,  which were
highly correlated with absorbence and thus
masked any relationship to DOC.

Shoreline  structure  indices  were  not
strongly  correlated   with  any   water
chemistry  variable  in  the   data   set.
Although there appeared  to be a  rough
negative   correlation   between    total
shoreline vegetation ground cover and
total P concentrations,  correlations were
insignificant.  This result  has  not been
controlled for  season, i.e. reservoir
draw down.
 Cluster Analysis

       One of  the  objectives  of  this
research was to develop  classes  or  a
typology for Pacific Northwest reservoirs to
aid in developing nutrient criteria.  Two
approaches were tried; visual inspection of
anion/cation chemical composition, and an
overall cluster analysis.  To investigate
anion/cation   composition,   trilinear
diagrams were used that show the relative
%  composition  of  various ions  on  a
triangular plot.  For both anions and base
cations, there were no apparent clusters of
lake types and this doesn't appear to be a
useful approach for developing a reservoir
typology

To develop clusters from the reservoir data
for  each Nutrient  Ecoregion, a principal
components analysis (PCA) was used to
identify the major components in the data.
These  components or factor scores  for
each site were then clustered.  Results of
the PCA  are very dependent on what
variables are included.  It's also desirable
to have non-redundant variables.  It was
also desirable to select a set of variables
that  coyer   a   range  of   reservoir
characteristics that are not correlated with
one another.  Five variables were chosen
for  the PCA analysis for the Xeric West
(Table  7). Potassium and alkalinity were
selected to represent ionic strength.  Both
were  used  as they represent different
processes in the Xeric West and were not
correlated with each other. Water clarity
was represented by three separate factors:
Abs325, turbidity and chlorophyll.

The resulting PCA had 4 factors or axes
that each explained more than 10% of the
variance and these 4 site factor  scores
were  used  for  the  subsequent  cluster
analysis. All together, these 4 components
explained 98% of the variance in the 5
variables. In looking at the correlation of
the original 5 variables to the axis scores,
axis 1   can  be  considered  an  ionic
strength/transparency  axis,  axis  2  is
related to alkalinity,  axis 3  is related to
chlorophyll, and  axis 4  is a turbidity axis.
We included chlorophyll as a variable in an
attempt to separate algal  and nonalgal
turbidity.

From the cluster analysis of the four axes
scores, four clusters were identified from
the resulting dendrogram (Fig.  16).  Box
and whisker  plots  showing  how  the
clusters differ on reservoir variables are
useful in assigning cluster attributes .

Cluster 1 -   Low ionic  strength,  high
transparency,  low  absorbence,   low
turbidity.

Cluster 2 -    Moderate  ionic strength,
moderate   transparency,  moderate
                                   Page 11 of 21

-------
absorbence, moderate turbidity due either
to algae or particulate matter.

Cluster 3 -  Moderate ionic strength, low
transparency due to non-algal turbidity.

Cluster 4 -  High ionic strength, very high
absorbence, very low transparency due to
algal and/or nonalgal turbidity.

We found no components for reservoirs in
the WFM.  The magnitude of total P in
these reservoirs ranged from 4 to 61 ug/L
(Appendix 1).  Base cations and  total P
were   all  correlated  in  this   Nutrient
Ecoregion, suggesting that a common

mechanism explains the concentrations of
total P in these reservoirs (Table 6).  The
highest total P concentrations were found
in  reservoirs   (e.g.   Lake  Simtustus,
Toketee, Willow Creek) that have very
four PCA axes, and the amount of variance
explained by each PCA axis.  Cumulatively, the
four axes explained 98% of the variance in the
5 variables.
Variable
Potassium
Alkalinity
Abs325
Turbidity
Chlorophyll
% variance
Axis
1
0.52
0.03
0.50
0.55
0.41
58%
Axis
2
0.31
0.89
-0.28
-0.15
0.11
24%
Axis
3
0.18
0.20
0.45
0.05
-0.85
14%
Axis
4
-0.32
0.15
-0.41
0.81
-0.20
3%
high  natural  sources   of   phosphorus.
Excluding  these  reservoirs,  the  other
reservoirs in the WFM all had total P
Table  7.    Variables  used   in  principal
   Reservoir Nome

