Accepted Manuscript The value of enduring environmental surrogates as predictors of estuarine benthic macroinvertebrate assemblages Michelle D. Wildsmith, Fiona J. Valesini, Samuel F. Robinson PII:
S0272-7714(17)30181-6
DOI:
10.1016/j.ecss.2017.08.006
Reference:
YECSS 5561
To appear in:
Estuarine, Coastal and Shelf Science
Received Date: 14 February 2017 Revised Date:
1 July 2017
Accepted Date: 6 August 2017
Please cite this article as: Wildsmith, M.D., Valesini, F.J., Robinson, S.F., The value of enduring environmental surrogates as predictors of estuarine benthic macroinvertebrate assemblages, Estuarine, Coastal and Shelf Science (2017), doi: 10.1016/j.ecss.2017.08.006. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
D M A
ACCEPTED MANUSCRIPT 1
The value of enduring environmental surrogates as predictors of estuarine benthic macroinvertebrate
2
assemblages
3 Michelle D. Wildsmith*1, Fiona J. Valesini2, Samuel F. Robinson2
4
Oceanvision Environmental Research Pty Ltd, c/o Challenger Institute of Technology, 1 Fleet Street,
6
Fremantle, Western Australia, 6160. 2
7
Centre for Fish and Fisheries Research, School of Veterinary and Life Sciences, Murdoch University,
8
Murdoch, Perth, Western Australia, 6150.
SC
9 10
*Corresponding author. Email:
[email protected], Tel: +61 430 203 162, Fax: +61 (0)8 92381332
11 Abstract
M AN U
12
RI PT
1
5
13 14
This study tested the extent to which spatial differences in the benthic macroinvertebrate assemblages of a temperate microtidal estuary were ‘explained’ by the enduring (biophysical) vs non-enduring (water and
16
sediment quality) environmental attributes of a diverse range of habitats, and thus the potential of those
17
environmental surrogates to support faunal prediction. Species composition differed significantly among
18
habitats in each season, with the greatest differences occurring in winter and spring and the least in summer. The
19
pattern of habitat differences, as defined by their enduring environmental characteristics, was significantly and
20
well matched with that in the fauna in each season. In contrast, significant matches between the non-enduring
21
environmental and faunal data were only detected in winter and/or spring, and to a lesser extent. Field validation
22
of the faunal prediction capacity of the biophysical surrogate framework at various ‘test’ sites throughout the
23
estuary showed good agreement between the actual vs predicted key species. These findings demonstrate that
24
enduring environmental criteria, which can be readily measured from mapped data, provide a better and more
25
cost-effective surrogate for explaining spatial differences in the invertebrate fauna of this system than non-
26
enduring criteria, and are thus a promising basis for faunal prediction. The approaches developed in this study
27
are also readily adapted to any estuary worldwide.
AC C
EP
TE D
15
28 29
Key words: enduring surrogates, estuary; infauna; habitat-faunal relationships; benthic ecology; faunal
30
prediction.
1
ACCEPTED MANUSCRIPT 31
1
Introduction
32 33
Benthic macroinvertebrates play a crucial role in the functioning of estuarine ecosystems. They are a major component of estuarine food webs and, through their burrowing and feeding activities, play important
35
roles in nutrient cycling (Hutchings, 1998; Constable, 1999). They are also widely known to provide excellent
36
indicators of aquatic environmental quality (e.g. Roach & Wilson, 2009; Pelletier et al. 2010; Warwick &
37
Somerfield, 2010; Whomersley et al. 2010). Many estuarine studies worldwide have examined spatial
38
relationships between these fauna and select environmental variables, e.g. salinity, dissolved oxygen
39
concentration, sediment characteristics and current velocity (Edgar & Barret, 2002; Ysebaert & Herman, 2002;
40
Ysebaert et al. 2002; Teske & Wooldridge, 2003). This knowledge has been used to determine key
41
environmental drivers of faunal structure, support understanding of ecosystem function, and guide estuarine
42
management and ecological health assessments (Constable, 1999; Hirst, 2004; Borja et al. 2007; Muxika et al.,
43
2007; Valenҫa & Santos, 2012; Robertson et al. 2016). More recently, such knowledge has been extended to
44
enable the prediction of how the distribution of key species may change under anticipated environmental
45
scenarios (e.g. Gogina & Zetler, 2010; Reiss et al. 2011; Cozzoli et al. 2014). Comparatively few estuarine
46
studies, however, have examined relationships between the full benthic invertebrate community and suites of
47
environmental variables that define integrated habitat types, for the purpose of then using those habitats as
48
surrogates to predict faunal characteristics. Such predictive capabilities have diverse management applications,
49
including anticipating those habitats which are most important for assemblages or species of interest, informing
50
shoreline or within-estuary development proposals, and guiding conservation planning (Ysebaert et al. 2002;
51
Thrush et al. 2003; Ellis et al. 2006; Banks & Skilleter, 2007).
EP
TE D
M AN U
SC
RI PT
34
Several workers in coastal and estuarine environments have proposed that habitats identified at local
53
scales (i.e. 10-100s metres) and defined using enduring environmental criteria (i.e. biophysical attributes that
54
undergo little or no natural change over time, e.g. site aspect, fetch, bathymetry, areal cover of substrate types)
55
rather than highly dynamic attributes (e.g. salinity, dissolved oxygen concentration, current velocity, sediment
56
grain size), are likely to provide more useful, practical and cost-effective surrogates for understanding and
57
predicting faunal distributions (Roff & Taylor, 2000; Banks & Skilleter, 2002; 2007; Roff et al. 2003; Skilleter
58
& Loneragan, 2003; Hume et al., 2007; Valesini et al. 2003, 2010). This not only reflects the fact that enduring
59
biophysical attributes can typically be measured accurately from mapped data rather than requiring high-
60
resolution field measurements, but also that they provide a consistent habitat framework irrespective of the
AC C
52
2
ACCEPTED MANUSCRIPT 61
many temporal shifts that occur in these highly dynamic environments. The latter is an important feature of
62
enduring habitat classification schemes, and reflects the concept that the underlying surrogate variables are
63
expected to capture and maintain a pattern of relative habitat differences that persists over time, regardless of the
64
specific non-enduring environmental variables that cause those spatial differences at any one time point.
65
One approach for classifying estuarine habitats that meets the above criteria is that developed by Valesini et al. (2010), which was applied to the nearshore waters of various estuaries in south-western Australia
67
but is adaptable to any system. This approach, which used a broad suite of 13 enduring environmental variables,
68
detected a far greater number of habitats than are often recognised in many studies of faunal-environment
69
relationships in estuaries, and further established that each of those habitats differed significantly in their
70
environmental composition, thereby representing distinct as opposed to perceived habitats. The pattern of spatial
71
differences among the enduring habitats was also well correlated with that defined by a suite of temporally-
72
variable water quality characteristics that are traditionally used to assess drivers of faunal change in estuaries,
73
e.g. salinity, temperature and dissolved oxygen. Moreover, this study also developed a method for assigning any
74
unclassified site to its most appropriate habitat on the basis of its enduring environmental attributes, and outlined
75
the potential to extend this into faunal prediction following robust habitat-faunal correlations.
M AN U
SC
RI PT
66
76
The main aim of this study was to test the extent to which spatial differences in the nearshore benthic macroinvertebrate community in the Swan Estuary, an urbanized system in south-western Australia, were
78
related to the enduring habitat types identified by Valesini et al. (2010) in that system, and thus the potential of
79
that habitat classification framework to forecast faunal composition at any unsampled site. We further compared
80
the extent to which any faunal differences among habitats were ‘explained’ by enduring vs non-enduring (water
81
and sediment quality) environmental variables, and thus which provides a better basis for faunal prediction. The
82
specific study objectives were as follows.
AC C
EP
TE D
77
83
1.
84
Determine whether the composition of the nearshore benthic macroinvertebrate assemblage throughout the Swan Estuary differs significantly among habitats (sensu Valesini et al. 2010) and,
85
if so, which species best characterise each habitat type.
86
2.
Test whether the pattern of differences among habitats, as defined by their faunal composition, is
87
significantly correlated with that defined by their enduring environmental attributes, and thus
88
whether spatial differences in the latter provide a sound basis for predicting the former.
89
3.
90
Test whether the pattern of any faunal differences among habitats is significantly correlated with differences in suites of non-enduring water and sediment quality variables, and compare the extent
3
ACCEPTED MANUSCRIPT 91
to which these non-enduring vs enduring environmental variables provide a better basis for faunal
92
prediction.
93
4.
Based on the outcome of Objective 3, assess the ability of the best surrogate environmental
94
framework to predict the characteristic benthic macroinvertebrate species at a range of ‘test’
95
(validation) nearshore sites in the Swan Estuary.
97
2
Materials and methods
2.1
Study area
RI PT
96
99
SC
98
The Swan Estuary is a permanently-open, wave dominated estuary on the lower west coast of Australia
101
(32.055º S, 115.735º E) that is ~ 50 km long, up to 4 km wide and has a surface area of ~55 km2 (Brearly, 2005;
102
Fig. 1). It is a shallow (typically < 5 m deep) drowned river valley system comprising a narrow entrance channel
103
which opens into a wide central basin, then a second smaller basin, and is fed by two tributaries, the Swan and
104
Canning rivers (Fig. 1). The estuary drains a large catchment of 126,000 km2 (2,100 km2 of which lies on the
105
coastal plain) that contains 75% of Western Australia’s population (Swan River Trust, 2009). The region
106
experiences a Mediterranean climate of hot dry summers (maximum mean temperature of ~31.7° C in February)
107
and cool wet winters (maximum mean temperature of ~18.4° C in July), and has a moderate to low rainfall
108
(mean ~728 mm yr-1) of which 70% occurs from May to September (Bureau of Meteorology, 2016). Tides along
109
this coast have a mean spring range of only 0.4 m and are predominantly diurnal (Department of Defence,
110
2003).
