&EPA
United States
Environmental Protection
Agency
Environmental Research
Laboratory
Corvallis OR 97330
EPA-600'3-78-087
September 1978
Research and Development
Procedures for
Quantitative Ecological
Assessments in
Intertidal Environments
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7, Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the ECOLOGICAL RESEARCH series. This series
describes research on the effects of pollution on humans, plant and animal spe-
cies, and materials. Problems are assessed for their long- and short-term influ-
ences. Investigations include formation, transport, and pathway studies to deter-
mine the fate of pollutants and their effects. This work provides the technical basis
for setting standards to minimize undesirable changes in living organisms in the
aquatic, terrestrial, and atmospheric environments.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/3-78-087
September 1978
PROCEDURES FOR QUANTITATIVE
ECOLOGICAL ASSESSMENTS IN INTERTIDAL ENVIRONMENTS
by
J. J. Conor and P. F. Kemp
School of Oceanography and Marine Science Center
Oregon State University, Corvallis, Oregon 97331
Grant No. R805018 01
Project Officer
Richard C. Swartz
Newport Field Station
Marine and Freshwater Ecology Branch
Corvallis Environmental Research Laboratory
Newport, Oregon 97365
CORVALLIS ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U. S. ENVIRONMENTAL PROTECTION AGENCY
CORVALLIS, OREGON 97330
-------
DISCLAIMER
This report has been reviewed by the Corvallis Environmental Research
Laboratory, U.S. Environmental Protection Agency and approved for publication.
Approval does not signify that the contents necessarily reflect the views and
policies of the U.S. Environmental Protection Agency, nor does mention of trade
names or commercial products constitute endorsement or recommendation for use.
11
-------
FOREWORD
Effective regulatory and enforcement actions by the Environmental Protection
Agency would be virtually impossible without sound scientific data on pollutants
and their impact on environmental stability and human health. Responsibility for
building this data base has been assigned to EPA's Office of Research and Develop-
ment and its 15 major field installations, one of which is the Corvallis Environ-
mental Research Laboratory (CERL).
The primary mission of the Corvallis Laboratory is research on the effects of
environmental pollutants on terrestrial, freshwater and marine ecosystems; the be-
havior, effects and control of pollutants in lake systems; and the development of
predictive models on the movement of pollutants in the biosphere.
This report provides procedural guidelines developed for the special require-
ments of quantitative ecological investigations in intertidal benthic rocky and
sedimentary environments.
A.F. Bartsch
Director, CERL
111
-------
ABSTRACT
Procedural guidelines developed for the special requirements of quanti-
tative ecological investigations in intertidal benthic rocky and sedimentary
environments are given in this report. Community and population characteris-
tics of intertidal macrofaunal and macrophyte assemblages and their relation
to possible effects of artificial disturbances are emphasized. The sampling
guidelines are applicable to assessments of chronic or episodic impacts, pre-
and post-impact baseline studies and long term monitoring programs. Quanti-
tative estimation of natural spatial and temporal variations and distinguish-
ing them from the effects of perturbations are identified as major problems
in all intertidal zone pollution investigations.
Methods for developing appropriate and efficient sampling designs through
consideration of program objectives and comparisons of relative yields and
costs or other constraints are given. Sampling designs stratified by tidal
level are recommended and methods for allocating sampling points among strata
are explained. Practical aspects of site levelling, surveying and sampling
point location are discussed for different shore types as are methods of
securing quantitative samples from these biotopes. Photographic methods are
recommended for their relative efficiency in non-destructive quadrat sampling
on rocky shores.
Specific sampling program development steps are outlined for tide flats,
sandy beaches, rocky shores and gravel and cobble shores. A discussion on
deriving and analyzing data from sampling programs emphasizes the use of
various types of information and data analysis methods in intertidal pollution
assessment studies.
IV
-------
CONTENTS
Page
Foreword iii
Abstract iv
Sections
I. GENERAL INTRODUCTION 1
II. SAMPLING PROGRAM DESIGN CONSIDERATIONS
Introduction 5
Stratification 9
Systematic Sampling 12
Random Sampling 14
Sample Unit Size Selection 15
Sample Size Selection 16
Replication in Space 18
Temporal, Replication and Periodicity 19
Subsampling 20
Levelling, Surveying and Sampling Point Location 21
III. COLLECTING AND PROCESSING SAMPLES
Introduction 27
Destructive Sampling Methods
Sedimentary shores 29
Coring devices
Handling core samples in the field
Sieving methods
Flotation and staining
Trenching and Linear Sampling 38
Rocky Shores 39
Sample Preservation and Storage 40
Measurements on Samples 41
Non-destructive Sampling on Rocky Shores
In-situ Estimation of percent cover 42
In-situ Counts and Measurements 44
Photographic Methods 45
-------
IV. SAMPLING PROGRAMS FOR SPECIFIC SUBSTRATES 54
Tidal Flats 54
Sand Beaches 60
Rocky Shores 64
Gravel and Cobble Shores 75
V. DATA ANALYSIS
Introduction 77
Standard Ecological Information 77
Diversity Measures 80
Similarity Measures 83
Advanced Data Analysis 84
Recruitment and Productivity 85
VI. REFERENCES CITED 90
VI
-------
ACKNOWLEDGMENTS
We thank Dr. R. C. Swartz for the opportunity to participate in this
series of reviews on pollution related ecological sampling in the marine
environment. He showed great patience and understanding when reality proved
to be different from plans. We acknowledge with our thanks the contributions
of James Kopp and Barbara Rosene during the literature searching stages of
this study, and Ruth Phinney, who read an early draft.
VII
-------
SECTION I
GENERAL INTRODUCTION
"Although much descriptive material on sea shore ecology has been
published, there is no general work dealing comprehensively with sampling
techniques" N. A. Holme, 1971.
The objective of this study was to assemble the best available published
procedures for quantitative ecological investigations in marine intertidal
benthic environments, applicable to evaluating existing or potential pollu-
tion effects. Early in the study, literature .search showed that the state-
ment quoted above from the IBP handbook on methods for the study of marine
benthos (Holme § Mclntyre, 1971), continues to be correct. There have been no
comprehensive original studies devoted to devising or testing methods applic-
able to intertidal habitats nor reviews critically evaluating published
methods for even a major portion of this marine environment. Most of the
numerous quantitative intertidal sampling techniques found were devised for
limited purposes and published as parts of papers in the very large,
diverse and widely scattered primary literature on intertidal ecology. Many
methods in current use for studies of the impact of human activities on inter-
tidal environments are found only in unpublished reports of limited circulation.
They consequently have not been critically evaluated in the open scientific
literature and their validity remains untested.
Because of the nature of the literature and the incompleteness or lack
of testing of the methods found there, merely assembling the existing pub-
lished procedures would not have met the need for a comprehensive guide to
intertidal sampling methods. Instead, the methods found were expanded upon
from the wider literature on quantitative ecological sampling principles and
procedures. The evident lack in many intertidal studies of a sound, objective
basis for the design, type and number of samples taken has led us to the
inclusion of much material on how to devise rigorous sampling programs,
rather than extensive and explicit directions on how to execute plans from
the literature, which were originally designed for specific or unique require-
ments. Adaptation of contemporary methods and standards in quantitative
ecological work to designs meeting the special requirements for work in
intertidal habitats required a synthesis. The resulting procedures recom-
mended here were compiled from parts of the publications of many workers. It
-------
was impossible or unwieldy to supply attributions to all parts of these
recommended designs, but the lack of specific citations does not imply a claim
of originality by the compilers of this report for the ideas used herein.
Intended uses and users. These guidelines to intertidal sampling are in-
tended for application in marine ecological impact assessments, pre- and post-
pollution baseline surveys and long term monitoring programs aimed at
detecting and forecasting potential and actual impacts on the intertidal zone.
They are intended for use as procedural guides by project managers and scien-
tific personnel who must design and perform such studies but who may have had
no prior experience in intertidal work and who lack familiarity with the
original literature in intertidal ecology.
Limits of coverage. Procedures given here are primarily for the determination
of macrofauna and macrophyte community and population composition and
abundance characteristics, relatable to pollution effects. The methods are
intended for the reliable estimation of conditions at the time of sampling,
with emphasis on distinguishing natural spatial and temporal variation from
levels of difference due to perturbations, by quantitative means which can
meet rigorous critical challenge. Very large scale, overall survey tech-
niques like that of Ellis (1966] are not considered.
The important ecological properties of rates of primary and secondary
production in the intertidal, also susceptible to pollution effects, are
covered mostly by reference to literature. Meiofauna, microalgae and
bacteria, also of great ecological importance, require field sampling methods
different from those given here.
The procedures given are specifically for rocky shores and shores of
unconsolidated sediments, both flats and open coast beaches. They may be
adaptable to salt marshes, mangrove swamps, coral reefs and other distinct-
ive intertidal environments, but the special problems in these situations
require additional methods not specifically covered.
Omitted also are methods for the measurement of physical and some
biological features of these shores well covered elsewhere. Methods for
sediment analysis, phytobenthos primary production and energy flow measure-
ments in particular are dealt with well in the IBP handbooks on methods for
study of the marine benthos (Holme and Mclntyre, 1971] and for the measure-
ment of primary production (Vollenweider, 1974). Also excluded are methods
for the study of biological interactions, another important aspect of inter-
tidal communities potentially sensitive to pollution effects. For rocky
shores, some methods for detecting and measuring biological interactions are
given in Connell (1972, 1974), Dayton (1975) and Paine (1974). Predation
effects on intertidal soft bottom macrofauna was similarly examined by
Woodin (1974) and by Reise (1977).
We have omitted entirely a description of a quasi-quantitative method
which has been used for pollution assessments on rocky intertidal areas in
Great Britain, because we strongly recommend against its use. This method
was intended to overcome the time and effort requirements resulting from
-------
intertidal community richness combined with the high degree of spatial hetero-
genity characteristic of these shores. The quantification of organisms is
by scale estimates of numerical density or percent cover. The abundance
scales used are given in quantitative terms but their application is by
subjective estimates. The results are inherently not comparable in time or
space by quantitative statistical methods and operator differences cannot be
resolved by testing the resulting data. When such data is converted to
histograms in publication, an impression is given of quantitative objectivity
which is illusory and subject to serious misinterpretation. This method, its
abundance scales and associated wave exposure scale is given in the following:
Ballantyne, 1961; Moyse and Nelson-Smith, 1963, 1964; Nelson-Smith, 1977; Crapp,
1971 and Baker, 1976 a and b".
Objectives of pollution related intertidal ecological studies. Field studies
of pollution impacts on intertidal habitats will range from assessments of
catastrophic episodic events like oil spills, through studies on the degree
and geographic extent of lesser, chronic impacts to baseline and monitoring
programs in unaffected areas. The goal of the latter category is the
early detection of potential future alterations or of estimating potential
impacts for regulation of permissable activities.
Common to all of these program types should be the objective of adequate
quantitative definition of the state of intertidal communities at the sites
of interest so that impact effects are distinguished from natural variations
when comparisons are made with previous conditions at the sites or contempor-
aneous conditions at control sites demonstrably free of impact.
For this quantification, the sampling study design and the methods used
must be scientifically rigorous so that the resulting data and conclusions
can bear critical scrutiny. Such studies will be under scientific review by
cognizant governmental agencies as well as many other bodies with scientific
capability, often under conditions of contested hearings or judicial review.
Guidelines for the design of sampling programs and the selection of quanti-
tative methods suitable for meeting the objective defined above are provided
in this compendium. These methods will be suitable for: a. assessing the
ecological effects of acute, single episodes of disturbance or pollution by
comparison to control sites, b. the establishment of quantitative baseline
information on critical or sensitive ecological areas to be used in detecting
the onset of pollution impacts, c. monitoring studies designed to detect
both progressive changes from chronic pollution and recovery processes follow-
ing abatement of ecological stress.
Programs in the above areas should employ the following criteria during
planning. Study designs and methods must be capable, of quantitatively
differentiating between impact sites and non-polluted areas, with adequate
definition of natural spatial variations or heterogenity and also natural
temporal fluctuations. They must be capable of discriminating natural long
term and seasonal changes from progressive alterations due to chronic or
increasing impacts. Comparative temporal studies.should result in a data
base on rates and sequences of degradation and recovery processes suitable
for predictive use.
-------
Special features of intertidal sampling
Accessibility. The planner of intertidal sampling programs is confronted
with severe time constraints on access to this region, imposed by the diurnal
and seasonal variations in the time and level of low tide. Both the frequency
and duration of tidal exposure changes down the tidal gradient. On the Pacific
coast of North America for example, only on about one-third of the days in a
year does the tide level at low tide approximate the mean lower low water
level (0.0, MLLW). Accessibility to most of the vertical tide range on a
single low tide exposure is consequently limited to relatively few days and
sampling at different levels on different days becomes a practical necessity.
The lowest low tides not only produce exposure of the lower levels but also
longer available working times at higher levels. Since the duration of ex-
posure of low levels is short as well as infrequent compared to upper levels,
priority must be given to lower level work. Inevitably other intervening
factors interfere with sampling schedules for the lower levels, such as high
storm surf which may make lower levels inaccessible at times set for winter
seasonal sampling.
Biotic stratification or zonation. The major spatial feature of the
distribution of intertidal species is their discreet vertical limits within
the tidal range. This results in an overall stratification or zonation of
the community with respect to tidal height, because of the approximate co-
incidence of many groups of these vertical limits. The ultimate cause of
zonation limits and the large number of factors modifying them on rocky shores
is the subject of a large literature not reviewed here, which identifies both
biotic and physical causes and interactions between the two. The many
variations in intertidal zonation are one of the major sources of natural
spatial variation on the large scale between sites, which must be adequately
accounted for in pollution assessment, especially in the selection of control
sites. While most attention has been devoted to zonation on rocky shores,
discrete vertical species limits are also characteristic of beaches and other
sedimentary-shores, but are not apparent from surface inspection alone.
Since vertical tide range limits of species and changes in their density
with tidal height are universal features of all shore types, it is recommended
that these known discontinuities and their direction be reflected in sampling
designs, and that the tidal heights of all sample units be carefully identi-
fied in any type of sampling design used.
In the face of known vertical biological stratification, any sampling plan
which utilizes horizontal distances alone for locating sampling points will
confound vertical and horizontal causes of differences between points unless
a vast number of sample units are used. Moreover, such designs severely
limit the usefulness of comparisons between sites or times.
Since on all types of shores the slope is not constant across the shore,
transects across shores, stratified by vertical height, are the most generally
applicable bases for sampling designs for the intertidal. As a rule, one foot
(0.3 m) vertical limits to strata offer ample resolution of population and
-------
community changes with vertical height. Allocation of sample units among
strata of unequal widths, resulting from slope differences, is dealt with in
later sections. A stratified sampling design, with strata parallel to the
shoreline and defined by tidal height contours or by obvious vertical limits
of species associations, also defines hypotheses about vertical distributions
and abundances within and between sites which statistical testing of the sample
data may refute or support. In turn, any analysis which, in order to make
comparisons with other sites or sampling periods, combines all data for all
levels within intertidal transects, will probably not reveal existing dif-
ferences between sites, because the within-transect variance produced by
combining data from different levels will be enormous. An appropriate
analysis will compare strata of similar tidal height from different sites.
An example illustrating the points above for rocky shores is the study
of Batzli (1969) who compared the relative biomass of the species populations
and species composition in two transects of tide pools stratified by tidal
height. While tide pools tend to minimize differences between heights because
they hold water at low tide; Batzli found that species number, biomass and
its distribution among species, and equitability as an index of the distribu-
tion of relative abundances of species were all related to the tidal height
of the pools. He further noted that this distribution along the natural
environmental gradient of stress resembled that resulting from pollution
stress gradients. A similar study by Glynn (1965), showed similar community
changes with tidal height over a vertical distance of approximately one meter.
II. SAMPLING PROGRAM DESIGN CONSIDERATIONS
INTRODUCTION
The role of field sampling in ecological studies is to produce unbiased,
reasonably precise estimates of population parameters without having to census
entire populations. An appropriate sampling design permits detection, within
specified limits of accuracy and precision, of changes in population or
community parameters. The selection of a sampling design is complex. Methods
have been developed to aid the selection of the most efficient design through
comparison of relative costs and yields. A discussion of the factors affecting
selection of a sampling design may be approached by considering the nature of
the environment to be studied.
The intertidal substrate ranges from mud to sand, pebble, cobble, or
consolidated rock, and grades or mixtures between these. Each substrate has
unique properties which influence the sampling process. Common to all sub-
strates is heterogeneity and the gradient of intertidal elevation. Generally,
the shoreline is broken into sections by natural or man made boundaries such
as headlands or jetties. A problem often presented in ecological investigations
is to characterize the assemblage of organisms within a bounded section of the
shore. Differences between areas or changes in one location through time can-
not be.interpreted unless they can be distinguished from the spatial and
temporal variations which are normally present at the locations. An important
-------
part of field sampling, therefore, is quantification of spatial and temporal
heterogeneity.
Changes in environmental factors or in the density of organisms may
occur over a range of distances or time intervals, which may be loosely
referred to as the spatial or temporal "scale" of variation. At a given lo-
cation, several scales of variability may be present. The degree of spatial
and temporal variation present at a range of scales must be identified to
accurately assess differences between samples.
The number of scales on which sampling must be applied requires clearly
defined terminology. The following terms will be used.
Sampling location, site or area: a section of the shoreline delimited
by natural or man made boundaries so as to be easily distinguished.
Transect: any representative part of the sampling location, usually
extending from the high water line to the low water line, or the
full width of the intertidal zone.
Stratum: divisions of the transect usually by elevation, which are
more variable between than within themselves.
Common shore terminology is not standardized and can be confusing because
three, not two, dimensions are the references to points on a shore. The
terms used here are as follows.
Width or horizontal depth of the shore: the distance on a shore from
the uppermost high tide line, directly across the beach slope toward
the lowermost tide or water line.
Tidal height, vertical level or tidal level: the elevation of a point
in the intertidal zone of the shore relative to sea level.
Length of the shore, or the alongshore directions: the horizontal
distance on a shore parallel to the uppermost high tide line and
lowermost low tide line. The topographic contours representing
vertical tidal elevations run along the length of a shore, demarking
slope changes across its width.
Each transect is one representative part of the sampling location,
regardless of the number of cores or quadrats taken within the transect. If
the sampling location can be assumed to be nearly uniform on a scale of
meters (the usual scale of transects), then one randomly positioned transect
would give a reasonable representation of the location. However, if the
location is more heterogeneous, then transects in different parts of the
sampling location may produce different conclusions. In most cases lack of
foreknowledge forces two or more transects to be selected in any given location,
unless the location is very limited in area. The determination of the number
-------
of transects to be sampled will be discussed in a later section, following
discussion of the basics of the selection of sample unit and sample sizes.
Positioning of transects is usually arbitrary, or randomized along a line much
longer than the transect is to be wide.
A transect can vary from a line of potential places to put a sample unit,
to a wide belt within which sample units are positioned according to various
designs. The width of the transect is a function of the available homogenous
area. When possible, transects are usually from twenty to thirty meters in
width. In the rocky intertidal the widest space available, free of inconveni-
ent changes in tidal height, outcrops or deep channels may be only one or
two meters wide, and the transect is correspondingly narrower.
The transect width should be responsive to the suspected important scale
of variation at the given location. When variations in population density or
environmental factors occur on a scale of meters, then the transect should be
meters wide. When the only major variation is on a scale of less than one
meter, the transect may be reduced to a line of quadrat-positions. However,
if no prior information is available on possible scales of variability, tran-
sects should not be reduced to lines of sample units. Specific recommendations
will be made for the various substrates in later sections.
The following terms will be used repeatedly in the remainder of the
discussions.
Sample unit = a single quadrat or core, which defines the area or
volume within which the measurements on organisms are to be taken.
Sample unit size = the area or volume of the sample unit.
Sample size = the number of sample units in one sample.
Sample = a representative selection of the total population of organisms,
which is used to estimate population and community parameters within
the area in which the sample was taken. A transect may be sampled
using one set of sample units (one sample). If this sample is
allocated to strata within the transect, each stratum has a separate
sample; the collection of samples within strata still represent
only one sample of the transect.
Two major types of sampling are commonly applied to the intertidal.
These are systematic and random sampling. The former spaces sample units
(quadrats) evenly in the area of the transect, or within each stratum of the
transect. The latter randomly places the sample units over the area of the
transect, or in each stratum of the transect. The fundamental procedure in
either case is unchanged for line or belt transects. The principal advantages
and disadvantages of the two methods as summarized by Cochran (1977) are:
1. Systematic sampling is simpler in execution. A single sample unit
position must be randomly selected. All other sample units are
-------
automatically positioned according to the size of the area and the
desired number of sample units. Random sampling, in contrast,
requires that each sample unit have a random position assigned to it.
Both the selection of positions and their location in the field are
usually more time consuming than for systematic sampling.
2. In some cases, especially with a small sample size, the greater
spread of a systematic sample can result in a better estimate of
the mean than that from a random sample of similar size.
3. Periodicity or linear trend in the variable under study, such as the
abundance of species, will often result in over or under estimation
of the mean of the variable, depending on the degree to which the
systematic sample is spaced coincidentally with the period of the
study variable or on its position relative to the central value of
a trend in the variable. Further discussion of this problem is found
in Kelley (1976).
4. There is no reliable method for the estimation of the variance of the
mean of a single systematic sample. At least two systematic samples
must be taken within each transect, or within each stratum, for the
estimation of the variance of the mean. For k independent systematic
samples, the estimate of the variance would have k-1 degrees of
freedom. A single random sample, in contrast, has n-1 degrees of
freedom for the estimate of variance, where n is the number of
sample units. The estimate of the variance tends to be erratic when
based on a small number of degrees of freedom. Systematic sampling
can be prohibitive when a large enough number of samples are taken
to yield an adequate estimate of the variance. In general, random
sampling is superior to systematic sampling for the type of study
required in the assessment of pollution effects, in which the variation
of the mean number of individuals of each species is important.
Further discussion of this topic may be found in Cochran (1977).
One of the most difficult parts of the development of a sampling
program is the selection of the sample size, which is closely linked tq
the selection of the size of the sample unit. The size of the quadrat
used must reflect the nature of the species present at the sample location.
Important large or rare organisms may be missed by a small quadrat. Large
quadrats, however, are generally more expensive in time and effort involved
in the sampling and sorting process. For each quadrat size which is
adequate to sample the species present at a given site, the cost and
efficiency are calculated. It may be that a smaller quadrat is less
costly than the number of larger quadrats required to produce the same
precision of estimation, but that the larger quadrat returns a greater
number of species. The criteria for selection of an appropriate sample
unit size and sample size are discussed in detail in later sections.
For the application of the formulae which are used to determine the
relative efficiency of sample units and sample sizes, some prior information
on the sample location is necessary. This information is normally obtained
-------
either from previous work in the same or similar locations, or from a pre-
liminary sample. The preliminary sample is necessarily inefficient, because
it is not optimal for the given conditions. The key part of the information
required to produce an efficient sampling program is the variability of the
area. The initial survey must gain a reasonably precise estimate of the
variability of the area at the time of the survey, and thus should be larger
than the final program rather than smaller. Since the initial survey is not
repeated, the loss of time or other expense is more than compensated for by
the efficiency of the final program.
In following sections the terms mean and variance will be frequently
used. In most cases, intertidal studies will involve a number of species,
each with an associated mean density and the variance of that mean. In methods
such as the determination of sample size, or the allocation of sample units
to strata, a single mean and variance is used. The calculations could be
repeated separately for each of the species for which the mean and variance
are known. In practice, the calculations are made for each of the species
which seem interesting, and various averaging methods ai*e used if the results
for different species are widely discrepant. An alternative is to accept a
loss in the efficiency of sampling some species in exchange for a greater
ability to sample others. Since the methods are approximate, fairly crude
means of averaging produce sampling programs which are quite able to detect
changes at the desired levels of accuracy and precision. In the following
sections the problem of dealing with several species, and therefore multiple
independent variables, will be elaborated as necessary.
STRATIFICATION
Stratification is the process of division of an area such as a transect
into sub-areas which are presumed to be more variable between themselves than
areas within each of the sub-areas. Strata are sampled independently of each
other, and the information can be used to compare strata or to form a better
estimate of the mean density and variance of species within the entire transect.
Excellent examples of the application of stratification procedures in marine
benthic work are the studies of Brinkhuis (1976) and Saila et al. (1975),
which may be consulted as models.
In the intertidal, the most common method is to stratify by either uniform
intervals of elevation, or by obvious discontinuities in the community of
organisms. When the transect is stratified by either method , the estimate
of the variance of the mean density of any species over the entire transect
is superior to that which would result from an unstratified sample. If the
communities within the strata are greatly different, the strata may be treated
as separate populations and their species compositions and densities considered
separately.
Within any given transect, the sample size n will be defined by the
sampling plan. The n units must be allocated to strata. The simplest
procedure for allocation is to allocate sample units in proportion to the area
of the stratum. The area of the stratum can be represented as the total
-------
number of possible contiguous positions for sample units; that is, the area
is some multiple of the area of the sample unit to be used. If the total
stratum area is N, , then the number of sample units allocated to the stratum
is given by
n, = n Nh = nW, , W, = Nh = stratum weight
h -IT" h h 5T
where N is the area of the entire transect expressed as a multiple of the
sample unit size. Since the relative areas of strata are compared and used
to weight their allocations, any means of expressing their areas could be used
(eg- m2). The purpose of using the sample unit size as the unit of measurement
will be discussed below.
This allocation method gives an unbiased estimate of the mean density of
any species for the entire transect, when the means within strata are pooled.
The pooled estimate of the transect mean density for any given species is
the weighted average of the stratum means,
N
or the equivalent expression,
^tran - Vl + V2 + ' * ' + \ *h
Since the sample units have been allocated proportionally, the stratum
weight W, can be expressed in terms of the stratum sample sizes (the allocations
to strata),
^l + n2?2 * '" * nh >^h
n
This is true simply because W, = N, /N = n, /n. The estimate of the variance
of the mean for the entire transect is given by
Here the finite sampling correction -^— is included. The sampling fraction
f is given'byn/N. If the ratio of the total area of the sample units in the
sample to the total area of the transect is less than 0.10, then the sample
can be assumed to be drawn from an infinite population, and the talcing of one
sample unit does not measurably affect the probability of taking another sample
unit at another position. If the ratio exceeds 0.10, the population is effect
ively finite. The finite population correction adjusts for the bias intro-
duced by sampling without replacement of units in a finite population. If N
is expressed in terms of multiples of the sample unit size, then the sampling
fraction is actually the number of sample units in the sample divided by the
number of units which could have been contiguously placed in the transect.
10
-------
The formula for the variance given above uses the true variances of the
strata means. In practice, the sample variance estimates the true variance
and the formula is
s2(ytran) =Vi + w2s2 + ••• +Vh
assuming that the finite population correction is not needed.
The method of proportional allocation produces a suitable unbiased
estimate of the transect mean and variance for each species, but does not
maximize the efficiency with which the mean and variance in each stratum are
estimated. Cochran (1977) describes methods for allocation which optimize
the estimation of stratum means and variances, with comparison of strata as
the primary goal.
A second problem is that proportional allocation does not take into
account differences in the variability of a species in different strata. An
optimal allocation scheme would assign more sample units to strata where the
variance was higher, even if the areas of strata were equal. Estimates of
the variance within strata are rarely available from preliminary sampling,
so this more refined method would seldom be applied in the intertidal. The
Neymann allocation method, one of those which account for differences in
variance, is detailed by Cochran (1977). Cochran also discusses the inclusion
of differences in the cost of sampling between strata, but this would seldom
be important in field studies of the intertidal. In most cases the sampling
method will be the same in all strata, and the cost per sample unit is constant
for all strata.
A potentially serious problem in stratification is the inaccurate
estimation of stratum area. If strata are greatly different in area from the
estimated size, then stratification can actually decrease the effectiveness
of the sampling program. In the intertidal, such errors could arise by
assuming that strata are equal in area because they are equal in the interval
of tidal height. Changes in the slope of a shore result in different areas
for equal intervals of elevation. The actual areas of strata should be used
rather than assuming the strata are of equal area. Final adjustments to
allocations may be necessary in the field, following surveying and actual
measurement of stratum areas.
A separate allocation design could be computed for each species. If
there is no major discrepency between the allocation schemes produced from
different species data, a simple arithmetic average of the allocations for
each stratum should be sufficient. If some species produce allocations which
differ greatly from others, a decision must be made concerning the importance
of those species in the sampling program. If they are important, they can
be given greater weight in the averaging of the allocations to a given stratum.
