Hydrobiologia 440: 379–391, 2000. M.B. Jones, J.M.N. Azevedo, A.I. Neto, A.C. Costa & A.M. Frias Martins (eds), Island, Ocean and Deep-Sea Biology. © 2000 Kluwer Academic Publishers. Printed in the Netherlands.
379
Detecting anthropogenic disturbance in an environment with multiple gradients of physical disturbance, Manukau Harbour, New Zealand J.I. Ellis1 , D.C. Schneider2 & S.F. Thrush1 1 National Institute of Water & Atmospheric Research Ltd., P.O. Box 11-115, Hamilton, New Zealand Fax: +64-7-8560151. Tel: +64-7-8567026 (ext 849). E-mail:
[email protected] 2 Ocean Sciences Centre, 4 Clark Place, Memorial University of Newfoundland, St. John’s, Newfoundland, Canada A1C 5S7
Key words: sampling design, macrobenthic community, gradient analysis, organic enrichment, physical disturbance
Abstract Demonstrating spatial or temporal gradients of effects on macrobenthic communities can be a useful way of providing strong empirical evidence of natural or anthropogenic disturbance. Gradient designs for environmental assessment are sensitive to change for point source data, enabling the scale of the effects of a disturbance to be readily identified. If the spatial scale that is sampled from the point source is adequate, problems of selecting control sites can be avoided. However, sources of spatial variation in macrobenthic communities, which are not related to the impact, can confound the use of gradient designs. This can occur if the natural spatial structure overlaps that of the gradient and cannot be identified either as a location or environmental covariable. The ability to detect point source impacts using a gradient design against natural spatial variability was tested using benthic macrofaunal data collected from Manukau Harbour, New Zealand. Treated sewage wastewater is discharged into the northwest area of the Manukau Harbour. Sandflats in the vicinity of the outfall are also subject to physical disturbance from wind-waves and strong tides. Ordination techniques and the testing of a priori predictions were used to try and separate the relative effects of organic and physical disturbance on the benthic communities. While the occurrence of other environmental disturbances along a gradient of anthropogenic disturbance makes interpretation of community pattern more difficult, the use of a gradient sampling layout, ordination analysis and the testing of a priori predictions enabled impacts of the anthropogenic and natural environmental disturbances to be interpreted. Gradient designs, therefore, provide a method of assessing complex impacts that operate over broad spatial and temporal scales.
Introduction The development of sampling designs for impact assessment has been based on a stratified design where random samples are collected within strata (treatment and controls). These Before After Control Impact (BACI) designs are based on the principle that if two locations (control and impact) are monitored before an anthropogenic disturbance, the impact location will show a different pattern after the disturbance than will the control location. The evolution of stratified designs has been considered in full previously (Underwood, 1992, 1993). Ellis & Schneider (1997) highlight problems when stratified BACI designs are applied to detect point
source disturbances where the scale of the disturbance is unknown. If a stratified BACI design is applied to point source disturbances, placement of control sites becomes problematic as the control areas must be far enough away that they will not be affected by the potential disturbance, and yet close enough that the areas are comparable. Furthermore, the ability to assess the spatial scale of a disturbance with time is reduced by sampling an impact and a control stratum rather than quantifying the gradient of contamination (Ellis & Schneider, 1997). When a contaminant disperses with distance from a point source, gradient designs can be more sensitive to change than randomised placement of samples. Gradient designs also enable the scale of a disturbance to be readily identified and, given
380 sufficient spatial extent, a gradient layout avoids the problem of selecting adequate control sites. Another advantage of gradient designs is that they lend themselves to constructing mechanistic models that generate testable predictions of attenuating effects with increasing distance. Gradient designs, however, assume that natural spatial patterns that occur will not confound the ability to detect an anthropogenic disturbance. The presence of large salinity gradients, sedimentary particle size gradients or other environmental gradients associated with an anthropogenic disturbance can distort the pattern of ecological change observed. Pearson & Rosenberg (1978) also noted that in estuaries, or other areas where there are salinity gradients, organisms may be exposed to the combined stress of organic enrichment and reduced and/or fluctuating salinities. Separation of the relative effects of these two factors is often difficult, even more so when other environmental factors (e.g. variation in tidal flow, sedimentary structure, temperature, etc.) also influence the species distribution. Furthermore, as the scale of a study or environmental impact increases, complex interactions or encountering spatial variation not related to the impact are more likely to occur. This is an issue for most sampling designs. For example, broad-scale environmental impacts mean block control impact designs are likely to have increased difficulties in finding adequate control sites and are more likely to encompass a broader range of changing environmental variables. Gradient designs are more likely to encompass spatial variation not related to the impact. Many environmental impacts are occurring over increasingly large scales. For example, runoff due to changes in land use, dredging and modification of coastlines have resulted in alterations in rates of sedimentation and the extent of coastal areas that are depositional environments. In recent years, there has been growing recognition of the potential for changes in marine ecosystems as a consequence of commercial fishing resulting in broad-scale changes to continental shelf regions (Dayton et al., 1995; Jennings & Kaiser, 1998; Watling & Norse, 1998; Auster & Langton, 1999). BACI designs have been proposed to detect changes in single response variables, which are particularly useful for documenting localised impacts. However, in order to study broad scale disturbance, it is imperative that we supplement these techniques with time-series data as well as by developing new concepts and statistical tools (Thrush et al., 1999; Ellis et al., 2000). Demonstrating spatial or temporal gradi-
