OPTIMUM MACROBENTHIC SAMPLING PROTOCOL FOR DETECTING POLLUTION IMPACTS IN THE SOUTHERN CALIFORNIA BIGHT S T E V E N E F E R R A R O , R I C H A R D C. SWARTZ, FAITH A. COLE and W A L D E M A R A. DEBEN U.S. Environmental Protection Agency, Pacific Ecosystems Branch, ERL-N, Hatfield Marine Science Center, 2111 S.E. Marine Science Drive, Newport, OR 97365-5260, U.S.A. (Received: March 1992; revised: March 1993)

Abstract. The optimum macrobenthic sampling protocol [sampling unit, sieve mesh size, and sample size (n)] was determined for detecting ecologically important pollution impacts in the Southern California Bight, U.S.A. Cost, in laborator~ processing time, was determined for samples obtained using fourteen sampling units (0.005-0.1 m ~ surface area) and two sieve mesh sizes (1.0 and 0.5 mm). Statistical power analyses for t-tests of means were performed to estimate the minimum sample size (nmin) needed to reliably (c~ = 0.05, 1 - f l > 0.95) reject the null hypothesis of no difference between a reference and both a stimulated and a degraded station on twelve measures of community structure. The optimum sampling protocol for detecting impacts was determined as that with the lowest total c o s t Xnmin o n most measures. Five replicate, 0.02 m 2 × 5 cm deep, 1.0 mm mesh samples per station could reliably distinguish reference from impacted conditions on nine or ten measures of community structure at less than one quarter of the cost of the standard samplin~ protocol of 5 replicate, 0.1 m 2, 1.0 mm mesh samples per station. About 5 replicate, small (< 0.1 m'), 1.0 mm mesh samples per station may often be optimal for detecting important structural changes in macrobenthic communities with naturally high species richness and abundance.

1. Introduction The optimum macrobenthic sampling protocol [sampling unit (area × depth), sieve mesh size, and sample size (n)] meets the study's objective(s) at minimum cost. Two basic types of benthic studies are faunal surveys and quantitative studies (Elliott, 1983). The purpose of faunal surveys is to characterize communities in terms of the number of species, species composition, faunal abundance, etc. Usually no hypotheses are tested. Cumulative species and recruitment curves (plots of the cumulative number of species and the relative number of new species, respectively, in successive samples) are simple graphic techniques used by some investigators to determine how many sampling units are needed to characterize a community (Holme, 1964). Collection of most of the species present is indicated when the curves approach an asymptote. Five replicate, 0.1 m 2 samples per station have generally been found to be a practical choice for macrofaunal surveys when taking into account ship-time costs and the diminishing returns of additional samples (Longhurst, 1959; Lie, 1968). 'Macrofauna' is usually operationally defined as the animals retained by a 1.0 or 0.5 mm mesh screen (Eleftheriou and Holme, 1984; Rees, 1984; Kingston and Riddle, 1989). EnvironmentaI Monitoring and Assessment 29: 127-153, 1994. Q 1994 KluwerAcademic Publishers. Printed in the Netherlands.

128

STEVEN R FERRARO ET AL.

The purpose of quantitative studies is estimation or statistical inference. The optimum sampling protocol for a quantitative study depends upon the populations of interest and the objective(s) of the study (Eberhardt, 1978; Gonor and Kemp, 1978; Green, 1979; McIntyre et al., 1984; Bachelet, 1990). How large a sampling unit should be is related to the distribution of the populations under investigation. There are statistical and sometimes practical advantages to taking many small sampling units as compared to fewer larger sampling units when sampling populations with contagious distributions (Green, 1979; Elliott, 1983; Heltshe and Ritchey, 1984). The optimum n for parameter estimation is the minimum n needed to achieve a desired level of precision of a sample mean (Elliott, 1983). The optimum n for statistical inference is the minimum n needed to test a hypothesis at a given significance level (a = the probability of rejecting a true null hypothesis) and a given statistical power (1 - / 3 = the probability of accepting a false null hypothesis) (Cohen, 1977; Tort and Shea, 1983). The minimum n for parameter estimation and the minimum n for statistical inference are not necessarily equivalent and they should not be confused (Millard and Lettenmaier, 1986). It is improbable that the sampling requirements of a quantitative study to detect differences between communities will be as much as those needed to thoroughly characterize the communities. McIntyre et al. (1984) and others, however, have suggested collecting a minimum of five replicate, 0.1 m 2 samples per station when the optimum quantitative macrobenthic sampling protocol is unknown, and many investigators have summarily adopted the five replicate, 0.1 m 2 samples per station protocol in quantitative macrobenthic studies. The statistical power of a test depends on oz, sampling reliability (e.g., standard error), and the magnitude of the difference one wishes to detect, i.e. the effect size (Cohen, 1977): o~ is usually set by the investigator to 0.01 or 0.05. Reliability is dependent on the statistical model and is a function of n. To estimate statistical power the investigator must specify the effect size by way of an alternative hypothesis. Rotenberry and Wiens (1985) point out that a major hindrance to the use of power analysis in ecological studies has been the reluctance of investigators to posit unknown effect sizes since "there is no conventional methodology for estimating a priori the magnitude of an [effect size] for ecological purposes". The question of what constitutes an important effect size raises the issue of scientific significance. Simpson et al. (1960) distinguish statistical from biological significance noting that for the latter "there is no test but intuition, experience, and intelligence". Fairweather (1991) gives a good overview of statistical power analysis and discusses its relevance to environmental monitoring. Preliminary field studies to determine the optimum sampling protocol are often recommended (Gonor and Kemp, 1978; Green, 1979; Malley and Reynolds, 1979), yet there are few published papers which address the issue of optimum sampling unit, sieve mesh size, and n for quantitative benthic studies. Downing (1979) predicted the sampling unit and n needed to obtain a desired level of precision for mean population densities of macrobenthic organisms in lakes and large rivers using

