Vol.

81: 197-204, 1992

MARINE ECOLOGY PROGRESS SERIES

Mar. Ecol. Prog. Ser.

Published April 21
- -

Spatial variation in soft-sediment benthos

D. J. Morrisey, L. Howitt, A. J. Underwood, J. S. Stark
Institute of Marine Ecology, Zoology Building A08, University of Sydney, NSW 2006, Australia

ABSTRACT V a r ~ a t ~ o(patchiness)
n
In the d ~ s t r ~ b u t i oofn organisms and other env~ronmentalvanables
exists at different spatial scales This patchiness has Important implicat~onsfor comparative and
descr~pt~v
studies
e
of d~stnbutionand abundance because ~tcomplicates comparisons of abundance at
the largest s p a t ~ a scales
l
Although the existence of patchiness has been recognised for a long time, it
has not been adequately addressed in most studies of m a n n e soft s e d ~ m e n t swhich are often confounded (or pseudoreplicated') because of a lack of appropriate spatlal replicat~on Spatial vanation In
the distnbution of soft-sediment macrofauna In Botany Bay, Australia 1s described using a nested
hierarchical sampling design Significant vanatlon was detected at spatial scales from 10 m to 3 5 km
Implications of patchiness for environmental sampllng and monitoring and the means of overcoming
associated problems are discussed The present study concerns the abundance of macrofauna but the
conclusions are general and are relevant to s t u d ~ e sof other vanables such as pollutants

INTRODUCTION

The patchy distnbution of benthic organisms in

marine soft sediments has been recognised for a long
time (Barry & Dayton 1991, McIntosh 1991). The distributions of other sediment-related variables, such as
pollutants (Krumgalz et al. 1989, Luoma & Phillips
1988) and sediment particle-size (e.g. Stanley & Swift
1976), are also likely to be heterogeneous.
Causes of patchiness in the distributions of benthic
invertebrates have been examined in several studies.
Physical environmental factors, such as water depth
and movement and sediment type, are believed to
determine large-scale patterns of distribution (e.g.
Thorson 1957, Gray 1974, Warwick & Davies 1977,
Barry & Dayton 1991). Within these patterns, however,
spatial heterogeneity exists at various scales, forming a
mosaic of patches. It has been suggested that heterogeneity is maintained by a variety of disturbances and
other biotic and abiotic factors (Thrush et al. 1989,
Barry & Dayton 1991).
Studies of the distribution of fauna, pollutants and
other variables in soft sediments often involve
sampling at widely-spaced locations (for example,
kilometres or tens of kilometres apart). Such studies
include those concerned with monitoring for environmental impacts of human activities. A common
0 Inter-Research/Pnnted In Germany

practice in these surveys is to collect a number of replicate samples (or, in some cases, only one sample) at
each of several locations or stations (e.g. Swartz et al.
1986, Elefthenou & Basford 1989, Hornung et al. 1989,
Duineveld et al. 1990, Newel1 et al. 1991; see review
in Butman 1987). There are, however, important problems of interpretation of the data arising from this
practice.
The spatial scales of patchiness in the variables
being measured a r e not often known before sampling
is done. Consequently, patchiness at any spatial scales
between that of the sampling units (small scale) and
the locations sampled (large scale) will not be revealed
by the sampling design. The within-location variation
has not been adequately estimated by the replicate
samples, preventing valid comparisons among locations (the data are 'pseudoreplicated': Hurlbert 1984).
The bigger the difference of scale between one level
of sampling (e.g, replicate grab samples) and the next
(e.g. locations), the bigger will be the possible range of
scales of patchiness that will be ~nterposed.Examples
where comparisons between locations several lulometres apart are made on the basis of replicate samples
collected at only one place in each location are given by
Pearson (1975),Botton (1979) and Swartz et al. (1986).
Examples at a smaller scale (hundreds of metres) are
given by Larsen (1979) and Newel1 et al. (1991).

