Altered fish community and feeding behaviour in close proximity to boat moorings in an urban estuary

Altered fish community and feeding behaviour in close proximity to boat moorings in an urban estuary

Marine Pollution Bulletin 129 (2018) 43–51 Contents lists available at ScienceDirect Marine Pollution Bulletin journal homepage: www.elsevier.com/lo...

1MB Sizes 0 Downloads 1 Views

Marine Pollution Bulletin 129 (2018) 43–51

Contents lists available at ScienceDirect

Marine Pollution Bulletin journal homepage: www.elsevier.com/locate/marpolbul

Altered fish community and feeding behaviour in close proximity to boat moorings in an urban estuary

T



Brendan S. Lanhama,b, , Adriana Vergésb,c, Luke H. Hedgea,b, Emma L. Johnstona,b, Alistair G.B. Poorea,b a

Evolution and Ecology Research Centre, School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney, NSW 2052, Australia Sydney Institute of Marine Science, Sydney, NSW 2088, Australia c Centre for Marine Bio-innovation, School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney, NSW 2052, Australia b

A R T I C L E I N F O

A B S T R A C T

Keywords: Boat moorings Fish Artificial structures Fine-scale Urbanization Estuaries

Coastal urbanization has led to large-scale transformation of estuaries, with artificial structures now commonplace. Boat moorings are known to reduce seagrass cover, but little is known about their effect on fish communities. We used underwater video to quantify abundance, diversity, composition and feeding behaviour of fish assemblages on two scales: with increasing distance from moorings on fine scales, and among locations where moorings were present or absent. Fish were less abundant in close proximity to boat moorings, and the species composition varied on fine scales, leading to lower predation pressure near moorings. There was no relationship at the location with seagrass. On larger scales, we detected no differences in abundance or community composition among locations where moorings were present or absent. These findings show a clear impact of moorings on fish and highlight the importance of fine-scale assessments over location-scale comparisons in the detection of the effects of artificial structures.

1. Introduction Growing human populations are causing a wide range of impacts on marine habitats in urban areas. Overfishing, pollution, recreational boating and ocean sprawl have all led to habitat degradation in urbanized coastal environments (Whitfield and Becker, 2014; Heery et al., 2017). Nearshore ecosystems are frequently altered by the addition of artificial structures (Dugan et al., 2011; Heery et al., 2017), resulting in large shifts in the composition and function of associated communities (Dafforn et al., 2015). The replacement of natural marine habitats with artificial structures such as breakwaters, piers and docks, marinas, jetties, pilings, pontoons, seawalls and boat moorings has resulted in large changes in the physical structure of marine habitats (Heery et al., 2017). These hard structures are often added to locations of low structural complexity, such as soft sediment habitats (Vaselli et al., 2008; Airoldi et al., 2015). The communities inhabiting artificial structures commonly differ from natural substrates due to both biological and physical processes that can differ between the two habitat types (Bulleri and Chapman, 2010; Dafforn et al., 2012). Overwater structures decrease benthic light availability (Able et al., 1998; Glasby, 1999) and reduce the growth and

percent cover of macrophytes in soft sediments (Heery et al., 2017). They can alter local hydrodynamics (Perez-Ruzafa et al., 2006) and increase sediment pollution (e.g., metal contamination from antifouling paints associated with marinas, Dafforn et al., 2009; Airoldi et al., 2015). Altered habitat conditions frequently reduce species diversity (Bacchiocchi and Airoldi, 2003) and create favorable conditions and ‘stepping blocks’ for invasive species (Airoldi et al., 2015). In some circumstances, however, artificial structures can increase the structural complexity of habitats, thus increasing the surface area for settlement of sessile organisms (Perkol-Finkel et al., 2012). Through mimicking natural substrates, artificial structures with high physical complexity can be used to mitigate habitat loss in degraded systems (Pister, 2009; Bulleri and Chapman, 2010; Wetzel et al., 2014). Detecting the ecological impacts of artificial structures frequently involves large-scale comparisons of modified habitats with areas that lack artificial structures. In highly urbanized estuaries, however, shorelines can be so heavily modified that very few locations lack artificial structures (Dafforn et al., 2015). Therefore, detecting the impacts of artificial structures will rely on assessments on smaller scales (i.e., with increasing distance from the structures), moving away from traditional location scale contrasts (Hedge et al., 2017). For example, breakwaters

⁎ Corresponding author at: Evolution and Ecology Research Centre, School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney, NSW 2052, Australia. E-mail address: [email protected] (B.S. Lanham).

https://doi.org/10.1016/j.marpolbul.2018.02.010 Received 23 November 2017; Received in revised form 30 January 2018; Accepted 4 February 2018 Available online 10 February 2018 0025-326X/ © 2018 Elsevier Ltd. All rights reserved.