 Alexander Reservoir
 Ch icknhoniiny
 Barry Reservoir
 Mud  Late
 Grastne re Reservoir
 Upper Rock Creek Reservoi
 Pridov Reservoir
 Ben  Ross Reservoir
 Snnd  Creek Reservoir
 Bryant Ml. Reservoir
 Block Canyon  Reservoir
 Monn  Creek Reservoir
 Daniels Reservoir
 Scoo tenev Reservoir
 Nes ton Creek  Reservoir
 Stone Reservoir
 Blue  Creek Reservoir
 Kern  Reservoir
 Little Blue  Creek Rese rvo
 Trail Storage Pond
 Bray  Lake
 Cedar Creek  Reservoir
 JC Boyle Reservoir
 Thief Valley  Reservoir
 (No r risp r i ngs  Reservoir
 Muddy Creek  Reservoir
 Paddock Valley Reservoir
                                                concentrations less that 40 ug/L.  If there
                               2
                                            6
                                                        10
                                                              12
                                                                    H
                                                                          16
                                                                                 18
                                               FI ex i bIe-B eta Distance
 Figure 16. Dendrogram of the cluster analysis of the four PCA axes for the 27 reservoirs in the
 Xeric West showing the 4 cluster used in subsequent analysis.
components analysis, their correlation with the
are significant differences among these
                                      Page 12 of 21

-------
reservoirs with respect to landscape, we
weren't able to discern them with  in our
data set.
Reservoir Categories

Developing a  typology or classes is an
essential first step for developing nutrient
criteria for reservoirs.   Based  on water
quality data  for each  of the  clusters
described in section 5,  we  propose a
reservoir typology for the Xeric West with
three categories.

Category  1 -  Low ionic strength,  high
transparency,   low  absorbence,   low
turbidity, low chlorophyll.

Category  2  - Moderate ionic strength,
moderate  transparency,  moderate
absorbence, moderate turbidity due either
to algae or particulate matter.

Category 3 - High ionic strength, very high
absorbence, very low transparency due to
algal and/or nonalgal turbidity.
Categories  1  and  2 are the  same  as
clusters 3 and 4, respectively. Category 3
is a combination of clusters 1 and 2 (Fig.
21).

Cutoff  criteria   for  these  types  were
selected to preserve the original clustering
as much as possible.  Thus, this typology
preserves the segregation observed in the
original  clustering  but has  very simple
parameters used  to define  the types.
Reservoir typing could be done with a
simple field visit and  measurements of
conductivity,   turbidity,  Secchi  and
chlorophyll. These measurements can all
be made fairly inexpensively.
Characteristics of Reservoir Categories in
the Xeric West

There  was  no  apparent  relationship
between   reservoir   category  and
geographic location (Fig. 17). However, all
but two of the reservoirs in the Xeric West
(Scooteney in Columbia Basin and Thief
Valley in Blue Mountains) were in the Level
III  Northern Basin and  Range Ecoregion.
Therefore  if  additional  reservoirs  are
sampled in the Columbia  Plateau  this
absence of apparent  geographic effect
may not hold. Conversely, classification of
reservoirs  by  Nutrient  Ecoregion  may
prove to be effective as an initial method to
stratify reservoirs by location for purposes
of developing nutrient criteria.

The  three  types  of  reservoir  thus
determined show differences in terms of
total P, total N Secchi but not chlorophyll
concentrations (Fig. 18-21).