OPTIMUM MACROBENTHIC SAMPLING PROTOCOL

129

the empirical relation: sample v a r i a n c e (82) ~--- f (mean density, sampling unit), obtained from a large data set. Downing's 8 2 " mean model has been used in subsequent studies to predict the sampling unit and n needed to obtain a desired level of precision for population densities of stream benthos (Morin, 1985), marine benthos (V6zina, 1988), freshwater epiphytes (Downing and Cyr, 1985), and biomass of freshwater macrophytes (Downing and Anderson, 1985). General s 2 • mean models have practical limitations (Downing, 1980; Taylor, 1980; Taylor, 1980; Riddle, 1989), they are not formulated for many measures of community structure (e.g., species richness), and important effect sizes must be specified before they can be used to predict optimum sampling protocols for impact assessments. Some studies have applied power analysis or minimum variance optimization methods to determine optimum sampling designs for universe estimation models (Cuff and Coleman, 1979) or point estimation models to detect differences in time or space (Gray, 1971; Saila et al., 1976; Michael et al., 1981; Bemstein et al., 1984; Science Applications and EcoAnalysis, 1984; Mar et al., 1985; Bernstein and Smith, 1986; Jacquez and Rohlf, 1986; Millard and Lettenmaier, 1986; Ferraro etal., 1989). Of these, only Michael et al. (1981) and Ferraro et al. (1989) consider the effects of using different sampling units, sieve mesh sizes, and n in analyses of macrobenthic community structure. This paper addresses three common macrobenthic sampling questions: How large should my sampling unit be? What sieve mesh size should I use? And how many replicate samples should I take? Our specific objective was to determine the least costly sampling protocol capable of reliably detecting ecologically important pollution effects on macrobenthic communities at 60 m water depth in the Southem Califomia Bight. Macrobenthic community stimulation and degradation has occurred in the Bight as a result of organic enrichment and chemical pollution from the Los Angeles County Sanitation Districts (LACSD) sewage-industrial discharge (LACSD, 1981; Swartz et al., 1985, 1986; Stull et al., 1986; Ferraro et al., 1991). Our approach was to revisit and collect five or six replicate macrobenthic samples using fourteen sampling units and two sieve mesh sizes at three stations along the LACSD pollution gradient representing reference, stimulated, and degraded conditions. Effect sizes of interest were determined by the magnitude of the difference observed between the reference and impacted stations. Cost, in laboratory processing time, was determined for each sampling unit, sieve mesh size, and station, and statistical power analyses for t-tests of means (Cohen 1977; Zar, 1984; Krebs, 1989 ) were performed to estimate the minimum n (nmin) needed to reliably (a = 0.05, 1 - / 3 _> 0.95) reject the null hypothesis of no difference from reference conditions on twelve measures of community structure. The optimum sampling protocol for a given measure and reference versus impacted station comparison was determined as that with the lowest total cost x nmin (Clarke and Green, 1988; Ferraro et al., 1989). A robust solution for optimality was sought by identifying sampling protocols which were optimal on most measures.

130

STEVEN P. FERRARO ET AL.

• ... :.:.:),:i..;::).;.c.)v.;;....

,o'3..

o

KILOMETERS

%

:..~...... •

=

~ ~ " 0.10; mns = 0.05 < P < 0.10;** = P < 0.01.

Number of Years Stations

2

3

4

5

6

7

8

9

10

Fig. 7. Power analysis results for Station R versus S and R versus D using the impact effects design. Shaded elements in 4 x 3 matrices indicate measures (see Figure 2) with a high probability (1 - / 3 > 0.95) of being detected significantly (c~ = 0.05) different in 0.1 m 2 samples.

C o r e samples o f 5 c m depth were less costly than 10 c m deep core samples (Table I) and they usually w e r e as or m o r e effective in discriminating reference f r o m stimulated and d e g r a d e d conditions (Figures 3 - 6 ) . S a m p l e s 8 c m deep were less costly and h a d equal or greater statistical p o w e r to detect m o d e r a t e differences in m a c r o b e n t h i c c o m m u n i t y structure than either 15 c m or 2 0 - 2 5 c m deep samples in P u g e t S o u n d , W a s h i n g t o n (Ferraro et al., 1989). Since m o s t species and individuals in u n p o l l u t e d s o f t - b o t t o m benthic c o m m u n i t i e s live in the u p p e r 4 - 1 0 c m (Sanders, 1960; W o o d i n , 1974; Word, 1976; Ankar, 1 9 7 7 ) a n d since animals tend to