2.2
Identification of habitat types
AC C
112
EP
111
TE D
M AN U
100
113
The procedure for quantitatively identifying the suite of 18 significantly different habitats present
114
throughout the shallows (≤ 2 m deep) of the Swan Estuary is detailed in Valesini et al. (2010). In brief, this was
115
achieved by firstly measuring 13 enduring environmental criteria (Table 1), reflecting either (i) proximity to
116
marine and riverine water sources, (ii) exposure to wave activity or (iii) substrate/submerged vegetation cover,
117
at 101 sites spanning the estuary, with each site representing waters in a 100 m radius of a point on the shore.
118
These criteria were chosen for their widely recognised influences, either directly or indirectly, on the spatio-
119
temporal distribution of estuarine benthic invertebrates (Valesini et al. 2010). The first group of enduring
120
variables was intended as a surrogate for the many dynamic water and sediment physico-chemical attributes that
4
ACCEPTED MANUSCRIPT typically vary spatially throughout an estuary due to differences in marine vs riverine water inputs, such as
122
salinity, water temperature, dissolved oxygen concentration, turbidity and sediment grain size and organic
123
matter content. The second group reflected the exposure to waves generated by local winds and the impact of
124
local bathymetry on waves as they approach the shore, while the third encompasses the nature and extent of
125
benthic habitat structure. With regard to the latter, it is recognised that submerged vegetation often exhibits
126
seasonal and/or interannual changes in biomass and composition, but the overall area (percentage cover) at a
127
site, which is the unit of measurement used in this study (Table 1), often does not change markedly over these
128
time scales.
129
RI PT
121
SC
All criteria were measured at each site from mapped data in a Geographic Information System (GIS). These data were then subjected to a hierarchical agglomerative clustering analysis and Similarity Profiles test
131
(SIMPROF Type 1; Clarke et al. 2008), which identified the optimal and significantly-different (P < 0.01)
132
‘breaks’ in the dataset by determining those points in the clustering procedure at which further subdivision of
133
sites was unwarranted (i.e. no significant internal ‘structure’). Eighteen homogenous groups of sites were
134
identified, which were considered to be discrete habitats. Habitats were labelled A (most distinct) to R (least
135
distinct) based on the dissimilarity level at which they separated from the remainder in the cluster analysis.
M AN U
130
136
A subset of these habitats (A, C, F, G, I, J and M) were chosen for benthic macroinvertebrate sampling (Fig. 1). These seven habitats were selected as (i) several of the remainder precluded effective operation of the
138
sampling gear due to their rocky substrates or very steep banks (B, D, H, K, L, O, P and R) and (ii) they
139
represented a wide range of the environmental diversity throughout the estuary. Habitat A comprised eight sites
140
in the uppermost tidal portion of the Swan River and had the greatest quantity of snags (submerged tree
141
branches) and intertidal reeds, but no submerged vegetation. This habitat had a narrow wave shoaling margin,
142
but was highly sheltered from wave activity due to its limited fetches in all directions (Table 1). Habitat C
143
contained 17 sites in the middle to lower Swan and Canning rivers. It also had very little submerged vegetation,
144
but was more exposed to wave activity than A given its greater fetches and slightly steeper slope. Habitat J,
145
represented by only three sites in lower Swan and Canning rivers, had moderate direct, northerly and easterly
146
fetches but small to no fetches in other cardinal directions, a very wide wave shoaling margin and small amounts
147
of submerged vegetation (mainly Gracillaria comosa). Habitats F (9 sites) and G (12 sites) comprised vast
148
shallow flats in the main basin, had the greatest direct fetches of all habitats and contained moderate amounts of
149
submerged vegetation (mainly Halophila ovalis). These two habitats were distinguished mainly by their fetches,
150
with F having very large easterly and limited westerly fetches, while the opposite was true for G. Habitat I in the
AC C
EP
TE D
137
5
ACCEPTED MANUSCRIPT lower main basin contained moderate amounts of submerged vegetation and rock, had moderate direct,
152
southerly and westerly fetches but limited northerly and easterly fetches, and a shallow sloping substrate (Table
153
1). Habitat M, located in the estuary channel, was characterised by small-moderate fetches in most directions, no
154
northerly fetch, the narrowest wave shoaling margin and most steeply-sloping substrate of all habitats, moderate
155
amounts of rock and relatively large diverse stands of submerged vegetation (e.g. Zostera sp., Heterozostera sp.,
156
H. ovalis. G. comosa, Chaetomorpha linum, Ulva spp., Enteromorpha sp. and Cystoseria trinodis; Table 1).
RI PT
151
157 158
2.3
159
Field and laboratory techniques
SC
Replicate samples of the benthic macroinvertebrate community and measurements of various nonenduring water and sediment attributes were collected at two randomly-chosen sites from each habitat, except
161
for F and I at which only a single site was sampled due to logistical constraints at the start of the study (Fig. 1).
162
Sampling was undertaken during the day in the last month of each Austral season in 2005 (i.e. February,
163
summer; May, autumn; August, winter; November, spring), with collection of the replicates at each site being
164
staggered over a two-week period to improve representativeness of seasonal conditions and reduce
165
pseudoreplication at the seasonal scale.
166 2.3.1
Benthic macroinvertebrates
TE D
167
M AN U
160
Five randomly-located replicate sediment cores were collected at each site in each season at water
169
depths of ~0.5-1 m using a corer that was 11 cm in diameter, 10 cm deep and had a surface area of 96 cm2.
170
Samples were wet-sieved through a 500 µm mesh and all retained material immediately preserved in 5%
171
formalin buffered in estuary water. All invertebrates in each sample were removed from the sediment under a
172
dissecting microscope, identified to the lowest possible taxon and counted.
173 174
2.3.2
175
AC C
EP
168
Non-enduring environmental variables
At each site on each sampling occasion, three replicate measurements of salinity, water temperature
176
(°C), dissolved oxygen concentration (mg L-1) and pH were recorded at the bottom of the water column using a
177
Yellow Springs International multi-parameter hand held meter. Three randomly-located cores of sediment were
178
also collected using a corer that was 10 cm deep and 10 cm2 in area, and used to determine mean grain size,
179
particulate organic matter content (POM) and the depth (to the nearest 0.5 cm) of the redox transition layer, i.e.
6
ACCEPTED MANUSCRIPT 180
the sediment depth at which conditions change from oxic to anoxic. Three additional sediment cores were also
181
collected to quantify sedimentary chlorophyll a (mg g-1), which were immediately stored on ice then frozen.
182
In the laboratory, the sediment cores for grain size and POM analysis were dried at 80 °C for 24 h, weighed to the nearest mg, ashed at 550 °C for 2 h and then reweighed. The ashed sediment weight was
184
subtracted from the dried sediment weight to determine the percentage contribution of POM in each sample
185
(Heiri et al. 2001). Each ashed sample was then wet-sieved through a 63 µm sieve to remove silt and clay
186
particles, dried and then weighed. This dry weight was subtracted from the ashed sediment weight to determine
187
the amount of silt and clay in each sample (Heiri et al. 2001). The remaining sample was then wet-sieved
188
through nested meshes that corresponded with the Wentworth scale of grain-size distribution, i.e. ≥2000,
189
1000≥2000, 500≥1000, 250≥500, 125≥250 and 63≥125 µm (Wentworth, 1922). Sediment retained on each
190
mesh was dried, weighed and converted to a percentage of the total dry weight to determine the contributions of
191
each grain size fraction (Folk & Ward, 1957). Grain size distributions at each habitat in each season were
192
examined and found to be typically uni-modal (data not shown). Sedimentary chlorophyll a was measured in
193
low light conditions using the acetone extraction method described by Parsons et al. (1984).
M AN U
SC
RI PT
183
194 2.4
196 197
The following data analyses were carried out using the PRIMER v7 multivariate statistics package (Clarke & Gorley, 2015) with the PERMANOVA+ add-on module (Anderson et al. 2008).
198 2.4.1
Spatial differences in benthic macroinvertebrate species composition
EP
199
Statistical Analyses
TE D
195
200
The benthic macroinvertebrate species abundance data was initially dispersion-weighted (Clarke et al. 2006) to downweight the contributions of highly variable species, then square-root transformed to balance the
202
contributions of highly abundant and less abundant species (Clarke et al. 2014). A Bray-Curtis similarity matrix
203
was constructed from this pre-treated data, then subjected to a preliminary three-factor site(habitat)×season
204
Permutational Multivariate Analysis of Variance test (PERMANOVA; Anderson et al. 2008) to determine if the
205
site replicates could be pooled to represent habitat type. Site was treated as a random factor, while habitat and
206
season were treated as fixed. The null hypothesis of no significant group differences was rejected if the
207
significance level (P) was ≤0.05, and the components of variation values (COV) were used to gauge the
208
importance of any significant effects. As this test showed that site differences were relatively unimportant but
209
that habitat and habitat×season differences were not (see Results), the site replicates were pooled for habitat and
AC C
201
7
ACCEPTED MANUSCRIPT subsets of the above Bray-Curtis matrix containing data for each season were then separately subjected to one-
211
way Analysis of Similarities tests (ANOSIM; Clarke & Green, 1988) to more fully explore habitat differences in
212
the faunal communities. The null hypothesis was the same as above and the extent of any significant differences
213
was gauged by the R-statistic (Clarke & Green, 1988). To illustrate the nature of significant habitat differences
214
in faunal composition, the distance among centroids was calculated for each habitat×season group then
215
subjected to non-metric Multidimensional Scaling ordination (nMDS).
RI PT
210
A shade plot (Clarke et al. 2014) was then used to ascertain which species best characterized the fauna
217
in each habitat×season group. Only those species accounting for >5% of the pre-treated averaged abundances in
218
at least one group were included. Species (y-axis) were ordered according to a group-average hierarchical
219
agglomerative cluster analysis of a resemblance matrix defined between species as Whittaker’s index of
220
association (Legendre & Legendre, 1998). A SIMPROF test (Type 3; Somerfield & Clarke, 2013) was also
221
applied to identify those points in the clustering procedure at which no significant structure could be detected.
222
Samples, displayed on the x axis, were ordered by habitat then season.