For natural water bodies, once a typology
has  been  established,  nutrient  criteria
would  be established  by looking at the
distribution of nutrient levels in a set of
undisturbed of reference systems  of a
given category.  That would provide a
scientifically defensible  picture of what
natural nutrient concentrations ought to be
for  setting criteria.     For reservoirs,
however, there is no way to have natural
reference sites.  Reservoirs by their very
nature are artificial human built systems so
the reference site concept doesn't apply.
Thus, setting nutrient criteria for reservoirs
will be especially problematic.

One method to define nutrient  criteria is
based upon  a percentile distribution of
scores of all systems in each category
(EPA, 2000). Recent data suggest that in
the Pacific  Northwest the  median-75%
interquartile  is an appropriate  range for
total P values in  lakes and reservoirs
(Vaga  and Herlihy 2004).   Percentile
scores and medians for nutrient  criteria
variables for each  reservoir category are
listed in  Table 8.   Using the percentile
approach,  total P  nutrient  criteria  for
Category 1  reservoirs would be <33 ug/L ,
<238  ug/L  in Category 2 reservoirs  and
<686   ug/L in  category  3 reservoirs.
Similarly, criteria for Secchi depth would
be 2.1,0.70 and 0.15 meters, respectively.
It is evident that the high variances of total
N   and   chlorophyll   preclude   any
determination   regarding   meaningful
assignments of criteria using this method.
Whether a larger sample  size would
provide more  precise  determinations of
these   parameters  among   reservoir
categories  is  problematic.   Category 1
reservoirs are clearly relatively oligotrophic
compared  with the  other  categories.
                                   Page 13 of 21

-------
Chlorophyll values in Category 2 reservoirs
indicate a variability  based upon season
but water transparency is sufficient to allow
for   development  of  significant   algal
biomass. Category 3 reservoirs evidently
have  the   potential   for  significant
accumulation of algal  biomass  but  that
potential can  be limited  by nonalgal
turbidity.  In  this study we attempted to
differentiate among the effects of color,
absorbence,  algal  turbidity and nonalgal
turbidity. It is evident that in Category 3
reservoirs we were unable to accomplish
this differentiation due to the fact that we
did  not have data on  Total  Suspended
Solids.  In future studies it is imperative to
collect Total  Suspended Solids data to
explicitly differentiate the effects between
algal  and  nonalgal  effects  on  water
transparency.  At all events, the results
presented here provide a useful guide to
future  investigations  regarding reservoir
nutrient dynamics in the Pacific Northwest.
Discussion

The magnitude of the median values for
total  P, total N,  Secchi  and chlorophyll
compare favorably with published values
(Table 9).  Total P values were roughly two
times those reported earlier (Vaga  and
Herlihy 2004).  This is reasonable, since
Vaga & Herlihy included a preponderance
of natural lakes versus reservoirs and thus
their values would be expected to be lower
than   for  reservoirs  alone.   A  similar
relationship  exists   for  Secchi   and
chlorophyll, i.e. the reservoirs appearto be
more  eutrophic.   Interestingly total  N
values reported here appearto be one-half
of the  lake+reservoir  study cited  above.
These data suggest that reservoirs are
relatively nitrogen poor as compared  with
natural lakes but the reason for this  is
unclear. However, these values for total P
and total N suggest that  natural lakes  in
the WFM and XW have N:P ratios of 38.7
and   13.2,   respectively.  In  contrast
reservoirs in the WFM and XWhave ratios
of 9.7  and  3.9, respectively.   Thus
irrespective of Nutrient Ecoregion, nutrient
criteria for reservoirs  may  have to take
total N into account.

Reservoirs typically are relatively nutrient
rich during the first years after construction
due  to leaching of nutrients  from the
recently flooded  soils  (Thornton et al.,
1990). Reservoir age did not appear to be
a factor in this study.  Reservoirs in the
WFM were constructed between 1910 and
1989 (average age 56 years) and those in
the XW between 1895 and 1970 (average
age   64   years).     In  neither  Nutrient
Ecoregion were any water quality/nutrient
parameters correlated with reservoir age,
suggesting  that  any   effects  due   to
construction are no longer factors in the
nutrient dynamics of these reservoirs.