If not, they can be given reduced weight in the averaging. More complex
methods to produce an optimal allocation are possible and some of these are
described by Cochran. In practice, the allocation-«of sample units to strata
11
-------
is sufficiently crude that moderately large discrepencies between the
allocations from different species data can simply be averaged.
SYSTEMATIC SAMPLING
For simplicity, assume that the transect has been selected and is to
be sampled without division into strata. The transect area may be expressed
as N units of the same area as the chosen sample unit. The transect may range
from a line of potential positions for the sample units to a larger area
within which sample units are positioned. In systematic sampling, the units
which are sampled are spaced evenly over the area. The spacing is simply
N^n, where n is the number of sample units in the sample. If k = N/n, every
k unit would be sampled to give a total of n units in the sample. In
practice, k is often not an integer, so the nearest integer value to N/n is
used.
The process of selection for the sample of n units is:
1. From the area of the transect and the area of the sample unit, the
value of N = total area is determined. Given the value of n
sample unit area
required for the program contemplated (see section on sample size
selection), the value of k is determined to the nearest integer.
In a simple example, suppose that-the transect is a line of positions
.5 m wide (sample units are .25 m ) and 100 m long. There are 200
possible contiguous positions in which to place a .5 x .5m sample
unit. If the sample size is to be 10 units, the value of k =200/10 =
20. Every 20th position should be used, or positions 20 x o.5 m =
10 m apart.
2. If the transect is a line of possible positions, the first unit to
be sampled is selected from the N possible positions at random.
Every ktn unit in both directions from the first unit position is
also sampled. In the example above, every 20th position from the
first in both directions is sampled; or equivalently, sample units
are positioned at 10 m intervals from the first unit. The purpose
of starting at some random position rather than at the first position
in the line is to ensure that the sample is unbiased. If k is not
an integer, the last few units might not be sampled if sampling is
begun at the first unit. A random start allows any sample unit
position to be sampled.
Note that if the distance separating chosen sample unit positions
is used rather than counting off every position, small units can
be readily located in a long line. Compare the difficulty of
locating 10 positions in a line of 1000 possible positions, to
simply measuring off distances and sampling at equal intervals.
3. If the transect is an area rather than a narrow line of possible sample
unit positions, the transect may be divided into adjecent rows of
potential sampling unit positions extending the width of the transect.
The unit positions can be numbered from one to N along adjacent rows.
A random position i is chosen, and positions i+k, l+2k, i+3k, and
12
-------
so on, and also positions 1-k, l-2k, l-3k, ..., would be sampled.
A disadvantage to this method is that in the direction parallel to
the waterline, the units are sampled at fairly long intervals,
whereas in the perpendicular direction the rows are adjacent. The
pattern of sample units could be so extreme as a set of columns of
sampled positions:
X X X X X
X X X XX
X X X X X
Obviously, the scale of variation measured in the two directions
would be different. For most purposes, it is better to have the
rows of sampling positions separated by the same distance as separates
the units horizontally, forming a grid of squares:
x
The mean of a systematic sample is calculated in the usual manner. There
is no unbiased method for the estimation of the variance of the mean of a single
systematic sample. Cochran (1977) gives approximations for the variance of a
single sample when the parent distribution follows certain forms, but these are
not likely to be encountered in ecological studies. In goneral, at least two
systematic samples are required within each transect in order to estimate the
variance. The variance of a set of systematic samples is calculated according
to the usual definition of the variance, where the means of entire samples
become the observations yi on the variable, and the mean of those sample means
becomes the variable mean Y.
1 k - 2
V(Y_V3 = 7 t (y. - Y)" for k samples.
sy Ki=1 i
The estimate of variance will have k-1 degrees of freedom; since k is usually
small, the estimate will be somewhat more erratic than the estimate of variance
derived from a single random sample with n-1 degrees of freedom.
If the transect is stratified, then the same procedure is followed within
each of the strata. At least two systematic samples would'be taken within each
stratum in order to estimate the variance in the strata, from which the transect
variance is calculated. Each sample should ideally have a sample size equal to
the number of units allocated to that stratum, but this would result in doubling
of the total number of sample units per transect if the minimum of two samples
per stratum were taken. The alternatives are to allow each^stratum sample to use
13
-------
half the allocation for that stratum, or to halve the number of strata and
permit each sample to have about the full allocation of one stratum. The first
alternative results in a loss of precision due to inadequate sample size, and
the second a loss of precision due to inappropriate stratification. If strata
have been selected on an arbitrary height interval basis, then halving of the
number of strata is probably less detrimental to the quality of the sampling
program than halving the sample size within strata. If the strata are based
on very obvious discontinuities, the best solution is to reduce the total number
of transects in the program in order to maintain the quality of the sampling
within the remaining transects.
RANDOM SAMPLING
The simplest case is when random sampling is used in an unstratified
line transect. The transect has N possible positions in which a sample unit
could be placed. A sample of n of the total possible N positions is to be
selected. In random sampling, every one of the n units is positioned randomly
and independently of the others, except that no position is sampled twice.
In a line transect, the possible positions are numbered from 1 to N and n
of these numbers are randomly chosen and sampled.
When the sample unit is small, and the transect area large, it is difficult
to identify every possible position by number. The usual method is to use a
subset of the total possible positions by imposing a grid over the area and
selecting a random number of points on the grid. For example, suppose that a
small core of 5 cm diameter is to be used, and the transect is 30 m wide and
100 m long. There are about 1.5 x 106 possible positions. Instead of using
this large number, a large subset is defined. The transect is divided into
0.5 m or 1 m intervals in both directions, forming a grid with either 12,000
positions or 3,000 positions. For easier location of sample unit positions,
two random numbers are used to define each position to be sampled. For 1 m
spacing of the grid, a random number, x, between 1 and 100 is chosen for the
first coordinate, and a second random number, y, between 1 and 30 defines
the second coordinate and sample unit position. One simply samples x units
from the top of the transect, and y units from the starting side of the transect
To speed the sampling process, all random positions are selected in advance,
and the positions along each row of the grid which are to be sampled are
listed. Sampling then consists of measuring to the row, and sampling at
predetermined distances along the row.
It is essential that the interval of the grid is such that a large number
of positions are available, and the sample size n is never more than 10% of
the possible number of positions.
When the transect is stratified, the same process is used within each
of the strata. Random sampling does not require more than one sample per
stratum, so the sample size within each stratum is the allocation to the stratum
14
-------
SAMPLE UNIT SIZE
A sample unit is a known area or volume of the sampling location,
enclosed by any device or bounded by any other means. A core is a common
sampling unit, but imaginary lines around a space 25 m on a side equally
define a sampling unit. In a few cases, only parts of the sampling unit area
are sampled; these are subsamples of the basic sampling unit, because their
position in the transect or in the stratum is determined by the position of
the sample unit.
Two major criteria are applied to the selection of an appropriate sample
unit size. The simpler is the maximization of the number of species retrieved
by the sample. As the sample unit size increases, the sample comprises a
greater part of the total area, and the probability of collecting all of the
species present in the area increases. The same qualitative effect occurs
when the sample size, or number of sample units, increases. Typically, as the
sample unit size increases, there is an initially rapid increase in the number
of species recovered in samples of the same size, followed by a levelling off
at some specific sample unit size. Further increase in the sample unit size
above this size yields relatively less of an increase in the number of species
recovered, although for any practical range of sample unit sizes, the number
of species recovered will continue to increase slowly with the sample unit
size. The optimal sample unit will be some unit near to or larger than the
size at which the levelling off occurs, by the criterion of maximization of
species recovered.
The second criterion is the minimization of the variance of the mean.
Most organisms are found to be patchily distributed in the intertidal. It
is well known that as the sample unit size changes, the variation in the numbers
of a given species between sample units also changes. This property has
been used to estimate the size of patches for a given species. As the quadrat
size approaches the scale on which patchiness occurs for a given species,
the variance of that species will increase. At quadrat sizes greater than
the patch scale, the variance again decreases. At some point, the scales of
quadrat and patches are so different that the sample may appear to have been
taken in a random distribution of that species. At this quadrat size, the
variance will be nearly equal to the mean (a characteristic of random distri-
butions) , and the estimate of variance will be minimized. Effectively, use
of an appropriate sample unit size can decrease the measured variation by
eliminating the component of variation which results from patchiness. Since
the comparison of locations and detection of differences is largely dependent
on the variance of the means of the locations, unnecessary addition to the
variance are disadvantageous. Of course, for two such locations to be compared
it follows that the quadrat sizes used in both locations must be equal.
Quadrats of different size vary in the cost of taking the sample and
processing it. In addition to maximization of the* number of species and
minimization of the variance, most programs must consider the economics of
sampling, and choose a sampling design which is most efficient monetarily
within the bounds of the scientific requirements. Cochran (1977) gives a
15
-------
method for the selection of the optimal sample unit size which includes both
the cost of each sample unit in use, and the variance of the sample when that
sample unit size is used. The "relative net precision" of a unit of a given
size is given by M2/C2S2, where M is the relative size as compared to the
smallest unit, C is ¥he relativeucost as compared to the cheapest unit, and
S2 is the variance associated with the unit of size u. The best unit is the
oHe with the highest relative net precision. Of course, each species has a
separate variance, so that a different set of relative net precision values
would be calculated for each species. If different unit sizes are indicated
when the data for several species are used, then some compromise must be made.
The simplest method is to sum the relative net precision values for each
quadrat size over all species of interest, and choose the unit size with the
highest sum. If some species are possibly more important than others,
weighting of their relative precision values would be appropriate.
The above method does not take into account the number of species returned
by each size of quadrat. The selection of the most efficient sample unit
size should be restricted to the range of sizes which satisfy the condition that
they recover a large part of the species present in the sampling location.
Smaller quadrats may prove to be more economical, but the usefulness of the
sample will be limited if the community is not adequately represented.
As was the case for allocation of sample units to strata, preliminary
information is required to evaluate the best sample unit size. The initial
survey should include sampling with a range of quadrat sizes, each quadrat
size used to take an independent random sample in an appropriate area. All
samples should be of the same size (have the same number of units). When
the approximate range of sizes possible have been determined from the plot
of species richness against sample unit size, the efficiency formula above,
or an equivalent, should be applied to determine which sample unit size has
the best yield in terms of low cost and minimal variance. The range of sizes
to be tried in the preliminary survey can be determined by examination of
sample unit sizes used under similar conditions. The relative cost of each
unit will be determined in application during the initial survey.
In some cases, a few species will be so different in size or abundance
that a completely different sample unit will be required* Large starfish,
for example, might be sampled by units as large as 100 m . The remainder of
the species in the area would be sampled using some much smaller quadrat
size. Species such as starfish are potentially important to the community
and should not be ignored because they are rare. In general, very large
fauna often have an important influence on the community which may not be
reflected in the total numbers or biomass of these large species.
SAMPLE SIZE
The sample size is the number of sample units in the sample taken within
a transect. The total sample size may be allocated among strata. Each stratum
sample then has its own sample size, which is the allocation assigned to it.
16
-------
In determining the total number of units to be taken, however, the sample
size of the entire transect is considered first.
The selection of a non-arbitrary sample size has several purposes. The
precision of estimation of the mean density of any species increases (the
variance decreases) as the number of sample units increases. At the same time,
the accuracy of the estimate of the true population mean increases. The
total cost of the sample also increases, so that it is desirable to select a
sample size which will yield an estimate of the true mean which has the desired
levels of accuracy and precision, but which is no larger than absolutely
necessary. The required sample size can be calculated for each transect or
location, and the total cost for each sample can be estimated. Under a given
budget, the total number of samples which can be taken at a location, or the
number of locations which can be examined given a specified number of transects
per location, can be determined. Unless the budget is unlimited, this sort
of cost estimation is essential. As a general rule, the sample size within
a transect should not be less than that indicated by the methods described
below. If cost limits the total sampling program, then it is better to
eliminate entire transects from the proposed sampling program and retain the
quality of the samples taken at the remaining transects, than to sample a
larger number of locations in an inadequate manner. Later sections will
expand the problems of developing the final program in detail.
A general formula for the optimal sample size for specified precision
and accuracy may be derived from the definition of D, the index of precision.
D is defined as the ratio of the standard error to the mean,
D- (s
If the tolerable error of estimation of the mean is 20%, then D = 0.2. For
a given value of D,
This specifies the minimum required sample size to obtain a set precision of
estimation of the mean density. If the variance were 100 and the mean 10 for
some given species, then for a tolerable error of estimation of 20% of. the
mean, the required sample size would be n = 25. On the average, a sample of
25 units in the same location would have a standard error of the mean equal to
about 20% of the mean. D may also be expressed as the relative error in
terms of percentage confidence intervals about the mean. For example, one
might specify that the tolerable 95% confidence limits about the mean in
future samples should be no more than -40% of the mean. D is now 0.4, and
the formula is multiplied by the value of Student's t for the desired
probability level.
17
-------
For 95% probability level of D, t is about equal to 2.0, so for D = 0.4,
the formula reduces to simply
2
n = 25 ^
x
Zimmerman and Merrel (1976) used a modified formula (Snedecor and Cochran,
1967) to include a specified level of probability of detection of changes in
the mean density at any given level. The modified formula is simply
D x
where the value of Y incorporates both the degree to which the mean must change,
and the probability of detection of that magnitude of change.
This formulation is more suitable for. any study in which changes in
density are expected to be important as indicators of pollution. The degree
to which the formula increases the recommended sample size over that in the
original formula depends on the magnitude of change and on the desired
probability of detection. For example, if the variance were 100, the mean
10, the magnitude of changes to be detected - 40%, and the probability of
detection 95%, the required+s ample size would be 162. The original formula
for estimation of the mean - 40% specifies 25 as the sample size. A probability
of 80% of detection of a 40% change in the mean reduces the recommended
sample size to 98. In most cases, a smaller sample size will have to be used
for practical limitations, and larger changes in the mean, or lower probability
of detection, will have to be substituted. An excellent example of the use of
procedures for determining sample size is the study of Livingston et al. (1976).
REPLICATION IN SPACE
The previous sections have discussed the aspects of the total sampling
program which relate to sampling in a given part of the sampling locations,
the transect. The transect, as mentioned in the introduction, is a single
part of a location which varies between parts. To be certain that the sampling
area is adequately characterized, at least one replicate transect must be
taken. The exact number of replicates depends on the likelihood that the
sampling location will be internally variable between transects. If the
sampling area is extensive, more replicates will probably be needed. Each
transect is a complete sampling program as specified by the methods described
in earlier sections, with a chosen sample size, sample unit, and stratification.
The sample size and sample unit area both affect the number and type of
organisms returned in the sample, and should not be changed between transects
of the same dimensions. The stratification of the transects should be the
same, since the location was defined as at least superficially homogeneous
internally. If necessary, the transects could be stratified differently. If
so, then the strata probably could not be compared between transects, but
the overall transect means could be compared. If strata are selected by
-------
biotic discontinuities, which are located in the same relative positions
but at different absolute heights in different transects, a comparison of
these strata between transects would be valid even if the height intervals
differed.
Since the sampling location has been selected on the basis of apparent
homogeneity, one might expect that two transects would be sufficient. In
some cases, large scale differences can occur even though not superficially
obvious, and more transects would be necessary to characterize this scale of
variation. The most comprehensive programs rarely have more than 4-5 transects
in a given location, and in most cases fewer are adequate.
Relatively few studies have examined the variability between sampling
locations, probably because of the extensive sampling program required. Each
site would be sampled at two or more transects, and practical limitations
would usually limit the sampling program to not more than two sites of any
given type of substrate. At least two sampling locations of each type would
be sampled to be certain that the results of sampling at one were not a
product of unusual conditions at that location.
The essence of the above comments is that at every level, replication is
essential if the conclusions reached from sampling are to be general. A
sampling program which examines a single sampling location in great detail
cannot confidently be used to state conclusions about that entire category
of shoreline types. A single transect is an insufficient basis for definite
conclusions concerning the remainder of the sampling area. The development
of the final program must be a series of compromises, in which the scales
of variation of greatest interest are examined. If it is important to gain
information about all of the shoreline types in a diverse region, then the
degree of variability within each location may have to be less well character-
ized in order to have the time and resources available to sample in many locations.
The variation between transects of a given location is least likely to be of
importance. The variation within a transect is most likely to be important,
so that transect sample sizes should rarely be reduced. Variations from one
type of shore to another may be of great or little importance, depending on
the objectives of the study.
TEMPORAL REPLICATION
In addition to the variation which occurs over distances from centimeters
to many meters, the abundance of species may change through time from natural
causes. These fluctuations are at least as important as spatial variation in
interpretation of possible pollution effects. Many species reproduce seasonally
and would normally have large fluctuations in abundance. Single samples at
one* time would potentially over- or under-estimate the average yearly abundance,
and changes in the abundance of species could not be separated from possible
normal temporal fluctuations without information on these fluctuations. An
absolute minimum frequency of sampling would be twice each year, in order to
sample at the probable times of greatest and least density. Many species,
19
-------
however, vary from the usual patterns of spring-summer highs and winter lows,
so that four times per year is preferable. At each time of sampling, the
program is repeated in its entirety if possible.
A related problem is error resulting from short-term variability in
apparent abundance, particularly diel periodicity. Many organisms follow
diel patterns of activity and are more easily observed during specific parts
of the day or night. While it is not practical to sample at all times of
day for every sampling run, at least one sample or series of samples should
include a check on variation in the recovery of species at different times
of the day. In some cases, the time of sampling may have to be changed to
accommodate the diel behavior patterns of potentially important species.
Finally, the major characteristic of the intertidal is simply the
alternating exposure and immersion of the substrate. At high tide, one can
always expect that a number of motile species not present at low tide will
be an active part of the community. Since sampling is largely restricted
to low tide, this regular alteration of the observable community will seldom
be quantifiable. A number of studies have qualitatively examined the
possibility that major members of the community are absent during the sampling,
by seining or other means of qualitatively sampling demersal or mobile
epibenthic organisms.
SUBSAMPLING
If a large sample unit size has been selected, it is sometimes impossible
to sample the entire unit. Generally, a large unit will have been selected
because it either yields a sufficient number of the species present, or
because it has the property of reducing the variance. In the former case,
it is the size of the total unit which controls the number of species
returned by the unit, so that only by sampling the entire unit can one recover
the desired number of species. If the unit was selected because it reduced
the variance, then in some cases it may be possible to subsample the unit and
retain most of the effectiveness of the unit. Subsampling consists of
selecting a random subset of equal-area portions of the unit. If the sample
unit can be divided into equal areas or equal volumes (in the case of cores),
totalling M possible subunits, a random selection of m of the M subunits is
made, and the sampling retrieval of organisms proceeds only within the m
subunits. A hazard of subsampling is that if the subunit areas correspond
to the scale of patchiness over short distances, then the variance associated
with that patchiness will be reflected in the estimate of the variance between
primary sample units which is derived from the information on the subunits.
For maximum retention of the effectiveness of the sample unit, subunit areas
should be selected which are different from the areas of those units in the
preliminary sample survey which had high variances. In some cases this is
unavoidable, and a loss of effectiveness of the primary sample unit must be
accepted.
Estimation of the means and variances of primary sample units is as
follows:
20
-------
1. To find an estimate of the mean of subunits within a single primary
sample unit, calculate the mean per subunit using the usual method
of estimation of a mean.
2. The variance of the subunit mean within the sample unit is calculated
in the usual manner for estimation of the variance of a random
sample, with m-1 degrees of freedom.
3. The simplest estimate of the mean per subunit of all primary units,
which is the mean per subunit for the sample, is
f = 1/n C/j + y2 + ... + yn)
where y. is the mean per subunit within the primary sample unit i,
and there are a total of n primary sample units in the sample. The
estimate requires that the estimate of the mearf per subunit within
each primary unit be unbiased, but it does not require that the
subunit sample size be constant for all units.
4. The variance of the mean per subunit. for the entire sample is given
by
= 1-n/N 2 1-m/M 2
- — s, + - — s_
1 mn 2
.
where s = the variance among primary units of the mean per subsample
unit, and s^ = the variance among subunits of tho mean per subunit:
2 n,-. =2 2 n m , . . -.,2
si = ? Cyi - y) $2 = z z Cyij - yi)
i n-1 i j n(m-l~)
"f-L ft*
where y. . is the value of the j subunit in the i unit, y. is
the meanjof the subunits in the itn unit, and y is the mean per
sample unit over all units, given above.
Further details of the use of subsampling, including methods for selection
of the optimum value of m, and use of subunits of unequal size, may be found
in Cochran (1977), Venrick (1971) and McAlice (1971).
LEVELLING, SURVEYING AND SAMPLING POINT LOCATION METHODS
For locating sampling point heights in any type of sampling design, and
for stratification of transects, the tidal elevation of the various parts of
the site or transect must be determined and both the transect or sampling
area and specific places inside it must be marked, permanently in most cases.
Site location and transect establishment.
Criteria for selection of sampling areas or sites will reflect the
requirements of the particular program and should be well established early.
21
-------
In pollution related studies sites will be selected for various reasons,
such as: because they represent typical, unaltered examples of prevalent
habitat types in the region; because of the possibility or existence of
degrees of pollution impact at specific places regardless of habitat type;
or because they have been determined to be of ecological importance and must
be monitored for any signs of alteration or to determine the effectiveness
of pollution abatement efforts, or to follow events before, during and
after episodes of chronic pollution at sites not identified and studied
previously.
Site selection will therefore be determined by program objectives;
the variability of the site and the sampling design to be used will determine
the extent of the alongshore distance or width of the transect. At the site,
the general location for the transect is selected in a relatively homogenous
and representative area, and a reference line, identifying the top of the
transect is established. If the area is sufficiently homogenous, a reference
line longer than the transect width may be laid out and by the use of a
randomly picked number, a starting point for the segment to be used as the
transect may be selected. Reference lines should be placed at the extreme
upper reach of waves at high tide if possible, to minimize damage to permanent
markers. On rocky shores at the base of cliffs, a point on the extreme
backshore reach of the intertidal bench should be used.
Levelling, stratifying and marking transects.
By the use of a transit or a sighting compass and temporary range poles,
the reference line should be oriented as parallel as possible to the low
tide waterline; this will probably also be parallel to the contours of
other tidal levels and result in approximately rectangular strata within
the transect.
All sampling points on a shore must be referred to a permanent reference
point of known tidal height, located by a marker in the supratidal just above
the reference line or on the reference line. The height of the reference
point itself is preferably determined by comparing its level to a U. S. Coast
and Geodetic Survey marker, or a secondary monument established from USGS
levels, in the vicinity. Alternatively, its level may be compared to water
levels at the predicted time of low tide on calm days, with the time and
level of the tide corrected for the site from corrections given in the tide
tables issued by the U. S. National Ocean Survey. If a USGS marker is to
be used and one can be found near the site, this should be done before
transect establishment and the number designation of the marker sent to the
Survey, with a request for the height of the USGS marker with reference to
mean sea level and a correction for chart datum (local 0.0 tidal level). A
standard surveyor's tripod mounted transit, preferably of the self-levelling
type, is the best instrument for bringing the level of a distant benchmark
down to the permanent marker at the top of a site. Bring the level down to
the site with transit and levelling rod and then back to the benchmark,
to adjust for accumulated error.
22
-------
The permanent reference point for beaches and flats can be marked by
driving a metal rod or pipe several feet into the top of the beach, nearly-
level with the surface, and the top of the marker used as the reference
level. A metal or fiberglass fence post, driven in deeply with a capped
length of iron pipe of a diameter sufficient to.slip easily over the post,
can be placed a short distance from the reference marker as a witness post
for easy relocation of the marker should it become obscured or buried by
debris.
On rocky shores, markers must be affixed into the rock by some means
suitable to the rock hardness and friability. Marks made on the rock with
picks or chisels are usually initially obvious but under coastal conditions
weather into obsurity remarkably fast. Permanent markers should be installed
in holes drilled with star drills, or masonry drills and a portable self
powered electric drill, and plastic or lead lag shields used to secure stain-
less steel or bronze screws and washers in the holes. Epoxy putty which sets
under water, or concrete, may be used as an aid in securing the lag shields
in the holes. In some soft rocks which do not fracture easily, a powder
actuated stud gun can be used to shoot masonry studs into the rock, or
hardened concrete or masonry nails may be hand driven, with a perforated
metal disc threaded on to the stud or nail and held by the head. In very
hard, fracturable rock like basalt, no holes can be easily drilled nor can
studs or nails be driven in. A larger hole is made with a heavy chisel and
hammer,and cement or epoxy putty is used to fix a lag shield or large bolt
in the rough depression. All of the above methods are also used to place
subsidiary markers, such as those for tidal levels or previously used quadrat
locations, if they are to be permanently marked. Intertidal markers on any
beach type will require maintenance. Avoid placing markers on rocks in
pre-existing cracks; they are places of previous weathering and weakness.
In locations where no bench mark is available, the height of a permanent
reference point is determined relative to water level at the site. Tidal
heights vary only slightly over long distances and the tide tables have
corrections for many points along coastlines and estuaries, adjusting both
time and tide height for these points. The time and level correction for the
place nearest the study site may be used. If there is reason to believe
that this is insufficient, a site correction should be determined before
transect levelling proceeds. Local tide staffs or tide gauges are invaluable
for this but are not likely to be available at remote locations. On calm
days with minimal wind and surf, tide observations on rocky shores can be
made in meander channels freely connected to the sea where water level is the
least affected by waves. On sandy beaches of the open coast, a levelling rod
is hand held in shallow water, or a temporary tide staff is erected by
sinking a marked rod into the sand just below the level of the low tide to
be used. The midpoint between the crests and troughs.of waves where the rod
is held or fixed is estimated and this is used as tide level.
Using either natural landmarks on rocky shores or the rod method described
for sandy shores and flats, water line observations are started about half
an hour before the predicted time of low tide for the nearest places with
23
-------
predictions available, and observations continued for the same period after
the time. The actual time, and on rocky shores or fixed staffs, the level,
of the lowest point during the observations is the time and level of the
low tide for that place. This provides a specific low tide time correction
for the site, which is then verified by repeating the procedure at several
other low tides at the site. Height corrections are seldom needed, and require
that the type of observations described be repeated for a set of landmarks
or a temporary tide staff at high and low levels over a longer period. Using
the predicted tide time and level and any necessary site correction, water
level at the time of a specific low tide is now used to determine the height
of the reference point. Transit or hand level methods are used to compare
the water level reached under calm conditions at the time of low tide with
the reference point on several low tide periods. This is most conveniently
done by placing the tripod or hand level position at a midpoint on the beach
and then observing the levelling rod starting just before the predicted time
of the lowest point, to determine the height of the instrument relative to
water level. The instrument is then turned in place and the instrument height
determined relative to the reference point, yielding the height of the refer-
ence point relative to the low tide water level.
Much of the required levelling within the intertidal for laying out
transects and heights within them is adequately done with a surveyor's hand
level, either the simpler Locke type or the Abney Level-inclinometer type,
held against an extra levelling rod at eye level to avoid fatigue and uneven
footing effects. The distances involved in intertidal levelling can be kept
down to short stretches and the rapidity of hand level operation permits
accuracy to be quickly checked by repeating the levelling. A sighting, or
Brunton, compass will keep the lines sufficiently straight, if the eye will
not suffice over short distances. On rocky shores, the potential greater
accuracy of a tripod mounted transit is offset by the much greater time
required to set up and lay out a transect, because of the small horizontal
irregularities and great vertical changes over small distances as well as the
difficulty of moving the heavier, larger instrument about. On the smooth
surface of flats and beaches, the greater horizontal distances and the
efficiency of levelling many points from one position make the transit the
preferred instrument.
For all types of levelling, the horizontal distances between the points
of determined height are measured to reconstruct the slope and horizontal
distances between the known vertical intervals and so that a more accurate
estimate of the area of each stratum in the transect can be made.
Once the reference point level is determined, the transect is laid out
using a levelling rod and a transit or surveyor's hand level to establish a
center line or two side boundary lines for the transect, perpendicular to
the top reference line. Regular vertical increments, preferably of one foot
(0.3 m), to be used as strata limits, are marked along the lines with stakes
in beaches or by the rock marking methods described. If permanent markers
are not installed, the levels on rocky shores may be easily relocated by
using the reference marker to re-level the transect every sampling period,
24
-------
but profiles of rocky shores are stable and permanent markers once placed will
subsequently save time. On open coast sand beaches, strong seasonal changes
in profiles are a normal annual event and'the transect must be re-levelled
and strata width remeasured at every re-sampling period. Temporary, wooden
stakes are adequate to mark the vertical contour intervals on beaches. Flats
are more stable and while some long term changes are to be expected, those
in very protected situations may be permanently marked with deeply driven
metal rods which should not protrude above the flat higher than needed for
relocation. Permanently marked tide flat levels can be quickly checked for
change at each sampling period and complete re-levelling avoided.