ents of effects on macrobenthic communities can be a useful way of providing strong empirical evidence of effects where the scale of disturbance is unknown (Ellis & Schneider, 1997). Determining broad-scale effects may be more difficult, however appropriate gradient designs can enable such disturbances to be assessed. Hence, we have tested the utility of these designs to detect an anthropogenic impact against natural spatial variability when multiple gradients of disturbance occur. Data of benthic macrofaunal abundance from the vicinity of the Manukau Harbour sewage outlet, New Zealand, were collected to test this assumption. Manukau Harbour is a dynamic and rigorous environment (Turner et al., 1995). This harbour is subject to physical wind-wave and tidal disturbance (Commito et al., 1995; Thrush et al., 1996), as well as the anthropogenic disturbance created by the sewage outlet. The organic enrichment gradient decreases with distance from the outfall while, in contrast, the wind-wave disturbance increases with distance (Bell et al., 1998). Adequately assessing spatial and temporal variability across environmental gradients requires consideration of the sampling grain, interval and extent. A trade off between replication to describe site variability versus sampling more distances requires careful consideration preferably from what is known about the scale of the pattern or process, e.g. from a pilot study (Thrush et al., 1994; Legendre & Legendre, 1998). Canonical correspondence analysis can be used to separate effects of multiple gradients. Another method is to make a priori predictions (Hewitt et al., 1996) of the expected community pattern as a response to the various gradients. The use of ordination techniques and the testing of a priori predictions were used to try and separate the relative effects of organic and physical disturbance on the benthic communities. The aim of this paper is, therefore, to determine how effective gradient designs are for detecting an anthropogenic disturbance in an environment with natural physical gradients operating over similar spatial scales to those of potential environmental effects. Predictions Organic enrichment Pearson & Rosenberg (1978) reviewed 47 publications of effects of organic enrichment and pollution on marine benthic communities. A consistent pattern of faunal changes observed along a gradient of increasing organic input to marine sediments was observed.
381 Close to a disturbed area, high densities of small, fast growing and rapidly colonising opportunistic species reach high abundance. Moving further from the disturbed area, a transition zone occurs, which is still dominated by opportunistic species although not at such high densities. At the end of the successional trajectory, a diverse assemblage dominated by suspension feeders and large and slow growing burrowing organisms is found. Further studies of effects of organic enrichment on marine communities support the community change described by Pearson & Rosenberg (1978) (Mirza & Gray, 1981; Essink, 1984; Whitlatch & Zajac, 1985; Pearson et al., 1986; Weston, 1990); however, Pearson & Rosenberg (op. cit.) also note that in areas where there are other natural gradients this species distribution may be altered. The Manukau is a physically dynamic environment. Physical disturbance is known to affect benthic community composition (see Prediction 4). Hence, we did not expect to see a diverse assemblage comprised of large slow growing species at the far sites that were affected by wind-wave disturbance. We, therefore, predicted the following: (1) The effects of organic enrichment on community structure will be most pronounced in the vicinity of the outfall and decrease progressively with distance from the discharge source due to mixing and dilution. (2) Consistent patterns to the faunal changes associated with an organic enrichment gradient occur. High numbers of a few small opportunistic (tolerant) species will occur adjacent to the discharge source. A maximum in the number of species will be reached with distance from the outfall, however, due to physical disturbance, we did not expect the far sites to be dominated by large and slow growing organisms. (3) A decline in suspension feeding organisms and an increase in deposit feeding organisms will occur as organic input increases. Physical wind-wave disturbance The hydrodynamic regime (tidal currents and waves) largely determines the sedimentary characteristics of an area. Hydrodynamics determine the nature of the bottom substratum, influence the stability of the sediment, and affect the nature of the food supply for benthic organisms (Sanders, 1958; Warwick & Uncles, 1980). In shallow-water coastal systems, environmental factors such as bottom topography, sediment characteristics and hydrodynamic processes play a definite role in structuring benthic communities and producing patchiness through a variety of mechanisms. Biotic interactions have traditionally been con-
sidered in the context of ‘static’ physical factors such as bottom topography and roughness, sediment characteristics and tidal inundation and exposure which in turn are largely determined by hydrodynamic processes (Bell et al., 1997). The importance of hydrodynamic variables such as current velocities, bed shear stress and wind-wave activity have also been recognised as influencing larval settlement and postsettlement transport (Grant, 1983; Commito et al., 1995), availability of particulate food resources (Warwick & Uncles, 1980) and sediment stability of the substratum (Warwick et al., 1991). Quantifying the linkages and interactions between physical factors and benthic community structure is difficult and continues to be debated (Hall, 1994; Raffaelli et al., 1994). However, significant differences in density, diversity and species composition of benthic communities have been correlated with various physical factors most of which can be related to exposure (Eleftheriou & Nicholson, 1975; Dexter, 1983, 1984). The relationship between the number of species observed and the degree of exposure to wave action is evident, with increases in species richness occurring with decreasing exposure to wave action (Dexter, 1992). Density, diversity and dominance of polychaetes relative to crustaceans also increase along this gradient. Similar correlations between physical parameters and faunal attributes have been noted in other comparisons of sandy beach communities (McLachlan et al., 1981; Dexter, 1983; Brown & McLachlan, 1990). Therefore, we predicted: (4) Species diversity, total abundance and species richness is reduced with increasing exposure to wave action.