M AN U
SC
216
223 224
2.4.2
Matching habitat patterns between the faunal and environmental data RELATE was used to test whether the pattern of relative differences (resemblances) among habitats, as
226
defined by their faunal assemblages, matched that defined by their (i) enduring environmental, (ii) water quality
227
and (iii) sediment quality attributes. This test was thus used to correlate, for each season, a Bray-Curtis
228
similarity matrix constructed from the pre-treated habitat averages of the faunal data, with three complementary
229
Manhattan distance matrices constructed either from the enduring, water quality or sediment quality data. Pre-
230
treatment of the enduring data is described in Valesini et al. (2010), while transformations and extent of any co-
231
linearity for the non-enduring water and sediment quality variables were determined from Draftsman plots. Data
232
transformations required were as follows: redox depth (square-root), POM (log[n+1]) and mean sediment grain
233
size and chlorophyll a concentration (fourth-root). No pairs of variables had a correlation exceeding 0.95. The
234
water and sediment quality variables were then each normalized to place all on the same (dimensionless) scale.
235
The null hypothesis of no similarity in spatial pattern was rejected if P ≤ 5%, and the extent of any significant
236
match was determined by the Spearman rank correlation coefficient (ρ). Comparisons of habitat patterns in the
237
faunal and environmental data were illustrated by subjecting each of the above matrices to nMDS ordination.
AC C
EP
TE D
225
238
The Biota and Environment matching routine (BIOENV; Clarke & Ainsworth, 1993) was then used to
239
ascertain whether a greater correlation between the faunal and each of the non-enduring environmental matrices
8
ACCEPTED MANUSCRIPT could be obtained with just a subset of the latter variables, rather than the full suite used in the RELATE tests.
241
BIOENV was thus used to maximise the possibility that the non-enduring variables may provide a better match
242
with the fauna than the enduring variables, and thus err on the conservative side before endorsing the use of the
243
latter. The null hypothesis, rejection criteria and test statistic were the same as for RELATE. Note that the
244
matrices employed in these tests were constructed from the site rather than habitat averages to maximise the
245
number of samples in the reference (faunal) matrices, and thus minimize the likelihood of BIOENV finding a
246
well-matched subset of water or sediment variables by chance. Significant matches were illustrated by
247
subjecting the relevant Bray-Curtis matrices derived from the faunal data to nMDS ordination, then overlaying
248
the selected water or sediment variables as circles of proportionate sizes (‘bubble-plots’).
249 250
2.5
251
Validating the faunal prediction approach
SC
RI PT
240
M AN U
The benthic macroinvertebrate fauna was sampled at five additional nearshore sites in the Swan Estuary that had not previously been sampled during the field regime for the original study, thus providing an
253
independent data set to assess the faunal prediction capacity of the most useful surrogate environmental
254
framework (Fig. 1). Four replicate samples were collected at each of these ‘test’ sites during summer and winter
255
of 2014/2015 using an Eckman grab that sampled an area of 225 cm2 and to a depth of 15 cm. Field and
256
laboratory processing of these samples was the same as described above. Although this sampling method
257
differed from the corer used in the current study, all species abundances were adjusted to the same density
258
(individuals 0.1 m-2).
259
TE D
252
EP
Given that the enduring biophysical variables used to define habitats provided the best environmental surrogate for characterising the spatial differences in the faunal assemblage (see Results), measurements for the
261
13 enduring variables listed in Table 1 were recorded for each of the five test sites, then used to assign those
262
sites to their most appropriate habitats using the habitat prediction approach developed by Valesini et al. (2010).
263
This latter approach comprised a novel application of LINKTREE (Clarke et al., 2008), a non-metric
264
multivariate regression tree technique. The level of agreement on the species that best characterised the original
265
(‘benchmark’) vs test site representatives of the selected habitats and seasons was determined by comparing
266
complementary shade plots constructed from the two data sets. These plots were constructed using the same
267
methodology as described above.
AC C
260
268 269
3
9
Results
ACCEPTED MANUSCRIPT 270 271
3.1
272
Benthic macroinvertebrate assemblages Sampling of the benthic macroinvertebrate assemblages at seven habitats throughout the Swan Estuary in
2005 yielded 314 944 individuals, following adjustment of the species abundances to individuals per 0.1 m2.
274
The assemblage comprised 69 species from seven phyla and six classes, of which the Polychaeta (32 species),
275
followed by the Crustacea and Bivalvia (12-13 species), were the most speciose and comprised most
276
invertebrates, i.e. ~50, 20 and 13%, respectively (Appendix A). The most speciose habitat by far was the
277
channel habitat M (47 species), while the least speciose was A in the upper estuary (21 species). Habitat G in the
278
middle reaches contained the greatest mean density of individuals (2 033 per 0.1m2), while the smallest was
279
found at A (457 per 0.1 m2). For the sake of brevity, a detailed description of species differences among habitats
280
is not given here, but is provided in the following subsection.
282
3.1.1
283
M AN U
281
SC
RI PT
273
Habitat differences in benthic macroinvertebrate assemblage composition The preliminary site(habitat)×season PERMANOVA test for differences in benthic macroinvertebrate
composition showed that while all terms were significant (P=0.001), habitat differences were far more important
285
than site differences, with the COV value for habitat being ~1.5 times greater than that for any term involving
286
site (Results not shown). Additionally, examination of pairwise differences between sites in the same habitat
287
showed they were either not significant or almost always less than those between a site from another habitat in
288
~70% of cases. Subsequent spatial analyses of the faunal assemblages were thus undertaken at the broader level
289
of habitat, pooling the representative site replicates, and also separately for each season given the habitat×season
290
interaction.
EP
TE D
284
One-way ANOSIM showed that invertebrate composition differed significantly among habitats in each
292
season (P=0.1%), and that the overall extents of those differences were greatest and moderately high in winter,
293
followed by spring (Global R=0.665 and 0.564, respectively; Table 2c-d). Significant faunal differences
294
occurred between most pairs of habitats in all seasons except summer, when overall differences were the
295
smallest (Global R=0.354; Table 2a). The extents of the faunal differences among habitats in each season are
296
illustrated on the centroid nMDS ordination plot shown in Fig. 2, and the species most responsible for causing
297
these differences are summarised in the shade plot in Fig. 3.
AC C
291
298
Assemblages at the upper estuary habitats A and C differed markedly from those at all other habitats in
299
winter and spring (R > 0.8 in most cases; Table 2c and d). While assemblages at A and C were also significantly
10
ACCEPTED MANUSCRIPT different from each other, the extent of those differences was low to moderate (R=0.121-0.586). Such trends are
301
illustrated in Fig 2, in which samples from A and C in these seasons lay relatively close to each other and were
302
separated to the greatest extent (i.e. longest trajectories) from those at other habitats, and particularly I and M in
303
the lower estuary. In both winter and spring, habitat A was typified mainly by the polychaetes Scoloplos
304
normalis, Simplisetia aequisetis and to a lesser extent by Boccardiella limnicola, and also by the bivalves
305
Arthritica semen and Arcuatula senhousia (Fig. 3). However, these species, and particularly the first two, were
306
often more abundant at other habitats in these seasons. The fauna at C was also characterised by the above
307
species in winter and spring, as well as others such as the amphipod Paracorophium excavatum, which was
308
most abundant at this habitat. In contrast to these upper estuary habitats, the faunas at habitats I and/or M were
309
characterised by particularly high densities of the bivalve Sanguinolaria biradiata and polychaete
310
Pseudopolydora kempi, and also by Grandidierella propodentata, Capitella spp. and S. normalis in these
311
seasons (Fig. 3). The assemblages at the basin habitats F and G were also comparatively distinct from those at
312
most other habitats (aside from A and C) in winter, i.e. R=0.408-0.812 (Table 2c; Fig. 2). Their faunas were
313
typified by notably higher densities of Capitella spp. than any other habitat in this season, as well as by S.
314
normalis, S. aequisetis, P. kempi, G. propodentata, Corophium minor and S. biradiata. Differences in the mean
315
densities of this latter group of species contributed to compositional differences between F and G, and the
316
gastropod Batillaria australis attained its greatest densities at F (Fig. 3). The smallest significant habitat
317
differences during winter surprisingly occurred between J at the base of the Swan River and I and M in the
318
lower estuary (R=0.232-0.264; Table 3c), although their faunas were distinct in spring (R=0.560-0.631; Table
319
2d) and distinguished mainly by the lower densities of P. kempi, G. propodentata, S. biradiata and Capitella
320
spp.at J than I and M, with the opposite being true for species such as B. limnicola and A. semen (Fig. 3).
EP
TE D
M AN U
SC
RI PT
300
321
AC C
The faunas at A and C were also relatively distinct from those at most other habitats in summer and
322
autumn, but to a far lesser extent than in spring and especially winter (Table 2a and b). Thus, while the samples
323
from A and C in these seasons also tended to lie to one side of the ordination plot and adjacent to those in winter
324
and spring (Fig. 2), their degree of separation from other habitats, and especially J, F and/or G, was notably less.
325
During summer, habitats G and I were the most distinct from various others, with the largest differences
326
surprisingly occurring between G and the adjacent habitat F (R=0.882; Table 2a; Fig. 2). Several species were
327
prevalent at one or both of these habitats that were notably less abundant or absent from most others,
328
e.g. Oligochaete spp., Nematode spp., B. australis, Capitella spp. and S. aequisetis. The large differences
329
between G and F were mainly due to the notable lack of fauna at the latter habitat (Fig. 3). In autumn, the faunas
11
ACCEPTED MANUSCRIPT 330
at the upper estuary habitat C and lower estuary habitat M were generally the most distinct, not only from each
331
other (R=0.838) but also most other habitats (Table 2b; Fig. 2). Habitat C was typified by several species that
332
occurred in greater densities than at any other habitat in this season (e.g. P. excavatum, S. aequisetis, A. semen
333
and B. limnicola), while M had a comparatively depauperate fauna in which Heteromastus filiformis. was the
334
only species that was more abundant than at any other habitat (Fig. 3).
336
3.2
337
RI PT
335
Matching habitat patterns between the faunal and environmental data
RELATE showed that the pattern of differences among habitats, as defined by their enduring
environmental characteristics (Table 1), was significantly (P=0.1-0.3%) and moderately to highly correlated
339
with that in their faunal composition in all seasons, with the greatest match occurring in spring (ρ=0.745) and
340
the least in autumn (ρ=0.501; Table 3). Such findings indicate that the enduring habitats identified in the Swan
341
Estuary provide sound surrogates for spatial differences in the benthic macroinvertebrate fauna. This is further
342
illustrated by the similarities in the arrangement of habitats on the nMDS ordination plots in Fig. 4, which have
343
been derived from habitat averages of the enduring environmental data (Fig. 4a) and faunal composition in each
344
season (Figs 4b-e).