Despite the known effects of wave exposure on the vertical limits of
intertidal organisms on all types of shores, the only objective physical
reference available for comparing levels on shores at different sites or at
different times, is standard tidal height, especially in- studies emphasizing
spatial and temporal variations and their causes. For this reason, the
vertical increments bounding strata in transects should be regular increments
related to the tidal reference level of 0.0 rather than the height of the
reference point itself. This will result in a standard stratification
which will be identical at all sites, with, for example, a stratum bounded
by the +2 ft. (+0.6 m) and the +1 ft. (+0.3nO above MLLW contour levels
present in each transect. Strata bounded by any arbitrary height intervals
are adequate for work within a specific transect if no comparisons are to
be made to conditions elsewhere.
Open coast beaches and many rocky shores (disregarding microrelief)
have uniform slopes for long horizontal distances along the shore, resulting
in parallel tidal contours. In these cases, stratification is easily
achieved at the time of sampling by stretching a rope or surveyor's tape in
a straight line across the transect width, between the markers for a given
tidal height. On some rocky shores, and commonly on tidal flats, the contours
at a given tidal height may meander significantly within the width of the
transect. Stratification of tidal fiats, especially in transects on the
order of 300 ft. or 100 meters wide, requires the establishment of a number
of points of the same height at each vertical interval rather than only two
on each side of the transect. Depending on the degree of irregularity,
several points of the same height should be found between the two located on
the side boundaries. These should be marked with tall temporary stakes, and
a sighting compass, transit or the eye used to establish a single straight
line, approximating their average position, across the transect and perpen-
dicular to the sides and the end positions staked on the sides. A strongly
curved stratum boundary would make the use of rectangular coordinates to
randomly locate sample points within the stratum very difficult.
Once the sides of the transect or a center line are located by marking
the strata boundaries at equal vertical intervals, the location of the sample
points requires that the dimensions of each stratum be measured. These
measurements will also permit the slope of the transect to be reconstructed
accurately. Because of variations in degree of slope down the beach to
waterline, the depth of strata will vary within the transect; upper strata
25
-------
and lower strata will be narrower than those at mid-tide level because of the
characteristic mid-tidal action of waves. Both measurements and location of
random points are done with surveyor's tapes, preferably plastic impregnated
cloth or fiberglass tapes which neither stretch nor corrode when wet with
sea water. Metal tapes are quickly destroyed on sandy and rocky shores by
abrasion and corrosion.
The linear dimensions of the strata are also used to calculate their
actual area, and this is then immediately used in the field to make the
final allocation of samples units to each stratum. Location of random points
is easily achieved by laying a tape down the longest boundary of a stratum,
and using it to locate one of the coordinates from which the second is located
by measuring with another tape or pacing at right angles to the first tape.
Only the second tape need be moved for each pair of coordinates, and the
first tape can be left in place and used for the stratum above and below it,
if it is laid on the vertical interval line between strata. For field use,
the estimated total number of sample units to be used in the transect is
used to calculate the amount of area represented by one core sample or
quadrat, for the entire transect, or for individual strata if sample size
has been independently estimated for each stratum planned. This may now be
used to allocate samples to strata on the basis of their actual area calcu-
lated from the measurements of their dimensions, as described elsewhere.
Basically, the area to be represented by one sample unit is divided into
the total area of each stratum to yield the number of points to be sampled.
The area of the usual quadrat or core used is sufficiently close to one
square foot or 0.1 square meter in transects on the order of 100 ft. or 30 m
wide, that the foot markings or the 0.1 meter intervals on standard surveyors
tapes can be directly used to locate points. The maximum length and width of
each stratum in feet or 0.1 m intervals are used as the limits when selecting
the coordinates for random point location from a random number table.
On intertidal sand and mud flats, and on some rocky intertidal benches,
certain strata, particularly the lower ones will be long and narrow in their
dimensions across the width of the transect because of the relatively steep
slope of lower intertidal areas. This arrangement maximizes the possibility
of the rare event of the chance clustering of the random points entirely
within one region or one end of the stratum. Dividing very narrow strata
down the middle and randomizing within each half minimizes this chance, with
half the sample unit allocation being assigned to each side. Alternatively,
if a set of randomly drawn coordinate pairs clusters the sample points closely,
the entire set may be rejected and another entire set of random pairs drawn
but do not attempt to "adjust" the original set. Extensive use of this
practice, which rests on a subjective judgment, will result not in randomi-
zation, but at best excess spacing of samples to some unknown but biased degree.
Retention of the coordinates used in randomization in each stratum,
and serially numbering sample units within the stratum in the order of the
use of the coordinates as well as giving each unit a stratum and transect
designation will completely locate and identify each individual sample unit.
26
-------
III. COLLECTING AND PROCESSING SAMPLES
Introduction
The most common approach to sampling of intertidal biota is recovery
of actual specimens in a quantitative sample, from which the density, biomass,
diversity, and other information, can be estimated. Since such sampling
removes a portion of the biota which are present in the field, it may be
referred to as destructive sampling, whatever the actual recovery method.
An alternative to destructive sampling is the acquisition of similar data in
the field using methods which do not disturb the community.
Two major arguments may be presented against destructive sampling and
for non-destructive sampling. The first is the fundamental one of attempting
to minimize the damage done to an area in the name of science. The second
relates to the information gained from such sampling, because repeated
sampling of an area could in many instances be biased by the effects of
previous sampling. These effects are most noticable when the species sampled
are long-lived, with poor recruitment capability, and low density. A prime
example is the removal of predatory starfish, which may have low recruitment
capacity compensated for by a long lifespan, and which have a major effect
on the community as top carnivores. Other sensitive species which may operate
as keystone species in the community are urchins and predatory gastropods.
Removal of such species from an area, or any substantial reduction in density,
can alter the rest of the community in such a manner that change due to
other events may be undetectable.
Species which have a high reproductive capacity may also be sampled
under circumstances which bias results. An example might be removal of a
set of quadrats primarily containing Barnacles. The individuals present in
a given position may represent the results-of several years changing conditions,
in a succession from early colonizing species, to species which overgrow
others, and finally to especially large individuals of those species which
reach an escape in size from their predators. Several years might elapse
before the patch created by the destructive sampling stabilizes in a form
similar to that which was removed. During this time, it would be difficult
to come to conclusions about the seasonal succession of species, recruitment
of new individuals, or other problems near the disturbed area. While the
sequence of succession varies with the community, evidence suggests that re-
turns to the original intertidal community may often require years. Monitoring
studies of a single area run the risk of distorting their own conclusions.
There are conditions, however, under which destructive sampling is not likely
to affect future sampling.
Unconsolidated substrates are characterized by frequent or continual
natural alteration of the substrate. High-energy environments are periodic-
ally shifted by increased wave energy and erosion of sediment, or decreased
energy and deposition. Low-energy environments usually experience continual
deposition, and in the case of mudflats often experience alteration of channels,
bioturbation, and other substrate disturbances. The fauna of sedimentary
27
-------
shores are usually mobile to the extent that they can reposition themselves
in the sediment, if not freely move around. Minor disturbances such as coring
will be quickly filled in by the surrounding organisms, so that the effects
of sampling are undetectable in a short time. The reduction of populations
of certain species which are slow to replace individuals can potentially affect
either a local density, or alter the community composition. It is not likely
that the typical sampling program would noticeably affect communities in
sedimentary environments.
Rocky shores, in contrast, have a stable substrate. Communities differ
greatly in their ability to fill in gaps or patches caused by sampling.
Mussels, for example, can reposition to fill a small patch. Barnacles must
recruit to replace a patch of removed individuals. In general, a more
complex assemblage of species will require a longer period to re-establish
the community pattern which was observed at the time of removal, than locations
where a monoculture dominates. Even monocultures, as stated earlier, may
be the result of a succession rather than an immediately reached pattern.
Sampling should never assume that the effects of previous sampling are
negligible, and previous quadrat positions should be marked and not be used
again except to observe the sequence of succession.
Non-destructive sampling is potentially applicable under any circum-
stances where the species to be sampled are visible, measureable, and un-
affected by the sampling procedure used. Most of the sessile macrofauna on
rocky shores may be sampled without removal, as is frequently done for
macroalgae. Quadrats are placed in the usual manner, and individuals counted
and measured in situ for some suitable dimension which can be converted into
an estimate of biomass. In some cases, it may be possible to sample some
species in sand, mud, or gravel sediments, provided that either some super-
ficial sign of individuals is available (such as openings to burrows), or
the individuals may be recovered and replaced without damage (such as cockles,
sand dollars, and other shallowly buried species).
Non-destructive sampling is inadequate to sample small and mobile species.
When diversity estimation is the major goal of the study, some destructive
sampling will be essential. Non-destructive sampling is particularly suited
to studies of the species which contribute a majority of the biomass in many
rocky intertidal communities, including barnacles, mussels, starfish, urchins,
limpets, tubeworms, macrophytes, anemones, and others. When non-destructive
sampling is used to estimate the size of populations of these species, a
minimum of destructive sampling precedes, to establish the conversion ratios
from numbers and linear size to biomass.
An important advantage of non-destructive sampling is that several of
the possible methods, particularly photography, allow very large samples to
be taken. These techniques have useful applications when the time in the
laboratory is unlimited, but that in the field is very limited by seasonal
weather, number of low tides in the time available, inaccessible sampling
locations, or briefness of the exposure of lower areas in the intertidal.
28
-------
Other methods used in non-destructive sampling involve field measurements
which may take as much or more time than sampling by removal, and their
accuracy is usually low.
Destructive sampling methods for sedimentary shores.
Coring Devices
A corer is any device which removes a quantitative volume of sediment.
Most corers are designed to do so without excavation of the interior of
the corer. In moderately wet sediments, a small corer will usually provide
enough frictional resistance to the core (the sediment trapped in the tube)
to keep the core from falling out. Larger corers may be capped or otherwise
sealed at the upper end so that the vacuum effect at the top of the core
helps to hold the core in the tube when the corer is removed. In very wet
sediment, unless the pressure at the top and bottom of the corer is equalized,
suction at the bottom of the corer will make it very difficult to remove a
core intact. Loose, dry, or large-particle sediment is not retained well
by corers and must generally be excavated from the interior of the unit. In
compacted sand, it is often helpful to have the coring tube mounted on some
sort of handle so that more pressure may be brought to^bear on the corer.
An assumption underlying the use of coring devices is that the insertion
of the corer does not disturb the sediment and the organisms sampled. The
assumption is always violated near the walls of the corer, where frictional
resistance of the sediment and corer wall distorts the sediment during both
placement and .removal of the corer. The walls of corers should be as thin
as possible for this reason, within the limits of the required strength of
the corer. Small corers will have a proportionally larger area subject to
distortion than larger corers.
During placement of corers, there is a tendency for the sediment already
in the corer to act as a plug which,forces the sediment aside. This increases
as the core lengthens and frictional forces increase. Loose, wet.sediment
is pushed aside after relatively short distances, while it .takes./longer to
displace well-compacted sediment. Wet mud or sand must either be sampled
by using a larger corer and rotating or rocking the corer into place, by
sampling with a series of successively deeper cores in the same position, or
by countering the frictional resistance of the core in some manner. Some
corers produce a mild vacuum above the core which counters the friction of
the core and helps to drive the corer in deeper. Rocking or rotating a
large corer into position can be used to take a fairly deep core, but
caution must be taken not to disturb mobile organisms which may then evade
the corer. A vacuum assisted corer, or one which takes a successively deeper
series of cores, may be more reliable in these applications.
Virtually any thin-walled strong material can be used to construct
corers. Metal is acceptable if corrosion is avoided, as corrosion increases
the frictional resistance of the metal surface. PVC or other plastics are
excellent for smaller corers, but are expensive for larger diameters. Small
29
-------
PVC tubes are slick enough so that the upper ends can usually be sealed briefly
by hand or stopper during retrieval of the core. The core is then quickly
released on removal of the stopper. Corers of any material or design should
be tested on sediment of particle size and compaction similar to that at
the actual sampling location, as a seemingly small difference in the sediment
texture can mean the difference between a working and a useless corer.
Many pollution surveys combine sampling for chemical and biological
analyses to increase efficiency of field efforts. The choice of the material
used to make corers may depend upon whether more emphasis is to be placed on
organic or aetallic pollutants. The requirements for both chemical non-
contamination of the sample and effectiveness in securing the biological
sample should be considered early in the design phase of the study. For
example, aluminum is seldom of importance as a metallic pollutant in marine
environments and aluminum tubing has a number of desirable physical character-
istics as corer material while stainless steel which has even better physical
properties may add copper, chromium or other metallic contaminants to the sample
The following describes a selection of sampling corers which illustrate
corer designs. Many other possibilities exist, and the individual worker may
choose to develop a more convenient design.
1. The simplest corer is a tube of sufficient length to penetrate the
required distance and still leave room for a good handhold. Small
corers used in firm sediment can easily be taken by hand pressure
alone, and the simplicity of such corers has the advantage of
rapid preparation and replacement of corers. The leading edge of
the tube should be beveled on the outside to cut through the
sediment without forcing sediment into the tube. Gentle pounding
can be used to drive in tubes if the sampled species are unable to
evade the disturbance.
2. A simple variation on the above design is to mount the corer on
a handle. The usual handle is a simple T shape. The handle is
hollow, and open to the top of the coring tube and to a small hole
at the top of the handle which is used to seal off the coring tube
during removal of the corer from the sediment. A crossbar at the
top of the coring tube allows foot pressure to be applied to a strong
part of the device rather than the more fragile tube itself. Some
designs permit interchangeable tubes of different size, or removal
of the handle from the tube for easier emptying of the corer.
3. A more complex design is suited for sampling in very wet sediment.
Two corers are placed side-by-side, one acting as the corer and the
other to equalize pressure. The first corer is placed into the sedi-
ment and sealed at the upper end. The second is forced into the
sediment as a thin rod, the lower end capped by a plug connected to
a rod inside the tube. When the second corer (the vent'] has reached
the same depth as the first, the inner rod is used to force the plug
30
-------
out, so that the vent is now a hollow tube extending from the surface
to the bottom of the first core. The first corer may now be with-
drawn without collapse of the walls of the hole, and without suction,
because the vent has equalized the pressure at the top and the bottom
of the first corer. After removal of the first corer, a longer tube
can be inserted into the same hole if desired. The sample is extruded
from the corer by a fitted plunger, so that it can be sectioned if
desired. The corer barrel acts as a resonator, so that frictional
binding of the core to the tube can be heard as a change in the sound
of the corer, and the corer is then stopped, removed, and a longer
corer used to continue to the maximum depth.
4. Satisfactory simple coring tubes may be made of lengths of aluminum
irrigation pipe, which is strong and has very smooth, thin walls.
Corrosion is easily prevented in this pipe by cleaning the corer
after use. Corers of this tubing, varying in diameter from three
inches (7.5 cm.) to six inches (15 cm), have been used sucessfully
in a variety of sandy and muddy sediments. The corers are constructed
from a two foot (0.6 m) to three foot (0.9 m) length of tubing, with
a plate welded across one end to close it. A braced handle of
small diameter thick walled aluminum pipe is welded across the plate
and a small hole in the plate near the handle permits air to escape
when the tube in pushed into the sediment. The tube is of sufficient
length so that the operator can insert it into the sediment from a
standing position, bringing body weight to bear on the handle by
leaning forward. The hole in the top plate is closed by thumb
pressure while withdrawing the corer from the sediment to aid in
retention of the core. Plastic tape or paint on the outside of the
barrel marks the depth the corer is to be inserted into the sediment.
Corers as large as those described above can be used to their full
length, but may lose the core if they are pulled up vertically.
The cores may slip out of the large tube because there is insufficient
suction and frictional resistance to overcome the pressure difference
from surface to the end of the corer.This difficulty can be over-
come very simply. The corer is inserted as vertically as possible,
but pushed over horizontally before being withdrawn'.* -With '"the air
hole closed, the corer is slowly pushed sideways until the sediment
breaks upward and the end is near the surface and can be grasped.
The corer is then pulled free of the hole nearly horizontally. In
packed sand, or stiff mud, a small shovel can be used to dig down
near the end of the corer, equalizing the pressure and permitting it
to be more easily tilted sideways for removal.
5. A coring type sediment sampler with a barrel diameter of about 7.5
inches (18.5 cm), and a closing plate is described by the Finnish
IBP-PM group (1969), illustrating some of the "possibilities for
the use of a large sampler based on the corer design.
31
-------
6. Smaller macrofauna and meiofauna of intertidal sedimentary shores
can be easily sampled with small piston corers made from 50 ml.
or larger plastic disposable syringes by cutting off the bottom,
or needle end of the barrel. The resulting tube is strong and thin-
walled, and the plunger is used to both aid in insertion of the
corer and to retain suction on the core as the corer is withdrawn.
The cores are kept intact, and may be slowly extruded and sectioned
for vertical distribution studies. The relative cheapness of plastic
syringes permits many corers to be available for storing cores
temporarily until processing.
7. Samples for the analysis of sediment characteristics are often
taken separately from those for organisms. Thin walled plastic
tube, used as corer liners in deep sea coring devices, in short
lengths makes useful sediment sample corers. The most convenient
sizes of this tubing is from one to two inches in diameter (2.5 to
5 cm).
8. If deeper, large macrofauna must be sampled and the substrate does
not permit a clean hole to be dug because of collapse of the wall,
a large sheet metal caisson can be used. This is any large cylindri-
cal or square container with straight sides and open ends, with handles
attached to the sides at one end so that it can be pulled out after
use. A caisson must be large to encompass enough area for an
adequate sample of larger, rarer burrowers like bivalves and thall-
asinid shrimps. It must also be large enough in the cross-sectional
dimension to permit the operator to work within it. A caisson
0.3 m to 0.5 m in diameter and 0.5 m or more long will meet both
requirements. The caisson is sunk into the substrate as far as it
will conveniently go to start the sampling, and a trowel or scoop
is used to dig out the sediment inside. The decrease in inside
friction will gradually allow the caisson to be sunk to the required
depth as excavation proceeds. This work should be done at a time when
the flat or beach has had an opportunity to drain at low tide, or the
depth in the sediment at which the caisson will fill with water faster
than it can be bailed will be too high for a deep sample to be taken.
The sediment and water within the caisson must be sieved in the field;
a coarse sieve of a mesh size of 0.5 to 1.5 cm is usually adequate.
This is a laborious procedure and the time requirements will preclude
many samples from being taken unless sufficient personnel are available.
In beaches of fine sand, the drained sand at low tide is often
sufficiently compacted so that a hole with firm walls can be dug.
The area is measured out on the surface and carefully dug out with
shovels, maintaining the wall dimensions as well as a standard depth.
In quiet embayments, diver-operated suction pumps have been used at
high tide over flats to sample a standard area to depth, within a
quadrat frame or ring, with the suspended material being passed through
a large hose to a sieve or mesh bag. Underwater holes do not collapse
32
-------
and the relative crudeness of the method is offset by its independence
from the tides and its relative speed compared to hand digging holes
at low tide, and sieving the excavated sediment.
Handling core samples in the field
Transects on both beaches and flats will require a relatively large
number of core samples, of heavy wet sediment, which must be moved about. In
order to use the available time at low tide for taking the cores, it is
impractical to sieve each core as it is taken. A practical method is as
follows. First, the beach or flat is surveyed and levelled, and the random
or systematic position of every core to be taken marked with, a light lath or
marker tall enough to be easily seen. The coring may start before the entire
beach is surveyed if there is sufficient personnel.
For efficient use of the available time at low tide a team of two persons
is required for core sampling. One person operates the coring tube, while
another holds open heavy plastic bags large enough to hold the core sample
and also fills out the remaining details on prepared labels. The corer
operator puts the end of the corer into the bag, releases the air hole and
slips the core out. The other operator inserts the label, adds formalin if
it is to be used, and ties the bag shut securely. The bagged core samples
are left next to the places they are taken from until at least the lower
strata of a transect are completed. The surveying and sample point location
team can begin to gather up the bags, carrying them in buckets to the top of
the beach, or using a light cart with bicycle wheels, transfer them to a
staging area.
Beach and tide flat sediments drain at low tide and there is virtual
immobilization of most of the species within the cores. If the cores are
kept out of the sun, most forms will survive for long periods in bagged cores.
Such cores may also be transported to the laboratory and stored under refrig-
eration for more than 24 hours in good condition.
If sieving is to be done on the beach, it may be done at high tide,
between low tide sampling work. The bagged samples may be left just at
waterline to keep them at sea temperature. Beach sieving, has the advantage
of an abundant supply of water, as does sieving on a dock. Samples accumulated
at the top of the beach may also be picked up by small boat at low tide to
minimize carrying distances.
The above presupposes that sieving will be done while the organisms are
in .good condition, and that it can be accomplished before degredation. Time
constraints may require that the sampling program complete a large amount
of coring and produce too many samples to sieve within 24 hours after they
are taken. In this case, or by choice, .the entire core sample may be preserved
intact, with formalin. For a given sieve mesh, more and smaller animals are
retained in some taxonomic groups, when sieving is done on preserved material.
Entire cores, 15 cm. in diameter and 20, cm. long can be sucessfully preserved
inside thick walled plastic bags, and after securely tying the bags closed,
the samples may be temporarily stored within the bags until they are transported
33
-------
to the laboratory for sieving. For success by this method, adequate amounts
of formalin must be added and mixed into the sediment sufficiently. The
latter is relatively easy in sand, and more difficult in mud. For fine sand,
a pore space of approximately 25% of the volume may be taken as the water
volume within the core. The core volume is calculated, an estimate made of
the pore space, and sufficient concentrated formalin solution added so that
when diluted by the pore water, the result will be a 10% formalin solution.
Do not attempt to add 10% formalin to cores; the result will be formalin of
a greater dilution and a large amount of fluid standing above the sediment in
the container, adding to the weight and hazard of handling. After tying the
bag closed, the bag is shaken to mix the contents well. If the plastic bags
develop leaks, they may be slipped into a second bag to contain the leakage.
Much of the hazard and difficulty of handling concentrated formalin in
the field can be obviated by the use of a portable garden sprayer to dispense
the formalin. A plastic or metal sprayer, with built-in pressure pump and
shoulder strap is filled with two gallons (7.6 liters) of concentrated
formalin and taken to the field closed and unpressurized. When needed, the
tank is pumped, and the formalin dispensed through the application pipe with
the atomizer removed. The core is placed into the heavy-duty plastic bag,
which is left open, the metal pipe is thrust into the middle of the core and
the valve opened, squirting formalin under pressure into the sediment and
mixing it. A previous test of the amount of formalin delivered during a time
count is adequate to assure that sufficient is used in each bag. This method
permits formalin to be used in the field without any pouring or spilling and
the formalin under pressure is discharged only within the sediment and so
does not splash out of the bag. After formalin is added to the bag, it is
tied closed, and shaken to further mix the contents. For further sealing,
electricians plastic tape may be tightly wrapped around the tied and twisted
neck of the bag.
Separating organisms from sediments
Sieving methods
The separation of the organisms contained within sediment samples must
be done with the same degree of care and precision used in sample program
planning and execution. If the program is to-quantify both the meiofaunal
component as well as that conventionally referred to as macrofauna, the
sample will have to be either double from the outset or split for processing
because of the different requirements. The macrofauna is conveniently
separated from the sediment by passing the sample through a mesh that retains
the organisms larger than the mesh size and passes the sediment and all
smaller biota through the mesh. Sieving is then a secondary subsampling
procedure for the primary sample and should be so treated.
The efficiency of this sampling is a function of the size and precision
of the sieve mesh as well as its geometry. Today there are both conventional
34
-------
meshes of woven metal or other flexible material such as nylon available and
also sheets of plastic materials with precision holes of different sizes,
spacing and density. All of the geometric parameters of the effective sieve
mesh determine retention properties and these should not be ignored, as they
affect different species according to their own shape.
Continual use gradually deforms the mesh of a sieve by:impact, clogging
and many other hazards of use; it is therefore important to start with a
good sieve and inspect its condition during the course of a long program.
Cheap substitutes for precision meshes, such as window screening may be a
necessity of the moment in emergencies, but have great variability of mesh
dimensions, distort readily and have a short life. The probability'of an
organism of a given shape and size passing through a round or square aperture
of the same maximum dimension differs slightly but probabilities for both
differ greatly from that for a rectangular mesh. Mesh geometry and its
effect on capture efficiency is discussed at length in the literature on
plankton nets.
For almost all intertidal work, sieves with a mesh size of 0.5 mm
diameter for round holes or internal side dimension for square woven mesh is
adequate to retain most of the macroorganisms. A safer method of sieve
selection for an unknown biota is to test sieves of different mesh size for
the number of species .they retain, when the preliminary sampling is done, by
nesting sieves of different mesh. The advantage of using the largest
satisfactory mesh possible is the reduction in sieving time ,that results with
using larger meshes. However, DeBovee et al. (1974) point out that the assump-
tion that a given sieve mesh size will be adequate to retain a constant
number of species in a given fauna may be violated if there are major seasonal
changes in the size of organisms. This is exactly the effect of seasonality
of reproduction, and one of the objectives of pollution related work is to
identify ,the cause of such variability. The size .class at which juveniles
enter the sample is obviously controlled by the mesh size used. If a sieve
mesh larger than 0.5 mm is used, periodic checks on the degree of retention
should be made to ensure that approximately the same percentage of total
species is retained throughout the study. Recruitment studies mayjrequire
two scales of sampling and two sieving routines, if the requirements for
adult sampling prove.very different from those for juveniles (George;1964).
The retention differences for a given sieve mesh .size between live and pre-
served animals of the same species has already been mentioned.
The most readily available sieves manufactured with precision meshes
are the standard geological sieves made principally for particle size analy-
sis and available in a graded series of mesh sizes, For strength, durability
and resistance to corrosion, the-mesh of the sieve should be matte of stain-
less steel cloth in a bronze, brass; or stainless steel body. Meshes made of
brass or bronze distort more quickly and special care must be used for the
prevention of corrosion. Both standard cylindrical 8 inch (20 cm) and 12
inch (30 cm) diameter geological sieves are available; the larger size sig-
nificantly reduces sievingitime,/ Geological sieves nest tightly and a large
35
-------
mesh sieve, with a 0.5 to 1 cm mesh size can be nested over 1 mm and 0.5 mm
mesh size sieves to catch coarse debris and rock fragments which tend to
clog the finer mesh. Sediments containing gravel may require an even coarser
top sieve to screen off the rocks and prevent them from damaging smaller
animals caught on the finer screens.
Precision mesh metal screening is available commercially and sieves can
be made of a size, shape and mesh suitable for particular uses. Metal screens
should be carefully braised or soldered into the frame so that there is no
space at the junction of screen and frame where small animals can become
wedged; even some commercial geological sieves may need to be sealed at this
place. Sieves can also be made of precision nylon monofilament cloth, mounted
in frames of plastic, for pollution studies where samples must be kept free
of heavy metal contamination prior to chemical analysis. Nylon sieves distort
readily and care must be taken not to load such sieves with heavy quantities
of sediments.
When washing sediments through sieves, do not place sediment samples in
a sieve and wash them through with a stream of water from above; this shreds
both live and preserved organisms and forces some through the mesh. Sediment
samples are either first suspended in a large volume of water which is then
flowed through a sieve, or they are placed on the sieve and washed by immersion.
Sediments can be manually suspended by gently stirring the sample into a
large container of sea water and with continued agitation, pouring the sus-
pension slowly through the sieve set to minimize the force. Mechanical means
for washing large numbers of samples are described by Birkett and Mclntyre
in Holme and Mclntyre (1971)and by Pedrick, 1974, who described a non-metallic
device suitable tor pollution studies.
Manual washing by immersion is done by placing the sample or parts of it
in the top of a nested set of sieves or a single sieve, partly immersing the
sieves in a bucket or tub of sea water almost to the rim of the sieve and
washing the sediment through with a gentle rocking and up-and-down motion.