Materials and methods Manukau Harbour is located adjacent to Auckland city on the west coast of the North Island of New Zealand. Treated sewage wastewater is discharged from a system of oxidation ponds into the north-west area of the Harbour. The Wastewater Treatment Plant discharges c. 3.9 m3 s−1 (Vant & Williams, 1992). Sampling sites were arranged along two transects taken with distance from the outfall (Fig. 1). Five sites were sampled on transect one, and six sites were taken on transect two. At each site, five core samples (10 cm diameter by 15 cm depth) were taken at 10 m intervals. Samples were sieved (500 µm mesh), preserved in 70% isopropyl alcohol and stained with 0.1% Rose Bengal. In the laboratory, macrofauna were sorted, identified to
382 To assess wind-wave disturbance, tidal current values, as well as wave orbital speeds at the bed, were calculated. This is because peak tidal currents in isolation are incapable of eroding bottom sediments, but in combination with near-bed orbital currents generated by only small wind waves, sediment transport can be initiated (Bell et al., 1997). Depth-averaged water velocity (cm/s) values for mean ebb and flood tides were generated from an oceanographic model of the harbour’s tidal regime (Bell et al., 1998). A numerical model WGEN (Black, 1997) was used to predict wave growth subject to wind stress. The model generates time series of wave height and wave period calculated from an input record of wind measurements or wind rose. The model can account for spatially variable bathymetry, shoaling, wave dissipation by bed friction and depth limited wave breaking. Linear wave theory was used to compute the orbital speed at the bed (Ub 3 , cm/s). Data analysis
Figure 1. (a) Manukau Harbour, North Island, New Zealand, showing the location of the Manukau sewage outfall. Dotted line indicates area of sand flat exposed at spring low tide. Wind rose data are for Auckland International Airport (Manukau Harbour); (b) position of the sampling sites with distance from the Manukau sewage outfall (T = Transect, S = Site).
the lowest possible taxonomic level, counted and preserved in 70% IPA. Surficial sediment (0–2 cm) was collected at the same location as each core sample. Replicate samples from a particular sampling location were pooled to assess sediment particle size and organic matter content. Sediment samples for particle size analysis were digested in 6% hydrogen peroxide for 24 h. Samples were then wet sieved (2 mm, 500 µm, 125 µm and 63 µm sieves) and each fraction was dried at 60 ◦ C in an oven and weighed. Organic matter was determined from ash-free dry weight. Samples were combusted for 6 h at 400 ◦ C.
Changes in macrofaunal abundance as a function of distance from the outfall, transect, replicate (within site variation) and interaction terms were analyzed using general linear models. Distance was treated as a continuous variable. The statistical package SAS was used to calculate p values under the assumption of independent residuals with equal variance. Residuals were plotted against expected values. If no association between the residuals and expected values were evident, the model was assumed to be an acceptable description of the data (Draper & Smith, 1981). The variability in macrofaunal community structure in relation to the changing wind-wave disturbance and organic loading gradients were examined using multivariate analyses. Data were analyzed using non-metric multidimensional scaling (MDS), correspondence analysis (CA), canonical correspondence analysis (CCA) and partial canonical correspondence analysis (PCCA). Both CA and non-metric MDS gave similar results, therefore, only MDS is presented. Non-metric MDS was performed on the species abundance matrices using PRIMER (Clarke, 1993). Canonical correspondence analysis and PCCA were used to identify relationships between the community structure and the environmental variables using CANOCO (Ter Braak, 1986, 1987). Environmental variables included the organic content, sediment grain size (% F
1 1 1 1 1 1 1
43.46 1.32 0.24 0.16 1.21 0.24 0.81
0.0001 0.25 0.6218 0.6911 0.2723 0.6222 0.3693
n = 3,795. SSTot = 400 070.6.
mean depth averaged flow speed of ebb and flood tides (cm/s). Another method to identify species responsible for changes in community structure is to calculate the magnitude and sign of spatial gradients. Analysis of Covariance (ANCOVA) for species abundance as a function of distance was used to test whether gradients were heterogeneous among species. A significant interaction term indicates whether species differ in pattern of changes in abundance with distance from the pollution source. Regression coefficients were calculated for all species and then ranked. A negative coefficient indicated tolerant species, a positive coefficient indicated sensitive species. Mid-ranking species showed little or no gradient relative to the contamination source. The statistical model to calculate species coefficient was Y = β o + β Dis + Dis + , where Y = benthic densities, Dis = distance and = residuals.
Results Analysis of variance indicated that distance from the sewage outfall was the only explanatory variable tested that was related to density of benthic macrofaunal organisms (Table 1). Changes in macrofaunal abundance for transect, replicate and interaction terms were not significant. Organic content of the sediment was elevated within 1400 m (T1/S1, T2/S1, T2/S2) of the outfall (Fig. 2). Sediment grain size indicated an increase in wind-wave disturbance with distance from the outfall (Fig. 3). Mud content (silt/clay fraction) was elevated within 1400 m of the outfall. These sites are
Figure 2. Change in sediment organic content as a function of distance from the Manukau Harbour sewage outfall. Table 2. Depth-averaged mean tidal currents and orbital speed at the bed for Manukau Harbour survey sites Transect Distance Peak ebb Peak flood Ub 3 (m) (cm/s) (cm/s) (cm/s) mean tide mean tide 1
800 1700 2900 3600 4200
25 30 19 25 26
16 36 30 27 23
9.1 14.0 7.6 16.0 5.0
2
800 1400 2300 2600 3300 4600
16 25 44 43 33 31
9 35 38 19 9 19
2.4 1.5 34.0 33.0 12.0 14.0
sheltered from the predominant wind-wave disturbance by Puketutu Island (Fig. 1b). The percentage of sand (125 µm – 2 mm) and tidal velocity (Table 2) increased for sites beyond Puketutu Island. Total abundance of benthic organisms decreased as a function of distance from the outfall (Fig. 4). The relationship between number of species and distance was not a simple linear function (Fig. 4). Species richness was reduced at sites 1 and 2 within 1400 m of the outfall. High numbers of a few tolerant species characterized these sites. There appeared to be a peak of high diversity at site 3, transect 2 (distance=2300 m). Diversity dropped to levels similar to sites adjacent to the outfall at the far sites (4200 m and 4600 m).