Reservoirs tend to  be long and narrow
resulting   in  a  gradients  in  physical,
chemical and biological parameters from
the  upstream end   to the  dam.   This
gradient can be divided into three general
zones,  e.g.   riverine,   transition   and
lacustrine  (Thornton,   et  al.,  1990).
Typically  the transition zone exhibit the
highest productivity  and  the lacustrine
portion of reservoirs  exhibit relatively low
productivity, since nutrients are removed
by algal settling in the transition zone.  In
this  study we  limited sampling to the
lacustrine areas of the reservoirs and more
specifically to the  forebay.   Thus our
estimates  of  nutrient  concentrations,
Secchi   and  chlorophyll   are   only
representative of the most nutrient poor
region of  the reservoirs.  Other zones  in
these reservoirs would in all probability
exhibit different nutrient dynamics.

In  addition to sampling location, season
also plays an important part in the be be
expected  to  have higher nutrients  and
chlorophyll  concentrations,   nutrient
dynamics   of reservoirs.   As  in  lakes,
reservoirs often exhibit a spring bloom  of
phytoplankton followed by a clear water
phase in  midsummer  and a secondary
bloom in the fall (Wetzel 2001).  In seven
WFM reservoirs total P  was <  10 ug/L
(Appendix 1,2).  These  low levels are
probably indicative of nutrient conditions in
these reservoirs but without further study
we cannot rule out the possibility of higher
concentrations  early in the year.   The
same holds for the six reservoirs in the XW
with  concentrations < 30 ug/L.

In   the Xeric  West  high  chlorophyll
concentrations were  not always observed
                                   Page 14 of 21

-------
 in   the  presence  of  high   total  P
 concentrations  (Appendix  1,  2).   For
 example,   in   Category   3  reservoirs
 Chickahominy and Alexander had over 600
 ug/L of total P and chlorophyll above 400
 ug/L (Fig.  11,  Appendix 2).   However,
 several reservoirs had over  100 ug/L  of
 total P (e.g. Upper Rock,  Priday, Barry,
 Grasmere, Warmsprings) but less than 10
 ug/L of chlorophyll. In several cases the
 presence  of  nonalgal  turbidity  (Upper
 Rock)  or absorbence due  to DOC (e.g.
 Barry,   Priday,  Grasmere)  may  have
 prevented  development of  high  algal
 biomass.  However, we cannot discount
 the fact that we may have missed an algal
 bloom  in these reservoirs.
                                         Reservoir
                                         Categories
                                          •  1
                                          •  2
                                          A  3
          Figure 17.  Location  of the three  reservoir
          categories in the  Xeric West.  There  is no
          apparent relationship between category and
          geographic location.
Table 8. Median and 75th percentile (Q3) values
for  nutrient  criteria  variables  in  the  three
proposed nutrient criteria reservoir categories, n
= number of reservoirs in category.
c
1
2
3
n
9
11
7
TP
Med(Q3)
25(33)
137(238)
344( 686)
TN
Med(Q3)
258(399)
483(578)
385(1930)
Chl-a
Med(Q3)
2.1(4.6)
27.9(76.9)
7.9( 460.6)
Secchi
Med(Q3)
2.10(4.4)
0.70(0.9)
0.15(0.3)
Table 9. Comparison of median (± 1 se) values
for total P, total N, Secchi and chlorophyll
observed in this study with those calculated from
Vaga and Herlihy (2004) for lakes and reservoirs
in corresponding Nutrient Ecoregions.
Variable/
Source
TP (ug/L)
This study
Vaga/Herl
TN (ug/L)
This study
Vaga/Herl
Secchi (m)
This study
Vsga/Herl
Chla (ug/L)
This study
Vaga/Herl
WFM XW
Mean