The material retained on sieves will consist of organisms, debris, coarse
sediment, shells and with muddy sediments, small compacted lumps of mud as
well as fine particles clinging to all these materials. Do not prolong
sieving in an attempt to thoroughly clean the sample. Retain all the material
by collecting it to one side of the sieve; Hold the sieve at an angle and
dip its bottom into water at sucessively steeper angles until the sample is
all against one edge. Pushing the sieve contents across the mesh can damage
small, delicate organisms by abrasion, so transfer the sample to storage con-
tainers with minimal handling. Use a brush and a fine stream of water
directed from the back of the sieve to move the sample over the side of the
sieve into the storage container, along with the sample label. Time is saved
in field work if the containers have been previously prepared by putting a
measured amount of undiluted formalin or other preservative in them. After
the sample has been transferred, immediately add more sea water to the final
level required for proper dilution and then shake the capped container to mix
the contents
36
-------
The use of nested sieves of graded sizes prevents clogging of the lower
finer mesh, but it also results in a type of geometric size sorting of the
organisms as well as faster sieving. Keeping sieve fractions separate for
each sample may be advantageous if this size fractionation is satisfactory
for the organisms of interest, but this is at the expense of at least
doubling the number of storage containers to be handled. Sieve separation of
organisms by size is affected by the shape of the species and its effectiveness
is very different for bivalves and polychaetes for example.
Samples sieved from muddy sediments will still contain fine particles
which may later interfere with sorting and greatly increase the time needed.
The preserved sample may be more thoroughly washed later by manually washing
it on a sieve finer than that used initially to ensure against .sample loss.
A more automatic washing apparatus will further save time; one is described
by Worswick and Harbour (1974).
Separating Organisms from Debris-Flotation and Staining
One of the better rapid methods of separation of small organisms from
debris in either field or laboratory is to float the lighter organisms free
of heavier debris using a medium of suitably high density. Among the media
suggested have been saturated solutions of sugars (Birkeland et al. 1976),
calcium chloride (Coleman and Hynes, 1970), magnesium chloride (Hummon, 1974),
zinc chloride (Mattheisen, 1960), and carbon tetrachloride (Whitehouse and
Lewis, 1966).
Small organisms can be floated free of shell and other debris after
collection, but the same essential method may also be applied as a sampling
method in the field. Ertl (1970) suggested a device which permitted the
retention of a pool of a suitable high-density medium on rough rocky surfaces.
Two metal or plastic cylinders are nested, the inner one about 2 cm smaller
in diameter and somewhat higher. The space between the two cylinders is
filled with Plasticine or some similar pliable material, which when pressed
against the rough surface seals the inner cylinder. Ertl then suggested
partial filling of the inner cylinder with water and pipette decanting of the
scraped organisms. The same device could be used to retain a higher density
medium which would float free small mobile organisms which can easily escape
normal quadrats.
Flotation methods are especially useful when the difference between the
density of the organisms and the debris is great, such as in sand with a
high shell content.
A second method'for separation of organisms from debris is staining.
Williams and Williams (1973)' examined the efficiency of many stains in sort-
ing of benthic marine samples, including gentian violet, eosin, fluorescein,
malachite green, iodine, fuchsin, methylene blue, rhodamine B, carmine, lignin
pink, rose bengal, and Lugol's iodine. Of all of these, they found that
rose bengal and Lugol's iodine were the best standard stains. Both are taken
up by the animals and the debris, if it was organic in nature, but animals
37
-------
concentrate rose bengal to a darker color than does organic debris, and
concentrate Lugol's iodine to a stronger orange color. They also tested
the effectiveness of stains which provide a dark background (counterstains)
and found that chlorazol black E provided the best counterstain for contrast
with the above primary stains. Suggested concentrations for the different
stains were 200 mg/liter for xanthene rose bengal, 1 gm/100 ml water of 70%
ethanol for chlorazol black E, and 1 gra iodine plus 2 gm potassium iodide
per 100 ml for Lugol's iodine. If water solutions are used, the stains
should be dissolved in distilled water. Primary stains should be added to
benthic samples in a ratio of about one part stain to four parts solid sample
and let stand for at least 24 hours. The counterstain is added to cover the
drained sample for only 30 seconds, then the sample is rinsed and sorted.
The primary stain can be left in the preservative solution for several weeks,
but the counterstain leaches out of the sample into any large volume of
liquid and restaining would be required.
Destructive Sampling - Trenching and Other Linear Sampling
Trenching provides a quantitative sample consisting of a series of
contiguous sample units taken in a line. It can be used only when the con-
sistency of the sediment permits excavation of a sample unit without physical
retention of the sediment walls, and is therefore primarily applied to
compacted, well-drained sand. The individual units are often rectangular
instead of square. There are two major disadvantages of trenching as opposed
to standard quadrat sampling. First, the amount of sediment removed is
considerable, and sieving must take place at the site. Quantitative sieving
of large samples requires considerable effort and manpower. Second, a
linear sample does not examine the heterogeneity of the beach parallel to the
waterline. Finally, a long, rectangular sample unit tends to obscure or
average discontinuities in the long-axis direction, so that sharp discontinui-
ties on sandy beaches would not usually be detected. The major advantage
of trench sampling is the ease of selecting sampling units (a trench is
essentially a systematic sample in which all units are chosen), and to some extent
the large size of the sample, and the greater depth that can be sampled into
the sediment.
A line sample in which all organisms within a certain distance of the
line are collected or counted is the rocky intertidal equivalent of the
trench sample. Again, horizontal heterogeneity is not examined by this
method except in the direction of the line.
In some cases, trenching or line sampling may be applied so that randomly
selected parts of each stratum are sampled. This is essentially a linear
random sample where the sample units are rectangular, and has the same
characteristics of any other linear random sample. The only difference is
the tendency to obscure discontinuities in the direction of the long
dimension.
38
-------
Destructive Sampling Devices - Rocky Shores
The unit of sampling on the surface of rocky shores is termed a quadrat,
a square or rectangular plot of any size, in which sample removal or in-situ
measurements are made. Most destructive sampling on rocky shores can readily
be done by using a quadrat frame to delimit the area to be sampled. While
the size and shape varies, frames are made so that they bound the entire
area to be sampled or have subdivision so that subsampling within the frame
is facilitated.
The most common quadrat frame used is a rigid square with flat sides of
wood or metal, enclosing a known area such as 0.25 m . The sides are
graduated at some convenient interval or string or wire is extended from the
sides in a grid which subdivides the quadrat. Quadrat'frames are further
described under photographic methods. The quadrat frame is laid on the
surface after the sampling point has been located by methods described else-
where. A uniform method of placement should be used; the random or systematic
point location method locates only a point. The frame can be centered on the
point, or one corner can be placed over the point, with the sides of the
frame oriented in a standard manner to the transect or stratum boundaries.
Sampling usually combines in-situ counts of subsamples of organisms badly
damaged by removal, such as barnacles and estimates of plant cover, with
total removal of all organisms from the entire quadrat or portions thereof.
It is highly recommended that all destructively sampled quadrats be photo-
graphed first by the methods described elsewhere, so that the intact appear-
ance is recorded and so that correlations can be established between removal
data and non-destructive sampling.
Organisms are removed from the rock surface by careful scraping with
strong knife blades or stiff putty knives with the ends bluntly rounded or
pointed and the edges ground into a bevel. Paint scraping devices, with
short blades bent at right angles tend to damage organisms more than longer
blades which can be manipulated easier. Algae with holdfasts inside the
quadrat are usually removed intact, but some standard, if arbitrary procedure
must be followed for the many organisms of great variety which are overlapped
by the frame edges. This is discussed further under photographic sampling.
All material removed from the quadrat constitutes the sample, which is most
conveniently handled by being placed with its label inside a plastic bag
which is securely tied closed. Intertidal biota of rocky shores can be
transported alive and stored in plastic bags under refrigeration of even deep
frozen until further processing can take place. Handling rigid containers or
formalin preservatives not only consumes sampling time on the shore but adds
to the hazards of work on slippery rocky surfaces.
Sampling in thick algal mats or mixtures of sessile animals and algae,
and sampling multilayered mussle beds requires special devices. The three-
dimensional network of spaces present in algal mats and mussle beds usually
has an abundant and diverse fauna, many of which are small and motile. If a
flat frame is placed on top of such a mat and then scraping is attempted, many
of these smaller motile forms will quickly escape. A partial solution is the
39
-------
use of a quadrat frame of metal, with vertical walls or sides and open at
top and bottom. Such a frame, either circular or square, can be pressed
down tightly against the mat, or forced into mussel beds to minimize the
escapement. The frame may also be of wood, or of metal, with a flat flange
on the lower side to which a gasket of flexible material such as foam neo-
prene rubber or foam plastic is attached, to improve the seal when the frame
is pressed down on algal mats or barnacle beds. A variant of this type of
sampling is to use a frame with such a gasket in conjunction with an agent
to kill or immobilize the small animals within. Glynn (1965) used such a
frame on the upper algal-barnacle mat on rocky shores, and improved the
retention of mobile forms by pouring hot sea water formalin solution into
the frame after it was placed. The hot formalin was carried to the site in
a thermos bottle, presumably one of steel rather than glass. While hot
formalin is effective, the resulting vapors may constitute both a hazard
and a significant time delay while the sample is scraped from the rock. Other
agents, with less irritating vapor would improve this method; alcohol,
acetone, or a mixture of either and dry ice (solid carbon dioxide) are
potentially more advantageous, as would be pressurized freon or ethylene
chloride, which would freeze or kill organisms by chilling them. Agents
such as diethyl ether might be effective by both chilling and narcotization.
Rounded or extremely irregular rock surfaces present special problems
also, if they cannot be avoided. Severe parallax problems would result in
trying to project a rigid, flat frame boundary onto a rounded surface, and
then making decisions as to what to sample beneath it. A flexible quadrat
frame can be used to avoid these problems as much as possible. A frame
made of a grid or network of rope or other flexible, relatively inelastic
material can be carefully applied to curved surfaces to minimize the area
distortion, and a satisfactory delimitation of the quadrat area obtained. On
extremely irregular surfaces the flexible quadrat frame may be quantitatively
superior to a rigid frame, because projection of the rigid frame boundary onto
the rounded surface could result in even more distortion of the area boundaries.
Sample Preservation and Storage
Samples from both sedimentary and rocky intertidal areas are usually
placed directly into preservatives soon after sieving or collecting. However,
in pollution related work where chemical analysis is also required, this may
not be possible or desirable. Such samples can be deep frozen until proces-
sing, at which time all material intended for chemical analysis can be
separated from the rest which should be placed in preservative as soon as it
has thawed.
Some advantages can be gained by narcotizing live material first before
preservation. If rocky shore samples are transported alive from the shore
or conditions permit live sieving of sediment samples, better specimens for
length measurements and for taxonomic analysis will result from narcotization.
Polychaetes usually constitute a large proportion of the macroinfauna from
sedimentary shores and can be abundant in some rocky shore samples. These
worms are much easier to work with if they have been narcotized so that they
preserve in a relaxed, undistorted state. Place the samples for 30 minutes
40
-------
to one hour in a solution of 7.5 percent magnesium chloride in fresh water,
or in 0.15 percent propylene phenoxetol in sea water (McKay and Hartzband,
1970), at a temperature near that of the environment.
Samples are usually preserved in 5 to 10 percent formalin in sea water,
and stored in 5 percent formalin or alcohol solutions. The undiluted stock
formalin is usually buffered with an excess of borax, calcium carbonate,
sodium acetate or hexamine to prevent the final solution from becoming
acidic. Gage (1977) has described an improved storage fluid for use after
fixation in 4 percent buffered formalin, consisting of 2.5% formalin, 1%
propylene phenoxetol, 10% propylene glycol and 86.5% deionized water, buffered
at pH 7.5. A conventional solution for storage after fixation in 10% sea
water formalin is 5% glycerine in 70% alcohol, either ethanol, methanol or
isopropanol. Large bodied species will significantly dilute the original
solution in which they are preserved and if formalin is the original preserva-
tive, such samples must be transferred to fresh solution for long term storage.
The determination of weights from preserved samples is significantly affected
by the extraction of soluble materials by both alcohol and formalin solutions.
At best, the storage time should be standardized to reduce the variation from
this source, but better data will be derived from fresh or frozen samples
taken for this purpose, as described elsewhere.
Algae in quantitative samples are well-preserved'by formalin-sea water
solutions initially, but in time colloids are extracted and the samples will
bleach if they are exposed to light. Algae in formalin kept in the dark are
satisfactorily preserved until they can be processed. Permananet voucher
specimens may be prepared from such formalin preserved algae by floating
them in water onto standard herbarium sheets in the usual way.
Measurements Derived from Destructive Samples
The minimum necessary data to be taken from each sample are the species
present, their relative abundances, the biomass or an estimate derived from
conversion factors using some measurement, and the variance of the abundance
of the species. Taxonomic procedures for species identification will not be
discussed here.
Biomass may be measured by the wet weight, the dry weight, the ash-free
dry weight, the carbon weight, or the caloric content,., The units chosen
should reflect the importance of the quantity of organic material to the
rest of the community, and ideally would be expressed in terms of the calories
which would be gained if the organism were eaten. In practice, it is generally
impossible to express the biomass with this level of refinement, and either
the ash-free dry weight or the dry weight are used..; The dry weight is the
weight of the tissue remaining after all water has been driven off.The
ash-free weight is that weight minus the weight of the inorganic ash which
remains after the tissue has been burned. Dry,weight may be converted to
ash-free dry weight if a conversion factor is calculated, but differences in
41
-------
the organic content with season, reproductive state, or the health of the
individual will introduce errors in the conversion. In general, the wet
weight of live or dead organisms gives undue emphasis to species which have
a high water content but little usable tissue, so wet weight should always
be converted to equivalent dry weight or ash-free dry weight. Wet weight
should not be used for species which have a variable amount of water retention,
such as anemones.
The wet weight is determined after sorting to species by blotting off
excess surface water and weighing. Dry weight is determined by drying the
species samples separately at about 70°C until successive weighing produces
the same weight of the sample. The ash-free weight of the sample is determined
by burning the sample in a muffle furnace at 500 C for 12 to 24 hours, weighing
the ash, and subtracting the remaining weight of ash from the weight of the
dried sample. This presumably represents the organic portion of the sample.
Note that organisms which have a large inorganic fraction will be given undue
emphasis by wet or dry weights. Shell should be removed if wet or dry weight
is to be used, as the shell contributes little to the higher trophic levels.
Samples should be weighed as soon as possible after preservation, as
lipids and other organic materials leach into the preservative and would
decrease the weight of organics in the sample. In many cases, it is better
to estimate the ash-free weight from conversion factors for numbers or meas-
urements to biomass, when the determination of ash-free weight would take too
much time and samples would have to remain unweighed for extensive periods.
These conversions should be calculated from freshly collected specimens for
the best estimate of a suitable conversion factor.
The species list, relative abundances, variances, and biomass of species
are sufficient data for most purposes. Various methods using this type of
information may be used to evaluate differences between sites or over time
at one site. In addition to these, the size-frequency data for the species
sampled can be highly informative. Individuals of all species where it is
practical to measure size are measured, and the frequency of intervals of
size are calculated. The smallest sizes are the newest members in virtually
all organisms, so that the size-frequency curves of a species allow estimation
of the recruitment to the species at the sampling site. This method is easier
to apply to species which have recognizable stages in their life histories,
and the frequency of individuals of each stage can be counted. In some species
it is possible to observe a seasonal peak in young individuals, and to follow
these through time, observing both the mortality of young and the increase in.
their biomass.
Non-destructive Sampling Methods
Percent Cover on Rocky Shores; Estimation in-situ
A standard method for the estimation of density of macroalgae is the per-
cent of surface area occupied by each species. The algal cover is usually
42
-------
divided into three layers: the canopy, understory, and primary layers.
Primary algae are closely associated with the surface, usually encrusting or
covering surface area to the exclusion of other species. Understory species
extend above the surface but are relatively short, and canopy species are
relatively long and overlie the understory and primary layers. Each of these
layers is considered separately in evaluating the percent cover of species.
The standard method for estimating the percent cover is:
1. A quadrat frame is placed in a selected position according to the
sampling design.
2. Within the quadrat frame, the percent cover of the canopy, understory,
and primary layers is estimated.
a. The percent cover of the canopy layer is estimated first. The
individual worker must decide whether to include or exclude
plants which hang over in the quadrat but have their point of
attachment outside the quadrat. Within each of the subdivisions
of the quadrat frame, the percent of the area covered by each
species of the canopy is estimated by eye. The average of these
subarea estimates is the percent cover estimate for that quadrat,
for each species in the canopy. An alternative method is to
place a transparent plastic sheet over the area, which has 100
or more randomly placed points marked. The alga under each
point is noted, and the percent of the total number of points
which each species is found under is the estimate of the percent
cover for that quadrat. The variance of point-counting is covered
elsewhere.
b. The percent covers of understory species are estimated next.
The canopy species can be removed and returned for estimation of
biomass in the lab, or they can be shifted aside. Understory
species are evaluated for their percent cover in the same manner
as canopy species. If the presence of canopy species affects
the position of understory species, then the positions of the
dots used in the second method above would possibly bias the
estimate of percent cover of primary and understory species,
if the same random set of points were used in all three layers.
If the random-dot method is used, a new set of point's should be
used for each layer. Rotation and reversal of the plastic sheet
provides eight possible sets of random points without,generation
of new sets of random numbers or making new sheets.
c. The process is repeated for the primary layer species, using
either method for percent cover estimation. The same method
should be used for all three layers.
In most studies percent cover estimates are treated as abundance estimates,
and the actual number of individuals which contribute to the percent cover
43
-------
in a given area is not estimated. Algal individuals are frequently difficult
to define, especially where the species is encrusting and individuals are
fused. Biomass is usually estimated in addition to the percent area covered.
The estimate of numerical abundance or of percent cover does not give infor-
mation on the amount of organic material bound in the populations of each
species, available to higher trophic levels as food. Algae may be thin and
broad, so that a small biomass corresponds to a large percent cover; or may
be thick and compact, so that the same biomass is present in a very small
area. Percent cover, like numerical abundance, is more suited for comparison
of the abundance of one species at several sites than for comparison of the
abundance of several species. Biomass, in contrast, is useful for either
sort of comparison. If species-specific in-situ surface area, or percent
cover to biomass conversions are determined from samples first estimated for
cover in the field, then removed for laboratory determination of the weight
of each species, the percent cover estimates are convertible to biomass equiv-
alents.
Counts and Measurements in-situ on Rocky Shores
Many species of potential importance are easily measured for size in the
field. Size measurements may be readily converted into estimates of biomass
if the relationship of linear size and biomass is reasonably consistent.
Species of this sort can usually also be counted in the field, since they
are large enough to be identified individually. Examples are barnacles
(see Conixell, 1961), mussels, anemones, starfish., and urchins. The process
of counting and measurement in the field is more time-consuming than removal,
so that application is limited to situations where there is sufficient time
or manpower to gather all the information over one or two low-tide periods.
When a linear measurement is to be used for conversion to biomass, some
experimentation with different dimensions will be necessary, in order to
select some dimension which is the most highly correlated with the biomass.
In many cases, previous work can be referred to for a given species. Destruct-
ive sampling to establish the conversion to biomass should always accompany
in-situ counts and measurements of size, even when previous work has established
conversions in other locations. Populations of the same species may differ
in their size-biomass relationship. The conversion factors should also be
checked during the study to ensure that the conversion factor has not altered
with seasonal changes in the population.
In-situ counts will not be appropriate for all species present at any
given site. If diversity is to be used as a major tool for detection of
changes in sites, destructive sampling must be used in order to sample the
small and mobile species, which cannot be counted or measured in the field.
Species for which in-situ counts and measurement are feasible may still be
sampled in this manner. The advantage to partial in-situ sampling is that the
larger species which may most readily be counted in the field .are often rare
or slow to reproduce, so that destructive sampling is more damaging to the
local population.
44
-------
It is important to make provisions for comparisons of studies done using
different measurement methods. For this reason, non-destructive sampling
must always be accompanied by a minimum of destructive sampling to establish
conversion factors for percent cover to biomass and numbers and linear size
to biomass. Destructive sampling for this purpose should be complete enough
so that the conversions can be specified with a known error for the entire
range of percent cover, numbers, or sizes found in samples. Even if the
biomass need not be known for the particular study using percent cover, in-situ
counts or measurements, the study will have lost much of its effectiveness if
it cannot be compared to other programs.
Photographic Methods
Introduction
When quantitative non-destructive and periodically repeated sampling of
the macroepibiota of rocky shores is required, photogrammetric techniques
offer many advantages over direct manual in-situ identification and enumeration.
Photography can also be usefully combined with destructive sampling to record
features of intact quadrats before removal of organisms. Undisturbed areas
reserved for photographic monitoring can be laid out parallel to transects
or areas where sampling with removal techniques is used. Photographic methods
offer particular advantages of speed to emergency sampling programs initiated
with short lead time in cases of catastrophic pollution events. When there
is no accompanying removal sampling program which would result in the retention
of specimens for identification and mensuration, voucher specimens should be
taken from the general site area, but away from the monitoring stations, for
taxonomic determination, size and wet, dry and ash free dry weight determinations,
for photograph calibration.
Major advantages of photographic methods are:
1. Minimal disturbance of the sampling area in monitoring indicator
sites over time.
2. Speed of sampling, resulting in either reduction of sampling time or
maximization of the number of quadrats sampled in the time available
at low tide, or before some impending event like the beaching of
an oil spill.
3. The photographic quadrat or sample may be very large compared with
those feasible by manual removal methods.
4. Avoidance of operator errors in data from in-situ counts, resulting
from: a. counting difficulties induced by the interference of bad
weather, b. inaccuracies resulting from fatigue,; distraction of
attention by the necessity for avoiding wetting and danger from heavy
surf and from the need for hurrying because of time constraints at
45
-------
low tide, c. data bias from operator decisions when parallax
problems occur with small organisms near the edge of counting frames
or their subdivisions.
The following types of data for macroepibenthic biota on rocky shores may
be obtained by photographic quadrat sampling.
1. The species present.
2. Density (frequency) per quadrat area for each species.
3. Percentage surface area of the substrate covered by algae and sessile
animals, including multi-layered algal assemblages.
4. Size-frequency per unit area data for each species with regular
morphological features measurable in linear dimensions such as
diameter or length.
5. By comparing data from photographic samples from different equivalent
places, or different times from the same place, estimates of spatial
and temporal variation in the above features.
6. When appropriate conversion factors have been determined for cover
or linear dimensions to weight or caloric content, measures and
comparisons of biomass weight per unit area and its changes with
time may also be estimated from data taken from scaled photographs,
without disturbing the sampling site.
The accuracy of a measured frequency of a species is dependent upon the
quadrat size used in sampling, the nature of the dispersion pattern of each
species and the size of individuals of each species. The ease of both
replication of quadrats of a given size and of either varying quadrat size or
producing a mosaic coverage of a larger area should allow photographic quadrat
sampling to be developed into a superior method for accurately determining
the degree of natural spatial variability in frequency related to dispersion
patterns.
Using the same equipment, photographic methods are adaptable to many
scales of area from quadrats on the order of a square meter for enumeration
of large biota to areas of 100 square centimeters for counting recently
settled barnacles. Photographic quadrat sampling can also be combined with
general photographic recording of landmark features of the site for later
relocation without altering the camera equipment.
Disadvantages of the photographic method are:
1. Identification may be a problem with similar species not distinguish-
able by external gross features if they co-occur at the site.
2. Minute species or those which hide under larger surface forms will
be missed in quadrats scaled for larger surface species.
46
-------
3. Enumeration or percent cover estimates for overlapped species in
thick layers is only possible by sequential photographs taken as
successive layers are lifted or removed.
4. Initial equipment purchase costs and the operating costs of film
purchase and processing may be greater than the costs of manual
in-situ identification and enumeration or of removal sampling and
the subsequent analysis of preserved samples.
Equipment and Materials Required
Many types and sizes of cameras can meet the few performance requirements
for field use in photographic quadrat sampling. Cameras must be easily carried
over rough and slippery rocks, readily operated when hand held and withstand
damp and salty field conditions without complicated protection and maintanence.
Large tripod mounted cameras lose many of the advantages of the method and
smaller hand held reflex cameras are generally more suitable. A major require-
ment is high quality bf the final image and hence high quality camera lenses
which can operate over a wide range of distances. The great variety of
35 mm single lens reflex cameras, with their "normal" or "normal to close up"
lenses are well suited to this work, and subsequent discussion implies their
use.
Two cameras mounted on a frame have been used for stereophotographic
methods or for simultaneous use of regular color and infra-red color film,
for detecting small algae and changes in the condition of algae. A single
camera is adequate for all other uses. Both color transparency film and
subsequent processing of good quality is essential for the best use of this
method. Color produces the contrast and definition required to identify
the image of many intertidal species and to distinguish individuals in groups
of organisms or from dark rock backgrounds. Color transparencies are readily
projected at variable sizes, which facilitates taking data from the film. For
some special uses, such as monitoring barnacle populations, color or black
and white negative film may be adequate or even advantageous. The main
advantage of prints, where negative film is applicable, is that they are
permanent records accessible without further equipment. The disadvantages
are that ordinary color negative film and printing results in an image some-
what lacking in resolution while bla.ck and white prints may lack the visual
contrast and definition required for dark, monochromatic species. The time
and expense required for print production may offer no cost advantages over
transparency material in the 35 mm film size. Prints of adequate tonal range
would lend themselves to analysis by automatic image scanning equipment, a
procedure not yet applied in this field.
Electronic flash equipment is essential for adequate photography in the
rocky intertidal. Even in good daylight, flash is_required to eliminate
strong shadows and uneven lighting which degrades image definition. Light-
weight, easily carried automatic flash equipment with a high light output
and capacity for sustained use is preferred. Rocky shore organisms are often
47
-------
dark in color and contrast poorly; an adequate exposure on color transparency
film may require a flash duration or intensity two or three times that
adequate for conventional photographic subjects, especially when the rocky
background is also dark. Consequently, calibration of the flash equipment
under field conditions is necessary, even for automatic electronic flash
units. Electronic flash units with automatic flash duration or intensity
control are preferred to simplify operation in the field where distances
will constantly vary between camera and rock surface. Such camera to subject
distances will vary from about two meters for large quadrats suitable for
counts of starfish, large algae, large sea anemonies and the like, to a half
meter or less for small quadrats taken at close distances for enumeration
of barnacles and other small organisms. Electronic flash units attached to
the camera by an adequate length of cable produce successful exposures in
deep tide pools. The flash unit is held to one side, close to the water
and at a low angle of incidence to the camera.
Photographic sampling is most effective when done with a gridded sampling
frame to scale and define the quadrat space. No records of camera to surface
distance or special apparatus for scaling, holding the camera normal to the
surface or for corrections for scale and perspective on projection are
necessary when frames are used. The frame grid provides an automatic linear
scale in two dimensions as well as correcting area for changes in perspective
across the field of view. The optical superposition of grid and organism
images resolves all parallax problems.
Light quadrat frames can be made of welded rigid, flat aluminum stock,
with holes bored along the sides at carefully regulated distances for the
grid strings. Frames should be painted a bright color for contrast, such
as yellow which will contrast strongly with the background but which will
minimize glare on projection. Frames painted white will be so overexposed
at exposures required for the organisms that their projected brilliant image
will interfere with the vision of the enumerator.
The grid of the frame is produced by tightly stringing the frame with
fine, strong white, yellow or orange nylon or polypropylene string, which will
stand out sharply from the background. The thickness of the string for a
given frame size should be fine enough so that its image cannot obscure the
smaller objects of interest. The size of the quadrat frame and its grid
spacing varies with the intended use. We have found use for three frame
sizes, but a single frame will suffice for most work over a large range of
quadrat sizes. For use with a 35 mm camera, frames with side dimensions In
a 1:1.5 ratio will most efficiently fill the camera film frame. The consistent
use of an orientation mark on the frame edge to denote upward or seaward
direction permits the camera to be held with its long frame axis horizontally
or vertically as required for positioning for a particular quadrat. Frame
sizes proven useful for quadrats varying in area over several orders of
magnitude have been one meter square, strung at either 10 ox 20 cm intervals,
a rectangular frame, 60 cm by 90 cm, with a 10 by 10 cm grid and a small
frame with a fine string grid of 3 by 3 cm. The entire frame or a portion is used
depending on requirements.
48
-------
The primary data from photographic quadrats on 35 mm color transparency
film is obtained by making counts, identifications and measurements from the
projected image of the slides. A good high intensity projector and a high
reflectance screen is required for ease in working with large numbers of
photographic quadrats. Zoom projection lenses minimize changes in projector
to screen distances when adjustment between slides is necessary to produce
an easily used image size. Counting and measuring from the projected image
requires no special equipment. Hand count recorders permit the operator to
record without looking away from the image. White rulers with sharp graduations
or dividers are used for measurements. Counting may be facilitated by moving
the shadow of an object across the screen as counting proceeds. Counts may
be recorded for each grid unit on a numbered diagram of the frame grid. For
large quadrats photographed on near-vertical walls significant zonation changes
will be present over distances of a few tenths of a meter. Data from different
vertical horizons can be easily kept separate on such record diagrams, and
horizontal grid rows can be treated separately in data analysis. Photographic
transects of rock walls, made by successive quadrats placed at measured levels,
are very quick and efficient.