OPTIMUM MACROBENTHIC SAMPLING PROTOCOL

145

concentrate nearer the surface under polluted conditions (Pearson and Rosenberg, 1978), there is reason to believe that pollution impacts will often be manifest in the upper 10 cm or less. Obviously, if deep dwelling animals are either common or the target species, deeper samples may be needed. A pollution impact assessment which focuses on an important, more easily sampled component of a community which is most affected by pollution (usually the upper 10 cm or less in soft-bottom benthic communities) is likely to be cost-efficient and environmentally protective. Samples of 1 mm mesh were less costly to process than 0.5 mm mesh samples (Table I) and they usually were as or more effective in discriminating reference from stimulated and degraded conditions (Figures 3-6). The power of the diversity indices (Measures 5-8, Figure 2) to detect impacts was slightly greater in the 0.5 mm than the 1.0 mm mesh samples, primarily due to a more even distribution of individuals among species in the 0.5 mm mesh samples at Station R. The power of the same indices to detect benthic alterations in Puget Sound was also sometimes greater in 0.5 mm than 1.0 mm mesh samples (Ferraro et al., 1989). Optimum (maximum sensitivity and minimum cost) macrobenthic sampling protocols, which could detect pollution impacts on multiple measures, almost always called for collecting macrofauna with a 1.0 mm mesh screen both in the Southern California Bight (Table III) and in Puget Sound (Ferraro et al., 1989). The 1.0 mm mesh sieve retained 73% (302/416) of the species and 49% (42,368/85,840) of the individuals retained by the 0.5 mm mesh sieve in this study. Obviously, some useful population-level information may be lost when using a 1.0 mm mesh screen to separate macrofauna. But if most of the macrobenthic community is _> 1.0 mm, or if animals in the > 1.0 mm and the 1.0 to 0.5 mm sizeclasses respond similarly to pollution, the consequences on community-level-based impact assessments are likely to be small. Impact assessments on adult macrobenthic communities may be confounded and made unnecessarily costly if the animals in the 1.0-0.5 mm size-class are primarily ephemeral, patchily-distributed juveniles of larger species (Hartley, 1982; McIntyre et al., 1984), or large meiofauna (Reish 1959; Schwinghamer, 1981). If a habitat is dominated by small adult macrofauna and the response to pollution in the _> 1.0 mm size-class is not representative of the macrobenthic community as a whole, or if the objective is to determine effects on macrofaunal juvenile recruitment, a 0.5 mm or smaller mesh screen should be used (Rees, 1984; Bachelet, 1990). The sensitivity of the twelve measures to stimulation and degradation in this study is indicated by their order of entry into the power analysis matrices in Figures 3-6. Measures entering first from left to right for a given sampling unit are more sensitive. The relative sensitivities of the measures are listed for optimum sampling protocols with n _< 5 for increasing numbers of measures in Table III. The Infaunal and Dominance Indices and ophiuroid biomass were sensitive measures of both stimulation and degradation. LOgl0 (mollusc biomass +1) and log10 (abundance) were sensitive measures of stimulation, and lOgl0 (number of species) and 1-Simpson's Index were sensitive measures of degradation. The Infaunal Index

146

STEVEN P. FERRARO ET AL.

was also a sensitive measure of macrobenthic changes in Santa Monica Bay (Word et al., 1980), and Puget Sound (Ferraro et al., 1989). Number of species, abundance, and Shannon's Index were sensitive measures of macrobenthic changes at a station on Georges Bank (Michael et al., 1981), in zones around sewage outfalls in southern California (Bernstein et al., 1984), and at other locations off the California coast (Science Applications and EcoAnalysis, 1984). The two most sensitive measures of stimulation in this study [the lnfaunal Index and log10 (mollusc biomass +1)] were also the two most sensitive measures of moderate macrobenthic stimulation in Puget Sound (Ferraro et al., 1989). Rank order of the sensitivity of the eleven measures common to both studies, however, was not the same (Spearman's p = 0.15, P > 0.05). In the Puget Sound study, water depth was about 10 m, sediments were medium to fine sand, and the sediment at the stimulated station was only moderately contaminated with high molecular weight aromatic hydrocarbons (PT1 Environmental Services and Tetra Tech, 1988). In this study, water depth was 60 m, sediments were medium-fine to coarse silt, and sediment at the stimulated station was organically enriched and contaminated with a large variety of chemicals (Ferraro et al., 1991). We would expect greater similarity in the sensitivity of response measures in more similar habitats and under more similar pollution conditions. As more data become available on the sensitivity of various measures in different situations, it may be possible to predict the cause of benthic impacts from the sensitivity of a set of response measures (Ferraro et al., 1989). Optimum sampling protocols for increasing numbers of measures are presented in Table III for n _< 5. Other optima are possible with n > 5. We chose to limit to _< 5 in our evaluation of sampling protocols so that costs would not be appreciably shifted from laboratory to field work. In general, different optimum sampling protocols can be obtained by changing one or more of the following constraints: (1) the minimum number - or a specified subset - of the measures to be detected with a desired probability; (2) the maximum number of replicate samples; or (3) the maximum laboratory processing time. Constraint (1) requires a scientific decision, while (2) and (3) are resource constraints. Sampling protocols for which none of the elements in the 4 x 3 power analysis matrices are shaded in Figures 3-6 are inadequate for assessing pollution impacts at the study site since those sampling protocols could not reliably detect a difference as large as that between Station R and S or D on any of our twelve measures. Sampling is excessive when the ability to detect impacts does not appreciably improve with additional sampling effort. Very little is gained, for example, by collecting and processing more than 5 replicate, 0.02 m 2 (4 cores) x 5 cm deep, 1.0 mm mesh samples per station when assessing degradation at the study site, since even a substantial increase in n (e.g. to 10) would not increase the number of measures which can reliably detect degradation nor appreciably increase the power of the tests (Figure 5).