384
Figure 3. Change in sediment grain size with distance from the Manukau Harbour sewage outfall.
MDS grouped sites as a function of the organic and physical disturbance (Fig. 5). Sites with high organic content, low wind-wave disturbance (high mud content) and low tidal velocity were grouped together (T1/S1, T2/S1, T2/S2). Unpolluted sites with high physical disturbance (low organic content, high% sand) were grouped together. The first four axes of the initial unconstrained correspondence analysis accounted for 42% of the variation in the species site data. Figure 6 indicates the position of the sites on the first two axes. Figure 7 shows the position of the species on the first two axes. Partial canonical correspondence analysis suggests that variability was related significantly to organic loading and physical disturbance or sediment characteristic covariables (correlation = 0.9325, F-ratio = 14.01, P ≈ 0.00). Organic loading accounted for 22% of the variability in community structure. Species identified as occurring in mud-enriched habitats were: Heteromastus filiformis, Aquilaspio aucklandica, Arthritica bifurca, Boccardia syrtis, Torridaharpinia hurleyi, Nereidae and Halicarcinus whitei and so on (Fig. 7). Although these animals occur throughout the Manukau Harbour, they were found at higher densities close to the out-
Figure 4. Change in macrofaunal abundance and species richness with distance from the Manukau Harbour sewage outfall. Note that standard error bars cannot be seen as they were less than the symbol size.
Figure 5. Non-metric MDS ordination by station of the macrobenthos species abundance data for Manukau Harbour sewage outfall (T = Transect, S = Site).
fall. These groupings are consistent with the species groupings identified using covariance analysis or plots of sensitive and tolerant species as discussed below.
385
Figure 6. Canonical correspondence analysis of environmental variables (lines) and sampling sites (•) using macrofaunal abundance data from Manukau Harbour (T=Transect, S=Site).
Figure 7. Canonical correspondence analysis of environmental variables (lines) indicating species positions for Manukau Harbour. Species codes are provided in Table 3.
386 Table 3. Species information and strength of density gradient, as estimated by regression for Manukau Harbour (n = 3795) Species Heteromastus filiformis Aquilaspio aucklandica Arthritica bifurca Boccardia syrtis Torridaharpina hurleyi Owenia fusiformis Nucula hartivigiana Nereididae Halicarcinus whitei Nemertean Waitangi brevirostris Amphipod ii Helice crassa Cominella glandiformis Anthopleura aureoradiata Ostracod ii Austrovenus stutchburyi Notoacmea spp. Glycera americana Soletellina siliqua Nebalacea Scolecolepides sp. Methalimendon sp. Ophelliidae Mactra ovata Gastropod i Exosphaeroma chilensis Ostracod i Cyclaspis thomsoni Cirolana sp. Phoronid Halicarcinus cookii Scaleworm Ruditapes sp. Corophid Edwardsii Phyllodocid Syllid Euchone Aricidea Aminotrypan Felaniella zelandica Paraonid ii Exosphaeroma falcatum Zeacumantus lutulentus Hermit crab Goniada emerita Macroclymenella stewartensis Lepidodontidae Eteone durantiaca
H.f Aq.a A.b B T.h O.f N.h Ne Ha.w N W.b A.ii He.c C.g A.a O.ii A.s No G.a S.s Nb S M Op M.o G.a E.c O.i C.t C Ph H.c S R Co E Py Sy Eu Ar Am F.z P.ii E.f Z.l Cr G.e M.s L E.d
Average
Sum
Coefficient
Feeding Guild
30.38 16.89 9.76 10.55 3.55 8.89 6.40 1.51 1.11 0.96 2.25 0.33 0.36 0.25 0.58 1.65 2.51 1.53 0.25 0.85 0.20 0.05 0.44 0.15 0.16 0.07 0.11 0.16 0.09 0.02 0.11 0.09 0.04 0.04 0.04 0.04 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.04 0.73 0.02 0.02 0.05
1671 929 537 580 195 489 352 83 61 53 124 18 20 14 32 91 138 84 14 47 11 3 24 8 9 4 6 9 5 1 6 5 2 2 2 2 1 1 1 1 1 1 1 1 1 2 40 1 1 3
−0.02230 −0.01637 −0.00977 −0.00934 −0.00252 −0.00215 −0.00123 −0.00106 −0.00100 −0.00076 −0.00059 −0.00022 −0.00020 −0.00017 −0.00016 −0.00014 −0.00014 −0.00011 −0.00007 −0.00003 −0.00002 0.00000 0.00001 0.00002 0.00002 0.00002 0.00003 0.00003 0.00003 0.00004 0.00004 0.00004 0.00004 0.00004 0.00004 0.00004 0.00005 0.00005 0.00005 0.00005 0.00005 0.00005 0.00005 0.00005 0.00005 0.00005 0.00005 0.00006 0.00006 0.00006