Mar. Ecol. Prog. Ser. 81: 197-204, 1992

The problem of confounding discussed above can

be overcome by the use of nested sampling designs
(Green & Hobson 1970, Green 1979, Underwood 1981,
Andrew & Mapstone 1987). Each of a series of successlvely smaller spatial scales is nested within the scale
above. This provides an estimate of the contribution of
each scale to the total variation among samples within
the largest scale of comparison. Partitioning of the variances associated with each scale, using analysis of
variance (Underwood 1981), permits unconfounded
comparisons among mean abundances of organisms
(or other variables) at any of the chosen scales.
Here we examine variation in the fauna of soft sediments at a range of spatial scales, using a nested
sampling design. The study was done in Botany Bay,
New South Wales, Australia. Very few studies of
the fauna in soft sediments have been done using
sampling designs capable of unravelling the confounded patterns of abundance at different spatial
scales (notable exceptions are Green & Hobson 1970,
Phillips & Fleeger 1985, Jones et al. 1990). This account
therefore demonstrates the importance of nested
designs, with appropriate replication at each scale, in
studies of the distribution of variables such as abundances of orqanisms
and concentrations of pollutants.
It also illustrates the consequences of inadequate
sampling designs to the interpretation of differences in
abundance at large spatial scales in heterogeneous
environments.

BOTANY BAY

b
SIDESOFBAY

KURNELL

LOCATIONS

SI~E
PLOTS

YARRA
A

D
j

36

Fig. 1 ( a ) Botany Bay showing locations sampled: Kurnell A

and B, Yarra C and D. Water depth at all locations was 6 m.
(b) Hierarchical sampling design used in the study. Three
Plots were nested within each of 3 Sites (represented here
by lower-case letters], nested within each of 2 Locations
(upper-case letters), nested within each Side of the Bay. Three
replicate cores were collected at each Plot

METHODS

simply reflect the fact that they are different localities

Sampling design. The design incorporated 5 spatial
scales (Fig. l b ) , ranging from 1 m between replicate
samples to 3.5 km between the 2 sides of Botany Bay.
The largest scale represents the sort of distance at
which areas are sampled in many published studies of
distribution and environmental impact. Replication at
this distance (3.5 km) within each side of the bay was
not possible in Botany Bay, but this should be considered in other studies. Instead, replicate locations
(A, B and C, D in Fig. 1) were spaced at 1 km apart
on each side of the bay.
Use of these replicate locations allows unconfounded
comparisons of the 2 sides of the bay. For example,
consider the situation with a single control area and a
single putatively disturbed area 3.5 km away from it
(such as an area with chemical contamination). These
are to be sampled to detect any influence of contaminants and sampling is done in a relatively small
part of each area. If there are significantly large differences in abun.dance from any one location to any other
at, say, a distance of 1 km apart, any differences
between the 2 areas (control and contaminated) may

- not that one is contaminated. To detect effects of

contaminants, sampling must demonstrate more difference between the contaminated and control areas than
is found from location to location within them. There
are other problems with this example (Underwood
1991a) because of the lack of replication, but this is not
the topic addressed here.
It is typical of published accounts of sampling
organisms in sediments to take replicate samples
within short distances of each other relative to the
distance apart of the areas being sampled. This, of
itself, creates a small scale of sampling within an area.
Thus, samples taken a few metres or tens of metres
apart in 2 locations 1 km apart are really only in a small
patch in each location. Again, there may be variation
in abundances of organisms from patch to patch that
would confound any larger-scale differences from
location to location.
To examine the potential need for sampling at different scales and to obtain data for cost-benefit analyses
(Snedecor & Cochran 1967, Underwood 1981, Kennelly
& Underwood 1984, 1985) for the design of future sam-