OPTIMUM MACROBENTHIC SAMPLING PROTOCOL

147

Scientific judgement will vary on which measures are appropriate and how much evidence is needed to determine an ecologically significant difference. The reader may judge for him/herself which and how many of the measures should be used to meet a particular objective at the study site, and find, with the aid of Figures 3-6 and Table III, the optimum sampling protocol. When assessing pollution impacts, we recommend using a sampling protocol capable of reliably detecting important differences on most of our measures. Sampling protocols capable of detecting differences on multiple measures provide corroborative evidence of effects and reliable detection of smaller effect sizes on the more sensitive measures. 2 Our results show that the standard macrobenthic sampling protocol of 5 replicate, 0.1 m 2, 1.0 mm mesh samples per station is not optimal for reliable detection of significant differences on most of our twelve measures at this study site. Five replicate, 0.02 m 2 (4 cores) x 5 cm deep, 1.0 mm mesh samples per station could reliably distinguish reference from stimulated and degraded conditions on nine or ten measures at less than 1/4th the cost of the standard sampling protocol (Table III). Other studies have shown that 5 replicate, 0.1 m 2 grabs per station may not be optimal -for detecting important differences in macrobenthic community structure. Six replicate, 0.04 m 2 van Veen grab samples had essentially the same power to detect changes in number of species, abundance, and H ~as 6 replicate, 0.1 m 2 van Veen grab samples at a station on Georges Bank (Michael et al., 1981). Two to 4 replicate, 0.1 m 2 samples per station could detect (with P > 0.80) a 50% difference in number of species, abundance, diversity, and evenness at several nearshore locations along the southern and central California coast (Science Applications and EcoAnalysis, 1984). Five replicate, 0.06 m 2 × 8 cm deep, 1.0 mm mesh samples per station could reliably ( P _> 0.80) detect moderate differences from reference conditions on five measures of community structure in Puget Sound (Ferraro et al., 1989). The mean density of species and individuals was fairly high (> 40 species/0.1 m 2 and > 1000 individuals/0.1 m 2) in the present study and in the studies just cited. Caswell and Weinberg's (1986) simulations show that, in communities having a canonical lognormal distribution, n -~ 5 is sufficient to detect a 20% mean difference in number of species, H ~, evenness, and Simpson's Index when sampling units collect _> 10 species and _> 100 individuals. Available evidence, therefore, suggest that small (< 0.1 m 2) sampling units and n _~ 5 will often be sufficient to detect what many ecologists would consider important impacts on macrobenthic communities with naturally high species richness and abundance. Further research, including additional field studies and simulations of communities with other probability distributions, is needed to corroborate this conclusion and to determine optimal sampling protocols under other conditions (e.g., habitats with low species richness) and for different objectives. When in doubt, a preliminary study should be conducted to determine the optimum sampling protocol (Green, 1979; Ferraro et al., 1989). Impact effects tests indicated that loga0 (total biomass ÷ l ) , lOgl0 (crustacea biomass + 1), and ophiuroid biomass may not be suitable measures of stimulation,

148

STEVENEFERRARO ETAL.

and logl0 (abundance), log10 (total biomass +1), log10 (polychaete biomass +1), lOgl0 (crnstacea biomass + 1), and ophiuroid biomass may not be suitable measures of degradation at the study site (Figure 7). Log10 (abundance) was a poor measure of degradation in both the location effects (Figures 5 and 6) and impact effects (Figure 7) tests because of the small mean difference in abundance at the reference and degraded station. Biomass data is often problematic. Preservatives (Mills et al., 1982) and the presence of shelled organisms affect the accuracy of biomass measurements, and one or a few large organisms can skew the data and greatly increase the sample variance (Tetra Tech, 1985). High natural temporal variability (Table IV) makes ophinroid biomass a questionable measure of pollution effects even though the presence of ophiuroids is a good indicator of unpolluted conditions in the study area (Swartz et al., 1985, 1986; Scanland, 1987). Indicator taxa, such as ophiuroids, with naturally variable population densities, might be more meaningfully and efficiently sampled and analyzed as categorical (presence-absence) data in environmental impact studies (Green, 1979). The statistical model used in this study was a t-distribution for a priori comparisons between means. This point estimation model is appropriate for determining the optimum sampling protocol when an environmental impact must be inferred from temporal change alone or spatial change alone [main sequences 2 and 4 in Green (1979)]. Point estimation models require within-group (usually 'withinstation' in environmental impact assessments) replication to detect differences among groups. A universe estimation model, such as that used by Cuff and Coleman (1979) to make bay-wide estimates of population parameters, may have a different optimal sampling protocol than that of a point estimation model used to detect differences among stations within the bay (Green, 1980). It will always be most efficient, for example, to maximize replication at the highest level (e.g., stations within bay) and take single observations at each of the lower levels (e.g., grabs within stations) in a hierarchic analysis of variance design when the total number of observations is fixed, but within-group replication will always be needed to test for differences at lower model levels (Sokal and Rohlf, 1981). Population-level changes in indicator taxa and similarity indices are other commonly used measures of macrobenthic community response to pollution (Washington, 1984). Various graphical (Warwick, 1986; Clarke, 1990) and multivariate approaches (Clarke and Green, 1988; Warwick and Clarke, 1991) have also been recommended. We intend to follow-up this research with an investigation of the cost-efficiency of alternative macrobenthic sampling protocols using some of these other measures and approaches. Population-level responses to pollution, we believe, will generally be less reliable than community-level responses since populations of even common macrobenthic species tend to be highly spatially and temporally variable (Eagle, 1975; Nichols, 1985) [also see Pearson and Rosenberg (1978), Science Applications and EcoAnalysis (1984), and Caswell and Weinberg (1986) for other evidence and explanations]. Multivariate approaches, which are designed to detect differences in species composition, are often very sensitive, but