Surface deposit feeder, motile, non-jawed Surface deposit feeder, discretely motile, tentaculate Suspension feeder Surface deposit feeder, discretely motile, tentaculate Filter &deposit feeder, discretely motile, tentaculate Deposit feeder Carnivore, motile, jawed
Deposit feeder Carnivore
Suspension feeder Carnivore Deposit feeder Surface deposit feeder Non-selective deposit feeder
Surface deposit feeder, motile, non-jawed Carnivore, motile, jawed
Carnivore Carnivore Filter feeder, sessile, tentaculate Non-selective sub-surface deposit feeder
Non-selective sub-surface deposit feeder Surface deposit feeder Carnivore, dicretely motile, jawed Sub-surface deposit feeder Carnivore Continued on p. 387
387 Table 3. contd. Chitin Maldonidae Exogonidae Aglaphamus macroura Cominella fusiformis Xymene plebeius Amphipod i Polydora sp. Sphaerosyllis semiverrocosa Oligochaete Microspio sp. Diloma subrostrata Paraonid i Colurostylis lemurum Orbinia papillosa Trochodota dendyi Cossura sp. Macomona liliana Magelona dakini
Ch M Ex A.m C.f X.p A.i P Sh.s O M D.s P.i C.l O.p T.d Co M.l M.d
0.04 0.04 0.02 0.04 0.04 0.04 0.02 0.02 0.02 0.05 0.11 0.49 0.05 0.18 0.04 0.11 0.49 5.93 15.73
2 2 1 2 2 2 1 1 1 3 6 27 3 10 2 6 27 326 865
Another method of determining which species accounted for the change in community structure, is to rank density gradients from negative to positive (Table 3). The negative coefficients indicate the species most tolerant to the contamination gradient. Species identified as tolerant, with high abundance adjacent to the outfall in enriched areas, were Heteromastus filiformis, Aquilaspio aucklandica, Arthritica bifurca, Boccardia syrtis, Torridaharpinia hurleyi, Owenia fusiformis, Nereidae, Nemertean etc. Species identified as sensitive were Magelona dakini, Macomona liliana, Cossura sp., Trochodota dendyi, Orbinia papillosa, Colurostylis lemurum etc. Benthic organisms identified as sensitive or tolerant from the regression coefficients were plotted to investigate biological change associated with the enrichment and physical disturbance gradient. Figure 8 provides examples of three tolerant species identified using regression analysis. Heteromastus filiformis, Boccardia syrtis and Aquilaspio aucklandica occur at high densities within 1700 m of the outfall. These sites are those identified as being enriched with low physical disturbance. Figure 9 provides examples of two sensitive species, Magelona dakini and Cossura sp. These species occur at sites identified as having low organic content and high physical disturbance. The species identified as sensitive or tolerant to organic enrichment using density gradients are the same as those identified using ca-
0.00007 0.00007 0.00007 0.00007 0.00007 0.00007 0.00007 0.00007 0.00007 0.00007 0.00008 0.00008 0.00009 0.00010 0.00010 0.00011 0.00030 0.00053 0.00770
Carnivore Carnivore, motile, non-jawed Carnivore Filter & surface feeder Carnivore, motile, non-jawed Surface deposit feeder
Non-selective sub-surface deposit feeder Sub-surface deposit feeder, motile, non-jawed Surface deposit feeder Surface deposit feeder, discretely motile, tentaculate
nonical correspondence analysis. Where possible, taxa were allocated to feeding guilds principally following Fauchald & Jumars (1979) with further information provided by Pridmore et al. (1990). A change from suspension feeders to deposit feeders along a gradient of increasing organic enrichment was not observed (Table 3). Discussion Prediction 1 As expected, elevated levels of organic content were most pronounced in the vicinity of the outfall. However, a progressive decrease in the level of organic enrichment with distance did not occur. Wind-wave disturbance created a sharp boundary of enriched versus non-enriched areas, rather than the predicted gradient of attenuation to background levels. The plots of organic content and sediment grain size for Manukau Harbour indicate that elevated levels of organic enrichment occur within 1400 m of the outfall. These sites were sheltered from the predominant south-west wind-wave disturbance by Puketutu Island. Beyond Puketutu Island, the wind-wave disturbance and tidal currents increase. The hydrodynamic regime largely determines the sedimentary characteristics of an area (Sanders, 1958; Warwick & Uncles, 1980). Water
388
Figure 9. Change in abundance of two sensitive species with distance from the Manukau Habour sewage outfall.