23.0
11.3

213.2
437.5

3.4
4.5

4.6
2.0
se

13.6
2.4

54.3
63.6

0.5
0.2

3.1
0.5
Mean

109.0
62.5

430.9
825.0

0.7
1.6

7.5
4.0
se

43.7
14.4

46.3
156.4

0.3
0.4

27.8
1.8
The differences observed in reservoir total
P  and Secchi depth in  the  two different
Nutrient Ecoregions  reflects differences in
landscape. Land use in the WFM is roughly
          0,12 and 80% agricultural, range land and
          forest, respectively.   In the Great Basin
          and Range Ecoregion, where all but one of
          the XW reservoirs is  located, land use is
          agricultural =  0,  Range Land  = 83 and
          Forest = 10%.  These differences as well
          as differences in geology and weathering
          probably  account  for   the   observed
          differences in total P and  Secchi depth.
          The high correlation of Potassium and total
          P in the WFM (Fig. 14, Table 6) suggests
          a single relatively homogenous  source for
          phosphorus  in  these  reservoirs,  with
          relatively low effect of land use differences.
          The much weaker correlation observed in
          the  XW   is  likely  due   to   a  more
          heterogeneous  source.    Whether  this
          heterogeneity is  due to  land  use  e.g.
          fertilizers or geophysical processes,  or a
          combination of both, is unknown.
3- :
1
100 =
« -
= :
o
— 10-
a E
1 !
i-
i *
*?


6




                                                              Reservoir Category
          Figure 18. Distribution of total P values in the
          three reservoir categories in the Xeric West.
          There was a clear difference among  the
          categories as defined  by cluster analysis.
          Boxes  show  the  interquartile range  and
          median,   whiskers   the  minimum/maximum
Page15of^lues-

-------
 IB
 ."5
            1          2          3
                Reservoir Category

 Figure 19.  Distribution of total N values in the
 three reservoir categories in the Xeric  West.
 Distributions of total N overlapped more than for
 total P among categories.  Boxes show the
 interquartile range and median,  whiskers the
 minimum/maximum values.


Conclusions

This   study   suggests   that  Nutrient
Ecoregions  are   useful  as   an   initial
stratification variable for developing numeric
nutrient  criteria for reservoirs.   There
appears to be several lines of fruitful future
investigation to further the development of
nutrient criteria:

1) The minimum size of reservoir should be
determined that will be subject to criteria.
Small reservoirs,  particularly in the Xeric
West, are most likely to be subject to  draw
down during the summer to the level that
makes criteria development unnecessary.
2)  Conversely,  the  maximum  size  of
reservoir should be determined where the
relationship between measurable landscape
parameters and  nutrient concentrations in
the reservoir can be empirically determined.
For example,  Columbia  River reservoirs
likely have no need for development of such
criteria.

3) Greater resolution of nutrient dynamics in
reservoirs   in   the  Western   Forested
Mountains   should  be   undertaken   to
ascertain to what  extent these reservoirs
can be  considered as  a single category.
Such a determination may be made simply
by accounting for local anomalies, e.g. high
total P groundwater. It is also possible that
reservoirs  in  the  WFM  are  all  similar
enough to warrant placement into a single
category.

4) Greater resolution of nutrient dynamics
in reservoirs in the Xeric  West should be
undertaken to determine whether the three
categories suggested  in this  study are
sufficient to account for  all  reservoirs  in
this Ecoregion.  In this study  there was
only one reservoir in the Columbia Plateau
(Level III Ecoregion).  Vaga and  Herlihy
(2004) found that lakes/reservoirs had a
median  and   75%percentile  total   P
concentrations in the Columbia Plateau of
30 and 71  ug/L. In the Northern Basin and
Range (Level III Ecoregion) similar values
were reported  and 90 and 204 ug/L. Thus
it may not be appropriate to combine these
two   Level   III  Ecoregions  into   one
population.  In addition TSS rather than
chlorophyll  should  be   used   in  the
classification process.