OPTIMUM MACROBENTHIC SAMPLING PROTOCOL

149

the ecological significance of the results can be more difficult to interpret than the results of species-independent methods of community analysis (Warwick and Clarke, 1991).

Acknowledgements We gratefully acknowledge the assistance of Kathy Sercu, David Specht, Doug Eberhardt and Rim Fay (field work), Stu Eide and Bryan Coleman (computer support), and Janet Lamberson, Kathey Sercu, Anja Robinson and Jill Jones (laboratory work). We thank Roger Green, Jim Heltshe, John Armstrong and two anonymous reviewers for their comments on an earlier draft of the manuscript. Mention of trade names or commercial products does not constitute endorsement or recommendation for use. Contribution No. N-052 of the U.S. Environmental Protection Agency, Environmental Research Laboratory, Narragansett, RI, USA.

Notes 1 We use the term 'composite' to refer to the creation of sampling units by combining core samples from the same grab (e.g., ~ 6 cores) and the term 'pooled' to the summing of sampling unit data (e.g., 5 replicate, ~ 6 cores). 2 For example, the 5 replicate, 0.02 m 2 (4 core) x 5 cm deep, 1.0 mm mesh sampling protocol can reliably (o~ = 0.05, 1 - / 3 > 0.95) detect a difference in the Infaunal Index 2.7 times smaller than (i.e., 3/8ths) that observed between Station R and S.

References Ankar, S.: 1977, 'Digging Profile and Penetration of the van Veen Grabs in Different Sediment Types', Contrib. Asko Lab., No. 16, University of Stockholm, Sweden, 22 pp. Bachelet, G.: 1990, 'The Choice of a Sieving Mesh Size in the Quantitative Assessment of Marine Macrobenthos: a Necessary Compromise Between Aims and Constraints', Mar. Environ. Res. 30, 21-35. Bascom, W.: 1978, 'Life in the Bottom', South. Calif. Coastal Water Res. Proj. Annu. Rep. 1978, 57-80. Bascom, W., Mearns, A.J., and Word, J.Q.: 1978, 'Establishing Boundaries Between Normal, Changed, and Degraded Areas', South. Calif. Coastal Water Res. Proj. Annu. Rep. 1978, 81-94. Bavry, J.L.: 1987, DESIGN-POWER Statistical Design Analys& System, Scientific Software, Inc., 1369 Neitzel Road, Mooresville, Indiana. Bernstein, B.B. and Smith, R.W.: 1986, 'Community Approaches to Monitoring', Oceans '86: Proceedings of the National Symposium on Monitoring Strategies 3, 974-979. Bemstein, B.B., Smith, R.W., and Thompson, B.E.: 1984, 'Sampling Design and Replication for Benthic Monitoring', South. Calif. Coastal Water Res. Proj. Biennial Rep. 1983-1984, 21-35. Caswell, H. and Weinberg, J.R.: 1984, 'Sample Size and Sensitivity in the Detection of Community Impacts', Oceans '86: Proceedings of the National Symposium on Monitoring Strategies 3, 1040-1045. Clarke, K.R.: 1990, 'Comparisons of Dominance Curves', J. Exp. Mar. BioL Ecol. 138, 143-157. Clarke, K.R. and Green, R.H.: 1988, 'Statistical Design and Analysis for a 'Biological Effects' Study', Mar. Ecol. Prog. Ser. 46, 213-226. Cohen, J.: 1977, Statistical Power Analysis for the Behavioral Sciences, revised ed., Academic Press, Orlando, Florida, 474 pp.

150

STEVEN P. FERRAROET AL.