A quantitative photographic method more advanced than that described here
is available. Lundalv (1971) and Torlegard and Lundalv (1974) describe a
system developed for quantitative studies of the long term changes, including
pollution effects, on the epibiota of rock surfaces along the Swedish west
coast. A single camera and a reference frame are used to make stereopairs of
photographs which are then analyzed for all the types of data mentioned above.
Stereoscopy increases image definition and a microstereocomparator available
commercially is used for size measurements of great accuracy. As described,
the system was intended for underwater use in the shallow subtidal, but is
readily adaptable to intertidal use by simplification of the equipment modifi-
cations made for underwater use.
When long term monitoring programs or baseline programs with extensive
geographical coverage are being planned, evaluations should be made of the
cost and relative advantages of conventional sampling and manual in-situ data
gathering and of photographic techniques as the major sampling method to be
employed on rocky shores.
Photographic Procedures in the Field
The basic techniques for vertical level determination, establishment of
transects, stratification, location of sampling points and determination of
sample number are unchanged for photographic quadrat sampling. Camera distance
is varied to produce adequate size definition of the smallest organisms of
interest as well as to enclose the required quadrat^ area within the camera film
frame. Conflict in these requirements may mean that the over-riding requirement
of adequate image size will require a given quadrat to be covered by more than
one film frame.
The ease with which scale can be changed in photography permits monitoring
sites to be initially mapped in detail as a base for later periodic sampling
to follow natural or induced ecological change.
49
-------
For detailed transect coverage, a nested sampling technique may be used
by progressively photographing an area covered by a square meter frame for
larger, rarer species, and then by moving closer, sucessively smaller sections
of the grid, systematically or randomly selected, can be photographed for the
enumeration of smaller, abundant species. Repeating this coverage successively
at determined tidal heights will quickly produce a detailed sampling of the
transect designated.
Photographic coverage is so quick and inclusive of greater area that
correct ji priori estimation of the minimum number of samples necessary for
a given level of precision becomes much less important than when quadrats are
enumerated in the field. In practice, oversampling with subsarapling
later as required becomes a feasible routine.
After the quadrat location is determined, the sampling frame is placed
by some convention such as centering it or placing one corner on the location
point, with sides oriented to some constant feature such as the transect
boundaries, which are usually arranged to be as close to right angles to the
waterline as possible. The frame should be held as parallel to the rock face
as possible. Commercial rubber cords with hooks or lengths of rubber tubing
with a hook made of wire attached to each end can be hooked to the frame
edge and to irregularities in the rock face. This will temporarily hold
the frame in place and at the right position and permit one operator to both
place the frame and to operate the camera. If multiple layers of algae or
animals are present, sucessive layer removal and repeated photographs are
necessary.
Camera operation will not be described since it is so dependent upon
specific design. The use of electronic flash permits smaller lens openings,
resulting in greater depth of field and sharper images, as well as simplified
operation under adverse conditions. The operator must have previously
mastered the operation of the camera and flash equipment. We have successfully
obtained photographic quadrat samples from previously determined locations
at night by using a bright lantern held close to the quadrat for focusing
and thereafter moving the lantern well to the side to avoid glare in the
camera lens during exposure.
Record Keeping
In addition to the usual records of location, level and sample designation,
records must be made of each photograph. The sequential frame numbering on
roll film simplifies record keeping, but rolls must be carefully identified in
the field as they are used. It may be sufficient to number the film rolls
and record data for each film frame in each toll in the field as they are
exposed. Commercial color film processing usually permits individual film
rolls to be given an identifying number which is retained with the packaged
slides when they are returned. When general orientation photographs of the
sample site or region of a transect are included in each roll or on some rolls
in a group used at one time, landmarks permit the location of at least larger
photographic quadrats to be identified easily. This provides a means for
quickly relocating them in the field when resampling the same site.
50
-------
Each photographic sample may be labelled uniquely, at the time the ex-
posure is made. Pre-prepared paper card labels bearing an identifying sample
serial number or letter-number designator written clearly and large enough
to be read at the scale to be used can be placed within the field of view,
in a corner of the quadrat or on the edge of the sampling frame. If sample
numbers cannot be assigned beforehand, or quadrat sizes and hence scales are
planned to vary widely, designations can be assigned in the field and marked
on stiff cards or pieces of reusable plastic sheet with a wax glass-marking
pencil and similarly placed within the frame.
Obtaining Data from Photographic Samples
The processed color transparency is the permanent sample from which the
data is derived. The slide, with the image of the frame edges and its grid
superimposed upon that of organisms and background, is projected at any size
which permits ready identification and measuring, with sufficient image
brightness. The grid and frame provide a direct measure of surface area in
the quadrat and its subdivisions as well as a linear scale in two dimensions
for calibrating any size measurements taken from the projected image.
For photographs taken relatively close to near vertical or near hori-
zontal rock surfaces, where the camera can be held centered on the frame and
normal to the surface, corrections in size measurements for differences in
near and far dimensions due to perspective, is usually not necessary. The
camera cannot always be positioned over or opposite the center of the sample
area in some places on vertical walls and on the horizontal, especially when
the quadrat is on the order of a meter square. Appreciable perspective
effects will be apparent when the slides are projected, and projector manipu-
lation usually results in poor focus of one or the other end of the image.
Perspective differences will have no effect on counts made within grid units
and no correction is needed for differences in the image size of grid units
of equal actual area. When sizes are being measured from the projected
image of quadrats with appreciable perspective effect, corrections for successive
distances away from camera center raay be taken for each row of grid units,
from the projected dimensions of the grid sides at those distances.
Edge Effects in Counts
Whether entire photographs are used or areas within them bounded by parts
of the frame grid, an edge effect on the frequency per unit area data is
produced by operator decisions on the inclusion or exclusion in the count for
the unit.of individual organisms near the periphery of the unit. This effect
is greater for the rectangular 35 mm film frame than from an equal area
circular or square frame or portion of the photograph simply because of the
differences in the length of the edge relative to the enclosed area, as it
is for any rectangular quadrat. Simple solutions to this problem, uniformly
applied, reduce its significance greatly.
A decision can be made as to the proportion of an organism which must be
51
-------
within the area to be scored as included. For species with regular shapes,
exclusion of all individuals overlapped by a boundary by more than 50% of
the dimension used may be an adequate rule. This method requires individual
decisions, possibly subjective and liable to bias, and applies poorly to
species of indefinite shape and boundary such as some macroalgae.
Other solutions are applicable to both counting and point counting cases
where points fall so close to the edge of the projected image of an organism
that it cannot be decided if they are inside or outside because of image un-
sharpness. In point counting, these cases may all be scored as one-half to
avoid subjective bias. In frame or grid subdivision counts for density
estimates, all cases of actual or unresolved intersection with the frame border
or a grid subdivision line can be counted separately and only one-half of these
intersect- counts later added to the non-intersect count total. For density
counting, another method avoiding subjective decisions on the degree of over-
lap required to be inside the count area is to omit counts of all individuals
touched by two adjacent edges or boundaries of the frame or grid units and
include all others touched by the other two adjacent boundaries.
Data on Species Presence, Density and Sizes
Species identification, counts and measurements may be taken for all
organisms of interest in the entire quadrat area photographed. Preliminary
analysis may define the size of a subsample smaller in area than the total
quadrat; selection of a subsample of the grid units by systematic or random
means will produce the subsample, but subsampling must not confound real vertical
differences. Counts of individual grid units should be kept separate and
identified horizontally and vertically to permit later analysis of differences
due to the tidal gradient when applicable. They may also be used for estimating
the contribution of patchiness, on the scale of the quadrat grid units, to
the variance of mean density. For quadrats from horizontal surfaces with no
appreciable vertical tidal height difference, the frame grid provides a
ready means for using rectangular coordinates for random selection of individual
grid subunits to count. On more nearly vertical surfaces where there is an
appreciable difference in tidal height within the quadrat, the random selection
can be confined to units within horizontal rows.
Individuals may be so abundant and crowded in some quadrats that counting
proves difficult, even with subsampling. If operator confusion causes difficulty
in distinguishing previously uncounted individuals from those already counted,
the slide may be projected onto a large piece of white paper and each individual
scored with a mark using a crayon of contrasting color. The marks themselves
can then be counted, and crossed off as they are used.
Data on Percent Areal Cover
For plants and sessile animals like barnacles, space on rock surfaces may
be a limiting ecological resource. Data on the percentage of available sub-
strate space occupied by various species can be used to compare relative
52
-------
density or space dominance within species assemblages as well as to compare
spatial or temporal site differences in these features (Littler, 1971,
Littler and Murray, 1975). Species occurrence and percent cover data can be
very quickly derived from photographic quadrats by the point counting, or
intercept, method. In this technique, a regular or random array of super-
imposed points sample a fraction of the covered area for the presence of the
species of interest, with 'hits' being scored. The procedure may be automated
by superposition of a point array on the projection screen, a glass plate
placed in front of it or over prints, or a point array inked onto a glass
plate laid over photographic paper during enlargement. Few quantitative
intertidal studies have used point counting on photographs for data on areal
fractions (percent cover) of different species to its greatest advantage and
none appear to have considered the sources of error and variance produced in
application of this useful method. The mathematics of point counting methods
has been greatly developed in the field of quantitative microscopy in biology,
mineralogy and metalography. A detailed exposition of the statistics and
probability involved is given in DeHoff and Rhines (1968).
If a sufficiently large enough number of sample points is used in an
array superimposed on the photograph (or part thereof) of a quadrat, the
number of sample points falling within the image of individuals of a species
will be proportional to the area covered by that species, and the total number
of points in the array will be proportional to the total area covered by the
array. In practice, the number of hits for each species is expressed as a
percentage of the number of points in the array, by simply dividing, to
express the fractional area coverage, or percent cover.
Form of the Sampling Point Array
A regular or random array of points or single lines with randomized points
along them, are used to sample a fraction of the total quadrat area covered
by the images of the species of interest. A grid made by lines joining points
in a regular array may be used to effect visual separation of the field while
counting. A large projected image may be covered by a smaller point array
by systematically moving the projector so that successive parts of the image
are sampled. For both regular and random point arrays, the probability of
a point falling within the bounds of the image of a given species is equal
to the fraction of the total quadrat area occupied by the images of all individu-
als of that species present in the quadrat. The number of sample points falling
within the image of a species is proportional to the area covered by that
species, and the total number of points in the array is proportional to the
total area covered. The number of hits for a species is divided by the total
number of points superimposed on the quadrat image to give the percentage
area covered by the species.
The accuracy of percent cover estimates made by point counting methods is
determined by the number of observations, or points used, and the spacing of
the point array. This spacing, or point density also affects the time efficiency
of the method; more points and hence more counting is needed to get a good
estimate of the area occupied by smaller species than larger ones. A given
53
-------
grid size, or distance between points in the array used will have different
accuracies for large and small species, and for rare and patchily distributed
species. The grid spacing should be relatively small with respect to the size
of the smaller organisms to be included in the count; it will then be adequate
for the larger also. An optimal density of points would be roughly one per
individual in the area covered. In regular arrays, a spacing about equal to
the average projected image size of the smaller species will be adequate and
produce optimal accuracy for the time involved. A systematic array of points
will produce less variance due to the size of individuals than a random array,
but starting and stopping the point count must be random with respect to any
object in the field, when point grids are used.
It is the final projected image size of organisms that is related to the
spacing o'f points in a regular array. For a given size of photographic
quadrat, the projection distance and hence the final image size may be standardized
and a single grid of a specific spacing (point density) and total size or area
used. Any change in the final image size used with a particular grid will
result in estimates with a difference in variance, as would its use with photo-
graphs of different sized quadrats which require a different final magnification
when projected.
The number of points to be counted to give an accurate representation of
the smaller species can be adequately determined by trial. Project a typical
slide of a photographic quadrat enclosed by the frame size which will be used
routinely, and adjust the projector distance or zoom lens until the smallest
individual or species of interest can be clearly identified in the projected
image. Make an estimate of the average linear dimension of the image of this
species, and the length of one of the sides of a unit in the frame grid so
that future projection distances for this size of quadrat can be kept the
same. Make a grid of 100 points spaced at the average linear dimension
determined on a clear glass or plastic plkte to be placed flat against the
screen or on a piece of heavy white paper to be fastened to a flat support
and used as a screen. The total quadrat area will probably project larger than
this grid array. Project the quadrat on this grid of points and record all
point intercepts, or hits, for the test species, moving the quadrat image
across the grid if necessary, to repeatedly sample the quadrat. Using the
ratio of hits to total points used, plot the percentage area occupied by that
species against the increasing total number of points counted until the curve
levels off. This will estimate the number of points needed to be recorded
with that point density to get an accurate estimate of cover for the species.
This density and point number should be more than adequate for the larger
species present.
IV. SAMPLING PROGRAMS FOR SPECIFIC SUBSTRATES
Tidal Plats
Tide or mud flats are broad areas of slight slope which experience little
wave energy input, and consequently are composed of fine sediment particles.
54
-------
Typically, there are variations in the sediment characteristics, often as a
result of the drainage pattern of the flat. These are associated with irregu-
larities in the tidal height both parallel and perpendicular to the waterline.
Patches of organisms are formed in response to local topography rather than
regular series of bands characteristic of more uniform intertidal gradients.
In many locations rooted vegetation alters the settlement characteristics
locally, producing further heterogeneity on the flat. Rhoads (1974), Kay and
Knights (1975) Bloom et al. (1972) and Gray (1974) discuss the nature of
animal-sediment relationships in more detail.
These factors contribute to the production of a diverse, patchy environ-
ment which is never sampled under the assumption that xit is homogeneous.
Tide flats are always stratified in some manner, generally reflecting the
overall change in tidal height, but also recognizing local irregularities in
topography and biotic patchiness.
Stratification of tide flats begins with mapping of the tidal height
gradient. Where the slope is readily defined, strata may be bounded by
uniform changes in tidal height. The low slope allows small changes in
elevation to correspond to relatively large areas, so that it is more convenient
to restrict the size of strata by use of smaller changes in elevation than
is common in other types of shore. Major swells or basins may be disregarded,
excluded from the transect, included as part of the stratification by defining
such features as separate strata, or by combining them with other strata of
equivalent elevation. Obvious discontinuities, whether physical or biotic,
are always recognized. Eelgrass beds, for example, form a separate stratum
if continuous along a tidal height. If such beds are discontinuous along the
same tidal height, then it may be preferable to separate areas at that height
with and without eelgrass. If they are not separated, interpretation should
note that eelgrass is not actually found at the average density recorded for
the given tidal height. Proportional allocation of sample units follows
the identification of appropriate strata.
The large area of tide flats presents some unique problems. If nothing
is known about the area, it is difficult to gain information using a pre-
liminary random sample. Positioning of random sample units in a very large
area is impractical, and the systematic sample becomes useful as an exploratory
sampling method. The systematic sample is more likely to reveal discontinuities
in the communities of organisms present at different parts of the same flat,
especially when the discontinuities are not related directly to obvious
Physical changes, or when relatively sharp discontinuities occur over short
distances representing nearly unmeasureable changes in elevation. The results
of the systematic sample should lead to better stratification of the transects
in such cases. If the total number of sample units in the preliminary survey
are divided into two samples, then an estimate of the total variability of
the transect can be calculated for the estimation of the sample size of the
final program. Once the preliminary survey has defined the strata, the final
Program should be one of stratified random sampling. If the transect area is
too large for convenience, transects should be reduced in width to manageable
dimensions.
55
-------
The selection of the transect position is largely arbitrary. Transects
should appear at least superficially representative of the entire site, or
should be numerous enough to include the obvious variations in the nature of
substrate or biota. At least two transects are always necessary for purposes
of replication, and in the case of tide flats more are highly desirable.
Temporal variations in physical features are less important on mudflats
than sandy beaches, where seasonal changes in wave action can greatly affect
the beach slope and sand particle size, or rocky shores, where typically
large fluctuations occur in the macroflora and associated or dependent macro-
fauna. Tideflats do, however, experience considerable seasonal variation in
recruitment, and seasonal repetition of sampling is necessary (Woodin, 1974;
Reise, 1977; Green and Hobson, 1970).
Sampling Program Development
The steps in the development of a program for sampling of flats are:
1. Identify the number of sites which the sampling program should attempt
to examine. Define the absolute minimum number which must be examined.
2. Select a site for the preliminary survey. At this location, take
samples of at least five units each, randomly placed within an
area 30 m wide, extending from high to low water, with each of a
set of cores ranging from about 20 cm^ to .25 m^. Table I summarizes
core dimensions and sample sizes used in a selection of typical
sampling programs on tideflats. All cores should be taken to at
least 20 cm in depth, and sectioned into 5 cm or smaller sections
to determine depth to which the fauna extend.
3. Determine the optimal sample unit size in terms of the minimum core
size which returns most of the species present. For core sizes
above this minimum, calculate the relative net precision of each
to determine the most efficient coring unit.
4. Assume that at least two transects will be taken at each site, and
determine the number of sites which can be adequately sampled.
Sacrifice an entire site rather than sample several sites poorly.
5. At each of the sites selected for inclusion in the program, define
two or more transects of 30 m width. Separate transects by at
least 60 m in order to examine large-scale variability.
6. Survey the beach to establish the profile. Stratify the transects
according to the elevation and obvious discontinuities. Note that
drainage channels can be excluded from transects unless they are an
important feature of the area. Eel grass beds, because of their
importance to the estuarine fauna, should be included in the transects
as separate strata or in separate transects. Systematic sampling may
56
-------
be useful to establish where discontinuities are located. Depressions
and mounds may be treated separate!/ or included in strata of similar
elevation. Strata should be few enough that each stratum is allocated
several sample units.
7. Allocate the sample units to strata in proportion to the areas of
strata.
8. If possible, avoid sampling strata consecutively from top to bottom
or the reverse, to avoid time-dependent errors in sampling procedures.
Strata may be sampled in random order.
9. Samples may be sieved in the field or in the lab. Samples should be
preserved within 24-36 hours after collection. The smallest sieve
mesh should be no larger than 0.5 mm.
10. Sieved and preserved samples are floated, stained, and sorted to
species. Species are identified as closely as possible, counted,
and weighed for biomass. Dry weight is preferred, but wet weights
may be used if suitable conversion factors to dry weight are pro-
vided in the report. The size-frequency data should be examined to
estimate recruitment.
11. Sediment samples should be taken at the same time for particle size
analysis in each stratum. The organic content of sediment may also
be useful. This information should accompany the site description.
A description of meteorological conditions at the time of sampling
should also be supplied.
57
-------
Name Investigators
Atkinson, R.J., 1974.
TABLE I
SUMMARY OF SAMPLE AND QUADRAT SIZES - TIDEFLAT STUDIES
Species Studied
Goneplax rhomboides
Quadrat Size
100 m areas
Sieve Size
census of
burrows
Sample Size
several large areas
mapped by divers
Cassie, R.M., and A.D.
Michael, 1968.
community
.25 m to
6-10 cm
2.5 nun
5 systematic transects,
40 stations total
in
00
Heip, C., 1974.
Hibbert, C.J., 1976
Mclntyre, A.D. 1964,
Myren, R.T., and J.J.
Pella, 1977.
Stromgren, et al, 1973.
community
12 bivalves
meiobenthos
primarily
Macoma balthica
community
2
6 cm to
? depth
not given
4 cm to
7 cm depth
100 cm to
8 cm
.25 m to
unspec, depth,
or 86 cm to
30 cm
1 cm
.5 mm +
.076 mm
3.2 mm
6.4 iran
1 nun
64 cores in 8x8
array at 10 cm intervals
to test diversity index
4 transects, 20(t m
apart, sample-d at 100m
intervals with 2 qurulrats
per position
4-6 cores per station
each time, bcnthic
sublittoral area
One location, 6 transects,
5 strata, 4 sample
units per stratum per
sampling time.
-------
TABLE I, Continued
Name Investigators
Species Studied
Quadrat Size
Sieve Size
en
ID
Vader, W. 1977. Perforatella
rubiginosa
Ward, D.V., et al., Uca pugnax
1976.
Woodin, S.A., 1974. polychaetes
Hughes, R.N. and Community
J.C. Gamble, 1977.
. 06 m to
1 cm
.52 m2 to
18 cm
.05 m to
14 cm
0.25 m2 to
40 cm
41 cm to
14 cm
41 cm to
14 cm
.5mm
1 mm
2 ram (5 per
site)
1 . o mm
(4 per site)
0.5 mm (10
per site)
Sample Size
Plot positions not
given;
Observations of crabs
within defined unit,
counts averaged.
A single stratum of
one transect 70 m
wide, sample 6/yr;
also some sampling
in other strata
All taken at
random within
5x5 m square
at many locations,
tidal heights not
determined.
-------
Sand Beaches
Sandy beaches are typically higher wave energy beaches than are tideflats,
and have greater slope covering less area. The slope is usually nearly uniform
or changes in slope are easily detectable. The higher energy input reduces
heterogeneity parallel to the waterline, so that the tidal height gradient is
the dominant one. In most cases, stratification of sandy beaches is a matter
of division into equal intervals of elevation, so that strata are a series of
bands parallel to the shoreline.
Either line transects or belt transects may be applied to sand beaches.
A belt transect is better suited for detection of variation on a scale of
several meters, and is preferred on sandy beaches. The width of the transect
is typically on the order of 20-40 meters, and transects are usually separated
by about twice the width of each transect. This gives an excellent measure
of the variation on scales of several meters to hundreds of meters.
A method of sampling which is possible on sand beaches but not generally
elsewhere is trenching. This is essentially a line transect where a contiguous
series of sample units are excavated in sequence. It is possible because
the sand tends to be compact and does not collapse, so that trenches can be
quantitative in area and volume. The disadvantages are the same as for any
other form of the line transect; variation on a scale of meters is not detected.
High-energy sandy beaches, such as are found on open coasts, are frequently
subject to erosion of sand on a seasonal cycle. High wave energy input during
winter storms removes sand, while low energy conditions in the summer permit
deposition of sand. This strong seasonality in the physical nature of the
beach must be reflected in the sampling program by adequate replication
through the year.
Sampling Program Development
The steps involved in the development of a sampling program for sandy
beaches are:
1. Identify the number of sites which the sampling program should attempt
to examine. Define the absolute minimum number which must be examined.
2. Select one site to cqnduct a preliminary sampling. At this location,
take samples of at least five sample units randomly placed over a
transect of 30 m width, with each of a range of core sizes from
approximately 20 cm2 to .25 m^. Table II summarizes core dimensions,
sample sizes and number of transects used in a selection of typical
sand beach sampling programs. All sample units should be taken to a
depth of 20 cm or greater and sectioned into 5 cm or smaller sections
to determine the required depth of future cores.
60
-------
3. From the results of the samples, determine the smallest core size which
returns most of the species present. For all core sizes above this
minimum, calculate the required sample size and the relative net
precision to determine which is the most effective.
4. Determine the number of samples which can be taken when each sample
is of the required size for the optimal core, assuming that one sample
equals one transect. Assume that at least two transects will be
taken at each site, and determine the number of sites which can be
adequately sampled. Sacrifice an entire site rather than sample
several sites poorly.
5. At each of the selected sites, define two or more transects of 30 m
width extending from high to low water, and separated by at least
60 m along the shore, to examine the large-scale variability.
6. Survey the beach to establish a profile. Stratify the beach by
uniform intervals of elevation of I1 or 2', unless obvious discontinui-
ties suggest other strata. Determine the area of each stratum.
7. Allocate sample units from the total sample to each of the strata,
in proportion to the area of the strata. Place the allocated sample
units randomly within the strata, assigning every sample unit a
position independently of the others. If possible, avoid sampling
strata consecutively from top to bottom, or bottom to top of the beach,
to avoid time-dependent errors in sampling methods. Sample strata
in random order.
8. Either return the preserved or unpreserved samples to the lab and
sieve within 24 hours, or sieve the samples in the field. In either
case, unless preliminary sampling indicated otherwise, the smallest
sieve size should be 0.5 mm. Larger mesh sieves may be used to
remove debris.
9. Sieved samples are floated if necessary, stained, and sorted to
species. Species are identified as far as possible, and are counted.
The biomass of each species is estimated, preferably by dry weight
or ash-free dry weight. Recruitment data is obtained, particularly
if life history stages are identifiable. Wet weights can be used
if conversions to dry weight are made available.
10: In addition to the biological samples, samples should be taken for
sediment particle size and organic content analysis. A complete
description of the temperature, humidity, and other meteorological
data should be compiled.
61
-------
TABLE II
Investigators
Alheit, J., E. Naylor,
1961
SUMMARY OF QUADRAT AND SAMPLE SIZES - SAND BEACH STUDIES
Species Quadrat Size Sieve
Eurydice pulchra .125 m x 10 cm 2 mm in situ
Sample Si zc
Transect line sampled
at 10 stations, 70 cm
intervals. One transect.
KJ
Armstrong, J.W. et al.,
1976.
Baxter, R.E. 1971.
Birkeland, C. et al
1976.
Dauer, D.M., J.L.
Simon, 1973.
Holland, A.F., and
T.T. Polgar, 1976.
Hubbard, J.D., 1971.
miscellaneous
clams
miscellaneous
miscellaneous
miscellaneous
miscellaneous
.25 m , subs.
with 2 15-cm
cores, rest to
30 cm depth
minimum 20 linear
feet by 8" by
1 ft deep.
.1m if emmersed
.025 m2 if immersed
apparently to 2"
combined 5 cores
of .008 m2 area
per sample unit
.25 m x 25 cm
.25 m x 30 cm
1 mm for 5 beaches, 7 transects
cores, rest per beach. Each transect
6 mm in situ duplicate units at 0,3,6'
above MLLW. Also special
studies of different design.
range to Several beaches, one
1mm transect per beach. Trench
8" wide, 1 ft deep, at
least >j of each 1 vert. ft.
interval dug as quadrat.
.5 mm One beach, one transect,
three strata, 8 sample
units per stratum in
random array.
.5 mm One beach, one transect,
four stations per transect.
Five cores=one sample unit,
five units per station.
One sandbar, four stations
along center. One unit per
station. Replicated six
times through 1.5 yr.
Several beaches, one transect
per beach, contiguous series
quadrats each transect.
-------
Investigators
Hummon, W.D. 1974.
Kaiser, R.J. 1976.
Mattheisen, G.C.
1960.
Nybakken, J.W. 1971
Southward, A.J.
1965.
Zimmerman, S.T.,
and T.R. Merrel, 1976
Species
miscellaneous
Mya arenaria
miscellaneous
miscellaneous
miscellaneous
TABLE II, Continued
Quadrat Size Sieve
10 cm cores flotation
horizontally at
each depth interval,
unstated distance in.
"fine"
1x1.5 mm
in-situ
-1m x 31 cm
subunits within
20 m squares
.0625 m x 5 era
.25 m x ?
recommends .2-. 5 mm
.1-.25 m2 to
20 cm
.01
x 10-20 cm
Sample Size
Two beaches, one transect
per beach, 2/Tstations
per transect (d=width beach),
series of depths per station,
3 cores horizontally at each
depth.
Unstated beaches, transect
arbitrarily within clara
popln boundries, 3 sample
units within each plot.
Proposed plan only.
One beach, eight transects
at 10 m distance. Sample
units taken every 10 m
along each transect.
Unstated beaches, 5 transects
per beach, contiguous series
of quadrats.
Recommendations only,
several transects per beach
Apparently 1-3 transects
per beach, depending on
length and variability.
-------
Rocky Intertidal Shores
Introduction
Rocky shores typically have greater relief than sandy beaches or mudflats.
At a given site, local topography may range from shallowly sloping benches
to vertical rock faces. Wave exposure may range from very protected to highly
exposed on opposite surfaces of a single outcrop. Both of these gradients are
of importance in determining the nature of the community present. Extremely
exposed areas often have reduced diversity, with relatively few species adapted
to severe wave shock. Very protected areas may be diverse, or may be dominated
by virtual monocultures of an algal species, while areas of intermediate
expsoure can present the most diverse communities. The slope of the surface
affects the degree of wave shock, since benches tend to dissipate wave energy
more than vertical faces. The slope of the surface also affects the rate at
which water drains an area, which in turn controls both dessication and the
availability of food to filterers. Each combination of wave exposure and
slope can potentially support a somewhat different community, so that replication
of transects must reflect these physical factors. In the rocky intertidal,
replicate transects must share the same approximate degree of exposure and
slope.