5) Downstream effects. The  obvious high
concentrations of total P in the Xeric West
reservoirs  should   be   evaluated  with
respect to tailwater effects on downstream
habitats.

6) The value of N:P ratios, e.g. 3.9 N:P in
the Xeric West, should be evaluated with
respect to the need for  nitrogen  criteria
and  the   concomitant   effects   on
downstream systems.

7) Shoreline characteristics seem to be
of less importance  in determining nutrient
concentrations than  overall watershed
characteristics. We surmise that at least
for reservoirs that experience draw down,
the effects of season are more important
than shoreline characteristics at full pool.
Acknowledgments.  We would like to thank Peter
Leinenbach and George Gibson for reviewing an
earlier draft of this report.
                                    Page 16 of 21

-------
 i
                Reservoir Category

Figure 20.  Distribution of Secchi depth in the
three reservoir categories  in the Xeric West.
There  was  a clear difference among the
categories  as defined  by cluster analysis.
Boxes show  the  interquartile  range  and
median,  whiskers  the minimum/maximum
values.
p
§> 100,0 =
f
o.
o 10.0=
5 :
1,0-


6
L

r




                                                                    Reservoir Category
                                                  Figure 21.  Distribution of chlorophyll values in
                                                  the three reservoir categories in the Xeric West.
                                                  The  low  variability in Category 1 reservoirs is
                                                  due  to relatively  low nutrients.   Category 2
                                                  reservoirs exhibit a higher median and variance
                                                  due  to  seasonal  effects,  i.e.  low nonalgal
                                                  turbidity  but higher  total  P.   Category 3
                                                  reservoirs have low median but high potential
                                                  for algal growth. Boxes show the interquartile
                                                  range   and  median,   whiskers   the
                                                  minimum/maximum values.
                                       Page 17 of 21

-------
References

Baker,  J.R., D.V.  Peck, and D.W. Sutton (eds.).  1997.  Environmental Monitoring and
   Assessment Program Surface Waters: Field Operations Manual for Lakes.  EPA/620/R-
   97/001. U.S. Environmental Protection Agency, Washington DC.

Cooke, G.D. and  R.H. Kennedy.  1989.   Water Quality Management for Reservoirs and
   Tailwaters.  In:  Reservoir Water Water Quality and Management Techniques. Tech. Rep.
   E-89. U.S.  Army Corps of Engineers, Washington, D.C.

Emmons, E.E., Jennings, M.J., and C. Edwards. 1999. An alternative classification method
   for northern Wsconsin  lakes. Can. J. Fish. Aquat. Sci. 56: 661-669.

Environmental  Protection Agency.  2000. Nutrient Criteria Technical Guidance Manual Lakes
       and  Reservoirs,   EPA-822-BOO-001,  U.S.  Environmental  Protection  Agency:
       Washington, DC.

Kennedy,  R.H.  1999.   Reservoir  design  and operation:  limnological  implications and
   management opportunities.  In:  Tundisi, J.G.  and  Straskraba, M.,  eds.   Theoretical
   reservoir ecology and its applications.  Brazil and the Netherlands: International Institute
   of Ecology,  Brazilian Academy of Sciences, and Backhuys Publ.. pp. 1-28.

Kennedy, R.H.   2000.  Nutrient criteria: considerations for Corps of Engineers Reservoirs.
   Water  Quality  Technical Note.  U.S.  Army Engineer Waterways Experiment Station,
   Vicksburg,  MS.

Omernik, J.M.  1987.  Ecoregions of the conterminous United States.  Annals Assoc. Amer.
   Geogr. 77(1):118-125.

Perkins, R.G. and  G.J Underwood. 1999.  Gradients of Chlorophyll A and Water Chemistry
   Along an Eutrophic Reservoir with Determination of the Limiting Nutrient by In Situ Nutrient
   Addition. Wat. Res. Vol. 34 No.3, 713-724.