Figure 8. Change in abundance of three tolerant species with distance from the Manukau Harbour sewage outfall.
movement is one of the critical factors controlling the deposition of organic material at the sediment-water interface. In areas of strong currents, little deposition will take place. The predicted gradient of impact was, therefore, modified by sediment transport due to increased hydrodynamic factors for sites beyond Puketutu Island. Predictions 2 and 4 A gradient of community change was observed for Manukau Harbour. High numbers of a few small tolerant species such as Heteromastus filiformis, Aquilaspio aucklandica and Boccardia syrtis were recorded within 1400 m of the outfall. The Pearson & Rosenberg model (1978) suggests that a maximum in species richness occurs at a transitory distance from the discharge source. At Manukau Harbour, the number of species rose to 46 at 2300 m away from the outfall, compared with 20 species in the enriched zone. The
species diversity at the far wind-wave exposed sites was between 20 and 26 species per core, similar to the diversity recorded adjacent to the outfall. Rigorous hydrodynamic conditions capable of mobilising sediment can, ultimately, preclude the development of complex benthic communities with high densities of large, long-lived species. The most obvious impacts of hydrodynamic conditions on the macrobenthic communities are likely to arise from wind-generated wave activity that increases physical disturbance of the surface sediment (Turner et al., 1995; Dolphin & Green, 1997). Wind will also contribute to desiccation stress at times of low tide, with potential effects on the survival of some species (Turner et al., 1995). Less obvious impacts are the effects of sedimentary and hydrodynamic conditions on the relative success of larval, juvenile and adult transport and settlement (Turner et al., 1995). The lower species diversity recorded at the wind-wave exposed sites is consistent with previous studies of the effects of exposure on macrofaunal communities supporting Prediction 4 (McLachlan et al., 1981; Dexter, 1984; Dexter, 1992).
389 Prediction 3 The prediction that a decline in suspension feeding organisms and an increase in deposit feeding organisms occurs as organic enrichment increases was not supported. The ratio of filter feeders: deposit feeders did not change noticeably with distance from the outfall. Decreases in suspension feeders in enriched areas are predicted due to physical clogging of ciliary and siphonal mechanisms, an inability to withstand lower oxygen tensions, and the increasing sedimentary instability brought about by deposit feeders (Pearson & Rosenberg, 1978). In Manukau, the enrichmenttolerant species identified using spatial gradients were mainly deposit feeders. However, there was no noticeable increase in suspension feeders with distance from the outfall. Pridmore et al. (1990) highlight problems with inferring relationships between trophic structure and grain size in estuarine systems. Firstly, allocating animals to trophic groups can be difficult. Studies of some taxa have indicated that feeding behaviour is too variable to allow allocation to distinct groups. Some organisms are capable of switching their mode of feeding; examples of plasticity in feeding behaviour as a function of flow and sediment transport are provided in Fauchald & Jumars (1979), Pridmore et al. (1990) and by Snelgrove & Butman (1994). For example, many species of surface deposit feeders are now known to be facultative suspension feeders (Hughes, 1969; Buhr & Winter, 1977; Dauer et al., 1981), evidently in response to flow and elevated fluxes of suspended particles (Levinton, 1991; Taghon & Greene, 1992). Finally, in shallow and turbid environments, where surface sediments and associated diatom mats are frequently resuspended, differentiation between deposit feeders, suspension feeders and grazers etc. is further confounded.
Conclusions Canonical correspondence analysis and the estimation of spatial gradients were both effective at identifying species-environment relationships. Multivariate analysis of community composition demonstrated that 22% of the variability in community composition was attributable to organic enrichment, and that sites beyond Puketutu Island were influenced by physical wind-wave disturbance. Both methods identified similar sensitive and tolerant species driving community composition at the enriched versus far sites. The use of
a priori predictions was also effective at inferring relevant processes. A priori predictions about effects of organic enrichment and physical wind-wave disturbance on benthic community composition were supported. However, the a priori predictions of the scale of disturbance from enrichment, and the predictions of a change to suspension feeding communities at the far non-enriched sites, were not supported. Physical wind-wave disturbance modified the scale of enrichment with elevated organic content within 1400 m of the outfall. Sampling down spatial gradients enables indications of the scale of processes to be postulated and enables the role of particular processes in accounting for site-to-site variation to be tested. By ordering studies down gradients, an understanding of the changing role of small-scale processes within the increasing spatial extent can be gained (Thrush et al., 1999). This understanding could not be achieved by simple random sampling or by block designs. Demonstrating gradient trends through time can also address problems of a lack of temporal information. A lack of adequate baseline data or controls can affect the ability to provide strong empirical evidence of effects (Dayton et al., 1998). For example, appropriate benchmark data describing natural standards are generally not available or, in the case of environmental accidents, ‘before’ data sets cannot be collected. Wiens & Parker (1995) consider appropriate sampling designs and assumptions for analysing the effects of accidental environmental impacts. Because they have the potential to document both initial and subsequent recovery, the impact level by time and impact trend by time designs seem particularly well suited to the analysis of unplanned impacts. Designs that treat contamination as continuous variables have the potential to document contamination effects with greater precision and to detect non-linerarity’s in responses of the community (Wiens & Parker, 1995). Suter (1996) also recommends monitoring gradients or levels of exposure and matching in space and time the measures of effects to these gradients or levels. Hence, ordering studies down gradients enables both temporal changes and changes in processes with increasing spatial extent to be identified. While the occurrence of other environmental disturbances along a gradient of anthropogenic disturbance makes interpretation of community pattern more difficult, knowledge of the differences in the expected and observed patterns increases greatly ecological understanding of how to link patterns and processes. The
390 use of a gradient sampling layout together with testing a priori predictions enabled impacts of the anthropogenic and natural environmental disturbances to be interpreted. However, this is an area requiring further research. Efforts to include analysis of spatial patterns (density variations and habitat gradients) in both field experiments and survey programmes, and the collection of time series data and spatial pattern analysis of larger scale phenomena, may enable general mechanistic hypothesis to be generated and tested.