Identification of the physical characteristics of the site in the rocky
intertidal is obviously the most important part of the planning of sampling
in the field. Conditions of wave exposure are difficult to measure (Field,
1968; Jones and Demetropoulous, 1968), but in most cases a subjective comparison
of locations is sufficient and precise quantification is not necessary for
site selection. The slope of a transect may be measured by the usual means
described in the section on transect surveying. Minor differences are not
likely to be crucial, so that relatively crude comparisons of slope will suffice.
Two additional features which potentially affect the community composition
are the type of rock substrate, and the minor relief of the rock surface. The
hardness of the substrate will affect the success of boring and attachment,
while extensive sculpturing of the surface on a small scale provides small
crevices which can support many refuge or opportunistic species, a diverse
crevice fauna, or in the case of high-exposure areas, provide refugia for
species which could not maintain themselves on open surfaces. Replication of
transects should attempt to recognize these factors in addition to the previous
two. Dahl (1973) presents a method for the approximation of the actual
surface area of a convoluted rocky surface, which can be used to identify the
degree of relief of a surface. The method compares the type of relief sculpturing
to any of several theoretical surfaces and calculates an index expressing the
increase in surface area by convolution over the area of a similarly bounded
planar surface. The method can also be used to describe the larger-scale relief
of an entire site, should quantitative comparison of the physical (topographical)
variability be desirable. Harlin and Lindbergh (1977) discuss the importance of
small-scale surface relief to recruitment of macro-algae.
64
-------
A given transect should extend from high to low water extremes. Within
the transect, changes in the slope of the surface should be represented in
the stratification, so that strata do not include two areas of greatly different
slope. If a given transect crosses areas differing greatly in wave exposure,
then the strata should separate areas of different exposure.
The order of factors controlling the stratification of a transect is:
obvious zonation in the community, tidal elevation, slope, exposure to waves
and sun, substrate composition and relief. If possible, however, the exposure
and slope of the surface should be grounds for establishing separate transects
rather than separate strata within single transects. Two transects should not
be considered as completely replicating each other unless they are in areas
which are subject to about the same physical conditions. Since replication
within a site is important, a very heterogeneous sampling location may require
that several transects be sampled in order to provide replication of sampling
under given conditions. If only a limited representation of the area is
possible for practical reasons, sampling should be concentrated on the most
common conditions at that site. A site which can generally be categorized
as protected, for example, might not be sampled in a few areas which are
exposed to severe wave shock. Outcrops and their vertical faces may be dis-
regarded in areas where the major feature is a shallowly.sloping bench.
In most cases within a given sampling location, the spatial extent of
areas which are subject to the same conditions will be small. Transects on
rocky shores are not as wide as those on tideflats or sand beaches because
of greater physical heterogeneity. The width of a transect controls the spatial
scale at which variations in the biota are detected. Rocky shores are commonly
highly variable over shorter distances than unconsolidated shores, so measure-
ment over smaller distances is appropriate. In some cases, the available area
for a given transect may be so small that the transect becomes a single line of
sample units. The positioning of sample units in this case is the same as
described for larger-area transects. Within the smaller areas, the preliminary
survey will allow assessment of the required sample size for estimation of the
mean densities and associated variances of the biota. Table III summarizes
some information from a selection of rocky shore studies which may aid in
planning preliminary surveys.
In planning sampling programs for rocky shores, a decision must be made
whether the sampling plan will attempt to sample the total biota present in
all microhabitats. Because of the high degree of spatial heterogeneity and
resulting diversity of distinctive habitats occupied by specialized species
assemblages, an effort to sample all of these conditions will require very
extensive committments of time and resources of the sampling program.
Rocky shores have distinctive cryptic faunas occupying both physically
and biologically created spaces such as holes, crevices, undersides of boulders
and ledges, dense algal mats, algal holdfasts, and spaces below and between
sheet or mat-forming animals such as mussels and barnacles. Only that portion
of the cryptic fauna associated with larger surface forms will be sampled with
the surface macrobenthos. In some cases, like the faunal assemblages of kelp
holdfasts (e.g. Jones, 1972, 1973), the quantification of these assemblages
has been used to detect pollution effects, by special studies of them alone.
65
-------
For other rocky shore cryptic spaces, no methods are known for routinely
and quantitatively sampling the fauna of borers, nestlers and other occupants
of crevices and holes in rock. These habitat spaces themselves are extremely
erratic in their horizontal distribution at any tidal height, but both borers
and more passive crevice seekers show distinctive distributions in the vertical
tidal gradient (Kensler, 1967). Most of this fauna is specific for cryptic
spaces, and while some are found both in rock cracks and the spaces enclosed
by larger surface organisms, none are found both free on surfaces and in
crevices.
A sampling program which mixes data from samples of biota from open rock
surfaces and that from crevices will reduce the definition possible for either
habitat, and the resulting data will contain unnecessary variation. If
cryptic spaces are to be sampled, the variety existing in the study areas should
be divided into defined categories, each of which should be sampled separately
in a plan stratified with respect to tidal height. The erratic distribution
of those spaces not resulting from other larger organisms will probably pre-
clude sampling in transects, and quantification of both the amount of cryptic
habitat available and its biota will be difficult. The work of Dahl (1973)
will aid in planning such work, as will the example of Steam, Scoffin and
Martindale (1977) who applied Dahl's considerations to determining the actual
area of a fringing reef, on several scales of relief.
It must be recognized that sampling of the cryptic fauna requires the
destruction of the habitat being sampled, temporarily in the case of the
biologically created spaces, permanently in the case of spaces within the
rock itself. The cryptofauna are obviously unsuited for monitoring studies
since undisturbed sampling is impossible. This group of species raises the
question: how much of the total community or species assemblage of a place
must be sampled to characterize the community in contrast to exhaustively
cataloging it, especially for the purposes of early detection of pollution
effects and their regulation. The precision gained from restricting the work
to a dominant and clearly characterized segment, in this case the biota of
open surfaces, appears to be preferred to the detail resulting from inclusion of
additional microhabitats at the sacrifice of time and resources which could be
used or replication at some scale between the sample unit and the site.
Another distinctive minor habitat of rocky shores is comprised of tide-
pools of various types. On all rocky shores some depressions are found
which hold variable amounts of water at low tide. They are the consequences
of highly variable geological and physical processes and only those directly
produced by the actions of organisms, like the pools resulting from the shallow
boring actions of sea urchins, show any regularity of horizontal or vertical
distribution. Because of their highly irregular and unpredictable occurrence,
few species are exclusively tidepool specialists, unlike the fauna of crevices,
and found only in this microhabitat. Almost all of the biota of tide pools
located at any tidal height, save upper spray zone pools, are species charact-
eristic of the level at which they are found, mixed with species from lower
66
-------
levels, in effect displaced upward by continuous submergence and which are
tolerant of the other physical and biological characteristics of pools. Tide-
pool salinities and temperatures, as well as oxygen content may differ markedly
from the adjacent open sea and the conditions in the rest of the rocky shore.
If a sampling protocol of any type permits sampling points to fall with-
out distinction within pools and on draining surfaces, the resulting data will
again contain erratic variation in number and level of organisms. This may
reduce the capability of resolving differences otherwise detectable between
sites and levels. Tide pools can be sampled quantitatively with sufficient
resources, and because of the sharp definition of area possible may make goo'd
monitoring points for repetitive non-destructive sampling.^
Perhaps the simplest solution to the presence of pools in areas where
transects are planned is to limit the major program to quantitative sampling
on open, draining surfaces with a constant orientation to the open sea and
incoming waves. The resulting samples will be restricted to a more uniform
assemblage of open surface forms and their cryptic associates, with good
fidelity to tidal height and thus amenable for use in spatial and temporal
comparisons of variations of presence and abundance and their possible relation
to disturbance. Pools may be largely avoided by an arbitrary rule of excluding
surface samples from all pools of standing water deeper than 2 cm and having
a surface area approximately that of the sampling quadrat in use.
If tide pools constitute important habitats for the study or comprise a
significant amount of the surface area of sites under study, and time and
effort resources permit, a separate sampling program for pools alone may be
justified by the goals of the program. Replication of sampling within pools
of any significant depth should distinguish between the side walls and the
bottom as well as between different depths within the same pool. Replication
between pools requires that pool depth, surface area and volume be matched
or controlled in some way, as well as the tidal height of the pools to be
compared. Rocky shore irregularities make close replication at a given tidal
height of pools with the same physical characteristics difficult.
A serious effort to include in rocky shore surveys those fishes which do
not depart with the tide will require that tidepools, including those at the
base of boulders, be sampled. Because these small fishes are highly mobile,
cryptically colored and formed and seek shelter, such an effort will require
that -poisons, anaesthetics or total emptying of water be used. In either
case of the use of chemicals, a change is made on the biota of the pools. Total
removal of water by bailing or pumping, followed by exhaustive searching of
all crevices and algae in the pool is necessary if the sampling is to be non-
destructive. Destructive quantification with poisons"like rotenone or anaes-
thetics like MS-222 (tricane methanesulfonate) and Benzocaine (ethyl-p-amino-
benzoate) and draining and searching is possible. Any tide pool sampling should
be accompanied by data collection on pool vertical height, surface area, depth^
volume, length of actual margin and some measure of physical and biological
cover.
67
-------
Sampling Program Development
The steps in the development of a sampling program for the rocky inter-
tidal are:
I. Identify the minimum and desired sites to be sampled.
2. Select a typical site for the preliminary survey.
3. Identify, within this site, the gradient from exposed to protected
areas, and select a vertical face and a shelf area. If either gradient
is minor it may be ignored. The preliminary sampling should examine
a position which is intermediate between the extremes of exposure,
unless it is possible to sample both extremes. The major slope of
the area should be used as the preliminary transect location. If
both vertical and level surfaces are important, both should be sampled.
At each selected part of the sample location, a transect should be
defined from the upper to the lowest extension of the rock surface
which has about the same slope and exposure. In some cases a vertical
face may not extend the full range, or a shelf not extend to the high
water line. Each transect should be as wide as possible for the area
with the desired characteristics, to 20 m. Line transects should be
avoided.
4. At each transect, at least five sample units (quadrats) should be
taken at random positions in the transect, with each of a range of
sizes from approximately 400 cm2 to .25 m2. Table III will aid in
quadrat size selection. If there is a great difference between the
upper and lower intertidal, it may be more effective to separately
sample each half (to stratify) for a better estimate of the variation
of the site. Stratification may also be used for each obvious dis-
continuity, provided that each stratum has at least two sample units.
5. Each sample unit (quadrat) is sampled as follows: a) The percent cover
of macroalgae for each layer (canopy, understory, primary layers) is
estimated by either the point-intercept method or the grid method.
b) All macroalgae are removed and retained for biomass determination.
c) Large invertebrate fauna are removed and retained for biomass
and counting, d) The surface is scraped clear of small and attached
fauna for later counting and biomass. In general, subsampling should
not be done at this stage, since this information will be used to
determine the future sample unit size and sample size.
6. Determine the optimal sample unit size for the number of species re-
covered, for the various communities sampled in different transects.
For the quadrats above this minimum size, determine the relative
net precision for each sample unit size, and select the sample unit
which is most efficient. Since this may vary between transects, use
a conservative decision rule, selecting the sample unit which is most
68
-------
suitable in all parts of the sample location. Calculate the required
sample size for this quadrat. This will also differ between transects,
so choose the largest sample size determined and use this sample size
in all transects.
7. Based on the required sample size, determine the number of transects
which the program can support and determine the number of sites which
are to be sampled with at least two transects; more transects should
be allowed per site if there are significant differences in exposure
or slope in parts of a site,
8. At each site, select at least two transects which incorporate the
range of exposure and slope in the site. Each characteristic area
should be replicated if possible (two exposed transects, two pro-
tected, two vertical faces, or two shelf transects). The transects
should extend as far up and down as possible, and be as wide as possi-
ble under the same set of conditions.
9. Survey the transects. Strata are defined on the basis of height or
overriding biotic discontinuities. Allocate sample units propor-
tionally to the strata, and position them randomly in the strata.
Each sample unit is sampled as in the preliminary sampling. If a
large unit has been selected and time becomes limiting, subsamples may be
taken at random within the larger unit. In general, macroalgae and
larger fauna should be sampled in the entire frame, and if necessary
small fauna sampled in a subsample of at least 20% of each quadrat.
10. Samples are preserved in the field or returned to the lab for sorting
and preservation. Samples are floated, stained, and sorted to species.
Size fractionation may be obtained by sieving. The biomass of each
species is determined. Recruitment is estimated if possible.
11. Within areas where boring and crevice fauna seem important, after the
standard quadrats are scraped, one or two may be chiseled out to
examine boring fauna. Crevices may be selected at random for quali-
tative sampling. These are pried apart and the fauna floated and
sieved free of the usual accumulation of debris.
12. Meteorological conditions during sampling are provided in the site
description.
69
-------
-J
o
NAME INVESTIGATOR
Barr, L. 1971
TABLE III
SUMMARY OF INFORMATION ON ROCKY SHORE STUDIES
SPECIES QUADRAT SIZE
Strongylocentrotus
species
Birkeland, C.,
et alia, 1976
Birkeland, C.,
et alia, 1976
Birkeland, C.,
et alia, 1976.
Boalch, G.T.,
N.A. Holme,
N.A. Jephson,
JMC Sidwell, 1974
miscellaneous
Laurencia papillosa,
Littorina, Tetraelita
Chthamalus. Abietinaria
miscellaneous
diver sampling,
.25 ra2 quadrats
.125 m x 6-8 cm
.06 ni
.125 m
1m, subs.
with .01 m- or
with .25 ni
SAMPLE DESIGN
Five sites selected around an
explosion center, plus one control.
20 selected plots by random throws,
using weight to select position
without diver bias. Sites
selected with bias for density or
algal cover.
Intertidal reef shelf. Transect
perpendicular to shore, separated
into obvious zones of communities.
4 quadrats selected in random
method from 2-d field. Quadrats
chiseled to 608 cm.
Biomass variation of one species -
a narrow zone on above shelf. 36
(apparently) quadrats in random 2-d
array in zone, rejected if greater
than 50% other species or in pool.
Andesite rock bench sampled in
four zones of species, 4 sample
units per zone in random array.
Slate reef, three radiating
transects from center of reef, one
down reef center,two to low
water line. % algal cover by
length of tape intercepted by a
species along a line.
-------
TABLE III, Continued
NAME INVESTIGATOR
Chan, G. 1973
SPECIES
miscellaneous
Chow, V., 1976.
Clark, R.C. et al,
1973.
Littorina scutulata
Miscellaneous
Connell, J.H., 1961
Feare, C.J., 1970.
barnacles
Nucella lapillus
QUADRAT SIZE
1 m subs.2
with .01 m
15 x 15 cm
30 x 30 cm,
subs, 10 x 10 for
limpets+barnacles,
10 x 30 for mussels
.1 m
25 ft x 3 in -
for census, .25 m
for growth studies
SAMPLE DESIGN
Apparently one transect with many
strata. Each stratum a perpen-
dicular line transect 10 m long,
divided into contiguous 1 m^
quadrats, either subs, or total
sampled.
Apparently two transects, one
in a sheltered area and one in an
exposed area. Line or maybe belt
transects separated into .1526 m
strata, 5 quadrats per stratum.
15 sites sampled once per month
for 3 months, then once at 6 and at
9 months. Each site sampled at 1
quadrat position per each "life
zone" evident in perpendicular
transect. Photographed and counted.
Species list within 5 m of each
quadrat.
Single location, 8 positions
chosen as the first 8 with more
than 50/.1 m square. Photographed
for time progression.
One location, a sloping shelf.
Four transects^ 375 ft long and
10 ft apart, perp. to shoreline,
divided into 15 25' sections as
sample units. Growth studies used
4 .25m^ sample units in each of
seven (undescribed) strata. Sample
units of arbitrary size when pop.
was aggregated in winter.
-------
TABLE III, Continued
NAME INVESTIGATOR
Frank, P.W., 1975
Kennedy, V.S., 1976.
Lees, D.C., R.J.
Rosenthal, 1975.
Littler, M.M., 1971
SPECIES
Tegula funebralis
mussels, 3 spp
QUADRAT SIZE
100,900 or
2500 cm2
.1 m
SAMPLE DESIGN
Selected sites along coast. At
each site a grid set up, and
random quadrats with size
dependent on density collected
until 500 snails were obtained.
A contiguous series of quadrats
through the intertidal on a vert-
ical rock face, all mussels taken.
Suggests sizes and frequencies of sampling for different areas and species,
depending also on the proposed information to be obtained.
Coralline algae
.07 m'
-4
K>
Littler, M.M. and
S.N. Murray, 1975.
Menge, B., 1972.
Biacroorganisms
Leptasterias and
prey species - small
larger
large
probably
.07 m2
10 JiT .
100 cm ,
1600-3600 as
1 n.2
The area was divided into strata
of equal area by unspecified means
and a variable number of throws
used to establish random sample
unit positions. Strata about
20x140 m.
Five radiating transects from outfall
plus two control areas SO and 100 m
away. Controls randomly placed
perpendicular to shoreline. Two
seasons used, combined in analysis.
% cover by point-intercept method.
Unrecorded number of sites, 2-4
transects per site. A single set
of random marks along a line used
for all transects for quadrat
positions.
-------
TABLE III, Continued
NAME INVESTIGATOR
Menge, J.L., 1974.
Moore, P.G., 1972.
SPECIES
Acathina punctulata
and prey - small
large/rate
holdfast fauna
QUADRAT SIZE
7
24 m subs
.01-.16m2
. 36-lnT
one holdfast
Nybakken, J.W., 1971.
O'Clair, C.E. and
K.K. Chew, 1971.
Paine, R.T., 1971.
Reimer, A.A. 1976.
miscellaneous
macrofauna
small
Tegula funebralis
Tetraclita stalactifera
.25 m
.25 m
subs. 5 cm
.026 m *
or .1m
.125 m
SAMPLE DESIGN
A permanent quadrat marked at
one site and subsampled, until 100
individuals per species were
collected.
15 sites selected in uniform
intervals along coast, 5 holdfasts
selected at each as units of the
sample of each site. Holdfast
choice by a sort of nearest-
neighbor selection process from
the first suitable holdfast.
A_grid surveyed at each site, .25
m sampled at each station randomly
located on the grid.
Unrecorded number of transects
divided into contiguous quadrats
and strata on the basis of sub-
strate and algal cover. A single
plot at each stratum chosen
subjectively.
Single site chosen and separated
into sloping bench and non-bench
strata. 22 quadrats were placed
in entire area, not stated if prop.
allocation used. Random placement.
A single site was sampled three
times with four random quadrats
along the center of the zone of
the study population.
-------
TABLE III, Continued
NAMR INVESTIGATOR
Russell, G. 1973.
Sammarco, P.W.,
Levinton, and Ogden,
1974.
SPECIES
miscellaneous
Diadema antillarium
QUADRAT SIZE
.25 m2
.25 m
Sutherland, J.P., 1970. Acmaea scabra
Underwood, A.J., 1976. 4 species gastropods
Zimmerman, S.T.,
T.R. Merrel, 1976
miscellaneous
400 cm
.25 m
1/16 m
SAMPLE DESIGN
One location. Five quadrats were
placed at 1 vertical ft intervals
and % cover and fauna1 density
recorded.
Five patches of reef previously
cleared of Diadema were sampled
by 6 quadrats each, laid along
a compass direction at uniform
intervals.
Two zones identified. 21 quadrats
split 18 and 3 in the zones, every
2 months censused for presence of
marked individuals.
16 quadrats permanently marked at
one study location, censused for
target sp monthly for one year.
1-3 transects perpendicular to
shore, depending on fetch. 15-50 m,
2-3 lines; 75-200 m, 1 line. Quadrats
at regular intervals along transects.
Boulders by random selection.
-------
Gravel and Cobble Shores
Introduction
Open coast beaches, which are unconsolidated, but in which the particle
size ranges from fine gravel to boulders, are typically high-energy, harsh
locations characterized by relatively low diversity. The slope is generally
steeper than on a sandy beach, with distinct changes in slope reflecting sum-
mer and winter wave conditions. These features are more pronounced than on
sand beaches.
Few studies have been conducted on open coast gravel or cobble beaches,
but they may be thought of as the extreme case of high-energy sedimentary
shores, and do not present problems fundamentally different from those which
may examined in sandy beach environments. One such study is that of Paul
and Feder (1973), who sampled two gravel beaches and one mudflat in examining
the distribution of Protothaca staminea. The sampling design was the same
as that used for the mudflat, using a single transect per beach, and the
number of sample units was a function of the width of the beach.
Within embayments and estuaries, cobble beaches built at times of extreme
river floods or past episodes of glaciation may now be very stable; these
share features with both rocky and sandy shores. The stable surface rocks afford
attachment places for hard substrate species; their interstices and under-
surfaces have a fauna analagous to the crevice fauna of open shore rocks, and
finally there is an infauna within the mixed sediment below the surface. The
algae and animals of the surface, both sessile and sedentary forms, must be
sampled by the quadrat methods described for the open coast rocky shore. The
effect of increased surface area produced by the cobble relief may appreciably
affect percent cover estimates as well as density. Because of the great
variability possible in the size of the rock particles present, no one method is
applicable for these shores. It may be necessary to use sample units of quite
different area for epibiota and infauna estimation.
The basic sampling design on gravel to cobble beaches is the same as for
sandy beaches, in that transects of moderate width are defined, strata selected
on the basis of tidal elevation (and in this case, changes of slope), and
random sampling conducted within the strata. The number of transects will
usually be the same or slightly fewer, since the greater energy input tends
to reduce the heterogeneity of the substrate.
Sampling Program Development
The steps in the development of a sampling program for sedimentary sub-
strates of gravel and larger size are:
1. The sites to be sampled,including an absolute minimum number, are
identified.
75
-------
2. Select one site of each sediment type (gravel or cobble) as the
preliminary sampling site. At these locations, take samples of at
least 5 sample units randomly placed within a representative transect
about 30 m wide, with each of a range of sample unit sizes from about
400 cm^ to .25 m^ in area.
3. Each sample unit is taken by: a) random placement of the unit
b) removal of algal cover, if any, from the surface of the upper
cobbles c) excavation of the sample unit to a depth of at least
20 cm, and preferably 30 cm. Flotation may be more useful than sieving
in separating out organisms.
4. Determine the best sample unit size for retention of the majority of
the species. For the sizes above this minimum, calculate the relative
net precision and the associated minimum sample size, and select the
optimal sample unit.
5. Determine the number of sites which may be sampled, assuming that each
site has two or more transects. Delete entire sites rather than sample
some site insufficiently.
6. At each selected site, define the transects using a width of about
30 m. Transects should be separated by at least 60 m to estimate
large scale variability. Survey the transects, and stratify the
transects according to changes in the slope of the beach, or to uni-
form changes in the elevation of about 2'(60 cm). Determine the area of
each stratum.
7. Allocate sample units proportionally to the size of the strata. Place
sample units randomly within each stratum, and if possible sample strata
in random order rather than consecutively to avoid time dependent errors
in sampling. The sample units are taken in the manner described above.
Note that larger sediments will require that a series of sieves be used
to avoid damaging the finer sieves.
8. Samples are either separated and preserved in the field or returned to
the lab; normally field separation and preservation would be done.
Samples are floated free of debris, stained, and sorted to species.
Species are identified as far as possible and counted, and wet or
dry weight is measured. Sizes of individuals or numbers of life
history stages are noted if possible to evaluate recruitment.
9. Samples are taken at the same time for analysis of sediment particle
distributions, organic detritus content. A description of atmospheric
conditions is compiled for the sampling period.
76
-------
V. DATA ANALYSIS
Introduction
The design of any sampling program, including sample type, size, arrange-
ment, etc. depends upon identification of the question to be investigated and
the statistical data analysis technique to be used. Studies of specific pollu-
tion effects such as the effect of a pollutant issuing from an intertidal point
source can be planned around specific questions, such as the nature and extent
of the effect on intertidal community characteristics at distances from the
source or determining the distance and dilution at which no community effect is
detected. The appropriate data analysis methods are ones testing specific
hypotheses, and the sampling program is then designed to produce data meeting
the requirements of the specific analysis to be used.
Such a study design will differ from a baseline or monitoring study
intended to detect any changes originating from any of a variety of potential
perturbations of unknown probability. In this case the goal of the study be-
comes a general characterization of important and readily measured features of
the intertidal community. Requirements of design differs and data analysis
becomes a problem of classification and characterization rather than the
statistical testing of specific hypotheses.
Every study will have its own design and data analysis requirements and
no plan can be given which will meet all contingencies. As a first step in
any program development, the problem or question to be investigated should be
clearly stated in as specific a form as the case permits, and analysis methods
appropriate to the question selected before the final plan is formulated. The
details of the sampling design can then be specified to meet the requirements
of the problem and the intended method of data analysis. This will also pro-
vide a rational basis for any decisions on limiting the sampling to conform
to realities of time, budget or other limitations. It is probable that, as a
consequence of limits to the understanding of the potential- effects of per-
turbations of all types on the intertidal community, most studies will include
some requirements for data to be used in testing specific hypotheses as well
as in analyses like clustering techniques which test no particular hypothesis.
The following sections will briefly review different types of data
analysis, largely from the viewpoint of types of questions or general ecological
information each is most suited for in pollution related intertidal investi-
gations. References are supplied which permit the techniques to be adapted to
specific programs. Once the problem has been identified, this section will
illustrate how various techniques of data analysis can aid in its resolution.
The final sampling design should reflect the specific statistical requirements
of the techniques chosen for the study.
Standard Ecological Information
The minimum information which should result from any study of the communi-
ties present in various locations is a list of species present in each sampling
77
-------
location, the absolute abundance of each species in both numbers and biomass,
and the variability of the abundance over space and through time. These data
are generally adequate to compare the similarity of the communities present at
each location, and to evaluate changes in a community at one location through
time.
Comparisons are also possible on the single-species level, for differences
in the abundance of selected dominant species through time or in different
locations. The computation and comparison of means and variances are discussed
in great detail in many standard statistical texts, and will not be covered
here; one such suitable text is Snedecor and Cochran (,1967'}. Elliott [1971'}
provides a very useful description of the procedures used in the analysis
of benthic samples for streams, which are often exactly the same procedures
as would be applied to samples from the intertidal. Simple comparisons of the
mean density of species in different locations or at different times will
usually be quite straightforward; the sole caution would be to note that most
count data will have a roughly negative binomial or other clumped distribution,
and therefore transformation of the data will be necessary in order to use
parametric statistics. Elliott notes that samples over 30 units in size may
assume normal distributions, and provides suggestions for transformations for
small samples with contagious distributions. Both Elliott (1971) and Mclntyre
(1971) suggest the transformation Iog10(x+l), for contagiously distributed data.
In addition to the basic data, several other types of information can be
very useful; among these are the size-frequency or its equivalent, the age-
class frequency distribution. This is used to evaluate the recruitment of
young to the population and in some cases the recruitment success of past
years, when the absence of an age class suggests that a past year has been
stressed (assuming that year classes are separable, see Yong and Skillman,
1975; Hancock, 19 65 and Hancock and Urquart, 1965). When the frequency
distribution can be expressed as discrete intervals, a chi-squared test may
be used to compare the size-frequency distributions of two locations, or one
location at two instances.
The health of individuals may be of use in determining whether pollution
or other disturbances have affected populations. Some measures of health are
the reproductive capacity as measured by gonad weight or content, the weight
per individual, or the regression of weight on some independent measure of size,
such as length.
The basic data listed above may be summarized by various statistics which
may be termed community composition parameters. These measure characteristics
of the community as a whole. The most commonly used parameters are diversity
measures and measures of evenness or its complement, dominance. These are dis-
cussed in more detail in the following section. In general, diversity indices
are parameters which express the composition of the community,, in terms of the
species present, their relative abundances within the site, and in some cases,
the overall abundance of the entire assemblage at a site. Examples of the use
of diversity indices to describe marine communities are Fager (1972), Johnson
(1970) and Sanders (1968). Changes in the diversity of a community through
time may be assumed to indicate that the site is undergoing some fundamental
78
-------
change from a stable equilibrium of species, or is recovering from a disturb-
ance and is returning to the original or to a different stable point. The
direction of the change may be toward greater or lesser diversity; studies have
shown that diversity may increase or decrease when a site is disturbed (Zimmer-
man and Livingston, 1976; Grimes and Mountain, 1971). It is the change in
diversity rather than the direction of change which may serve as an indicator
of disturbance effects.