Ryder,  R.A.  1978. Ecological heterogeneity between north-temperate reservoirs and glacial
   lake systems due to differing succession rates and cultural uses. Verh. Internat. Verein.
   Limnol. 20:  1568-1574.

Straskraba, M.  J.G. Tundisi and A. Duncan  1993.  Comparative Reservoir  Limnology and
   Water Quality Mangement.

Thornton, K.W., B.L. Kimmerland F.E. Payne. Reservoir Limnology: Ecological Perspectives.
   Wley-lnterscience Publication, New York. 246pp.

USEPA.  1996. National Nutrient Assessment Workshop Proceedings.   USEPA Office of
   Water. EPA-822-96-004.

USEPA. 1998. National strategy  for the development of regional nutrient criteria.  USEPA
   Office of Water. EPA-8221-R-98-002.


USEPA. 2000. Nutrient Criteria Technical Guidance Manual: Lakes and Reservoirs. USEPA
   Office of Water. EPA-822-BOO-001.

Vaga, R.M. and AT. Herlihy. 2004. A GIS inventory Of Pacific Northwest Lakds/Reservoirs
   and an analysis of historical nutrient and water quality data. EPA 910-R-04-009.
                                  Page 18 of 21

-------
Vaga, R.M. 2005. Analysis of hydrologic and Water Quality Data for American Falls Reservoir,
   ID. EPA910-R-05-001. EPA 910-R-05-002.

Vaga, R.M. and AT. Herlihy.  2005.  A Classification of lakes in the Coast Range Ecoregion
   with respect to nutrient processing. EPA 910-R-06-003.

Willen, E., Hajdu, S., and Pejler, Y. 1990.  Summer phytoplankton in 73 nutrient poor Swedish
   lakes: classification, ordination, and choice of long term monitoring objects. Limnologica,
   20: 217-228.

Wetzel, R.G.  1990. Reservoir ecosystems: conclusions and speculations, pp. 227-238. In:
   K.W. Thornton, B.L. Kimmel, and F.E. Payne (eds.),  Reservoir Limnology: Ecological
   Perspectives. John Wiley and Sons,  Inc., New York,  NY.
                                   Page 19 of 21

-------
 Appendix 1. Nutrient and site data for Western Forested Mountain reservoirs.
NIDID
ID00337
WA00263
WA00264
WA00291
WA00170
OR00016
OR00748
OR10020
WA00183
OR00548
OR83030
WA00223
WA00021
WA00152
OR00550
OR00098
OR00579
ID00115
WA00153
OR00554
OR00212
Lake Name
Brundage
Bumping Lake
Clear Lake
Conconully
Diablo Lake
Fern Ridge
Galesville
Henry Hagg
Judy
Lake Simtustus
Laurence Lake
Leader Lake
Long Lake
Mayfield
N. Fork Clackamas
Ochoco
Prineville
Sage Hen
Skookumchuck
Toketee
Willow Creek
Latitud
45.0418
46.8733
46.6333
48.5583
48.7133
44.1150
42.8492
45.4917
48.4717
44.6933
45.4583
48.3617
47.8367
46.5033
45.2417
44.2983
44.1133
44.3255
46.7850
43.2647
42.4800
Longitud
116.1306
121.3000
121.2667
119.7450
121.1300
123.2917
123.1778
123.2139
122.1883
121.2300
121.6583
119.6967
117.8383
122.5900
122.2817
120.7250
120.7800
116.1941
122.7167
122.4194
122.4433
Chl-
2.4
0.1
1.1
7.5
1.8
36.9
5.7
6.0
1.3
34.8
7.5
9.8
3.5
4.6
0.1
2.1
21.2
1.4
2.8
0.4
18.8
Secc
4.2
7.8
6.9
3.7
7.3
0.3
4.1
3.3
3.0
1.5
3.4
3.3
5.2
3.4
4.6
3.2
3.8
4.7
4.2