Morrisey et al.: Benthic spatlal variation

pling programmes, 2 other scales (hundreds a n d tens

of metres apart) were examined in the present study.
In each Location, 3 Sites (areas ca 50 m in diameter)
100 m apart were randomly selected In each Site, 3
replicate Plots (about 2 m in diameter) 10 m apart were
randomly chosen. Finally, in each Plot, 3 replicate
cores (see details below) were sampled, 1 m apart.
Sites and Plots were chosen to represent the spatial
scale of replication in different published studies. Most
studies would sample the equivalent of the 2 Sides of
the Bay with individual replicate samples spaced at
distances of hundreds of metres (i.e. Sites), tens of
metres (i.e. Plots) or smaller distances apart.
Sampling methods. Samples were collected during
3 consecutive days in September 1990 by divers using
plastic core tubes (10 cm diameter, 10 cm deep). Pilot
tests had shown that the precision of the estimates of
fauna1 numbers obtained with this size of core was
acceptable (SE/mean = 0.1; Elliott 1977). Numbers of
animals collected with 20 cm deep cores were no
greater than with 10 cm deep cores, justifying the
depth of core used. Samples were fixed with 7 %
formalin to prevent deterioration before extraction of
the fauna. They were sieved through a 0.5 mm mesh.
Identification of the fauna was to a relatively coarse
level; families in the case of polychaetes, sub-orders or
higher taxa for other groups. This level of resolution
was appropriate given the great taxonomic uncertainty
of many infaunal animals in southeastern Australia.

Eight of the taxa in the samples were chosen a s

being consistently present a n d representing different
relative abundances (small, medium and large; Fig. 2).
These, plus the total numbers of individual animals
and the total numbers of taxa, were each analysed by
4-factor nested analyses of variance. All factors (Side
of Bay, Locations, Sites and Plots) were random.
Homogeneity of variances was checked using
Cochran's test and heterogeneity was removed by
log ( X + 1) or ( X + 1)' transformations.

RESULTS
For the number of taxa, the number of individuals
a n d the 8 separate taxa analysed, differences in mean
abundance were observed at all of the spatial scales
examined (Tables 1 & 2; Fig. 2). The pattern of differences a t various spatial scales is not the same for each
taxonomic group (Table 2). For example, whilst the
mean number of polychaetes varied at spatial scales of
Location, Site a n d Plot, the component families showed
different patterns. Of the more abundant families,
syllids were patchy in abundance at small spatial scales
of 10's and 100's of metres (i.e. among Plots and Sites)
whereas numbers of spionids were significantly greater
at Kurnell than at Yarra (km's apart) a n d no other
spatial scales were significant. Sabellids were significantly variable at the scales of Locations and Plots.

Table 1. Summaries of analyses of variance for selected taxa

Source of
variation

df

1
Sides of Bay
Location(S)
2
Site (L(S))
8
Plot (Si(L(S))) 24
Residual
72

Source of
variation

df

1
Sides of Bay
Location (S)
2
Site (L(S))
8
Plot (Si(L(S))) 24
Residual
72

Source of
variation

df

Sides of Bay
1
2
Location (S)
8
Site (L(S))
Plot (Si(L(S))) 24
72
Residual

a . Total no. taxa

Mean Sq. F
p
389.12 0.54 >0.50
722.46 11.45 <0.01
63.12 2.39 <0.05
26.37 3.61 <0.001
7.31

e . Spionids
Mean Sq. F
p
14421.33 47.43 <O 05
304.07 0.80 >0.40
379.96 1.95 >0.05
194.51 1.56 >0.05
124.93
i. Caprellids2
p
Mean Sq. F

29.33
35.75
4.13
0.95
0.36

0.82
8.66
4.36
2.63

>0.45
<0.01
<0.005
<0.001

b. Total individuals'
MeanSq. F
p

2181.26
697.44
236.63
77.69
11.77

3.13
2.95
3.05
6.60

>0.20
>0.10
<0.02
<0.001

f . Sabellids'
MeanSq. F
p

c . Total polychaetes2
Mean Sq. F
p
23.65
19.52
3.10
0.75
0.15

1.21
6.30
4.12
5.15

>0.30
<0.05
<0.005
<0.001

g . Cirratulids2
Mean Sq. F
p

d Sylllds2
Mean Sq. F
3.09
15.41
4.57
0.55
0.21

0.20
3.37
8.37
2.57

p
>0.65
>0.05
<0.001
<0.005

h. ~ m p h i p o d s '
Mean Sq. F
p

147.42 12.01 >0.05

12.28 14.76 <0.005
0.83 2.30 >0.05
0.36 2.05 <0.05
0.18
j. Bivalves
MeanSq. F
p

8.33 0.07
124.81 7.02
17.77 2.01
8.84 1.81
4.90

>0.80
<0.05
>0.05
<0.05

'