Cuff, W. and Coleman, N.: 1979, 'Optimal Survey Design: Lessons from a Stratified Random Sample of Macrobenthos', J. Fish. Res. Board Can. 36, 351-361. Downing, J.A.: 1979, 'Aggregation, Transformation, and the Design of Benthos Sampling Programs', J. Fish. Res. Board Can. 36, 1454-1463. Downing, J.A.: 1980, 'Precision vs. Generality: a Reply', Can. J. Fish. Aquat. Sci. 37, 1329-1330. Downing, J.A. and Anderson, M.R.: 1985, 'Estimating the Standing Biomass of Aquatic Macrophytes', Can. J. Fish. Aquat. Sci. 42, 1860-1869. Downing, J.A. and Cyr, H.: 1985, 'Quantitative Estimation of Epiphytic Invertebrate Populations', Can. J. Fish. Aquat. Sci. 42, 1570-1579. Eagle, N.A.: 1975, 'Natural Fluctuations in a Soft Bottom Benthic Community', J. Mar. Biol. Assoc. U.K. 55, 865-878. Eberhardt, L.L.: 1978, 'Appraising Variability in Population Studies', J. Wildl. Manage. 42, 207-238. Elefthefiou, A. and Holme, N.A.: 1984, 'Macrofauna Techniques', Methods for the Study of Marine Benthos, 2nd edition, N.A. Holme and A.D. McIntyre, Eds., IBP Handbook No. 16, Blackwell Scientific Publications, Oxford, U.K., pp. 140-216. Elliott, J.M.: 1983, Some Methods for the Statistical Analysis of Samples of Benthic Invertebrates, 2nd ed., Sci. Publ. No. 25, Freshwater Biological Association, Ferry House, U.K., 156 pp. Fairweather, P.G.: 1991, 'Statistical Power and Design Requirements for Environmental Monitoring', Aust. J. Freshwater Res. 42, 555-567. Ferraro, S.P. and Cole, EA.: 1990, 'Taxonomic Level and Sample Size Sufficient for Assessing Pollution Impacts on the Southern California Bight Macrobenthos', Mar. Ecol. Prog. Ser. 67, 251-262. Ferraro, S.P., Cole, EA., DeBen, W.A., and Swartz, R.C.: 1989, 'Power-Cost Efficiency of Eight Macrobenthic Sampling Schemes in Puget Sound, Washington, USA', Can. J. Fish. Aquat. Sci. 46, 2157-2165. Ferraro, S.P, Swartz, R.C., Cole, F.A., and Schults, D.W.: 1991, 'Temporal Changes in the Benthos along a Pollution Gradient: Discriminating the Effects of Natural Phenomena from SewageIndustrial Wastewater Effects', Estuarine Coastal ShelfSci. 33, 383--407. Gallardo, V.A.: 1965, 'Observations on the Biting Profiles of Three 0.1 m2 Bottom-Samplers', Ophelia 2, 319-322. Glass, G.V., Peckham, P.D., and Sanders, J.R.: 1972, 'Consequences of Failure to Meet Assumptions Underlying the Fixed Effects Analysis of Variance and Covariance', Rev. Educ. Res. 42, 237-288. Gonor, J.J. and Kemp, EE: 1978, Procedures for Quantitative Ecological Assessments in Intertidal Environments, EPA-600/3-78-087, U.S. Environmental Protection Agency, Corvallis Environmental Research Laboratory, Corvallis, Oregon, 104 pp. Gray, J.S.: 1971, 'Sample Size and Sample Frequency in Relation to Quantitative Sampling of Sand Meiofauna', Proceedings of the First International Conference on Meiofauna, N.C. Hulings, Ed., Smithsonian Contributions to Zoology, No. 76, pp. 191-197. Gray, J.S.: 1981, The Ecology of Marine Sediments. An Introduction to the Structure and Function of Bottom Communities, Cambridge University Press, Cambridge, U.K., 185 pp. Green, R.H.: 1979, Sampling Design and Statistical Methods for Environmental Biologists, Wiley, New York, 257 pp. Green, R.H.: 1980, 'Comment on Optimal Survey Design', Can. J. Fish. Aquat. Sci. 37, 296. Green, R.H.: 1984, 'Statistical and Nonstatistical Considerations for Environmental Monitoring Studies', Envir Monitg Assessmt 4, 293-301. Green, R.H.: 1987, 'Statistical and Mathematical Aspects: Distinction Between Natural and Induced Variation', Methods for Assessing the Effects of Mixtures of Chemicals, V.B. Voulk, G.C. Butler, A.C. Upton, D.V. Parke and S.C. Asher, Eds., Wiley, Chichester, U.K., pp. 335-354. Hartley, J.P.: 1982, 'Methods for Monitoring Offshore Macrobenthos', Mar. Pollut. Bull. 13, 150-154. Heltshe, J.E and Forrester, N.E.: 1985, 'Statistical Evaluation of the Jackknife Estimate of Diversity when Using Quadrat Samples', Ecology 66, 107-111. Heltshe, J.E and Ritchey, T.A.: 1984, 'Spatial Pattern Detection Using Quadrat Samples', Biometrics 40, 877-885. Holme, N.A.: 1964, 'Methods of Sampling the Benthos', Adv. Mar. Biol. 2, 171-260.