0.7
Max
8.7
11.0
17.4
6.8
48.5
6.5
31.7
11.2
6.8
42.0
12.5
8.0
16.7
27.5
33.2
23.6
34.0
9.4
29.5
7.6
7.9
TP
10
8
21
26
4
36
10
17
9
61
17
37
110
9
14
30
39
9.5
440
58
36
TN
82


279
55
182
116
93
192
164
61
403
864
56

227
177
99
173
58
167
NIDID = National Inventory of Dams ID code
Chl-a = Chlorophyll-a  (ug/L)
Secchi = Secchi Depth (m)
Max Depth = Maximum Reservoir depth at time of sampling (m)
TP = Total Phosphorus (ug/L)
TN = Total Nitrogen (ug/L)
                                      Page 20 of 21

-------
Appendix 2. Nutrient and site data forXeric ecoregion reservoirs.
NIDID
ID00060
OR00108
ID00136
ID00282
ID00194
ID00042
OR00344
ID00045
OR00228
ID00006
ID00190
OR00559
OR00181
ID00193
ID00285
OR00569
OR00377
ID00250
OR00369
ID00010
WA00566
ID00007
OR00592
ID00239
OR00157
OR00082
ID00074
Reservoir Name
Alexander
Barry
Ben Ross
Black Canyon
Blue Creek
Bray Lake
Bryant Mt.
Cedar Creek
Chickahominy
Daniels
Grasmere
JC Boyle
Kern
Little Blue Creek
Mann Creek
Mud Lake
Muddy Creek
Paddock Valley
Friday
Sand Creek
Scooteney
Stone
Thief Valley
Trail Storage Pond
Upper Rock Creek
Warmsprings
Weston Creek
Latitude
42.6446
42.0950
44.5235
43.9302
42.3097
43.0344
42.1050
42.2237
43.5450
42.3455
42.3555
42.1231
42.9183
42.2917
44.3918
42.2117
42.1950
44.1983
42.3400
44.2025
46.6667
42.0680
45.0150
43.0553
42.6867
43.5850
42.1302
Longitud
1 1 1 .6960
119.5317
116.4446
116.4357
116.1807
114.8760
121.3283
114.8786
119.6133
112.4424
115.9039
122.0439
118.7817
116.1203
116.8928
119.7150
120.5200
116.5980
119.9000
111.6144
119.0333
112.6942
117.7783
115.3196
119.3050
118.2083
112.1260
Chl-
519.
2.9
2.2
2.6
2.2
76.8
10.2
64.0
460.
4.8
4.4
58.3
27.9
18.8
7.5
29.3
176.
222.
4.9
0.9
12.7
4.6
14.8
5.5
7.9
4.4
2.8
Secc
0.2
0.2
2.5
4.2
0.9
0.7
0.8
0.7
0.2
4.5
0.3
0.2
0.7
0.8
1.7
0.2
0.5
0.3
0.4

1.1
1.3
1.1
0.6
0.2
1.4
4.7
Max
15.7
4.0
12.0
21.0
5.8
2.8
1.1
7.8
3.6
9.8
5.0
8.9
5.8
6.3
14.4
2.7
6.4
4.5
3.8
2.9
11.0
6.7
9.1
6.0
3.7
11.3
4.7
TP
686
344
22
22
108
137
23
151
887
42
201
252
73
109
26.5
456
325
110
201
25
46
23
225
104
340
238
33
TN
243
385
118
93
509
371
598
468
193
258
367
123
274
292
123
453
579
929
365
268
792
151
483
514
289
251
399
NIDID = National Inventory of Dams ID code
Chl-a = Chlorophyll-a  (ug/L)
Secchi = Secchi Depth (m)
Max Depth = Maximum Reservoir depth at time of sampling (m)
TP = Total Phosphorus (ug/L)
TN = Total Nitrogen (ug/L)
                                      Page 21 of 21

-------