Data transformed to ( X + l)" before analysis

'Log ( X + l ) transform

Mar. Ecol. Prog. Ser. 81: 197-204. 1992

200

Table 2. Results from the spatial var~ationstudy in Botany

Bay. Spatial scales at bvhlch significant variation exists
among replicates are shown -: differences not significant;
'significant differences ( p < 0.05)
Taxon

Spatial scale
Sides of Bay Location Site

Total no. taxa

Total no. individuals

Plot

Large abundance:
Polychaetes
Amphipods
Medium abundance:
Syllids
Spionids
Sabellids
Caprellids
Small abundance:
Cirratulids
Bivalves

There were more significant differences at the

smaller spatial scales of Sites (7 of the 10 analyses) and
Plots (8 out of 10) than at the larger spatial scales of
Locations (5 out of 10) and Sides of Bay (2 out of 10).

These spatial scales were not, however, equally important in terms of the magnitude of the variations
observed.
The contribution of each spatial scale to the total
variance was calculated for each of the taxa analysed
(Table 3 ) . Interpretation of variance components is
not straightforward and should be done with caution
(Underwood & Petraitis 1991). For example, the size of
the residual variance in the analysis of variance for
each taxon will affect the sizes of the contributions of
the different spatial scales to the total variance. In the
present study, the proportion of the total variance
contributed by the residual variances differed among
the different taxa by a factor of 4 (10 to 4 5 %; Table 3).
This makes it difficult to compare the relative importance of different spatial scales across taxa. The
components of variance do, however, indicate that
there is considerable residual (i.e. within-Plot) variation for most taxa. This suggests that patchiness exists
at smaller spatial scales than Plots.
The variance components also confirm the relative
importance of the Sites scale in the distribution of
syllids and the Sides of Bay scale for spionids, sabellids
and cirratulids. They indicate, too, that in some instances certain scales are important even though they
were not detected as being significant by the analysis

total no, taxa

sabellids

total individuals

cirratulids

polychaetes

amphipods

site
location
side of bay

e
B

KURNELL

syllids

caprellids

spionids

bivalves

~k
D

YARRA

site
location
side of bay

e
B

KURNELL

k
D

YARRA

Fig. 2. Mean (k SE) numbers 01 taxa or animals per core (n = 3) in each Plot. There are 3 Plots in each of 3 Sites in each of
2 Locations in each Side of the Bay. The category 'arnph~pods'excludes caprellids, which are shown separately

Morrisey e t al.. Benthic spatial variation

Table 3 Variance estimates derived from the analyses of variance for selected taxa (calculated from untransformed data
according to Underwood 1981; negative estimates are assumed to be zero)

Source of
variation

Total
Total
Polychaetes
no. taxa no. ind.

0
Sides of Bay
24
Location (S)
4
Site (L(S))
Plot (Si(L(S))) 6
Residual
7

25035
6508
25059
43095
26130

0
3240
650 1
1394
1882

Syllids Spionids Sabellids Cirratulids Amphipods Caprellids Bivalves

0
394
1236
414
590

261
0
21
23
125

of variance. For example, the apparent difference

between Sides of Bay for sabellid polychaetes in Fig. 2
was not significant in the analysis of variance. This is
likely to be due to the significant difference among
Locations within Sides of Bay and the consequent very
small power of the test for the difference between
Sides of Bay (1 and 2 degrees of freedom). The power
is, in fact, 0.34, with probability of Type I error set at
p = 0.05. Comparison of the variance components
shows the component for Sides of Bay to be relatively
large (Table 3). Similarly, the large differences among
Plots in the mean numbers of amphipods obscures the
importance of variation at larger spatial scales. Again,
the large variance component for Sides of Bay indicates that the differences across the Bay were not
detected because of the lack of power.
The object of this study was to examine variances at
different spatial scales, rather than means associated
with the particular areas sampled (hence the random
choice of these areas). Some consistent patterns do,
however, emerge from consideration of the means and
variances. As discussed below, such patterns may
provide useful pointers for future work. For example,
at Location D, Yarra Bay, Site k had greater numbers
of individuals in all 5 comparisons among Sites at
this Location where differences were significant
(shown by SNK tests on mean number of taxa, total
numbers of individuals, total numbers of polychaetes,
syllids and caprellids; Fig. 2, Table 2 ) . These differences corresponded with the presence of tubebuilding chaetopterid polychaetes at Site k, as discussed below.
DISCUSSION