OPTIMUMMACROBENTHICSAMPLINGPROTOCOL

151

Jacquez, G.M. and Rohlf, EJ.: 1986, 'Problems in the Variance Analysis of Nine Environmental Monitoring Variables: Determining the Number of Samples Needed to Detect a Change in Mean of 50%', Oceans '86: Proceedings of the National Symposium on Monitoring Strategies 3, pp. 974-979. Kingston, P.E and Riddle, M.J.: 1989, 'Cost Effectiveness of Benthic Faunal Monitoring', Mar. Pollut. Bull 20, 490--496. Krebs, C.J.: 1989, Ecological Methodology, Harper and Row, New York, 654 pp. Lie, U.: 1968, 'A Quantitative Study of Benthic Infauna in Puget Sound, Washington, USA, in 1963-1964', Fiskeridir. Skr. Ser. Havunders. 14, 229-556. Longhurst, A.R.: 1959, 'The Sampling Problem in Benthic Ecology', Proc. N.Z. EcoL Soc. 6, 8-12. Los Angeles County Sanitation Districts (LACSD): 1981, Ocean Monitoring and Research Annual Report 1980-1981, Los Angeles County Sanitation Districts, Whittier, California, 384 pp. Magurran, A.E.: 1988, 'Ecological Diversity and Its Measurement', Princeton University Press, Princeton, New Jersey, 179 pp. Malley, D.F. and Reynolds, J.B.: 1979, 'Sampling Strategies and Life History of Non-Insectan Freshwater Invertebrates', J. Fish. Res. Board Can. 36, 311-318. Mar, B.W., Lettenmaier, D.P., Homer, R.R., Richey, J.S., Palmer, R.N., Millard, S.P., MacKenzie, M.C., Vega-Gonzalez, S., and Lurid, J.R.: 1985, 'Sampling Design for Aquatic Ecological Monitoring, Volume 1: Summary Report', Electric Power Research Institute Project 1729-1, Department of Civil Engineering, University of Washington, Seattle, Washington, USA. (Available from Research Reports Center, Box 50490, Palo Alto, California 94303). Mclntosh, R.P.: 1967, 'An Index of Diversity and the Relation of Certain Concepts to Diversity', Ecology 48, 392-404. Mclntyre, A.D., Elliott, J.M., and Ellis, D.V.: 1984, 'Introduction: Design of Sampling Programmes', Methods for the Study of Marine Benthos, 2nd edition, N.A. Holme and A.D. Mclntyre, Eds., IBP Handbook No. 16, Blackwell Scientific Publications, Oxford, U.K., pp. 1-26. Michael, A.D., McGrath, R.A., and Long, C.D.: 1981, 'Benthic Grab Comparability Study', Final Report, Contract No. AA851-CT1-69, U.S. Department of the Interior, Bureau of Land Management, Washington, D.C., 33 pp. + Appendix (Available from Atlantic OCS Regional Office, Office of Minerals Management Service, 1951 Kidwell Drive, Vienna, Virginia 22180). Millard, S.P. and Lettenmaier, D.P.: 1986, 'Optimal Design of Biological Sampling Programs Using the Analysis of Variance', Estuarine Coastal ShelfSci. 22, 637-656. Miller, R.G.: 1974, 'The Jackknife - a Review', Biometrika 61, 1-15. Mills, E.L., Pittman, K., and Munroe, B.: 1982, 'Effect of Preservation on Weight of Marine Benthic Invertebrates', Can. J. Fish. Aquatic Sci. 39, 221-224. Morin, A.: 1985, 'Variability of Density Estimates and the Optimization of Sampling Programs for Stream Benthos', Can. J. Fish. Aquatic Sci. 42, 1530-1534. Nichols, F.H.: 1985, 'Abundance Fluctuations Among Benthic Invertebrates in Two Pacific Estuaries', Estuaries 8, 136-144. Odum, E.P.: 1985, 'Trends Expected in Stressed Ecosystems', BioScience 35, 419-422. Pearson, T.H. and Rosenberg, R.: 1978, 'Macrobenthic Succession in Relation to Organic Enrichment and Pollution of the Marine Environment', Oceanogr. Mar. Biol. Annu. Rev. 16, 229-311. Pielou, E.C.: 1966, 'The Measurement of Diversity in Different Types of Biological Collections', J. Theor. Biol. 13, 131-144. Pielou, E.C.: 1975, Ecological Diversity, Wiley, New York, 165 pp. PTI Environmental Services and Tetra Tech: 1988, Everett Harbor Action Program: Analysis of Toxic Problem Areas, TC-3338-26, Final Report, Appendices A, B and E (Available from PTI Environmental Services, 3625 132nd Ave. SE, Suite 301, Bellevue, Washington 98006). Rees, H.L.: 1984, 'A Note on Mesh Size Selection and Sampling Efficiency in Benthic Studies', Mar. Pollut. Bull. 15, 225-229. Reish, D.J.: 1959, 'A Discussion of the Importance of the Screen Size in Washing Quantitative Marine Bottom Samples', Ecology 40, 307-309. Riddle, M.J.: 1989, 'Precision of the Mean and the Design of Benthos Sampling Programmes: Caution Advised', Mar. BioL (Berl.) 103, 225-230.

152

STEVENEFERRARO ETAL.