345
M AN U
SC
338
In contrast, habitat differences in the fauna were only significantly correlated with those in the nonenduring water quality variables in winter and spring (P=0.2 and 2.5%, ρ=0.681 and 0.514, respectively; Table
347
3) and sediment variables in spring (P=1.2%, ρ=0.571). These RELATE correlations were weaker than the
348
comparable ones between the faunal and enduring environmental data in almost all cases. BIOENV also only
349
detected a significant match between the faunal and water quality data in winter and spring (P=1%), with
350
improved correlations using data for only salinity and/or temperature (i.e. ρ=0.732 and 0.874 in winter and
351
spring, respectively; Table 3). BIOENV found a significant and moderate match between the faunal and
352
sediment matrices only in spring (P=3%, ρ=0.692, respectively) using data for organic matter content (Table 3).
AC C
EP
TE D
346
353
Spatial relationships between the invertebrate composition and magnitude of the water or sediment
354
variables selected by BIOENV are illustrated by the nMDS ordination and associated bubble plots in Fig. 5. In
355
both winter and spring, the marked faunal differences between the upper estuary habitats A and C and
356
particularly the lower estuary habitats I and M were well reflected by spatial differences in salinity, with
357
averages of 3.5-7.5 in the former habitats grading to 21.8-27.5 in the latter (Fig. 5a and c). Average water
358
temperatures at A during winter were also notably lower than those at any other habitat throughout the estuary
12
ACCEPTED MANUSCRIPT 359
(13.8 vs 15.1-16.9 °C; Fig. 5b) and in spring, organic matter content was greater at habitats A and/or C than all
360
others (Fig. 5d).
361 362
3.3
363
Validating the faunal prediction approach Given that the enduring environmental framework provided a better and/or more consistent spatial
surrogate for the faunal assemblage than the non-enduring water or sediment attributes, each of the five new
365
‘test’ sites at which invertebrates were sampled in summer and winter 2014/2015 were assigned to their
366
respective habitats using their enduring environmental measurements and the habitat prediction approach of
367
Valesini et al. (2010). Two sites were classified as habitat A, two as habitat C, and one as habitat F (Fig. 1).
SC
RI PT
364
368
Comparison of the complementary shade plots constructed from the invertebrate data recorded at the original (‘benchmark’) vs test site representatives of habitats A, C and F in summer and winter generally
370
showed a good agreement in characteristic species (Fig. 6). In summer, 10 of the 14 species that characterised
371
the benchmark assemblage at C also characterised the test assemblage, with only two other taxa (Nematode spp.
372
and T. burchardi) typifying the latter but not former (Fig. 6). Similarly, at habitats A and F, six of the 9-10
373
species that characterised the benchmark assemblages also characterised the test site faunas, with two to four
374
species typical of the test sites but not the benchmark sites. In winter, six of the nine, seven of the nine and
375
seven of the 12 species that characterised the benchmark faunas at habitats A, C and F, respectively, also
376
typified the corresponding test site faunas, with two to six species typifying the latter but not the former (Fig. 6).
377 4
Discussion
4.1
Habitat differences in benthic macroinvertebrate assemblages
380
AC C
379
EP
378
TE D
M AN U
369
381
The composition of the benthic macroinvertebrate assemblages sampled throughout the Swan Estuary
382
differed significantly, and to a considerable extent, among the various nearshore habitats identified in that
383
system based on their enduring environmental characteristics (sensu Valesini et al. 2010). The most distinct
384
assemblages were typically found at the most environmentally-distinct habitats A and C in the upper estuary and
385
habitat M in the lower estuary, often reflecting relatively impoverished faunas in the former habitats and
386
speciose and/or abundant faunas in the latter. This finding is largely consistent with those in other studies of
387
estuarine infauna worldwide, and has often been related to the influence of strong spatial gradients and/or
388
variability in environmental factors such as salinity, dissolved oxygen concentration, sediment attributes and
13
ACCEPTED MANUSCRIPT food availability (e.g. Ellis et al. 2006; Whitfield et al. 2012). However, the current study, which focussed on
390
habitats identified using a more comprehensive and finer-scale approach than is traditionally used, detected
391
several faunally-distinct habitats that contravened this general trend. Habitats F, G and I, for example, which
392
were all located relatively close together in the main basin but differed considerably in their wave exposure and
393
substrate/submerged vegetation characteristics, maintained moderate to high faunal differences in most seasons.
394
Indeed, during summer, the largest faunal differences between any pair of habitats occurred between F and G.
RI PT
389
395
The uppermost habitat A was characterised by a limited number of species in all seasons, including the polychaetes Simplisetia aequisetis, Scoloplos normalis and Boccardiella limnicola and the bivalves Arthritica
397
semen and Arcuatula senhousia. The first two of these species also typified the assemblages at most other
398
habitats throughout the year, reflecting their proficiency at coping with variable environments. Simplisetia
399
aequisetis is highly fecund and omnivorous, feeding directly on plant material (Hutchings, 1998) and other
400
benthic invertebrates (Fauchald & Jumars, 1979; Stevens et al. 2006), while S. normalis can tolerate fresh to
401
marine salinities (Hutchings & Murray, 1984). Arthritica semen is also physiologically adept at tolerating
402
variable salinities and employs a range of life history strategies for coping with high river flow, such as multiple
403
reproduction events throughout the year, rapid growth, a short life cycle and brooding its eggs and larvae inside
404
the mantle cavity (Wells & Threlfall, 1982a, b). Arcuatula senhousia, an invasive species from Japan, also
405
employs some of these strategies (Slack-Smith & Brearly, 1987; Crooks, 2002). Such characteristics make these
406
species particularly suited to the physiologically-stressful conditions often found in the upper reaches of
407
estuaries (e.g. Platell & Potter, 1996; Kanadjembo et al. 2001), and it is relevant that habitats A and/or C
408
experienced relatively large seasonal changes in salinity (3.6-19.8 and 3.8-31, respectively), water temperature
409
(ca 14-27 °C) and, in summer, the lowest mean dissolved oxygen concentrations (3.3 mg L-1). The depth of the
410
sediment transition (oxic to anoxic) layer was often shallowest at these habitats (particularly A, i.e. <1 cm), and
411
likely reflects the relatively large quantities of particulate organic matter (3.2-4.1% vs <1.5 % at all other
412
habitats), which can prevent oxygen diffusion into the sediment. It is also noteworthy that the above polychaete
413
species were smaller in body size at A/C than other habitats throughout the estuary (pers. obs.), which may be
414
indicative of physiological stress given it is often observed in species exposed longer-term to anoxic sediments
415
and/or toxic sulfides (Hagerman, 1998; Lee & Lee, 2005).
AC C
EP
TE D
M AN U
SC
396
416
Various other taxa characterised the faunas at habitat C, such as Capitella spp. in most seasons and
417
Paracorophium excavatum, Grandidierella propodentata, Corophium minor, Sanguinolaria biradiata and B.
418
limnicola in autumn and/or winter. The first of these species was ubiquitous, characterising many other habitats
14
ACCEPTED MANUSCRIPT 419
but especially F, G and I (see below), whereas P. excavatum was generally restricted to the upper half of the
420
estuary (habitats A, C and J), a trend paralleled by other species of this genus in other Australian and New
421
Zealand estuaries (Ford et al. 2001; Chapman et al. 2002). This may reflect a physiological ‘preference’ for
422
upper estuarine conditions, including greater quantities of particulate organic matter given that P. excavatum is a
423
detritus feeder.
424
RI PT
The distinctiveness of the fauna at the basin habitats F, G and I was not only due to the greater numbers and/or consistency of occurrence of several widely distributed taxa such as S. aequisetis, S. normalis and
426
Capitella spp., but also several taxa that almost exclusively typified these habitats in particular seasons, e.g.
427
Oligochaete spp. and Nematode spp. at G in summer, the gastropod Batillaria australis and S. biradiata at I in
428
summer, and the polychaete Australonereis elhersii at I in autumn. The prevalence of the first three polychaete
429
taxa is likely to reflect, at least in part, the abundance of their plant-based food sources, given the considerable
430
quantities of macrophytes and/or sedimentary chlorophyll a present at these habitats. For example, Capitella
431
spp. is a sediment-ingesting deposit feeder, whose abundance has been shown by Platell & Potter (1996) to be
432
positively associated with seagrass biomass in Wilson Inlet, another south-western Australian estuary. The
433
higher water temperatures at F, G and I compared to the upper estuary habitats (e.g. mean winter values of ~17
434
vs 13 °C at habitat A) are also likely to be more conducive to the faster growth and thus reproduction of benthic
435
invertebrates. Differences in the abundance of particular food sources between F, G and I may partly explain
436
why some species exclusively typified or were more prevalent at just one of those habitats in select seasons. For
437
example, the greater mean concentrations of sedimentary chlorophyll a at I than F and G (~24 vs 15-17 mg g-1),
438
which probably reflects its lower wave activity and smaller mean sediment grain size, was paralleled in summer
439
by comparatively high densities of B. australis, which uses specialised feeding apparatus to scrape microalgae
440
from sediment grains (Hori, 2006).
AC C
EP
TE D
M AN U
SC
425
441
The lowermost habitat M was by far the most speciose, undoubtedly reflecting its relatively consistent
442
and/or high salinities (~24-36), temperatures (~16-24 °C) and dissolved oxygen concentrations (~7-9 mg L-1),
443
high degree of shelter and benthic structural heterogeneity. This habitat contained 22 taxa that were not found at
444
any other throughout the estuary, but seven of which were recorded by Wildsmith et al. (2005) in nearby coastal
445
waters. The presence of such ‘marine straggler’ species not only reflects the relatively marine conditions at this
446
habitat, but also its proximity to the ocean. Various other studies have similarly recorded more speciose benthic
447
infaunas in the lower than upper reaches of estuaries (e.g. Whitfield et al. 2012; Barros et al. 2014). Some of the
448
taxa that typified M occurred there exclusively (e.g. the polychaete Heteromastus filiformis.), while others were
15
ACCEPTED MANUSCRIPT 449
notably more prevalent at this habitat and/or I just further upstream, particularly compared to A and/or C where
450
they were rarely recorded, e.g. A. elhersii and P. kempi. Such distribution patterns suggest a greater preference
451
or requirement for more marine conditions.