Evenness and dominance indices, respectively, measure the degree to which
species are present in equal or in greatly disparate numerical abundances,
versus the case when one or a few species dominate. Like diversity indices, the
evenness or dominance of a site may increase or decrease following a disturbance,
depending on which species are susceptible and which are resistant. It is again
the change which serves to indicate that the disturbance has had an effect.
When two sites are compared by any of the community composition parameters,
the comparison is almost meaningless if the sites are not essentially replicates
in their potential species compositions (Jumars, 1974), at least for the
purposes of the type of studies discussed here. A shift from two replicate
sites being essentially identical in diversity, to one site having a greatly
different diversity, is a strong indication of disturbance at the site showing
the change in diversity. The same magnitude of change when the sites differ
greatly in environmental characteristics provides almost no information on
whether the change was due to normal or unusual causes. Selection of control
sites on the basis of similarity of environmental factors and superficial
similarity of species composition will often be adequate. However, innumerable
subtle environmental factors result in stable natural differences in community
composition and zonation on rocky shores; many examples are given in Lewis,
(1964). Control sites in this habitat type must be selected with care, and
preferably on the basis of preliminary surveys. More detailed comparisons
may be made using preliminary samples and computing any of several possible
measures of diversity, discussed in a later section.
Population and community statistics are greatly affected by various scales
of non-random dispersion. Fager (1972) determined that Sanders (1968)
rarefaction method for the determination of benthic diversity, for example,
was biased when aggregation properties differed between species. Numerous
papers have dealt with the effects of aggregation on estimates of mean
density; these are summarized in Pielou (1977). References to methods and
applications in the marine benthic environment will be found in Gage and Cog-
hill (1977) The interaction of sample unit size and the scale of heterogeneity
has been shown to materially affect the ability of a sampling program to
measure aggregation and the degree to which aggregation influences estimates
of community parameters (Lloyd, 1967; Morisita, 1962; Iwao, 1972; Grieg-Smith,
1964; Goodall, 1974).
Techniques used to evaluate the dispersion of marine organisms have
included grids or lines of contiguous quadrats (Gage and Coghill, 1977; Heip,
1975; Angel and Angel, 1967; Jumars e't al., 1977), paired cores randomly
positioned in the intertidal (Gardefors and Orrhage, 1968), distribution
79
-------
mapping with closely spaced sample units (Kenchington and Hammond, 1978;
Paine, 1963), spatial autocorrelation (Jumars et al., 1977) and cores with
increasing separation (Thistle, 1978; Hogue, 1977). Small scale distribution
mapping has been used in plankton studies (Mebe, 1970; Cassie, 1959).
Spectral analysis has been applied to plankton distributions in a few studies
where many samples were available (Denman and Platt, 1974).
Diversity Measures
The community composition parameters which collectively can be referred
to as diversity and evenness measures express three aspects of the composition
of a community: the species richness, or number of species; the relative abun-
dance of each (as evenness or dominance); and the absolute density of all
organisms at the site. Most express either the first and second, or the second
alone and a few express all three. Diversity is defined such that it increases
with species richness, evenness of relative density, and overall density.
Diversity indices fall into several categories by their source. These
may be termed the information group, derived from information theory; the
Simpson group, based on probability theory; the Mclntosh group, derived from
a continuum concept of communities; and a miscellaneous group of varied
origin. Within each of these groups, a set of indices can be defined which
express the species richness and the evenness components of diversity, and
the evenness component alone. A few cases also express the overall density of
organisms. A summary of the indices and the component or components of
diversity which each expresses is presented in Table IV. Not all indices are
included, as the number of available indices proposed is large (eg. Smith and
Grassle, 1977) and many are variants on other indices. Sources of information
on the meaning of many diversity and evenness measures are Pielou, 1975 and
1977. Before planning sampling for the use of any index, the exact composition
of the index, and the factors to which the index is responsive should be
clearly understood. Indices, even those which express the same components of
diversity, differ in their sensitivity to rare versus common species, and may
differ in the degree to which evenness or species richness control the value
of the index. The selection of a suitable index should be controlled by the
emphasis which the individual worker wishes to place on rare species, and
on the different aspects of diversity.
In general, diversity measures are applied to community studies for two
reasons. The first is to compare sites to determine the extent to which sites
are replicates, as discussed earlier. The second is to detect differences
between control and sites affected by disturbances, or between pre- and post-
disturbance stages at a single site. The assumption is that disturbance of
a site will affect the species composition or the relative abundance of species
in some manner detectable by changes in a diversity index. For example, if
a pollution event results in dominance of a single resistant species, while
most other species are severely reduced in number, the diversity of the site
should be reduced relative to earlier measurements, or to control areas which
are not in range of the pollution. However, it is also possible to have
increases in diversity following pollution, as might occur when a dominant
80
-------
species is susceptible to pollution, and reduction of the dominant's density
allows other species to increase in numbers. In such cases, the evenness and
species richness might both increase. Diversity indices calculated on the
basis of species richness and numerical density would not reflect changes in
the reproductive capacity or health of organisms, until these changes
influenced either species present or abundance of species.
A phenomenon which has been repeatedly observed (Dayton, 1975; Paine, 1974;
Loya, 1976) is the successive increase and decrease in diversity following a
disturbance. The apparent cause is removal of the dominant species, colonization
of ephemeral and fugitive species, and subsequent reestablishment of the domi-
nance of the major species. Pollution or other disturbances may be expected
to cause a sucessional pattern in the dominant species which will be reflected
in increases and decreases in diversity measures over long periods of time.
If, however, the community which was disturbed was not in equilibrium to begin
with, there may not have been dominant species, and changes in diversity
would be unpredictable. A change in a diversity measure cannot be inter-
preted as indicating a disturbance effect of a particular event unless support-
ing evidence indicates that the diversity was stable prior to the event, or
possibly was changing in some entirely different manner.. Diversity may also
change seasonally, as species are introduced and are unsuccessful, or as the
abundance of species increases due to influx of young individuals. Seasonal
patterns would have to be separated from changes due to disturbance.
81
-------
TABLE IV
DIVERSITY INDICES
Name
species richness
Margelef's index
log-normal indices
Brillouin's index
Shannon's index
Redundancy
Pielou's J
(evenness)
Lloyd's evenness
Simpson's index
" evenness
" redundancy
Hurlbert's index
MeInto sh's index
" evenness
" redundancy
Discussion
Margelef (1957)
Preston (1948)
Brillouin (1962)
Shannon and
Weaver (1963)
Shannon and
Weaver (1963)
Pielou (1975)
Lloyd, Zar, and
Karr (1968)
Simpson (1949)
Simpson (1949)
Simpson (1949)
Hurlbert (1971)
Mclntosh (1967)
Mclntosh (1967)
Mclntosh (1967)
Symbol
S
d
S*=SQ, a
B
H
H1
R
J
IT, r ,
X, or D
E or J
s s
R
s
PIE
MDI
J
m
R
m
Expressing: 1,2,5
1 only
1 only
1 and 2
1, 2, and 3
1 and 2
1 and 2
2 only
2 only
2 only
1 and 2
2 only
2 only
2 only
1, 2, and 3
2 only
2 only
v
1: Species richness
2: Evenness or dominance
3: Absolute density
82
-------
Similarity measures
A large number of similarity measures have been proposed. Many are
discussed in detail in the standard reference by Clifford and Stephenson (1975).
These may be divided into a limited number of categories by their derivation.
These are:
1. Similarity measures - ratios of shared attributed of two sites to
the total number of attributes possible in the comparison.
2. Association measures - measures of the correlation of attributes of
pairs of sites.
3. Euclidean distance - the distance separating two sites when the
species are axes and their abundances coordinates in the S-dimensional
space.
4. Diversity measures - compare the sites in such a way as to minimize
the within-group diversity as sites are successively fused.
5. Probability measures - estimates of the probability that two sites
are as similar as those under consideration, assuming that they came
from the same superpopulation.
The attributes of similarity measures (using the term in the general
sense rather than the restricted meaning favored by Clifford and Stephenson)
vary greatly. Some emphasize dominant species, others rare species. Several
can be used which require only species lists for two sites, assuming that the
same effort of sampling was applied to both sites. This sort of data would
be particularly easy to obtain by brief surveys, and therefore would be possible
to obtain in preparation for selection of control and replicate sites. Use of
simple similarity indices to select replicate sites would greatly increase the
probability that disturbance events would be detected, since true replicate
sites would have a greater probability of responding in the same manner to
disturbance.
Similarity measures are also commonly used to cluster sites. In essence,
clustering methods group sites according to the degree of similarity of sites,
either by successively more similar sites being grouped, or by successively
less similar sites being grouped. If some sites are disturbed and others are
not, at some level of the clustering, the disturbed and undisturbed sites should
fall into clearly separate clusters. Clifford and Stephenson (1975) discuss
several clustering strategies. Examples of the use of similarity indices to
cluster communities are given in Boesch (1977), Hummon (1974), Field and McFar-
lane (1968), Rosenberg (1976), Stephenson, et al (1971) and Littler and Murray
(1975). The last is specifically an attempt to evaluate the effects of sewage
outfall on rocky intertidal organisms, and is a good representative example of
the type of program which must be conducted to fully evaluate the effects of
disturbances on marine communities. Another good example of the application
of similarity analysis methods to the detection of pollution effects in the
83
-------
marine environment is the evaluation by Rosenberg (1976) of the response of
benthic marine communities to the closure of a sulfite pulp mill and the
cessation of waste liquor discharge.
Advanced Data Analysis Methods
Several numerical methods are available but not commonly used for the
assessment of disturbances, or the relation of species to environmental
factors. Ordination is a method of ordering the data in such a manner as to
permit identification of the major factors controlling distributions of species.
An example of the use of ordination is given in Hughes and Thomas (1971 a, b).
Clifford and Stephenson (1975) detail methods of ordination. Various methods
for ordination of benthic communities are evaluated by Erman and Helm (1971).
Canonical analysis is a method of relating two sets of data, in order to
identify aspects of the variability of both which appear to be related, and
which might be caused by the same factor, or which may be used to infer that
one set of variables caused some of the variation in the other set. Canonical
analysis is discussed in Clifford and Stephenson (1975); and in Cooley and
Lohnes (1971), who provide programs in FORTRAN IV for canonical analysis.
An example of the use of canonical analysis is given in Read and Renshaw (1976),
who compare beaches in different stages of pollution for the relationship of
species to environmental variables such as salinity and tidal height.
More familiar is the use of analysis of variance to compare the abundances
of species in different locations. The best example in the intertidal literature
at present is the study by Myren and Pella (1977), examining the natural
variability in the distribution of a bivalve at a beach subject to potential
oil pollution. The ANOVA methods used in this paper are not extremely advanced,
but are possibly beyond the immediate capabilities of the average worker. The
degree of refinement in this study is such that the probability of detection of
disturbances to the bivalve population is high.
A second method which is more familiar to the average worker is univariate
regression. This, like canonical analysis, is useful in comparing two sets
of data and finding correlations between variables. Any of a number of standard
statistical textbooks, such as Snedecor and Cochran (1967), are suitable
sources for details of linear regression methods. An excellent text is that
of Neter and Wasserman (1974), who discuss regression methods in great detail.
An example of the use of regression methods is given in Phillips (1976), who
examined the effects of metallic pollutants on mussel tissue weights as an
indicator of pollution effects.
Multivariate regression methods may be used instead of univariate regression
methods when the response of entire sets of variables to independent variables
is to be examined. The text by Cooley and Lohnes (1971) gives both a theoreti-
cal description of multivariate analysis, and examples and programs for
application of the methods.
84
-------
Allen (1971) examined the use of multivariate analysis in studies of
terrestrial algae. Moore (1973, 1974) used multivariate methods to examine
the kelp holdfast fauna of northeast Britain. Cassie (1972) provides a
computer program for multivariate statistical analysis. Additional information
on complex computer programming in ecology is given inDavies (1971). An
excellent guide to the detailed application of these methods in pollution
effect studies is the manual by Boesch (1977) in this report series.
Recruitment and Productivity
Recruitment
The relative abundance and number, of species, summarized in diversity
statistics, have been the most commonly used community parameters in the
assessment of the effects of pollution on marine organisms (e.g. Littler and
Murray, 1975). The assumption underlying the use of diversity measures in
assessing pollution effects is that the number of species or the relative
abundance of species will be affected by pollution within the period over
which the area is studied. In many cases, mature'individuals are much more
resistant to pollutants than larval or juvenile individuals. Studies which
ignore the effects of pollution on the rate of recruitment may underestimate
the magnitude of the damage.
There are two major ways in which the recruitment rate, may be affected.
The reproductive capacity of adults may be reduced by a decrease in the over-
all health of mature individuals or by some more specific effect. Theisuccess
rat;e of newly recruited individuals may also be decreased'if the larval or
juvenile stages are more sensitive to pollutants than are the adults. Either
of these can be important. A decrease in the successful recruitment of new
individuals to the population may not be detectable as a change in the popu-
lation density for some time afterward. The effects will be greatest when the
adults have a short lifespan and would normally be replaced by a new generation
on a yearly or seasonal basis. Long-lived species are more likely to recover
from pollution damage and reproduce normally. Obviously, the magnitude of the
effects of a pollution event will depend on the coincidence of the event and
the reproductive cycle. Toxins during the period of larval dispersal may
kill most of the larvae, while pollution in the non-reproductive season may
allow surviving adults to recover and reproduce normally. It is useful to know
in advance whether the dominant species reproduce yearly, and whether their
lifespan and susceptability to toxins make them:vulnerable to pollution at
critical periods of the year. Baseline studies should-attempt to evaluate the
recruitment patterns of the major species.
If size data are calculated, me recruitment of small and presumably young
individuals can be determined. Many species reproduce seasonally and show a
typical pattern of a seasonal peak in the smallest size classes. Comparison
to control areas or to previous data in the same areas should indicate whether
a given frequency of small individuals represents the normal recruitment to
the population. If the peak is followed through time, the decrease in numbers
85
-------
of individuals remaining from the initial recruitment can be compared to the
normal mortality rate, to evaluate whether young individuals are suffering
an abnormally high mortality. If recruitment is not strongly seasonal, there
may be a constant input of new individuals at some typical rate, and the observed
frequency of small individuals can be compared with the normal frequency at
control locations.
It is essential that evaluation of the recruitment rate recognize that
there are normal variations in the reproductive success of species. Some
species show patterns of relatively infrequent successful "sets", such as
mussels (Paine, 1974). The normal variation of recruitment must be separated
from the variation due to catastrophic events. If the recruitment to a control
area remains within the limits of normal variation, a reduction in the re-
cruitment to a study area is probably due to the event, but may also indicate
that the control area is not suitable (other causes may be producing the
reduction). If the recruitment to a control area is reduced by an amount similar
to the study area, then the reduction is probably not due to the pollution event
at the study area. If the history of variation at the study area is available,
a reduction in recruitment should not be considered unless it is greater than
the expected year-to-year variation. The significance of the reduction can
be evaluated by the usual statistical methods, given estimates of the varia-
bility in recruitment through time. If changes in recruitment success were
within normal limits, it would be necessary to prove that a control area did
not show the same reduction in order to prove that the reduction, though not
abnormally large, was due to some pollution event. This points out the value
of having both a monitoring history of a location, and control areas. The
difficulty of interpreting short-term impact studies because of long-term
variability is well demonstrated by Boesch et al. (1976") who review the dynamics
of estuarine benthic communities. The macrobenthic fauna of estuaries show
substantial dynamic seasonal and long-term variability caused both by life
history features and long-term habitat changes.
Productivity
It is important to separate estimation of the standing stock from esti-
mation of the production of an area (Allen, 1971) . The standing stock is the
biomass present at a given time. Single samples in time measure the standing
stock at that time. The production of an area is the amount of organic material
produced through growth and reproduction of individuals. It is possible to
have a high production rate and a low biomass, if the production is consumed
rapidly. The major effect of pollution could be to decrease the production
rate rather than the biomass, so that the effects on biomass would not be
immediately evident. Populations also go through normal fluctuations in biomass
as a result of seasonality of production. Deviations from the normal pattern
of changes in the production and the biomass present may be good indicators
of pollution effects. Bellamy et al. (1972) discuss the utility of productivity
data in pollution monitoring studies. The following are some examples of ways
in which pollution effects on production may be detected.
86
-------
Many species have highly seasonal recruitment and show strong fluctuations
in biomass and mean size of individuals. The fluctuations will be most pro-
nounced when the adults die after reproducing, such as occurs in some amphipod
species on west coast sandy beaches (Bosworth, 1976). A large part of the
production of these populations goes into reproduction during spring. In the
remainder of the year most of production goes to growth. Pollution effects
could be detected through decreases in the population in the following year
due to failure of the larval recruitment (larval deaths) or to diminished
reproductive capacity of adults (adult health), or perhaps due to a decrease in
the number of adults reaching reproductive size. These effects would be in
addition to the death of adults immediately following pollution events.
Pollution effects on growth could be examined through following the newly
recruited cohort of juveniles as they increase in size.
When the adults do not die after reproducing, a bimodal distribution in
size often results. The larger mode consists of older adults. The smaller
size mode represents newly recruited juveniles. The juveniles could be identi-
fied for growth rate estimates until their mode merged with the adult size mode.
Species which have continual reproduction often have a stable distribution
of age or sizes. The frequency of individuals in any given age or size class
is roughly constant, provided that conditions are stable. A pollution event
might be expected to alter the stable age distribution, if recruitment is
affected. Deviations from the typical distribution of sizes would be indicative
of disturbances to the population.
The second major component of production is growth. The growth of indi-
viduals can be decreased without death, reducing the biomass present without
changing the numerical abundance. It has already been implied that in some
cases, a reduction in the growth rate could reduce reproductive capacity of
the population by decreasing the number of individuals which reach reproductive
size. Ideally, growth rates can be measured for many species either directly
(see next section), or indirectly through following the average size of a year
class of newly recruited juveniles.
Health of individuals may also be affected. In most cases these effects
will be difficult to measure quantitatively in field sampling and data may
be more in the nature of qualitative observations on the appearance of collected
specimens. In some cases, the quantity of certain storage products such as
lipids may be useful measures of the health of individuals. Another measure
is the weight of reproductive tissue. Unusual behavior may often result from
poor health, but would seldom be quantifiable. The regression of weight per
individual on some independent measure of the linear size of individuals may
indicate if a species is strongly affected by pollutants, in that one might
expect that individuals of a given size may be of lesser weight than usual.
Measurement of Production
Methods for the measurement of production may be divided into long and
short term techniques. Methods which measure the production of either animals
87
-------
or plants over long periods generally measure the change in weight of the
individuals tested or of populations on the average. Short-terra estimates
usually test the rate of uptake of some substance over a period of hours.
Bittaker and Iverson (1976) compared the effectiveness and interpretation
of two relatively short-term methods, applied to the seagrass Thalassia testu-
dinum. The methods compared were the Wetzel (1964, 1965) adaptation of the
Steeman-Nielsen (1952) C14-uptake technique, and the Zieman (1974, 1975) method
for staple-marking of growing plants. The former method measures the in situ
uptake of radioactive carbon over a period of several hours, giving an estimate
of the rate of carbon particulate production over that period. The latter
method measures the rate of lengthening of blades over a period of four to six
days, and estimates the dry weight production of leaves (blades). Details of
the methods and numerous cautions on the application of the methods are given
in Bittaker and Iverson (1976) as well as the original sources. This paper
also provides an excellent bibliography for productivity measurement for sea-
grasses and for benthic macroalgae. Many of the methods available for the
measurement of macrophytic production are described by Round and Hickman (1971),
who consider the estimation of production from increases in area, volume, wet
weight, dry weight, ash-free dry weight, and length (as measured by stapling
or by punching holes). The problems of the various methods are discussed and
C14 methods are also described. Similar papers giving some methods for macro-
algal production are Westlake (1969) and Ott and Maurer (1977). Discussions
of productivity measurement for algae in special circumstances, such as in sands
or overlying mud, may also be found in Sournia (1976, sands), Steele and
Baird (1968, sands), and Leach (1970, mudflats). The seasonality of rocky
intertidal macroalgal standing crop, density, growth and reproduction, and
methods for their measurement are discussed in Hansen and Doyle (1976) and
Mathieson et al. (1976).
Over long periods, the simplest method for the estimation of production
involves estimation of the rate of intake of energy, and the sum of the rates
of energy expenditure. Given a feeding rate, the production of a population
may be summarized as the sum of growth, reproduction, and mortality. The
difference between the gross intake and the net production goes to maintenance
or is lost as excretion or egestion. An example of the construction of an
energy budget is given in Kay and Brafield (1973) for the polychaete Neanthes
virens. Other examples of the estimation of littoral macrofaunal production
are Hibbert (1976), Burke and Mann (1974), Warwick and Price (1975) and
Hughes (1970, a and b) on bivalve and gastropod production; and Paine (1971)
on production of a gastropod. Numerous other studies have examined aspects of
the total flow of energy in populations without construction of estimates of
energy flow and production rates. A total energy budget is necessary to
estimate the true production of a species. Crude estimates of the production
of populations can be derived from the increase in biomass of the population over
a period of time, without consideration of the rates of mortality and reproduction.
These estimates underestimate the amount of biomass produced by the population
by ignoring the biomass which was utilized by predators or scavengers, as well
as production in the form of gametes, which is also largely returned as food
for the rest of the community. In many cases, the time and economic limitations
88
-------
of a program will prevent complete definition of the energy flow in the dominant
species, let alone the entire community. These studies should attempt to
estimate the production of the major species as well as the available infor-
mation allows. Most programs should be able to document the change in standing
stock of major species, and estimate the recruitment to these species. The
mortality rate may not be available from the standard sampling program. Simple
estimates of mortality rates may be obtainable from the rates of loss of
marked individuals from .defined permanent quadrats, or from the difference
between the potential number of adults resulting from one season's recruitment
and the actual number of individuals which survive to adulthood. Ebert (1972)
gives methods for the estimation of growth and mortality from size data. The
studies of Nichols (1975, 1977) illustrate how dynamic aspects of production
can be measured from field sample data alone. Crisp (1975) and Mann (1969)
provide good short reviews of secondary productivity and conversion factors,
as well as production/biomass ratios for some littoral benthic invertebrates.
89
-------
VI. REFERENCES CITED
Alheit, J., and E. Naylor, 1961. Behavioral basis of intertidal zonation in
Eurydice pulchra Leach. J. Exptl. Mar. Biol. Ecol. 25: 135-144.
Allen, K. R., 1971. Relation between production and biomass. J. Fish. Res.
Bd. Canada 28:1573-1581.
Allen, T. F., 1971. Multivariate approaches to the ecology of algae on
terrestrial rock surfaces in North Wales. J. Ecol. 59:805-826.
Angel, H. H. and M. V. Angel, 1967. Distribution pattern analysis in a marine
benthic community. Helg. wiss. Meeres. 15:445-454.
Armstrong, J.W., C.P. Staude, R.M. Thorn, K.K. Chew, and R. E. Morris, 1976.
Intertidal beach studies. Research in Fisheries, Report Univ. of Wash.
Atkinson, R. J. A., 1974. Behavioral ecology of the mud-burrowing crab
Gonep 1 ax rhomboj^des. Mar. Biol. 25:259-252.
Baker, J. M. 1976a. Biological monitoring principles, methods and difficulties,
pp. 41-53, In Baker, J. M. (Ed.), Marine Ecology and Pollution, John Wiley
and Sons, New York.
Baker, J. M. 1976b. Ecological changes in Milford Haven during its history as
an oil port. pp. 55-67. In_ J. M. Baker, £d.) Marine Ecology and Oil
Pollution. John Wiley and Sons, New York.
Ballantyne, W. J. 1961. A biologically defined exposure scale for the comparative
description of rocky shores. Field Studies 1_(3): 1-19.
Barr, L., 1971. Studies of populations of sea urchins, Strongylocentrotus sp,,
in relation to underground nuclear testing at Amchitka Island, Alaska.
Bioscience 21:614.
Batzli, G. 0. 1969. Distribution of biomass in rocky intertidal communities on
the Pacific coast of the United States. J. Anim. Ecol. 3J?(5) :531-546.
Baxter, R. E., 1971. Earthquake effects on clams of Prince William Sound.
In_: The great Alaska earthquake of 1964. Biology. Committee on the
Alaska earthquake of the Divison of ^Earth Sciences, National Research Council
National Academy of Sciences, Washington, D. C.
90
-------
Bellamy, D. J., D. M. John, D. J. Jones, A. Starkie and A. Whittick. 1972.
The place of ecological monitoring in the study of pollution of the marine
environment. In: M. Kuvio (Ed.) Marine Pollution and Sea Life, Fishing
News Books Ltd., West Byfleet, Surrey, England.
Birkeland, C., A, A. Reimer, and J. R. Young, 1976. Survey of marine communities
in Panama and experiments with oil. EPA report 600/3-76-028, Naragansett
Laboratory.
Bittaker, H. F., and R. L. Iverson, 1976. Thalassia testudinum productivity:
a field comparison of measurement methods. Mar. Biol. 37:59-46.
Bloom, S. A., J. L. Simon, and V. D. Hunter, 1972. Animal-sediment relations
and community analysis of a Florida estuary. Mar. Biol. 13:43-56.
Boalch, G.T., N.A. Holme, N.A. Jephson, and J. Mi C. Sidwell, 1974. A re-survey
of Colman's intertidal traverses at Wembury, South Devon. Mar. Biol.
Assoc; U. K. 54_: 551-553.
Boesch, D. F., 1977. Application of numerical classification in ecological
investigations of water pollution. EPA-600/3-77-033, National Technical
Information Service, Springfield, VA.
Boesch, D. F., M. L. Wass and R. W. Vimstein, 1976. The dynamics of estuarine
benthic communities, pp. 177-196. In M. Wiley, (Ed.), Estuarine Processes,
Vol. I. Academic Press, New York.
Bosworth, W., 1976. Biology of the genus Eohaustorius (Amphipoda: Haustoriidae)
on the Oregon coast. Ph.D. Thesis, OregonSta'te University.
Brett, C. E., 1964. A portable hydraulic diver-operated dredge-sieve for
sampling subtidal roacrofauna. J, Mar. Res.22(2):205-209.
Brillouin, L. 1962. Science and Information Theory. 2nd ed. Academic Press,
New York.
Brinkhuis, B. H., 1976. The ecology of temperate salt marsh fucoids. I.
Occurance and distribution of Ascophyllum nodosum ecads.: Mar. Biol. 34:
325-338. ' '''
Burke, M. V. and K. H. Mann, 1974. Productivity and production: Biomass
ratios of bivalve and gastropod populations in an eastern Canadian estuary.
J. Fish. Res. Bd. Canada.; "31:167-177.
Cassie, R. M. 1959. Microdistribution of plankton. New Zealand J. Sci.
2^:398-409.
Cassie, R. M., 1972. A computer programme for multivariate statistical
analysis of ecological data. J.Exptl. Mar. Biol. Ecol. 10:207-241.
91
-------
Cassie, R. M., and A. D. Michael, 1968. Fauna and sediments of an intertidal
mudflat: a multivariate analysis. Exptl. Mar. Biol. Ecol. 2i\-27>.
Chan, G., 1973. A study of the effects of the San Francisco oil spill on marine
organisms. In_: Proceedings of the joint conference on prevention and
control of oil spills (EPA, API, USCG, Wash. D.C., 1973).
Chow, V., 1976. The importance of size in the intertidal distribution of
Littorina scutulata. Veliger 18:69-78.
Clark, R. C., et al., 1973. Interagency investigations of a persistent oil spill
on the Washington coast. In: Proceedings of the joint conference on
prevention and control of oil spills (EPA, API, USCG, Wash., D.C., 1973).
Clifford, H.T., and W. Stephenson, 1975. An introduction to numerical classifi-
cation. Academic Press, New York, N. Y.
Cochran, W.G., 1977. Sampling techniques; 3rd edition. 428 pp. John Wiley
and Sons, Inc., New York, NY.
Coleman, M.J., and H. B. N., Hynes, 1970. The vertical distribution of the
invertebrate fauna in the bed of a stream. Limn, and Ocean. 15:31-40.
Connell, J. H., 1961. The influence of interspecific competition and other
factors on the distribution of the barnacle Chthamalus stellatus. Ecology
42j_ 710-723.
Connell, J. H., 1972. Community interactions on marine rocky intertidal
shores. Ann. Rev. Ecol. Syst. 3_: 169-192.
Connell, J. H., 1974. Field experiments in marine ecology, pp. 21-54, In:
R. N. Mariscal (ed.), Experimental Marine Biology, Academic Press,
New York.