Rohlf, EJ.: 1982, BIOM a Package of Statistical Programs, Department of Ecology and Evolution, SUNY at Stony Brook, New York, 11974. Rotenberry, J.T. and Wiens, J.A.: 1985, 'Statistical Power Analysis and Community-Wide Patterns', Amer. Nat. 125, 164-168. Saila, S.B., Pikanowski, R.A., and Vaughan, D.S.: 1976, 'Optimum Allocation Strategies for Sampling Benthos in the New York Bight', Estuarine Coastal Mar. Sci. 4, 119-128. Sanders, H.L.: 1960, 'Benthic Studies in Buzzards Bay. III. The Structure of the Soft-Bottom Community', Limnol. Oceanogn 5, 138-153. Scaniand, T.B.: 1987, 'Succession and the Role of Ophiuroids as it Applies to the Marine Infaunal • Associations of Palos Verdes', Joint Water Pollution Control Plant 1988 Revised Application for Modification of Secondary Treatment Requirements for Discharges into Marine Waters, Volume II, Appendix F-5, 24 pp + 1 Table + 3 Figures. (Available from Los Angeles County Sanitation Districts, Whittier, California 90607). Schwinghamer, P.: 1981, 'Characteristic Size Distributions of Integral Benthic Communities', Can. J. Fish. Aquatic. Sci. 38, 1255-1263. Science Applications and EcoAnalysis: 1984, 'Analysis of Historic Benthic Data for Assessment of Long-Term Changes in Biological Communities in the Santa Mafia Basin and Western Santa Barbara Channel', Contract No. 14-12-001-30032, Minerals Management Service, Pacific OCS Region, Los Angeles, California, 84 pp. + Appendix. Shannon, C.E. and Weaver, W.: 1964, The Mathematical Theory of Communication, University of Illinois Press, Urbana, Illinois, 125 pp. Simpson, E.H.: 1949, 'Measurement of Diversity', Nature (Lond.) 163, 688. Simpson, G.G., Roe, A., and Lewontin, R.C.: 1960, Quantitative Zoology, revised ed., Harcourt, Brace and World, New York, 440 pp. Sokal, R.R. and Rohlf, EJ.: 1981, Biometry. The Principles and Practice of Statistics in Biological Research, second ed., W.H. Freeman and Company, San Francisco, California, 859 pp. Stull, J.K., Haydock, C.I., Smith, R.W., and Montagne, D.E.: 1986, 'Long-Term Changes in the Benthic Community on the Coastal Shelf of Palos Verdes, Southern California', Mar. Biol. (Berl.) 91, 539-551. Swartz, R.C., Cole, EA., Schults, D.W., and DeBen, W.A.: 1986, 'Ecological Changes in the Southern California Bight Near a Large Sewage Outfall: Benthic Conditions in 1980 and 1983', Mar. Ecol. Prog. Ser. 31, 1-13. Swartz, R.C., Schults, D.W., Ditsworth, G.R., DeBen, W.A., and Cole, F.A.: 1985, 'Sediment Toxicity, Contamination, and Macrobenthic Communities Near a Large Sewage Outfall', Validation and Predictability of Laboratory Methods for Assessing the Fate and Effects of Contaminants in Aquatic Ecosystems, T.P. Boyle, Ed., American Society fQr Testing Materials, Special Technical Publication 865, American Society for Testing and Materials, Philadelphia, Pennsylvania, pp. 152-175. Taylor, W.D.: 1980, 'Comment on "Aggregration, Transformations, and the Design of Benthos Sampling Programs'", Can. J. Fish. Aquatic Sci. 37, 1328-1329. Taylor, L.R.: 1980, 'New Light on the Variance/Mean View on Aggregation and Transformation: Comment', Can. J. Fish. Aquatic Sci. 37, 1330-1332. Tetra Tech: 1985, Recommended Biological Indices for 301 (h) Monitoring Programs, Final Report TC-3953-03, U.S. Environmental Protection Agency, Office of Marine and Estuarine Protection, WH-556M, Washington, D.C., 17 pp. Toft, C.A. and Shea, P.J.: 1983, 'Detecting Community-Wide Patterns: Estimating Power Strengthens Statistical Inference', Amer. Nat. 122, 618-625. V6zina, A.E: 1988, 'Sampling Variance and the Design of Quantitative Surveys of the Marine Benthos', Mar. Biol. (Berl.) 97, 151-155. Warwick, R.M.: 1986, 'A New Method for Detecting Pollution Effects on Marine Marcobenthic Communities', Mar. Biol. (Berl.) 92, 557-562. Warwick, R.M. and Clarke, K.R.: 1991, 'A Comparison of Some Methods for Analysing Changes in Benthic Community Structure', J. Mar. Biol. Assoc. U.S. 71,225-244. Washington, H.G.: 1984, 'Diversity, Biotic and Similarity Indices. A Review with Special Relevance to Aquatic Ecosystems', Water Res.18, 653-694.

OPTIMUM MACROBENTHICSAMPLINGPROTOCOL

15 3

Woodin, S.A.: 1974, 'Polychaete Abundance Pattems in a Marine Soft-Sediment Environment: the Importance of Biological Interactions', Ecol. Monogr. 44, 171-187. Word, J.Q.: 1976, 'Biological Comparison of Grab Sampling Devices', South. Calif. Coastal Water Res. Proj. Annu. Rep. 1976, 189-194. Word, J.Q.: 1978, 'The Infaunal Trophic Index', South. Calif. Coastal Water Res. Proj. Annu. Rep. 1978, 19-39. Word, J.Q.: 1980, 'Classification of Benthic Invertebrates into Infaunal Trophic Index Feeding Groups', South. Calif. Coastal Water Res. Proj. Biennial Rep. 1979-1980, 103-121. Word, J.Q. and Mearns, A.J.: 1978, 'The 60-Meter Control Survey', South. Calif. Coastal Water Res. Proj. Annu. Rep. 1978, 41-56. Word, J.Q., Striplin, P.L., and Tsukada, D.: 1980, 'Effects of Screen Size and Replication on the Infaunal Trophic Index', South. Calif. Coastal Water Res. Proj. Biennial Rep. 1979-1980, 123130. Zahl, S.: 1977, 'Jackknifing an Index of Diversity', Ecology 58, 907-913. Zar, J.H.: 1984, Biostatistical Analysis, second ed., Prentice-Hall, Englewood Cliffs, New Jersey, 718 pp.