Acknowledgements We thank Rob Bell for generating mean ebb and flood values, and John Oldman for generating orbital speed values. Thanks to Bob Whitlatch who provided invaluable comments on this work. We also wish to thank Vonda Cummings and Judi Hewitt for reviewing earlier drafts of this manuscript.
References Auster, P. J. & R. W. Langton, 1999. The effects of fishing on fish habitat. Am. Fisheries Soc. Symp. 22. Bell, R. G., T. M. Hume, T. J. Dolphin, M. O. Green & R. A. Walters, 1997. Characterisation of physical environmental factors on an intertidal sandflat, Manukau Harbour, New Zealand. J. exp. mar. Biol. Ecol. 216: 11–31. Bell, R. G., S. V. Dumnov, B. L. Williams & M. J. N. Greig, 1998. Hydrodynamics of Manukau Harbour, New Zealand. New Zealand J. mar. Freshwat. Res. 32(1): 81–100. Black, K. P., 1997. WGEN3DD, Wave generation model for enclosed water bodies and support software. University of Waikato. Brown, A. C. & A. McLachlan, 1990. Ecology of sandy shores. Elsevier Amsterdam. Buhr., K. J. & J. E. Winter, 1977. Distribution and maintenance of a Lanice conchilega association in the Weser Estuary (FRG), with special reference to the suspension-feeding behaviour of Lanice Conchilega. In Keegan, B. F. (ed.), Biology of Benthic Organisms. Pergamon Press, Oxford: 101–113. Clarke, K. R., 1993. Non-parametric multivariate analyses of changes in community structure. Aust. J. Ecol. 18: 117–143. Commito, J. A., S. F. Thrush, R. D. Pridmore, J. E. Hewitt & V. J. Cummings, 1995. Dispersal dynamics in wind-driven benthic system. Limnol. Oceanogr. 40(8): 1513–1518. Dauer, D. M., C. A. Maybury & R. M. Ewing, 1981. Feeding behaviour and general ecology of several spionid polychaetes from the Chesapeake Bay. J. exp. mar. Biol. Ecol. 54: 21–38. Dayton, P. K., S. F. Thrush, T. M. Agardy & R. J. Horman, 1995. Environmental effects of fishing. Aquatic Conservation: Mar. Freshwat. Ecos. 5: 205–232. Dayton, P. K., M. J. Tegner, P. B. Edwards & K. L. Riser, 1998. Sliding baselines, ghosts and reduced expectations in kelp forest communities. Ecol. Apps. 8: 309–322. Dexter, D. M., 1983. Community structure of intertidal sandy beaches in New South Wales, Australia. In McLachlan, A.
& Erasmus (eds), Sandy Beaches as Ecosystems. Dr W. Junk Publishers, The Hague: 461–472. Dexter, D. M., 1984. Temporal and spatial variability in the community structure of the fauna of four sandy beaches in southeastern New South Wales. Aust. J. mar. Freshwat. Res. 35: 663–672. Dexter, D. M., 1992. Sandy beach community structure: the role of exposure and latitude. J. Biogeog. 19: 59–66. Dolphin, T. J. & M. O. Green, 1997. Sediment dynamics of an estuarine ‘turbid fringe’. Pacific Coasts and Ports, 1997. Proceedings Vol 1. Centre for advanced engineering, University of Canterbury: 113–118. Draper, N. R. & H. Smith, 1981. Applied Regression Analysis. 2nd edn. Wiley, New York. Eleftheriou, A. & M. D. Nicholson, 1975. The effects of exposure on beach fauna. Cah. Biol. mar. 16: 695–710. Ellis, J. I. & D. C. Schneider, 1997. Evaluation of a gradient sampling design for environmental impact assessment. Envir. Mon. and Assess. 48(2): 157–172. Ellis, J. I., A. Norkko & S. F. Thrush, 2000. Broad scale disturbance of intertidal and shallow sublittoral soft sediment habitats: effects on benthic macrofauna. J. Aquat. Ecos. Health 7(1): 57–74. Essink, K., 1984. The discharge of organic waste into the Wadden sea - Local effects. Netherlands Inst. Sea Res. Pub. Ser. 10: 165– 177. Fauchald, K. & P. A. Jumars, 1979. The diet of worms: A study of polychaete feeding guilds. Oceangr. mar. Biol. ann. Rev. 17: 193–284. Grant, J., 1983. The relative magnitude of biological and physical sediment reworking in an intertidal community. J. mar. Res. 41: 673–689. Hall, S. J., 1994. Physical disturbance and marine benthic communities: life in unconsolidated sediments. Oceanogr. mar. Biol. ann. Rev. 32: 179–239. Hewitt. J. E., S. F. Thrush, V. J. Cummings & R. D. Pridmore, 1996. Matching patterns with processes: predicting the effect of size and mobility on the spatial distributions of the bivalves Macomona liliana and Austrovenus stutchburyi. Mar. Ecol. Prog. Ser. 135: 57–67. Hughes, R. N., 1969. A study of feeding in Scrobicularia plana. J. mar. biol. Ass. U.K. 49: 805–823. Legendre, P. & L. Legendre, 1998. Numerical Ecology. 2nd edn. Elsevier, The Netherlands. Levinton, J. S., 1991. Variable feeding behavior in three species of Macoma (Bivalvia: Tellinacea) as a response to water flow and sediment transport. Mar. Biol. 110: 375–383. Jennings, S. & M. J. Kaiser, 1998. The effects of fishing on marine ecosystems. Adv. mar. Biol. 34: 203–314. McLachlan, A., T. Wooldridge & A. H. Dye, 1981. The ecology of sandy beaches in southern Africa. Sth. african J. Zool. 16: 219–231. Mirza, F. B. & J. S. Gray, 1981. The fauna of benthic sediments from the organically enriched Oslofjord, Norway. J. exp. mar. Biol. Ecol. 54: 181–207. Pearson, T. H. & R. Rosenberg, 1978. Macrobenthic succession in relation to organic enrichment and pollution of the marine environment. Oceanogr. mar. Biol. ann. Rev. 16: 229–311. Pearson, T. H., A. D. Ansell & L. Robb, 1986. The benthos of the deeper sediments of the Firth of Clyde, with particular reference to organic enrichment. Proc. r. Soc. Edinburgh 90B: 329–350. Pridmore, R. D., S. F. Thrush, J. E. Hewitt & D. S. Roper, 1990. Macrobenthic community composition of six intertidal sandflats in Manukau Harbour, New Zealand. New Zealand J. mar. Freshwat. Res. 24: 81–96.