452 453
4.2
Surrogacy potential of enduring vs non-enduring environmental characteristics for predicting spatial patterns in the benthic invertebrate fauna
455
The pattern of habitat differences in the Swan Estuary, as defined by their enduring environmental
RI PT
454
characteristics, was significantly and well matched with that in their benthic macroinvertebrate assemblages in
457
each season. This biophysical habitat framework also provided a better and/or more consistent spatial
458
correlation with the fauna than the non-enduring water or sediment attributes, with significant matches for the
459
latter largely being restricted to winter and/or spring. Such findings provide robust ‘in-principle’ support for the
460
surrogacy value of the enduring habitats identified by Valesini et al. (2010) in characterising and, more
461
importantly, predicting the invertebrate species likely to typify any nearshore site in the Swan Estuary at any
462
time of year. This potential was further demonstrated by the outcomes of the field validation exercise, which
463
showed generally good agreement between the actual and predicted invertebrate species that typified the five
464
‘test’ sites based on their classified habitat type. These results suggest that no explanatory power is lost by using
465
enduring as opposed to the more traditionally adopted non-enduring criteria as environmental surrogates for
466
characterising faunal patterns, but rather that these broad biophysical criteria are capturing several other
467
influential elements of the estuarine environment beyond the specific suite of water and sediment variables that
468
were measured in situ. Moreover, given that measurements for enduring features can often be easily obtained
469
from mapped data sources, they are far more cost-effective to acquire than those for non-enduring features,
470
which need to be made in the field on a regular (often daily) basis.
AC C
EP
TE D
M AN U
SC
456
471
The above findings are promising, despite the fact that demonstration of this approach was based on
472
only a single year of faunal data to derive the ‘benchmark’ species characterising each habitat type. Clearly, the
473
robustness of the latter would be improved by integrating faunal data over several consecutive years to
474
accommodate natural inter-annual variability in the assemblages, as well updating the benchmark suite on a
475
periodic basis (e.g. every 10-15 years) to capture any significant changes in faunal composition related to
476
shifting baselines such as climate change or ongoing catchment development. Such limitations in this study are
477
likely to have contributed to the fact that, although the results of the field validation exercise were encouraging,
478
there were some species in the benchmark suite that were not typical of the test sites and vice versa. However, in
16
ACCEPTED MANUSCRIPT 479
almost all cases, the key species recorded at the “test” sites were also recorded in moderate numbers at the
480
corresponding “benchmark” habitats and some of the ‘mismatched’ species were also recorded at those habitats
481
although they contributed <5% to the assemblage abundances in their respective habitat x season groups.
482
Recent studies in marine reef environments have highlighted the benefits of using remotely sensed and geographic surrogates as predictors of benthic species distributions and assemblage diversity (Mellin et al. 2012;
484
Hill et al. 2014; Rees et al. 2014). However, other studies in soft sediment marine and estuarine environments
485
have tended to rely on non-enduring abiotic predictors of faunal distribution patterns which have produced poor
486
results (Stevens and Connolly, 2004; Shokri & Gladstone, 2013; Jackson & Lundquist, 2016). While such non-
487
enduring features tend to form the basis of individual species distribution models, which have become an
488
increasingly popular and accurate method of predicting species occurence in recent years (Gogina & Zetler,
489
2010; Reiss et al. 2011; Cozzoli et al. 2014; Meiβner et al. 2014), they require input on species tolerances and/or
490
known relationships with influential environmental variables. Thus, the practical application of such predictive
491
methods is limited to a small number of species and not suitable for forecasting species assemblages.
M AN U
SC
RI PT
483
492
Given the above-suggested improvements to the benchmark species for each habitat, management applications of the faunal prediction approach demonstrated here could include, for example, (i) a much
494
improved basis for estuarine conservation planning, by enabling the forecasting of the likely fauna and thus
495
quantification of the ecological value for any area of interest, (ii) helping guide decisions on where new within-
496
estuary developments (e.g. jetties, marinas) might best be located, based on the uniqueness or resilience of the
497
anticipated invertebrate fauna, and (iii) supporting understanding of estuarine ecological function through, for
498
example, providing supporting data for the development of food webs or other ecosystem models.
EP
TE D
493
499
Lastly, it is important to note that although this study has focussed on a single estuary, the habitatfaunal prediction approach presented here could be readily adapted to any estuary worldwide. Such adaptations
501
may include modification of the particular suite of biophysical variables used to characterise habitats (although
502
several of those employed by Valesini et al. 2010 are also likely to be relevant to many other estuaries), or
503
modification of the faunal sampling regimes at each habitat type to develop their benchmark suites of species.
AC C
500
504 505
Acknowledgements
506 507 508
The Australian Fisheries Research and Development Corporation (FRDC 2004/045) and Murdoch University are gratefully acknowledged for funding this research. We also thank Professors I. C. Potter, K. R.
17
ACCEPTED MANUSCRIPT 509
Clarke and R. M. Warwick for providing guidance and advice during the development of this project, and Dr P.
510
Hutchings for assistance with taxonomic identification.
511 512
References
513 Anderson, M. J., Gorley, R. N., Clarke, K. R., 2008. PERMANOVA+ for PRIMER: Guide to Software and
515 516
Statistical Methods. PRIMER-E, Plymouth, UK.
Banks, S. A., Skilleter, G. A., 2002. Mapping intertidal habitats and an evaluation of their conservation status in
517
Queensland, Australia. Ocean & Coastal Management 45: 485-509.
SC
518
RI PT
514
Banks, S. A., Skilleter, G. A., 2007. The importance of incorporating fine-scale habitat data into the design of an
519
intertidal marine reserve system. Biological Conservation 138: 13-29.
Barros, F., Blanchet, H., Hammerstrom, K., Sauriau, P., Oliver, J., 2014. A framework for investigating general
521
patterns of benthic b-diversity along estuaries. Estuarine Coastal and Shelf Science 149: 223-231.
522
Borja, A., Josephson, A.B., Miles, A., Muxika, I., Olsgard, F., Phillips, G., Rodriguez, J. G., Rygg, B., 2007. An
M AN U
520
523
approach to the intercalibration of benthic ecological status assessment in the North Atlantic ecoregion,
524
according to the European Water Framework Directive. Marine Pollution Bulletin 55: 42-52. Brearly, A., 2005. Swanland. Western Australia, University of Western Australia Press.
526
Bureau of Meteorology, 2016. Australian Climate Data Online.
TE D
525
527
Chapman, M. A., Hogg, I. D., Schnabel, K. E., Stevens, M. I., 2002. Synonymy of the New Zealand corophiid
EP
528
Fromhttp://www.bom.gov.au/climate/data/index.shtml?bookmark=200
amphipod genus, Chaetocorophium Karaman, 1979, with Paracorophium Stebbing, 1899:
530
morphological and genetic evidence. Journal of the Royal Society of New Zealand 32: 229-241.
531
Clarke, K. R., Ainsworth, M., 1993. A method of linking multivariate community structure to environmental
AC C
529
532
variables. Marine Ecology Progress Series 92: 205-219.
533
Clarke, K. R., Gorley, R. N., 2015. PRIMER v7: User Manual/Tutorial. Plymouth, PRIMER-E.
534
Clarke, K. R., Green, R. H., 1988. Statistical design and analysis for a 'biological effects' study. Marine Ecology
535 536
Progress Series 46: 213-226. Clarke, K. R., Chapman, M. G., Somerfield, P. J., Needham, H. R., 2006. Dispersion-based weighting of species
537
counts in assemblage analyses. Marine Ecology Progress Series 320: 11-27.
18
ACCEPTED MANUSCRIPT 538
Clarke, K.R., Somerfield, P. J., Gorley, R. N., 2008. Testing of null hypotheses in exploratory community
539
analyses: similarity profiles and biota-environment linkage. Journal of Experimental Marine Biology and
540
Ecology 366: 56–69.
541
Clarke, K. R., Tweedley, J. R., Valesini, F. J., 2014. Simple shade plots aid better long-term choices of data pretreatment in multivariate assemblage studies. Journal of the Marine Biological Association of the United
543
Kingdom 94: 1-16.
Constable, A. J., 1999. Ecology of benthic macroinvertebrates in soft-sediment environments: A review of
545 546
progress towards quantitative models and predictions. Australian Journal of Ecology 24: 452-476. Cozzoli, F., Eelkema, M., Bouma, T. J., Ysebaert, T., Escaravage V., Hermann, P. M. J., 2014. A mixed
SC
544
RI PT
542
547
modelling approach to predict the effect of environmental modification on species distributions. PLoS
548
ONE 9 (2): 1-14.
Crooks, J. A, 2002. Predators of the invasive mussel Musculista senhousia (Mollusca: Mytilidae). Pacific
550
M AN U
549
Science 56: 49-57.
551
Department of Defence, 2003. Australian National Tide Tables 2003. Australian Oceanographic Publication.
552
Edgar, G. J., Barrett, N. S., 2002. Benthic macrofauna in Tasmanian estuaries: scales of distribution and relationships with environmental variables. Journal of Experimental Marine Biology and Ecology 270: 1-
554
24.
TE D
553
Ellis, J., Ysebaert, T., Hume, T., Norkko, A., Bult, T., Herman, P. M. J., Thrush, S., Oldman, J., 2006.
556
Predicting macrofaunal species distribution in estuarine gradients using logistic regression and
557
classification systems. Marine Ecology Progress Series 316: 69-83.
558
EP
555
Fauchald, K., Jumars, P. A., 1979. The diet of worms: A study of polychaete feeding guilds. Oceanography and
559
AC C
560
Marine Biology: An Annual Review 17: 193-284. Folk, R.L., Ward, W. C. 1957. Brazos River Bar: A study in the significance of grain size parameters. Journal of
561 562
Sedimentary Petrology 27: 3-26. Ford, R. B., Thrush, S. F., Probert, P. K., 2001. The interacting effect of hydrodynamics and organic matter on
563
colonization: a soft-sediment example. Estuarine, Coastal and Shelf Science 526: 705-714.