This study shows that abundances of infauna in soft

sediments are patchy at a range of spatial scales, from
a metre up to several kilometres. It is also, of course,
possible that such heterogeneity exists at scales less
than 1 m. Volckaert (1987), using spatial autocorrelation analyses, detected patchiness in the distribution
of polychaetes in soft sediments at scales of < l 0 cm,
10 to 50 cm and >50 cm. Thrush (1986a, b) described

55
20
3
3
9

5
1
1
0
4

26539
0
7710
31011
17124

0
65
2
22
61

0
4
1
1
5

the effects on benthic communities of the pit-digging

activities of crabs and of decomposing accumulations
of seaweed. Both of these disturbances operated at
scales of less than a metre.
Factors influencing spatial heterogeneity in the
distribution of soft-sediment macrofauna have been
mentioned in the 'Introduction'. In a recent review,
Barry & Dayton (1991) consider large-scale (both
spatial and temporal) abiotic factors as defining broad
patterns of distribution. Within these patterns, other
processes operate continually at smaller temporal and
spatial scales to modify distributions and abundances.
These smaller-scale factors may be biotic, abiotic or
involve interactions between the two (Thrush 1991).
Many will operate at a range of scales, not necessarily
by the same process at each scale.
Patterns of variation in studies such as the present
one may provide pointers to the mechanisms causing
the observed variation. An example from our study has
been mentioned above. The larger numbers of taxa
and of individuals of several taxa at Site k compared
with Sites j and 1 at Location D, Yarra, coincided with
the presence of patches, up to several metres in diameter, of the tubes of chaetopterid polychaetes Mesochaetopterus sagittarius (unpubl. data). Chaetopterids
were also present at Yarra D, Site 1, Plot 35, and again
this coincided with greater numbers of individuals of
various other taxa compared with the other plots at this
site. Several other taxa, such as tanaids and eunicid
polychaetes, were more abundant at these sites. This
observation raises the possibility that the presence of
the tubes is in some way facilitating the presence of the
other animals.
The present findings have important consequences
for studies of the distribution of organisms in soft sediments, including those concerned with environmental
monitoring. The spacing among replicates in such
studies is generally of the order of a few metres, partly
depending on method of collection. With remote
sampling methods, such as grabs, the actual distance is
likely to be unknown and will vary with such factors
as water depth and water movement. Levels of the
measured variables in these replicates are, of course,

Mar Ecol. Prog. Ser. 81: 197-204, 1992

intended to be representative of levels over the location as a whole. The locations themselves are usually
spaced at distances ranging up to several kilometres.
Any spatial variation at levels between that of the
spacing of samples and the spacing of locations will be
compounded into the variation among locations in the
statistical comparison of locations. Scales of variation
smaller than the spacing of replicate samples will be
incorporated into estimates of variation among locations and among samples within each location. Thus,
such smaller-scale variations in mean abundances of
organisms will not cause errors in identification of
differences among locations. They will, however, lead
to loss of power in tests for such differences. This loss
of power will be particularly acute in situations where
organisms are distributed differently in different
patches within each location, giving rise to large variation about the mean of each location.
The problem of confounding is particularly pertinent
in studies where distributions of animals, pollutants or
other variables are extrapolated from means of replicate samples to whole locations for purposes of comparison. A common example would be for the purpose
of mapping distributions. As an example of such confounding, consider the data from the present study
concerning the distribution of numbers of polychaetes
(Fig. 2). These data showed no significant difference
between the 2 sides of the bay. Had sampling been
done only at the scales of Sides of the Bay and replicate
cores 1 m or so apart, comparisons between the 2 sides
would be spatially confounded. They might suggest,
for example, that polychaete numbers were larger at
Yarra than at Kurnell, no different, or smaller, depending on whether the replicate cores from Yarra had,
by chance, been collected from Plot 31, Plot 35 or Plot
34 respectively. The size of the variances of sample
means may give rise to caution in this respect, assuming that replicate samples have been collected.
Spacing of replicates may, however, be such that all
happen to lie within one of several different patches
within the location. In this case, among-replicate
variance will be small even though within-location
variance is large.
Comparisons for the same location over time (e.g.
Buchanan et al. 1974, 1978, Probert 1981, Buchanan
& Moore 1986, Swartz et al. 1986, Josefson 1987,
Shlllabeer & Tapp 1989) are also likely to be confounded by small-scale spatial variation, because the
samples will not necessarily come from the same type
of patch at each tlme of sampling (because softsediment sampling is invariably destructive and, in the
case of remote sampling, because of inaccuracy in
positioning the sampler).
The problems associated with the spatial scales
of sampling can be overcome by using the kind of