391 Raffaelli, D. G., A. G. Hildrew & P. S. Giller, 1994. Scale, pattern and process in aquatic systems: concluding remarks. In Giller, P. S., A. G. Hildrew & D. G. Raffaelli (eds), Aquatic Ecology: Scale, Pattern and Process. Blackwell Scientific Publications, London: 601–606. Sanders, H. L., 1958. Benthic studies in Buzzards Bay. I. AnimalSediment relationships. Limnol. Oceanogr. 3: 245–258. Snelgrove, P. V. R. & C. A. Butman, 1994. Animal–Sediment relationships revisited: cause versus effect. Oceangr. mar. Biol. ann. Rev. 32: 111–177. Suter, G. W., 1996. Abuse of hypothesis testing statistics in ecological risk assessment. Human Ecol. Risk Assess. 2(2): 331–347. Taghon, G. L. & R. B. Greene, 1992. Utilization of deposited and suspended particulate matter by benthic ‘interface’ feeders. Limnol. Oceanogr. 37: 1370–1391. Ter Braak, C. J. F., 1986. Canonical correspondence analysis: A new eigenvector technique for multivariate direct gradient analysis. Ecology 67(5): 1167–1179. Ter Braak, C. J. F., 1987. The analysis of vegetation–environment relationship by canonical correspondence analysis. Vegetatio 69: 69–77. Thrush, S. F, R. D. Pridmore & J. E. Hewitt, 1994. Impacts on softsediment macrofauna: the effects of spatial variation on temporal trends. Ecol. Apps. 4(1): 31–41. Thrush, S. F., R. B. Whitlatch, R. D. Pridmore & J. E. Hewitt, 1996. Scale-dependent recolonization: the role of sediment stability in a dynamic sandflat habitat. Ecology 77(8): 2472–2487. Thrush, S. F., S. M. Lawrie, J. E. Hewitt & V. J. Cummings, 1999. The problem of scale: uncertainties and implications for soft-bottom marine communities and the assessment of human impacts. In Gray, J. S. (ed.), Biogeochemical Cycling and Sediment Ecology. Kluwer Academic Publishers, Dordrecht, The Netherlands: 195–210.
Turner, S. J., S. F. Thrush, R. D. Pridmore, J. E. Hewitt, V. J. Cummings & M. Maskery, 1995. Are soft-sediment communities stable? An example from a windy harbour. Mar. Ecol. Prog. Ser. 120: 219–230. Underwood, A. J., 1992. Beyond BACI. The detection of environmental impacts on populations in the real, but variable, world. J. exp. mar. Biol. Ecol. 161: 145–178. Underwood, A. J., 1993. The mechanics of spatially replicated sampling programmes to detect environmental impacts in a variable world. Aust. J. Ecol. 18: 99–116. Vant, W. N. & B. L. Williams, 1992. Residence times of Manukau Harbour, New Zealand. New Zealand J. mar. Freshwat. Res. 26: 393–404. Warwick, R. M. & R. J. Uncles, 1980. Distribution of benthic macrofauna associations in the Bristol channel in relation to tidal stress. Mar. Ecol. Prog. Ser. 3: 97–103. Warwick, R. M., J. D. Goss-Custard, R. Kirby, C. L. George, N. D. Pope & A. A. Rowden, 1991. Static and dynamic environmental factors determining the community structure of estuarine macrobenthos in SW Britain: Why is the Severn Estuary different. J. appl. Ecol. 28: 329–345. Watling, L. & E. A. Norse, 1998. Disturbance of the seabed by mobile fishing gear: a comparison to forest clearcutting. Conserv. Biol. 12: 1180–1197. Weston, D. P., 1990. Quantitative examination of macrobenthic community changes along an organic enrichment gradient. Mar. Ecol. Prog. Ser. 61: 233–244. Wiens, J. A. & K. R. Parker, 1995. Analyzing the effects of accidental environmental impacts: approaches and assumptions. Ecol. App. 5(4): 1069–1083. Whitlatch, R. B. & R. N. Zajac, 1985. Biotic interactions among estuarine infaunal opportunistic species. Mar. Ecol. Prog. Ser. 21: 299–311.