564
Gogina, M., Zettler, M. L., 2010. Diversity and distribution of benthic macrofauna in the Baltic Sea Data
565
inventory and its use for species distribution modelling and prediction. Journal of Sea Research 64: 313-
566
321.
567
Hagerman, L. 1998. Physiological flexibility: a necessity for life in anoxic and sulphidic habitats. Hydrobiologia
19
ACCEPTED MANUSCRIPT 568
376: 241-254.
569
Heiri, O., Lotter, A. F., Lemcke, G., 2001. Loss on ignition as a method for estimating organic and carbonate
570
content in sediments: reproducibility and comparability of results. Journal of Paleolimnology 25: 101-
571
110.
572
Hill, N. A., Lucieer, V., Barrett, N. S., Anderson, T. J., Williams, S. B., 2014. Filling the gaps: Predicting the distribution of temperate reef biota using high resolution biological and acoustic data. Estuarine Coastal
574
and Shelf Science 147: 137-147.
575
RI PT
573
Hirst, A.J. 2004. Broad-scale environmental gradients among estuarine benthic macrofaunal assemblages of south-eastern Australia: implications for monitoring estuaries. Marine and Freshwater Research 55: 79-
577
92.
Hori, M. 2006. Intertidal surfgrass as an allochthonous resource trap from the subtidal habitat. Marine Ecology-
579 580
Progress Series 314: 89-96.
Hume, T.M., Snelder, T.H., Weatherhead, M., Liefting, R., 2007. A controlling factor approach to estuary
581 582
classification. Ocean and Coastal Management 50: 905–929.
Hutchings, P. A., 1998. Biodiversity and functioning of polychaetes in benthic sediments. Biodiversity and
583
Conservation 7: 1133-1145.
Hutchings, P. A., Murray, A., 1984. Taxonomy of polychaetes from the Hawkesbury River and the southern
TE D
584
M AN U
578
SC
576
585
estuaries of New South Wales, Australia. Records of the Australian Museum Supplement 3: 1-119.
586
Jackson, S. E., Lundquist, C. J., 2016. Limitations of biophysical habitats as biodiversity surrogates in the
587
EP
588
Hauraki Gulf Marine Park. Pacific Conservation Biology 22:159-172. Kanandjembo, A. N., Platell, M. E., Potter, I. C., 2001. The benthic macroinvertebrate community of the upper reaches of an Australian estuary that undergoes marked seasonal changes in hydrology. Hydrological
590
Processes 15: 2481-2501.
591
AC C
589
Lee, J. S., Lee, J. H., 2005. Influence of acid volatile sulfides and simultaneously extracted metals on the
592
bioavailability and toxicity of a mixture of sediment-associated Cd, Ni, and Zn to polychaetes Neanthes
593
arenaceodentata. Science of the Total Environment 3383: 229-241.
594
Legendre, P., Legendre, L., 1998 Numerical Ecology, Volume 24, Second Edition, Developments in
595
Environmental Modelling. Elsevier Science B.V., Amsterdam.
596
Meiβner, K., Fiorentino, D., Shnurr, S., Arbizu, P. M., Huettmann, F., Holst, S., Brix, S., Svavarsson, J., 2014.
597
Distribution of benthic marine invertebrates at northern latitudes- An evaluation applying multi-
20
ACCEPTED MANUSCRIPT 598 599
algorithm species distribution models. Journal of Sea Research 85: 241-254. Mellin, C., Parrott, L., Andrѐfouёt, S., Bradshaw, C., MacNeil, M., Caley, M., 2012. Multi-scale marine
600
biodiversity patterns inferred efficiently from habitat image processing. Ecological Applications 22: 792-
601
803.
602
Muxika, I., Borja, A., Bald, J., 2007. Using historical data, expert judgement and multivariate analysis in assessing reference conditions and benthic ecological status, according to the European Water
604
Framework Directive. Marine Pollution Bulletin 55: 16-29.
Parsons, T. R., Maita, Y., Lalli, C. M., 1984. A manual of chemical and biological methods for seawater
606 607
analysis. Oxford, Pergamon Press.
SC
605
RI PT
603
Pelletier, M. C., Gold, A. J., Heltshe, J. F., Buffum, H. W., 2010. A method to identify estuarine macroinvertebrate pollution indicator species in the Virginian Biogeographic Province. Ecological
609
Indicators 10: 1037-1048.
610
M AN U
608
Platell, M. E., Potter, I. C., 1996. Influence of water depth, season, habitat and estuary location on the
611
macrobenthic fauna of a seasonally closed estuary. Journal of the Marine Biological Association of the
612
United Kingdom 76: 1-21.
613
Rees. M. J., Jordan, A., Price, O. F., Coleman, M. A., Davis, A. R., 2014. Abiotic surrogates for temperate
614
TE D
615
rocky reef biodiversity: implications for marine protected areas. Diversity and Distributions 20: 286-296. Reiss, H., Cunze, S., König, K., Neumann, H., Kröncke, I., 2011. Species distribution modelling of marine
616
Roach, A. C., Wilson, S. P., 2009. Ecological impacts of tributyltin on estuarine communities in the Hastings
EP
617
benthos: A North Sea case study. Marine Ecology Progress Series 442: 71-86.
618
River, NSW Australia. Marine Pollution Bulletin 58: 1780-1786. Robertson, B. P., Savage, C., Gardner, J. P. A., Robertson, B. M., Stevenson, L. M., 2016. Optimising a widely-
620
used coastal health index through quantitative ecological group classifications and associated thresholds.
621
Ecological Indicators 69: 595-605.
622
AC C
619
Roff, J. C., Taylor, M. E., 2000. National frameworks for marine conservation - a hierarchical geophysical
623 624
approach. Aquatic Conservation: Marine and Freshwater Ecosystems 10: 209–223. Roff, J. C., Taylor, M. E., Laughren, J., 2003. Geophysical approaches to classification, delineation and
625
monitoring of marine habitats and their communities. Aquatic Conservation: Marine and Freshwater
626
Ecosystems 13: 77-90.
627
Shokri, M. R., Gladstone, W., 2013. Limitations of habitats as biodiversity surrogates for conservation planning
21
ACCEPTED MANUSCRIPT 628 629
in estuaries. Environmental Monitoring and Assessment 185: 3477-3492. Skilleter, G. A., Loneragan, N. R., 2003. Assessing the importance of coastal habitats for fisheries, biodiversity
630
and marine reserves: a new approach taking into account 'habitat mosaics'. Aquatic Protected Areas.
631
What works best and how do we know? Proceedings of the World Congress on Aquatic Protected Areas,
632
Cairns, Australia. Australia: ASFB. Slack-Smith, S.M., Brearley, A., 1987. Musculista senhousia (Benson, 1842); a mussel recently introduced into
RI PT
633 634
the Swan River estuary, Western Australia. Records of the Western Australian Museum 13: 225-230. Somerfield, P. J., Clarke, K. R., 2013. Inverse analysis in non-parametric multivariate analyses: distinguishing
636
groups of associated species which covary coherently across samples. Journal of Experimental Marine
637
Biology and Ecology 449: 261-273.
Stevens, T., Connolly, R. M., 2004. Testing the utility of abiotic surrogates for marine habitat mapping at scales
639 640
M AN U
638
SC
635
relevant to management. Biological Conservation 119: 351-362.
Stevens, M. I., Hogg, I. D., Pilditch, C. A., 2006. Evidence for female-biased juvenile dispersal in corophiid
641
amphipods from a New Zealand estuary. Journal of Experimental Marine Biology and Ecology 331: 9-
642
20.
Swan River Trust, 2009. Swan Canning Water Quality Improvement Plan. Swan River Trust, Perth.
644
Teske, P. R., Wooldridge, T. H., 2003. What limits the distribution of subtidal macrobenthos in permanently
TE D
643
645
open and temporarily open/closed South African estuaries? Salinity vs sediment particle size. Estuarine,
646
Coastal and Shelf Science 56: 1-14.
Thrush, S. F., Hewitt, J. E., Norko, A., Nicholls, P. E., Funnell, G. A., Ellis, J. I., 2003 Habitat change in
EP
647
estuaries; predicting broad-scale responses of intertidal macrofauna to sediment mud content. Marine
649
Ecology Progress Series 263: 101-112.
650
AC C
648
Valenҫa, A. P. M. C., Santos, P. J. P., 2012. Macrobenthic community for assessment of estuarine health in
651
tropical areas (Northeast Brazil): Review of macrofauna classification in ecological groups and
652
application of ATZI Marine Biotic Index. Marine Pollution Bulletin 64: 1809-1820.
653
Valesini, F. J., Clarke, K. R., Eliot, I., Potter, I. C. 2003. A user friendly quantitative approach to classifying
654
nearshore marine habitats along a heterogeneous coast. Journal of Experimental Marine Biology and
655
Ecology 56: 1-15.
656
Valesini, F. J., Hourston, M., Wildsmith, M. D., Coen, N. J., Potter, I. C., 2010. New quantitative approaches for
657
classifying and predicting local-scale habitats in estuaries. Estuarine, Coastal and Shelf Science 86: 645–
22
ACCEPTED MANUSCRIPT 658 659
664. Warwick, R. M., Sommerfield, P. J., 2010. The structure and functioning of the benthic microfauna of the
660
Bristol Channel and Severn Estuary, with predicted effects of a tidal barrage. Marine Pollution Bulletin
661
61: 92-99.
662
Wells, F. E., Threlfall, T. J., 1982a. Density fluctuations, growth and dry tissue production of Hydrococcus brazieri Tenison Woods, 1876 and Arthritica semen Menke, 1843 in Peel Inlet Western Australia.
664
Journal of Molluscan Studies 48: 310-320.
RI PT
663
Wells, F. E., Threlfall, T. J., 1982b. Salinity and temperature tolerance of Hydrococcus brazieri Tenison Woods,
666
1876 and Arthritica semen Menke, 1843 from the Peel-Harvey estuarine system, Western Australia.