Cooley, W.W., and P. R. Lohnes, 1971. Multivariate data analysis. John Wiley
and Sons, Inc., New York, NY, 364 pp.
Crapp, G. B., 1971. Monitoring the rocky shore, pp. 102-113, In: E. B. Cornell
(Ed.), The ecological effects of oil pollution in littoral communities.
Institute of Petroleum, London.
Crisp, D. J., 1975. Secondary Productivity in the Sea. pp. 71-89, in Productiv-
ity of World Ecosystems. Proc. of a Symposium. Nat. Acad. Sci. Washington,
166 pp.
Dahl, A. L., 1973. Surface area in ecological analysis: quantification of
benthic coral-reef algae. Mar. Biol. 23_(4) : 239-249.
Dauer, Di'-M., and J. L. Simon, 1973. Repopulation of an intertidal habitat
following defaunation. 2. General aspects of the first year of study.
Am. Zool. 13:1329.
92
-------
Dayton, P. K., 1975. Experimental studies of algal canopy interactions in a
sea otter dominated community at Amchitka Island, Alaska. Fish. Bull.
73_: 230-237.
Davies, R. G., 1971. Computer programming in quantitative biology, 492 pp.
Academic Press, New York.
DeBovee, F., J. Soyer, P. H. Albert, 1974. The importance of the mesh size for
the extraction of the muddy bottom meiofauna. Limn, and Ocean. Ji9:350.
DeHoff, R. T. and F. N. Rhines (Eds), 1968. Quantitative microscopy, 422 pp,
McGraw-Hill, New York.
Denman, K. L, and T. Platt, 1974. Coherences in the horizontal distributions
of phytoplankton and temperature in the upper ocean. Sixth colloquium on
Ocean Hydrodynamics, Liege, April, 1974.
Ebert, T. A., 1972. Estimating growth and mortality rates from size data.
Oecologia 11:281-298.
Elliott, J. M., 1971. Some methods for the statistical analysis of samples
of benthic invertebrates. Freshwater Biological Assoc. Sci. Publ 25,
Windermere Lab, The Ferry House, Ambleside, Westmorland.
Ellis, D. V., 1966. Aerial photography from helicopter as a technique for
intertidal surveys, Limnol. Oceanog. 11(2):299-50L.
Erman, D. C. and W. T. Helm, 1971. Comparison of some species importance values
and ordination techniques used to analyze benthic invertebrate communities.
Oikos 22:240-247.
Ertl, M., 1970, A quantitative method of sampling periphyton from rough
substrates. Limn, and Ocean. 15:576.
Fager, E. W., 1972. Diversity: a sampling study. The American Naturalist
106:293-510.
Feare, C.J., 1970. Aspects of the ecology of an exposed shore population of
the dogwhelk Nucella lapillus (L.). Oecologia 5:1-18.
Field, J. G., 1968. The turbulometer — an apparatus.for measuring relative
exposure to wave action on shores. Zool. Afric. 3{2):115-118.V
Field, J. G., 1971. A numerical analysis,of changes in the soft-bottom fauna
along a transect across False Bay, South Africa. J. Exp. Mar. Biol,
Ecol. 7/216-53.
Field, J. G., and G. McFarlane, 1968, Numerical, methods in marine ecology.
I. A quantitative similarity analysis of rocky shore samples"in?False
Bay, South Africa. Zool. Afric. 3^:119-138.
Finnish IBP-PM Group, 1969. Quantitative Sampling Equipment for the Littoral
Benthos. Int. Rev. Ges. Hydrobiol. 5£(2):185-193.
93
-------
Frank, P.W., 1975. Latitudinal variation in the life history features of
the black turban snail Tegula funebralis. Mar. Biol. 31:181-192.
Gage, J.D., 1977. Structure of the abyssal macrobenthic community in the
Rockall Trough, pp. 247-260. In_ B. F. Keegan, P. 0. Ceidigh and P. J. S.
Boaden (Eds.)., Biology of Benthic Organisms. Pergaraon Press, New York.
Gage, J. D. and G. G. Coghill. 1977. Studies on the Disperions Patterns of
Scottish Sea Loch Benthos from Contiguous Core Transects, pp. 319-337.
In: B. D. Coull (ed) Ecology of Marine Benthos Univ. South Carolina
Press, Columbia.
Gardefors, D., and L. Orrhage, 1968. Patchiness of some marine bottom animals;
a methodological study. Oikos 19: 311-21.
George, J.D., 1964. The life history of the cirratulid worm, Cirriformia
tentaculata, on an intertidal mud flat. J. Mar. Biol. Assoc. U.K. 44:47-65.
Glynn, P.W., 1965. Community composition, structure, and interrelationships in
the marine intertidal Endocladia muricata - Balanus glandula association
in Monterey Bay, California.Beaufortia, 12(148):1-198.
Goodall, D. W. 1954. Minimal area: a new approach. Rep. Int. Bot. Congress,
Paris, Sect. 7:19-21.
Gray, J. S., 1974. Animal-sediment relationships. Oceanog. Mar. Biol.
Ann. Rev. 12:223-261.
Green, R. H., and K. D. Hobson. 1970. Spatial and temporal structure in a
temperate intertidal community with special emphasis on Gemma gemma
(Pelecypoda: Mollusca). Ecology 51:999-1011.
Grieg-Smith, P. 1964. Quantitative Plant Ecology. 256 pp. Butterworths, London.
Grimes, C.B., and J. A. Mountain, 1971. Effects of thermal effluent upon marine
fishes near the Crystal River steam electric station. Fla. Dept. Nat.
Resources. Prof. Paper Series, No. 17.
Hancock, D. A., 1965. Graphic estimation of growth parameters. Jour, de Conseil.
29_:340-351.
Hancock, D. A. and A. E. Urquhart, 1965. The determination of natural mortality
and its causes in an exploited population of cockels (Cardium edule L.)
Fish. Invest. Ministry of Agri., Food and Fisheries, Ser. 2, 2£ (2):1-40.
Hansen, J.E., and W. T. Doyle, 1976. Ecology and natural history of Iridaea
cordata (Rhodophyta:Gigartinacea). Population structure. J. Phycol.
12(3): 273-278.
94
-------
Harlin, M. M., and J. M. Lindbergh, 1977. Selection of substrata by seaweeds:
optimal surface relief. Mar. Biol. 40:53-40.
Help, C., 1974. A new index measuring evenness. J. Mar. Biol. Assoc. U.K.
54_:555-557.
Heip, C., 1975. On the significance of aggregation in some benthic marine
invertebrates. Proc. 9th Env. Syrup, on Mar. Biol. pp. 527-538.
Hibbert, C. J., 1976. Biomass and production of a bivalve community on an
intertidal mudflat. J. Exptl. Mar. Biol. Ecol. 25:249-261.
Hogue, E. W., 1977. Spatial and temporal dynamics of a subtidal estuarine
gastrotrich assemblage. M. S. Thesis, Univ. N. Carolina, 37 pp.
Holland, A. F., and T. T. Polgar, 1976. Seasonal changes in the structure
of an intertidal community. Mar. Biol. 37:341-548.
Holme, N. A. and A. D. Mclntyre (Eds.), 1971. Methods for the study of the
marine benthos. IBP Handbook 16, Blackwell Scientific Pubs. Oxford and
Edinburgh, 334 pp.
Hubbard, J. D., 1971. Distribution and abundance of intertidal invertebrates
at Olsen Bay in Prince William Sound, Alaska, one year after the 1964
earthquake. In: The great Alaska earthquake of 1964. Biology. Committee
on the Alaska earthquake of the Division of Earth Sciences, National
Research Council, National Academy of Sciences, Wash., D. C.
Hughes, R. N., 1970. Population dynamics of the bivalve Scrobicularia plana
(da Costa) on an intertidal mudflat in North Wales. J. Anim. Ecol. 39:
333-356.
Hughes, R. N., 1970. An energy budget for a tidal flat population of the
bivalve Scorbicularia plana (da Costa). J. Anim, Ecol. 59:357-581.
Hughes, R. N. and M. L. H. Thomas, 1971. Classification and ordination of
benthic samples from Bedeque Bay, an estuary in Prince Edward Island,
Canada. Mar. Biol. 1.0:227-235.
Hughes, R. N. and J. C. Gamble, 1977. A quantitative Survey of the Biota of
Intertidal Soft Substrata on Aldabra Atoll, ^Indian Ocean. 279;527-555.
Phil. Trans. Roy. Soc. London, Ser. B.
Hughes, R. N., and M. L. H.Thomas, 1971. The classification and ordination
of shallow water benthic samples from Prince Edward Island, Canada.
J. Exptl. Mar. Biol. Ecol. 7:1-39.
Hummon, W. D., 1974. SH: A similarity index based on shared species diversity,
used to assess temporal and spatial relations among marine gastrotrichs.
Oecologia 3_:203-220.
Hurlbert, S. Hv, 1971. The nonconcept of species diversity: a critique and
alternative parameters. Ecology 52:577-586.
*
Iwao, S., 1972. Application of the m - m method to the analysis of spatial
patterns by changing the quadrat size. Ecol. 14:97-124.
95
-------
Johnston, C. S,, 1969. The ecological distribution and primary production of
macrophytic marine algae in the eastern Canaries. Int. Rev. Ges. Hydrobiol.
Hydrogr. 5_4:473-490.
Johnson, R. G., 1970. Variations in diversity within benthic marine
communities. Amer. Nat. 104:285-500.
Jones, D. L., 1972. Changes in the ecological balance of invertebrate communi-
ties in kelp holdfast habitats of some polluted North Sea waters.
Helgolander wiss. Meeresuntersuch. 25:248-260.
Jones, D. L., 1975. Variation in the trophic structure and species composition
of some invertebrate communities in polluted kelp forests in the North Sea.
Mar. Biol. 2^:351-365.
Jones, W. E. and A. Demetropoulos, 1968. Exposure to wave action: measure-
ments of an important ecological parameter on rocky shores on Anglesey.
J. Exp. Mar. Biol. Ecol. 2_(l):46-63.
Jumars, P. A., 1974. Two pitfalls in comparing communities of differing
diversities. Am. Nat. 108:589.
Jumars, P. A., D. Thistle, M. L. Jones, 1977. Detecting two-dimensional
spatial structure in biological data. Oecologia 28:109-125.
Kaiser, R. J., 1976. Ann. Report. DCS Assessment Project Razor clam habitat
survey, GOA, NOAA, pg. 17. In: Env. Assess. Cont. Shelf Vol. 7;
Fish, Plankton, Benthos, Littoral. NOAA and BLM.
Kay, D. G., and A. E. Brafield, 1973. The energy relations of the polychaete
Neanthes virens CSars). J. Anim. Ecol. 42:675-692.
Kay, D. G. and R. D. Knights, 1975. The macro-invertebrate fauna of the inter-
tidal soft sediments of Southeast England. J. Mar. Biol. Assoc. U. K.,
55J4):811-852.
Kelley, J. C., 1976. Sampling the sea, pp. 561-587, Li D. H. Gushing and
J. J. Walsh (Eds.), The Ecology of the Seas, Saunders, Philadelphia,
467 pp.
Kenchington, R. A. and L. S. Hammond, 1978. Population structure, growth, and
distribution of Lingula anatina (Brachiopoda) in Queensland, Australia.
J. Zool. 184_:63-81. ~
Kennedy, V. S., 1976. Dessication, higher temperatures, and upper intertidal
limits of three species of sea mussels in New Zealand, Mar. Biol. 35:127-137.
Kensler, C. B., 1967. Dessication resistance of intertidal crevice species as
a factor in their zonation. J. Anim. Ecol. 56:391-406.
96
-------
Leach, J. H., 1970. Epibenthic algal production in an intertidal mudflat.
Limn, and Ocean. 15:514-521.
Lees, D. C., and R. J., Rosenthal, 1975. Progress report: an ecological
assessment of the littoral zone along the outer coast of the Kenai
Peninsula. IN: Envtl. Assessment of the Alaska Coast, Principal
InvestigatorTs reports, July-Sept., 1975, Vol. 1. NOM and BLM.
Lewis, J. R., 1964. The ecology of rocky shores. English Universities
Press, London, 323 pp.
Littler, M. M., 1971 Standing stock measurements of crustose coralline algae
(Rhodophyta) and other.saxicolous organisms. J. Exptl. Mar. Biol. Ecol.
6_:91-99.
, and S. N. Murray, 1975. Impact of sewage on the distribution, abundance,
and community structure of rocky intertidal macroorganisms. Mar. Biol.
32:277-291.
Livingston, R. J., R- S. Lloyd, and M. S. Zimmerman, 1976. Determination of
sampling strategy for behthic macrbphytes in polluted and unpolluted
coastal areas. Bull. Mar. Sci. 26(4):569-575.
Lloyd,.Monte, 1967. Mean crowding. J. Anim. Ecol. 56:1-50.
Lloyd, M., J. Zar, and J. Karr, 1968. On the calculation of information
theoretical measures of diversity. Amer. Midi. Nat. 79:257-272.
Lundalv, T. L., 1971. Quantitative studies on rocky-bottom biocoenoses by
underwater photogrammetry. A methodological study; pp 201-208 IN:
Proc. 6th European Symp. Mar. Biol., Thalassia Jugoslavica 7^1):1-445.
Loya, Y., 1976. Recolonization of Red Sea corals affected by natural catastrophies
and man-made perturbations. Ecology J>7:278-289.
Mann, K. H., 1969. The dynamics of aquatic ecosystems. Adv. Ecol. Res. <6:1-81.
Margelef, R., 1957. information theory in ecology. Memorias de la Real Academia
de Ciencias y Artes de Barcelona, 23:573-449.
Mattheisen, G. C,, 1960. Intertidal zonation in populations of Mya arenaTJa.
Limn, and Ocean. 5_:381-388.
Mathieson, A. C., J. W. Shipman, J. R. O'Shea,.and R. C. Hasevlat. 1976.
Seasonal growth and reproduction of estuarine fucoid algae in New England.
J. Exp. Mar. Biol. 25J273-284.
McAlice, B. J., 1971. Phytoplankton sampling with the Sedgwick-Rafter cell.
Limn, and Ocean. 16:19-28.
97
-------
Mclntosh, R. P., 1967. An index of diversity and the relation of certain concepts
to diversity. Ecology 48:577-586.
Mclntyre, A.D., 1964. Meiobenthos of sublittoral mud. J. Mar. Biol. Assoc.
U.K. 44_:665-674.
, 1971. Design of sampling programs. IN: N. A. Holme and A. D. Mclntyre
(eds.)) Methods for the study of the marine benthos. IBP Handbook 16,
1971. Blackwell Scientific Publications, Oxford and Edinburgh.
McKay, C. R. and D. J. Hartzband, 1970. Propylene phenoxytol: narcotic agent
for unsorted benthic invertebrates. Trans. Amer. Micros. Soc. 8£:53-54.
Menge, B. A., 1972. Foraging strategy of a starfish in relation to actual
prey availability and environmental predictability. Ecol. Mono. 42;25-50.
Menge, J. L., 1974. Prey selection and foraging period of the intertidal
predaceous snail Acathina punctulata. Oecologia 17:295-516.
, and B. Menge, 1974. Role of resource allocation, aggression, and spatial
heterogeneity in coexistence of two competing intertidal starfish. Ecol.
Mono. 4£: 189-209.
Moore, P. G., 1972. The kelp fauna of Northeastern Britain. I. Introduction
and the physical environment. J. Exptl. Mar. Biol. Ecol. 15:97-125.
Moore, P. G., 1975. The kelp fauna of N. E. Britain. II. Multivariate classifi-
cation: turbidity as an ecological factor. J. Exptl. Mar. Biol. Ecol.
1^:127-163.
Moore, P. G., 1974. The kelp fauna of N. E. Britain. III. Qualitative and
quantitative ordinations, and the utility of a multivariate approach.
J. Exptl. Mar. Biol. Ecol. 16^:257-500.
Morisita, M., 1962. I. - index as a measure of dispersion of individuals.
Res. Popl. Ecol. 4_:l-7.
Moyse, J. and A. Nelson-Smith. 1963. Zonation of animals and plants on rocky
shores around Dale, Pembrokeshire Field Studies 1_(5):1-31.'
Moyse, J, and Nelson-Smith, A., 1964. Effects of the severe cold of 1962-63
upon shore animals in South Wales. J. Anim. Ecol., 35:185-190.
Myren, R. T., and J. J. Pella, 1977. Natural variability in distribution of
a population of Macoma balthica subject to potential oil pollution at
Port Valdez, Alaska. Mar. Biol. 41^:371-582.
Neter, J., and W. Wasserman, 1974. Applied Linear Statistical Models.
Richard D. Darwin, Inc., Homewood, 111. 842 pp.
98
-------
Nelson-Smith, A., 1967. Marine Biology of Milford Haven, The distribution of
littoral plants and animals: Field Studies 2^:435-477.
Nichols, W. H., 1975. Dynamics and energetics of three deposit-feeding benthic
invertebrate populations in Puget Sound, Washington. Ecol. Monogr. 45:57-82.
Nichols, W. H., 1977a; Dynamics and production of Pectinaria kofeni (Malm) in
Kiel Bay, West Germany, pp. 453-463 In: Keegan, B. W., P. 0. Ceidigh,
and P. J. S. Boaden (eds.J, Biology of Benthic Organisms, llth European
Symposium on Marine Biology. Pergamon Press. New York.
Nichols, W. H., 1977b. Infaunal biomass and production on a mudflat; San
Francisco Bay, California, pp. 339-357. In; B. C. Coull (Ed.), Ecology
of Marine Benthos. Univ. So. Carolina Press, Columbia.
Nybakken, J. W., 1971. Preearthquake intertidal zonation of three species at
Saints Bay, Kodiak Island, Alaska. IN: The great Alaska earthquake of
1964. Biology. Committee on the Alaska earthquake of the division of
earth sciences, National Research Council, National Academy of Sciences,
Wash., D. C.
O'Clair, C. E., and K. K. Chew, 1971. , Transect studies of littoral .macrofauna,
Amchitka Island, Alaska. Bioscience '21;661.
Ott, J. and L. Maurer. 1977. Strategies of energy transfer from marine
macrophytes to consumer levels: The Posidonia oceanica example, pp. 493-
502. In: Keegan, B. F.,.-P, 0. Ceidigh, and P. J. S. Boaden (eds),
Biology~bf Benthic Organisms, llth European Symposium on Marine Biology.
Pergamon Press. New York.
Paine, R. T., 1963. The ecology of the Brachiopod Glottidia;pyrimidata.
Ecol. Mono. 33_: 167-213.
Paine, R. T., 1971. Energy flow in a. natural population of the herbivorous
gastropod Tegula funebralis. Limn, and Ocean. 16:86-98.
, 1974. Intertidal community structure: experimental Studies on the
relationship between a dominant competitor and its principal predator.
Oecologia 15_: 93-120.
Paul, A. J., and H. M. Feder, 1973. Growth, recruitment, and distribution of
the littleneck clam, Protothaca staminea, in Galena Bay, Prince William
Sound, Alaska. Fish. Bull. 7i:665r677.
Pedrick, R. A., 1974. Nonmetallic elutriation and sieving device ±or bentnic
macrofauna. Limnol. § Oceanog. 19_(3):535-538.
Phillips, D. J. H., 1976. The common mussel Mytilus edulis as an indicator of
pollution by zinc, cadmium, lead, and copper. I. Effects of environmental
variables on uptake of metals. Mar. Biol. 58:59-69.
99
-------
Pielou, E. C., 1975. Ecological diversity. Wiley-Interscience, New York, NY.
Pielou, E. C., 1977. Mathematical Ecology. Wiley-Interscience Publ., Wiley
and Sons, New York. 384 pp.
Preston, F. W., 1948. The commonness and rarity of species. Ecology 29;254-285.
Read, P., and T. Renshaw, 1976. Organism and environment on polluted beaches:
a canonical correlation analysis. J. Applied Ecol. 14:31-42.
Reimer, A. A., 1976. Description of a Tetraclita stalactifera panamensis
community on a rocky shore of Panama. Mar. Biol. 55:225-258.
Reise, K., 1977. Predation pressure and community structure of an intertidal
soft-bottom fauna, pp. 513-519. In; Keegan, B. W., P. 0. Ceidigh, and
P. J. S. Boaden (eds), Biology of Benthic Organisms, llth European
Symposium on Marine Biology. Pergamon Press. New York.
Rhoads, D. C., 1974. Organism - sediment relations in the muddy sea floor.
Oceanogr. Mar. Biol. Ann. Rev. 12;263-300.
Rosenberg, R., 1976. Benthic faunal dynamics during succession following
pollution abatement in a Swedish estuary. Oikos 27:414-427.
Round, F. E., and M. Hickman, 1971. Phytobenthos sampling and estimation of
primary production. IN: N. A. Holme and A. D. Mclntyre (eds.), Methods
for the study of the marine benthos. IBP Handbook 16, Blackwell
Scientific Publications, Oxford and Edinburgh.
Russell, G., 1973. The litus line: a reassessment. Oikos 24:158-161.
Saila, S., R. A. Pikanowski and D. S. Vaughn, 1975. Optimum allocation
strategies for sampling benthos in the New York Bight. Estuarine and
Coastal Marine Science 42_:119-128.
Sammarco, P. W., J. S. Levinton, and J. C. Ogden, 1974. Grazing and control
of coral reef community structure by Diadema antillarium Phillipi:
a preliminary study. J. Mar. Res. 52:47-55.
Sanders, H. L., 1968. Marine benthic diversity: a comparative study.
Amer. Nat. 102; 243-82.
Shannon, C., and W. Weaver, 1963. The mathematical theory of communication.
Univ. of Illinois Press, Urbana, 111.
Simpson, E. H., 1949. Measurement of diversity. Nature 165:688.
Smith, W., and J. F. Grassle, 1977. Sampling properties of a family of diversity
measures. Biometrics 35:285-292.
100
-------
Snedecor, G. W., and W. G. Cochran, 1967. Statistical methods. 6th edition,
Iowa State Univ, Press, Ames, Iowa.
Sournia, A., 1976. Primary production of sands in the lagoon of an atoll and
the role of foraminiferan symbionts. Mar. Biol. 37:29-52.
Southward, A. J.,1965. Life on the seashore. Harvard Univ. Press, Cambridge,
Mass.
Steam, C. W., T. P. Scoffin, and W. Martindale, 1977. Calcium carbonate
budget of a fringing reef on the west coast of Barbados. Part I, zonation
and productivity. Bull. Mar, Sci. 27(5):479-510.
Steele, J. H., and I.E. Baird, 1968. Production ecology of a sandy beach.
Limn, and Ocean. 13:14-25.
Steeman-Nielsen, E., 1952. The use of radioactive carbon (C ) for measuring
the organic production in the sea. J. Cons. Perm. Int. Explorer.,
Her. 18_: 117-140.
Stephenson, W., W. T. Williams, and S. D. Cook,, 1971. Computer analyses of
Peterson's original data on bottom communities. Ecol; Mono. 42:387-408.
Stromgren, T., R. Lande, and S. Engen, 1973. Intertidal distribution of
fauna on muddy beaches in the Borgenfjord area. Sarsia. 53:49-70.
Sutherland, J. P., 1970. Dynamics of high and low populations of the limpet
Acmaea scabra (Gould). Ecol. Mono. 40:169-188.
Thistle, D., 1978. Harpacticoid dispersion patterns; implications for deep-
sea diversity maintenance. J. Mar. Res. 36_:377-397i
Torlegard, A.K.I, and T. L. Lundalv, 1974. Underwater Analytical System.
Photogramm. Engin. 40_(3)287-293.
Underwood, A. J,, 1976, Analysis of patterns of dispersion of intertidal
prosobranch gastropods in relation to macroalgae and rockpools. Oecologia
25:145-154.
Vader, W., 1977. Habitat and distribution of Perforatella rubiginosa in the
fresh water tidal region of the Scheldt estuary, Belgium. Hydrobiologia
52_:23-28.
Venrick, E, L., 1971. The statistics of subsampling. Limn, and Ocean. 16_:811-818.
Vollenweider, R. A., 1974. A manual on methods for the measurement of primary
production in aquatic environments. 2nd Ed. IBP Handbook No. 12. Blackwell
Scientific Pubs. Oxford and Edinburg 225 pp.
101
-------
Ward, D. V., B. C. Howes, D. F. Ludwig, 1976. Interactive effects of predation
pressure and insecticide (Temefos) toxicity on populations of the marsh
fiddler crab Uca pugnax. Mar. Biol. 55:119-126.
Warwick, R. M., and R. Price, 1975. Macrofauna production in an estuarine
mudflat. J. Mar. Biol. Assoc. U. K. 55:1-18.
Westlake, D. F., 1969. Estimating quality and quantity of biomass: Macrophytes
and (with R. G. Wetzel) Periphyton. IN: Vollenweider, R. A., ed., A
manual on methods for the measurement of primary productivity in aquatic
environments, including a chapter on bacteria. IBP 12, Blackwell
Scientific Publications, Oxford and Edinburgh.
Wetzel, R. G., 1964. Primary productivity of aquatic macrophytes. Vexh.
int. Verein. theor. angew. Limno. 15:426-436.
, 1965. Techniques and problems of primary productivity measurements in
higher aquatic plants and periphyton. IN: C. R. Goldman, ed., Primary
productivity in aquatic environments. Univ. of Calif. Press, Berkeley,
Calif.
Whitehouse, J. W., and B. G. Lewis, 1966. The separation of benthos from stream
samples by flotation with carbon tetrachloride. Limn, and Ocean. 11:118-120.
Wiebe, P. H., 1970. Small-scale spatial distribution in oceanic zooplankton.
Limn, and Oceanog. 15:205-217.
Williams, D. D., and N. E. Williams, 1973. A counterstaining technique for
the use in sorting benthic samples. Limn, and Ocean. 18:152-154.
Woodin, S. A., 1974. Polychaete abundance patterns in a marine soft-sediment
environment: the importance of biological interactions. Ecol. Mono.
44_: 171-187.
Worswick, J. M. and M. T. Barbour, 1974. An elitriation apparatus for macro-
invertebrates. Limnol. and Oceanog. 19(3):538-540.
Yong, M. Y. Y., and R. A. Skillman, 1975. A computer program for analysis of
polymodal frequency distributions (ENORMSEP), Fortran IV. Fish. Bull.
73_:681.
Zieman, J. C., 1974. Methods for the study of the growth and production of
turtle grass, Thalassia testudinum Konig. Aquaculture £: 139-143.
, 1975. Quantitative and dynamic aspects of the ecology of turtle grass,
Thalassia testudinum. IN: L. E. Cronin, ed., Estuarine research, Vol. 1.
Academic Press, New York, NY.
102
-------
Zimmerman, M. S., and R. J. Livingston, 1976. The effects of kraft mill
effluents on benthic macrophyte assemblages in a shallow bay system
(Apalachee Bay, Florida, USA). Mar. Biol. 32_:297-312.
Zimmerman, S. T., and T. R. Merrel, 1976. Baseline characterization, Littoral
biota, Gulf of Alaska and the Bering Sea. IN: Environmental Assessment
of the Alaskan continental shelf, Vol. 7 (Fish, Plankton, Benthos, and
Littoral), NOAA and BLM.
103
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/5-78-087
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Procedures for Quantitative Ecological Assessments
in Intertidal Environments
5. REPORT DATE
September 1978
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
J.J. Gonor and P.P. Kemp
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
School of Oceanography and Marine Science Center
Oregon State University
Corvallis, Oregon 97331
10. PROGRAM ELEMENT NO.
1BA608
11. CONTRACT/GRANT NO.
Grant R805018 01
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Research Laboratory—Corvallis
Office of Research and Development
U.S. Environmental Protection Agency
Corvallis. Oreaon 97330
13. TYPE OF REPORT AND PERIOD COVERED
final
14. SPONSORING AGENCY CODE
EPA/600/02
15. SUPPLEMENTARY NOTES
16. ABSTRACT
The report is an effort to assemble the best available published procedures for
quantitative ecological studies in marine intertidal benthic environments,
applicable to evaluating existing or potential pollution effects. Considerable
material on how to rigorously devise sampling programs is included. Methods and
procedures recommended are synthetic and compiled from a variety of sources.
Guidelines are intended for application in marine intertidal ecological impact
assessments, pre- and post-pollution baseline surveys and long term monitoring programs
intended for the detection and forecasting of potential and human impacts in the
interttdal.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIEHS/OPEN ENDED TERMS C. COS AT I Field/Group
18. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (This Report)
unclassified
21. NO. OF PAGES
112
20. SECURITY CLASS (Thispage)
unclassified
22. PRICE
EPA F
------- |