nested design described here. The scales can be

chosen arbitrarily and adapted to the objectives of the
particular study. Given the often large costs of collecting and sorting samples, it is usually appropriate to do
a pilot study to identify scales at which variation is
significant.
Spatial autocorrelation has been used to examine
patterns of distribution (Jurnars et al. 1977, Volckaert
1987, McArdle & Blackwell 1989, Thrush et al. 1989).
This technique allows description of small-scale spatial patterns, larger-scale studies being impractical because of the large numbers of samples required. It is
not, therefore, an alternative to the nested sampling
design used here for the identification of ranges of
spatial scales of variation. The nested design has an
important practical advantage over autocorrelation. As
mentioned, the nested design permits cost-benefit
analyses to be done and the results to be applied
to subsequent sampling. In studies of impacts, costbenefit and power analyses will be relevant to the
design of sampling to detect what are considered, a
pnori, to be biologically or environmentally significant
changes in the fauna (see references in Underwood
1981, Kennelly & Underwood 1984, 1985). One of the
aims of the present study was to enable us to estimate
the magnitude of effects that would be detectable in
experimental manipulations of the fauna in Botany
Bay.
While the present study has concentrated on the
distribution of macrofauna, other sediment-related
variables, such as pollutants, are likely to show similar
variation. Nested designs are therefore appropriate to
sampling these variables. Correlation among scales of
patchiness between different variables can then be
examined. These may suggest causal relationships, or
associations with common covariables.
Finally, it is important to note that the spatial patterns identified in this study are likely to change with
time. Interactive effects of space and time are probable; there is no reason to presume that temporal
trajectories of mean abundance will be exactly the
same for all scales of distribution of the organisms.
Short-term temporal variation will have similar effects
of confounding or reduction of power on longer-term
comparisons to the small-scale spatial variation considered above (Underwood 1991a, b).
In conclusion, much more care is needed in the identification of appropriate spatial scales for sampling
before conclusions are reached about differences in
abundances of organisms from one place to another in
soft sediments. Although this is supposedly w~dely
known, few studies demonstrate that appropriate
sampling has been done to unconfound the effects of
smaller-scale spatial differences when larger-scale
comparisons are made.