667
Journal of the Malacological Society of Australia 5: 151-156.
SC
665
Wentworth, C. K. 1922. A scale of grade and class terms for clastic sediments. Journal of Geology 30: 377-282
669
Whitfield, A.K., Elliot, M., Basset, A., Blaber, S. J., West, R. J., 2012. Paradigms in estuarine ecology: A
M AN U
668
670
review of the Remane diagram with a suggested revised model for estuaries. Estuarine, Coastal and Shelf
671
Science 97:78-90.
672
Whomersly, P., Huxam, M., Bolam, S., Schratzberger, M., Augley, J., Ridland, D., 2010. Response of intertidal macrofauna to multiple disturbance types and intensities – an experimental approach. Marine
674
Environmental Research 69:297-308.
675
TE D
673
Wildsmith, M.D., Potter, I. C., Valesini F. J., Platell, M. E., 2005. Do the assemblages of benthic macroinvertebrates in nearshore waters of Western Australia vary among habitat types, zones and
677
seasons? Journal of the Marine Biology Association of the United Kingdom 85: 217–232.
678
EP
676
Ysebaert, T., Herman, P. M. J., 2002. Spatial and temporal variation in benthic macrofauna and relationships with environmental variables in an estuarine, intertidal soft-sediment environment. Marine Ecology-
680
Progress Series 244: 105-124.
681
AC C
679
Ysebaert, T., Meire, P., Herman, P. M. J., Verbeek, H., 2002. Macrobenthic species response surfaces along
682
estuarine gradients: prediction by logistic regression. Marine Ecology Progress Series 225: 79-95.
23
ACCEPTED MANUSCRIPT
Average values for enduring and non-enduring variables at the seven habitats at which benthic macroinvertebrate fauna were sampled in the Swan Estuary. The unit of measurement and broad category to which each variable was assigned is also included. %C- % areal cover. Unit
Broad category
A
Location
42759
Direct modified effective fetch
m m
Exposure
69
North modified effective fetch
m
Exposure
60
East modified effective fetch
m
Exposure
South modified effective fetch
m
Exposure
West modified effective fetch
m
Exposure
Wave shoaling margin (distance to 2 m contour)
m
I
J
M
34787
13312
22614
7587
21729
3694
615
2430
4120
1411
1478
428
194
74
1023
76
1047
0
249
1607
0
10
632
390
43
413
1751
1631
1047
0
580
64.
518
171
3499
1051
345
132
Exposure
46
311
175
602
154
993
38
°
Exposure
0.5
1.3
1.7
1.8
1.5
0.8
6.0
Vegetation cover
%C
Substrate/SAV
0
0.3
34.6
43.0
44.4
14.6
50.2
Rock
%C
Substrate/SAV
0
3.8
0.4
2.3
14.9
0.8
13.8
Snags
%C
Substrate/SAV
5.0
1.5
0
0
0
0
0
Reeds Bivalve beds
%C %C
5.3 0
1.3 0.5
0 0
0 0
0 0
0 0
0 0
°C
20.0
20.7
20.3
21.6
20.5
20.6
20.5
mg/L
6.7
7.4
9.2
10.1
8.4
7.6
8.0
-
9.4
14.9
27.9
27.5
29.3
24.1
30.9
-
7.7
7.8
8.3
8.4
8.3
8.1
8.3
µm
460.4
571.6
282.9
343.8
225.1
443.7
410.1
mg/g
15.7
11.7
15.6
17.0
24.0
8.4
11.8
cm
1.3
2.3
3.5
5.1
3.1
4.2
3.7
Dissolved oxygen Salinity pH Mean grain size Sedimentary Chlorophyll a Depth of redox transition layer
TE D
Temperature
Substrate/SAV Substrate/SAV
EP
Non-Enduring variable
M AN U
62
Substrate slope
24
G
AC C
Midline distance
F
SC
Enduring variable
C
RI PT
Table 1
ACCEPTED MANUSCRIPT Table 2
R-statistic and/or significance level (P) values for global and pairwise comparisons in one-way ANOSIM tests of the benthic macroinvertebrate composition among habitats sampled in the Swan Estuary during (a) summer, (b) autumn, (c) winter and (d) spring 2005. Significant pairwise comparisons are in bold.
(a) Summer 2005; P=0.1%, Global R=0.354 F
G
I
0.083 0.060 0.799 0.752 0.188 0.493
0.054 0.464 0.188 0.066 0.257
0.882 0.840 0.064 0.225
0.695 0.488 0.400
0.524 0.006
(b) Autumn 2005; P=0.1%, Global R=0.450 C
0.425 0.479 0.602 0.457 0.252 0.539
0.504 0.609 0.761 0.392 0.838
F
G
I
M AN U
C F G I J M
A
J
RI PT
C
0.351 0.480 0.203 0.617
0.419
SC
C F G I J M
A
0.591 0.297 0.668
J
0.121 0.203
0.372
TE D
(c) Winter 2005; P=0.1%, Global R=0.665 C
F
G
I
J
0.586 0.873 0.833 0.846 0.713 0.900
0.966 0.942 1.000 0.699 0.941
0.396 0.588 0.408 0.515
0.812 0.490 0.660
0.232 0.225
0.264
AC C
EP
C F G I J M
A
(d) Spring 2005; P=0.1%, Global R=0.564
C F G I J M
25
A
C
F
G
I
J
0.121 0.754 0.896 0.930 0.766 0.947
0.684 0.854 0.863 0.588 0.926
0.059 0.360 0.160 0.364
0.271 0.410 0.395
0.631 0.312
0.560
ACCEPTED MANUSCRIPT
Matrix
Summer
Autumn
RI PT
Table 3 Rho values (ρ) and significance levels (P%) for RELATE and/or BIOENV tests of correlations between the faunal data and (i) enduring environmental, (ii) non-enduring water quality and (iii) non-enduring sediment quality data at the various habitat types in each season. Significant correlations are in bold text. The variables chosen by BIOENV are also provided for each significant correlation. Sal-Salinity; Temp-water temperature; DO-dissolved oxygen concentration; MGS-mean sediment grain size; RD-depth of redox transition layer; %POM-contribution of sedimentary particulate organic matter; Chl a- concentration of sedimentary chlorophyll a. Winter
Spring
BIOENV
RELATE
BIOENV
RELATE
BIOENV
RELATE
BIOENV
ρP%
ρP% /Variables
ρP%
ρP%/Variables
ρP%
ρP%/Variables
ρP%
ρP%/Variables
Enduring Variables: (see Table 1)
0.545 1.1
NA
0.501 3.4
NA
0.664 0.4
NA
0.745 0.2
NA
Water Quality Variables: (Sal, Temp, DO)
0.157 21.2
0.255 22
0.182 30.2
0.302 22
0.681 0.2
0.732 1/Temp, Sal
0.514 2.5
0.874 1/Sal
Sediment Variables: (MGS, RD, %POM, Chl a)
0.088 28.3
0.299 36
0.351 12.5
0.21018.6
0.371 41
0.571 1.2
0.692 3/%POM
M AN U
TE D EP AC C 26
SC
RELATE
0.696 6
ACCEPTED MANUSCRIPT
A/P
C %
136.5120.2
R
14.9
M
F %
1
274.5263.1 65.2
R
9.7
M 205.2130.6 117.9
39.658.4
Scoloplos normalis
90.6
9.9
2
65.1
2.3
6
Arthritica semenMo/B
68.077.4
7.4
3
344.5540.4
12.2
1
Desdemona ornataA/P
31.8146.0
3.5
4
5.218.9
0.2
14
Grandidierella propodentataAr/C
22.4105.9
2.4
5
63.880.7
2.3
7
6
36.8
Boccardiella limnicola Arcuatula senhousia
A/P
Mo/B
Corophium minorAr/C
18.5
39.3
2
7
17.474.0
1.9
2.1
25.5
0.9
104.2
156.3245.7
11
4.5
5
88.0
87.8
1.7
8
108.6116.1
6.8
2.8
5
8
104.2218.7
3.7
4
149.5176.1
154.0
24.0
6.5
1.9
8
1.2
9
1.1
10
293.5406.9
10.4
2
Capitella spp.A/P
6.316.5
0.7
11
56.076.7
2
8
324.5294.9
14.2
Sanguinolaria biradiataMo/B
6.034.7
0.7
12
31.572.0
1.1
10
62.079.1
2.7
13
3.6
9.6
0.1
17
0.3
1.6
<0.1
23
0.3
Prionospio cirriferaA/P Manerogenia maneroo
Ar/C
Caraziella victoriensisA/P G
Batillaria australis Spisula trigonella
Mo/B
Caraziella sp.2A/P Paranthura kunzeaAr/C Ficopomatus sp.
A/P
Marphysa sanguineaA/P Amphipod sp.34
Ar/C
27
3.1
15.0
0.3
14
1.03.9
0.1
15
0.83.6
0.1
16
7.622.7 1.6
<0.1 <0.1
0.6
4.9
0.1
16
0.3
0.53.3
0.1
18
0.31.6
0.8
0.5
3.3
0.3
1.6
<0.1
20
0.31.6
<0.1
20
0.1
18
7.639.5
0.3
12
26.4
0.1
16
1.85.2
0.1
18
5.9
<0.1
19
1.3
7.4 8.9
249.0454.2
4.8
5
135.4
1.2
9
79.9158.8 178.4790.2
8
2
175.5203.7
7.9
7.4
3
19
134.6
9.1
6.1
4
138.2
3.9
6
3.6
6
22.4103.2
0.8
10
3
239.3499.4
8.2
2
4.9
<0.1
27
0.4
11
6.3
0.1
18
3.9
5
170.6285.8
5.9
3
555.5
5.8
4
<0.1
17
2.1
1.4
8
85.9223.7
1
359.7
7
11.5
0.2
14
8.338.1
0.4
12
4.7
113.5
8
15.9
45.876.9
3.4
R
2.4
4
114.1
%
71.1123.5
2.3
0.5
M
1
0.8
169.8
118.5152.0
4.1
5
6
71.192.6
1.7
8
292.7331.9
8.7
2
52.165.2
2.3
8
112.2173.2
3.9
7
67.6
1.3
10
196.4
5
4
155.0
3.4
7
404.0
11.4
1
0.52.3
<0.1
20
0.52.3
<0.1
17
3.69.9
0.2
16
2.15.9
0.1
17
3.19.5
0.1
13
1.63.8
<0.1
14
3.410.1
0.2
17
11.523.6
0.4
13
17
13.6
11
3.3
<0.1
21
2.16.8
0.1
17
6.3
0.1
17
23
0.1
362.2
<0.1
5
R
9.4
9
20
1.6
6.6
163.0
%
209.6218.9
1.7
<0.1
15
39.189.9
6
M 6
36.763.2
23
5.1
2.7
R
4.5
1
9
0.3
7
303.2
%
12.2
0.7
6.3
152.6360.5
412.0379.0
2.3
13.3
M
3
2
15.145.4 0.5
2.2
R
M
8.9
12
44.8
7.9
J
361.5271.9
1
7
0.2
4.2
11
875.2
2
18.7
4.7
0.3
2.8
301.8316.0
TE D
Polydora websteriA/P
5.2
268.8359.3
4
73.3
EP
Paratanytarsus grimmii
Ar/I
19.0
6.8
AC C
Pseudopolydora kempi
A/P
33.6
3
I
%
323.7248.6
10.222.9
17.4
Paracorophium excavatumAr/C
M
2
0.8
291.8
80.2
R 9
91.3
Oligochaete spp.