Morrisey et al.: Benthic spatial variation

Acknowledgements. We thank G . Inglis, S. McCune and

J . Miller for help in collecting and sorting samples, and
K. Astles for help with everything, especially computing. This
study was supported by an ARC grant (to A.J.U.) and funds
from the Institute of Marine Ecology and the Research Grant
of the University of Sydney.
LITERATURE CITED
Andrew. N. L., Mapstone, B. D. (1987). Sampling and the
description of spatial pattern in marine ecology. Oceanogr.
mar. Biol. Ann. Rev 25: 39-90
Barry, J. P., Dayton, P. K. (1991). Physical heterogeneity and
the organization of marine communities. In: Kolasa, J . ,
Pickett, S. T. A. (eds.) Ecological heterogeneity. SpringerVerlag. New York. p. 270-320
Botton. M. L. (1979). Effects of sewage sludge on the benthic
invertebrate community of the inshore New York Bight.
Estuar. coast. Shelf Sci. 8: 169-180
Buchanan, J. B., Kingston, P. F., Sheader, M. (1974). Longterm population trends of the benthic macrofauna in the
offshore mud of the Northumberland coast. J mar. biol.
ASS.U.K. 54: 785-795
Buchanan, J . B., Moore, J. J . (1986). A broad review of
variability and persistence in the Northumberland benthlc
fauna - 1971-85. J , mar. biol. Ass. U.K. 66: 641-657
Buchanan, J . B., Sheader, M,, Kingston, P. F. (1978). Sources
of variability in the benthic macrofauna off the south
Northumberland coast, 1971-1976. J . mar. biol. Ass. U.K.
58: 191-209
Butman, C. A. (1987). Larval settlement of soft-sediment
invertebrates: the spatial scales of pattern explained by
active habitat selection and the emerging role of hydrodynamical processes. Oceanogr. mar. Biol. Ann. Rev. 25:
113-165
Duineveld, G. C. A., d e Wilde, P. A. W. J., Kok, A. (1990). A
synopsis of the macrobenthic assemblages and benthic
ETS activity in the Dutch sector of the North Sea. Neth. J .
Sea Res. 26: 125-138
Eleftheriou. A., Basford, D. J. (1989). The macrobenthic
infauna of the offshore southern North Sea. J. mar. biol.
ASS.U.K. 69: 123-143
Elliott, J . M. (1977). Some methods for the statistical analysis
of samples of benthic invertebrates, 2nd edn. Freshwater
Biological Association, Ambleside
Gray, J . S. (1974). Animal-sediment relationships. Oceanogr.
mar. Biol. Ann. Rev. 12: 223-261
Green, R. H. (1979). Sampling design and statistical methods
for environmental biologists. Wiley, Chichester
Green, R. H., Hobson, K. D. (1970). Spatial and temporal
structure in a temperate intertidal community, with
speclal emphasis on Gemma gemma (Pelecypoda:
Mollusca). Ecology 51: 999-1011
Hornung, H., Krom, M. D., Cohen, Y. (1989). Trace metal
distribution in sediments and benthic fauna of Haifa Bay,
Israel. Estuar. coast. Shelf Sci. 29: 43-56
Hurlbert, S. H. (1984). Pseudoreplication and the design of
ecological field experiments. Ecol. Monogr. 54: 187-211
Jones, G. P,, Ferrell, D. J., Sale, P. F. (1990).Spatial pattern in
the abundance and structure of mollusc populations in the
soft sediments of a coral reef lagoon. Mar. Ecol. Prog. Ser.
62: 109-120
Josefson, A. B. (1987). Large-scale patterns of dynamics in
subtidal macrozoobenthic assemblages in the Skagerrak:
effects of a production-related factor? Mar. Ecol. Prog. Ser.
38: 13-23

Jumars, P. A , Thistle, D . , Jones, M. L. (1977).Detecting twodimensional spatial structure in biological data. Oecologia
(Berl ) 28: 109-123
Kennelly, S. J . , Underwood, A. J . (1984).Underwater microscope sampling of a sublittoral kelp community. J . exp.
mar. Biol. Ecol. 76: 67-78
Kennelly, S. J., Underwood, A. J . (1985). Sampling of small
invertebrates on natural hard substrata in a sublittoral
kelp forest. J . exp. mar. Biol. Ecol. 89: 55-67
Krumgalz, B. S., Fainshtein, G., Sahler, M., Gorfunkel, L.
(1989).'Field error' related to marine sediment contamination studies. Mar. Pollut. Bull. 20: 64-69
Larsen, P. F. (1979). The shallow-water macrobenthos of a
northern New England estuary. Mar. Biol. 55: 69-78
Luoma, S. N., Phillips, D. J . H. (1988).Distribution, variability,
and impacts of trace elements in San Francisco Bay. Mar.
Pollut. Bull. 19: 413-425
McArdle, B. H.. Blackwell, R. G. (1989).Measurement of density variability in the bivalve Chione stutchburyi using
spatial autocorrelation. Mar. Ecol. Prog. Ser. 52: 245-252
McIntosh. R. P. (1991). Concept a n d terminology of homogeneity and heterogeneity in ecology. In: Kolasa, J.,
Pickett, S. T. A . (eds.) Ecological heterogeneity. SpringerVerlag, New York, p. 24-46
Newell, R. C . , Maughan, D. W., Trett, M. W., Newell, P. F.,
Selderer, L. J . (1991). Modification of benthic community
structure in response to acid-iron wastes discharge. Mar.
Pollut. Bull. 22. 112-118
Pearson, T. H . (1975).The benthic ecology of Loch Linnhe and
Loch Eil, a sea-loch system on the west coast of Scotland.
IV. Changes in the benthic fauna attributable to organic
ennchment. J . exp. mar. Biol. Ecol. 20: 1-41
Phillips, F. E., Fleeger, J . W. (1985). Meiofauna meso-scale
variability in two estuarine habitats. Estuar. coast. Shelf
Sci. 21: 745-756
Probert. P. K. (1981). Changes in the benthic community of
china clay waste deposits in Mevagissey Bay following a
reduction of discharges. J . mar. biol. Ass. U.K. 61: 789-804
Shillabeer, N.. Tapp. J. F. (1989). Improvements in the benthic
fauna of the Tees Estuary after a period of reduced pollution loadings. Mar. Pollut. Bull. 20: 119-123
Snedecor, G. W., Cochran, W. G. (1967). Statistical methods,
6th edn. University of Iowa Press
Stanley, D. J., Swift. D. J. P. (1976).Marine sediment transport
and environmental management. Wiley, New York
Swartz, R. C., Cole, F. A., Schults. D. W., DeBen, W. A. (1986).
Ecological changes in the southern California Bight near a
large sewage outfall: benthic conditions in 1980 a n d 1983.
Mar. Ecol. Prog. Ser. 31: 1-13
Thorson, G. (1957). Bottom communities (sublittoral or shallow
shelf).In:Hedgpeth, G. (ed.)Treatise on marine and palaeoecology, Vol. 1. Mem. Geol. Soc. Am., no. 67, p. 461-534
Thrush, S. F. (1986a). Spatial heterogeneity in subtidal gravel
generated by the pit-digging activities of Cancerpagurus.
Mar. Ecol. Prog Ser. 30: 221-227
Thrush, S. F (1986b). The sublittoral macrobenthic comrnunlty structure of an Irish sea-lough: effect of decomposing
accumulations of seaweed. J . exp. mar. Biol. Ecol. 96:
199-212
Thrush, S. F. (1991).Spatial patterns in soft-bottom communities. TREE 6. 75-79
Thrush, S. F., Hewitt, J. E., Pridrnore, R. D. (1989).Patterns in
the spatial arrangements of polychaetes and bivalves in
intertidal sandflats. Mar. Biol. 102: 529-535
Underwood, A. J . (1981).Techniques of analysis of variance in
experimental marine biology and ecology. Oceanogr. mar.
Biol. Ann. Rev. 19: 513-605

204

Mar. Ecol. Prog. Ser. 81 197-204, 1992

Underwood, A. J. (1991a). Beyond BACI: experimental

designs for detecting human environmental impacts on
temporal variations in natural populations. Aust. J mar.
Freshwat. Res 42. 569-587
Underwood, A . J (1991b). Spatial and temporal problems
with monitoring. In: Calow. P., Petts, G. E. (eds.) Rivers
handbook Vol. 2. Blackwell Scientific, Oxford (in press)
Underwood, A. J., Petraitis, P. S. (1991).Structure of intertidal
assemblages in different locations: how can local pro-

cesses be compared. In: Ricklefs, R. E., Schluter, D. (eds.)

Historical and geographical determinants of community
diversity. University of Chicago Press (in press)
Volckaert, F. (1987) Spatial pattern of soft-bottom Polychaeta
off Nova S c o t ~ aCanada.
,
Mar. Blol. 93: 627-639
Warwick. R. M,. Davies, J. R. (1977). The distribution of
sublittoral macrofauna communities in the Bristol Channel
in relation to the substrate. Estuar. coast. Shelf Sci. 5:
267-288

This art~clewas submitted to the editor

Manuscript first received: December 12, 1991

Revised version accepted: March 10, 1992