A/O
19.3
45.3
%
3
102.1
G
SC
M
Species
Simplisetia aequisetisA/P
S/D
M AN U
A P/C
RI PT
Mean density (M; individuals 0.1 m-2) and standard deviation (SD), percentage contribution to the overall mean density (%) and rank by density (R) of the benthic macroinvertebrate taxa recorded at each habitat sampled in the Swan Estuary in all seasons in 2005. Abundant taxa at each habitat type (i.e. those that contribute >5% to the overall mean density) are highlighted in grey. Each taxon has been assigned to its respective phyla and class (P/C), i.e. A- Annelida, Ar- Arthropoda, Mo- Mollusca, S- Sipuncula, PlPlatyhelminthes, N-Nematoda, Ne- Nemertea, P- Polychaeta, O- Oligochaeta, B- Bivalvia, G- Gastropoda, C- Crustacea, I- Insecta, U-Unknown.
Appendix A
12 16
54.4
1.0
3.2
<0.1
12.0
0.1
14
0.31.6
<0.1
0.52.3
<0.1
2.9
168.8
7.3
0.5
330.5
10.3
0.2
14
2.1
23
6.518.5
0.3
13
0.52.3
<0.1
30
20
2.14.2
0.1
18
1.06.6
<0.1
22
0.5
2.3
0.2
76.0
<0.1
17
4.7
ACCEPTED MANUSCRIPT
C %
R
M
F %
R
M
Platyhelminthes sp.Pl/U
0.84.9
<0.1
20
Microspio sp.A/P
0.53.3
<0.1
21
A/P
3.3
<0.1
21
0.31.6
<0.1
23
4.715.3 4.7
Orbiniella sp.
0.5
Tanais dulongiiAr/C
1.6
<0.1
23
Polydora socialisA/P
0.31.6
<0.1
23
Bivalve sp.3Mo/B
0.31.6
<0.1
23
Tritia burchardi.
Mo/G
Sphaeromatid sp.1
0.3
1.04.7
1.0
Ar/C
7.832.6
%
R
<0.1
18
0.2 <0.1
0.3
18
3.2
13
0.1
15
Venerupis crenataMo/B
1.63.8
0.1
16
0.5
2.3
<0.1
20
0.5
2.3
<0.1
20
Australonereis elhersii Opistobranchid sp.
Mo/G
Bivalve sp.1Mo/B
Decamastus sp. Fusinus sp.
A/P
Mo/G
Laturnula sp.
Mo/B
Sipunculan sp.3
S/U
Sphaeromatid sp.2Ar/C Heteromastus sp. Syllid sp.7
A/P
A/P
Nephtys graverii
A/P
Paranthurid sp.3Ar/C Phyllodoce sp. Bivalve sp.2
A/P
Mo/B
Rhyncospio sp.A/P
28
TE D
Gastrosaccus sorrentoensisAr/C
EP
Sipunculan sp.1S/U
AC C
Bivalve sp.4
Mo/B
66.7
R
0.1 <0.1
179.5
M
5.711.1
%
12
1.67.0
17
6.1
3.6
<0.1 0.1
R
13
11.4
0.1
12
4.2
0.1
11
27.641.8
0.8
10
1.04.7
<0.1
16
0.1
15
1.811.5
<0.1
16
0.52.3
<0.1
20
0.31.6
<0.1
23
1.6
<0.1
23 0.52.3
<0.1
17
0.52.3
<0.1
17
2.3
<0.1
17
0.5
M
M %
R
M
%
R
1.05.2
<0.1
22
14.624.6
14
9
2.616.5
0.3
J
1.6
M AN U
0.3
A/P
1.0
10
2.67.5
I %
5.232.9
Tellina deltoidalisMo/B
5.7
M
14
19.0
Nematode spp.
N/U
G
RI PT
M
Species
S/D
SC
A P/C
1.0
3.2
<0.1
20
2.9
0.31.6
1.4
<0.1
10
0.5
11
0.1
16
0.31.6
<0.1
34
0.31.6
<0.1
34
21.0
0.2
15
26.044.7
0.9
9
6.0
31.562.3
8.5
0.53.3
<0.1
30
1.05.2
<0.1
22
22 11.723.3
0.4
12
6.325.5
0.2
14
5.7
0.1
20
1.67.3
0.1
21
1.8
1.0
3.9
<0.1
22
1.0
6.6
<0.1
22
0.84.9
<0.1
27
ACCEPTED MANUSCRIPT
A P/C
M
S/D
C %
R
M
%
R
M
%
%
R
<0.1
27
Pseudopolydora sp. A/P
0.53.3
<0.1
30
3.3
<0.1
30
0.31.6
<0.1
34
1.6
<0.1
34
0.31.6
<0.1
34
1.6
<0.1
34
0.31.6
<0.1
34
1.6
<0.1
34
0.31.6
<0.1
34
0.3
1.6
<0.1
34
0.3
1.6
<0.1
34
0.31.6
<0.1
34
1.6
<0.1
34
0.31.6
<0.1
34
Ne/U
Lumbrinerid sp.A/P A/P
A/P
Syllid sp.9A/P A/P
Maldanis sp.A/P Cirriformia filigera Bivalve sp.5
A/P
Mo/B
Ostracod sp.3Ar/C Tanaid sp.
Ar/C
21
29
M
%
Total mean density
457
1412
Number of samples
40
40
18276
56492
AC C
Total number of individuals
R
0.5
0.3
0.3
0.3
0.3
22
25
22
22
47
1142
2033
1687
1107
1450
20
40
20
40
40
22836
81332
33740
44260
58008
EP
Number of species
TE D
Diogenid sp.Ar/C
29
R
M AN U
Magelona sp.
%
SC
Brania sp.A/P Syllid sp.8
M
M
0.84.9
Polydorella sp.
R
J
RI PT
M
I
M
Nemertean sp.
R
G
Sipunculan sp.1S/U
Species
%
F
ACCEPTED MANUSCRIPT List of Figures
Fig. 1
Map of the Swan Estuary showing the nearshore sites (black circles) and their assigned habitats (A, C, F, G, I, J or M; sensu Valesini et al. 2010) at which benthic macroinvertebrate fauna were sampled in each season in 2005. The location of the five ‘test’ sites (assigned to their habitats) used to validate the
RI PT
faunal prediction approach are also shown (white circles). Inset (a) shows the location of the Swan Estuary in Western Australia
nMDS ordination plot derived from the centroids (based on Bray-Curtis similarities) of the benthic
SC
Fig. 2
macroinvertebrate species composition in each habitat and season
Shade plot illustrating the pre-treated abundances of the most prevalent benthic macroinvertebrate
M AN U
Fig. 3
species in each habitat and season, with shading intensity being proportional to abundance. Species are ordered by a hierarchical cluster analysis of their mutual associations across habitat×season groups. Dashed lines in the dendrogram indicate species with significantly similar patterns of abundance, as
Fig. 4
TE D
detected by SIMPROF
nMDS ordination plots derived from the averages at each habitat of their (a) enduring environmental characteristics and (b-e) benthic macroinvertebrate composition in each season. Significance levels (P)
EP
and rho values (ρ) obtained from RELATE tests between the environmental and faunal data are also
Fig. 5
AC C
provided for each season
nMDS ordination plots derived from the average benthic macroinvertebrate composition at each site (coded for habitat) in a particular season. The magnitudes of the water or sediment quality variables selected by BIOENV are overlaid on each site as circles of proportionate sizes. Significance levels (P)
and rho values (ρ) obtained from the BIOENV tests are also provided.
Fig. 6
Shade plot illustrating the pre-treated abundances of the most prevalent benthic macroinvertebrate species at the habitats and seasons (‘benchmark species’, as per those recorded in the current study) that corresponded with the habitat classifications and field sampling regime at the five ‘test’ sites sampled
30
ACCEPTED MANUSCRIPT in 2014/2015. White circles denote those species that were prevalent in both the actual (test) and predicted (benchmark) data sets. Black circles denote those species that were prevalent in the actual but not benchmark data. Shading intensity is proportional to abundance, and species are ordered by a hierarchical cluster analysis of their mutual associations across habitat×season groups. Dashed lines in the dendrogram indicate species with significantly similar patterns of abundance, as detected by
AC C
EP
TE D
M AN U
SC
RI PT
SIMPROF
31
CE PT ED
M AN US C
RI
PT ED
M AN US
PT ED
M AN US C
AC C EP TE D
M AN US C
RI PT
PT ED
M AN US C
CE PT ED
M AN US C
R
ACCEPTED MANUSCRIPT
EP
TE D
M AN U
SC
RI PT
The potential for enduring surrogates to predict estuarine infauna was examined. Spatial differences in enduring surrogates consistently matched that in the infauna. Enduring surrogates better matched infaunal variation than non-enduring surrogates. Field validation showed promising prediction capability of non-enduring surrogates.
AC C
1. 2. 3. 4.