Marine Pollution Bulletin 129 (2018) 43–51

B.S. Lanham et al.

and marinas result in the accumulation of finer sediments due to the reduction of water flow in close proximity (< 10 m) to the modified habitats (Zanuttigh et al., 2005; Rivero et al., 2013). Boat moorings are ubiquitous within sheltered embayments in urbanized estuaries and are examples of widespread artificial structures with known ecological impacts to organisms in close proximity, especially seagrasses (Walker et al., 1989; Unsworth et al., 2017). Growing populations in coastal areas are leading to increased boating activity and pressures to expand or improve boating infrastructure (Whitfield and Becker, 2014; Mayer-Pinto et al., 2015). Most commonly, moorings consist of a large concrete block and a heavy anchor chain attached to a lighter chain or rope connecting to a surface buoy and the boat (known as ‘swing moorings’, Unsworth et al., 2017). Changes in wind and current direction move the boats and result in the anchor chain being dragged across the sediment. The chain scour removes benthic organisms (Walker et al., 1989; Harasti, 2016; Unsworth et al., 2017) and can change the physical and chemical composition of soft sediments in close proximity to moorings (Hedge et al., 2017). The negative effects of boat moorings on seagrasses are well established, with chain scour reducing seagrass cover and leaving a barren halo of unconsolidated sediments around the mooring block that prevents regeneration and leads to further habitat loss (Hastings et al., 1995; Demers et al., 2013; Unsworth et al., 2017). Moorings can also reduce the light availability for seagrasses through shading and the resuspension of sediments, further reducing seagrass cover (Orth et al., 2006; Waycott et al., 2009). Despite the expectation that changes to seagrass cover and sediment properties would affect other components of the marine ecosystem, few studies have examined how moorings may interact with other organisms (see Lynch et al., 2015; Serrano et al., 2016; Silberberger et al., 2016; Hedge et al., 2017). In this study, we quantify the relationships between the abundance, composition and feeding activity of the demersal fish community with distance to boat moorings on two spatial scales: within mooring fields and among locations. Fish are frequently associated with artificial structures (Wickham et al., 1973; Bohnsack, 1989), with many studies showing that jetties, artificial reefs, wharves, pontoons and breakwalls can support high abundances of some fish (e.g., Bohnsack and Sutherland, 1985; Rilov and Benayahu, 2000; Clynick, 2008; Folpp et al., 2013). These structures can support high abundances due to high concentrations of available food (Cresson et al., 2014) and mating opportunities (meeting point hypothesis, Freon and Dagorn, 2000). The potential exists for boat moorings to act as fish attractants, but the decline in seagrasses in close proximity to moorings could also deter fish (Edgar and Shaw, 1995). An understanding of how fish communities are affected by moorings in urban areas is needed given the important role they play as predators in soft sediments (Thrush, 1999). Any concentration of fish around artificial structures may alter predation pressure in nearby areas, and physical disturbance to the benthos may result in shifts in the distribution of fish utilizing soft sediment and seagrass habitats (Smith et al., 2011). To assess the impact of multiple small boat moorings on benthic fish assemblages we asked the following specific questions: (1) How does the abundance, diversity, composition and feeding behaviour of the fish community vary with distance from boat moorings? (2) Is the fish community best explained by distance from moorings or seagrass cover? (3) On larger scales, how does the abundance, diversity, composition and feeding behaviour of the fish community differ between locations with and without moorings?

million (Johnston et al., 2015; ABS, 2016). The estuary has many shallow embayments which contain extensive mooring fields (MayerPinto et al., 2015). Boating is an important recreational activity and there will be an estimated 22,000 recreational vessels registered in the Port Jackson catchment by 2020 and over 5,500 boat moorings within the estuary itself (Transport for NSW, 2013). Six locations were sampled in the outer harbour of Port Jackson, all within 5 km of the estuary mouth and subject to high tidal flushing while protected from the ocean (Fig. 1). Clontarf (33°48′18.9″ S, 151°15′07.0″ E), North Harbour (33°48′00.9″ S, 151°16′09.3″ E), Hunters Bay (33°49′39.6″ S, 151°15′15.7″ E) and Manly Cove (33°48′05.9″ S, 151°17′05.8″ E) each contain an extensive mooring field. The mooring density across these locations is 0.0024 moorings/m2, 0.0023 moorings/m2, 0.0019 moorings/m2, and 0.0022 moorings/m2 respectively. Quarantine Bay (33°48′48.7″ S, 151°17′08.2″ E) and Rose Bay (33°51′39.8″ S, 151°16′05.3″ E) have no boating infrastructure. The sampling locations were within a depth of 1 to 12 m and dominated by soft sediment habitats. Only the location at Manly Cove contains extensive seagrass beds of three species (Halophila australis, Posidonia australis and Zostera muelleri). Hunters Bay contains sparsely distributed H. australis and seasonal patches of Z. muelleri. At Manly Cove, 23 seagrass friendly moorings are installed across the location (Fig. S1). 2.2. Variation in fish communities and feeding behaviour with distance to moorings To test whether the presence of boat moorings is associated with altered fish communities and feeding behaviour, we used unbaited remote underwater video to survey fish within the four locations containing mooring fields. At each location, up to 20 cameras were placed at pre-determined sites within a 120 m diameter sampling zone, using a generalized random tessellation stratified design (GRTS, Stevens and Olsen, 2004). GRTS allows for fine-scale environmental sampling with sample sites evenly spread across the geographic space and with respect to environmental variables used as predictors. The predictor variables used here were distance from shore (as a proxy for depth), and the distance to nearest mooring. The positions of each mooring block on the seafloor were determined using high-resolution aerial imagery from Nearmap (www.nearmap.com.au). Where the mooring block could not be seen in the imagery due to high turbidity (30 of 147 moorings), the position of the mooring buoy at the surface was recorded at multiple time points and their centroid used as the position of the mooring block. The sampling sites were loaded onto a real time kinematic (RTK) GPS unit and cameras were dropped from a boat at each site. The GPS position of each camera was recorded on deployment to ensure the highest possible spatial accuracy (Fig. 1). Camera units consisted of a GoPro Hero 4 Silver Edition in an underwater housing mounted on steel stands positioning the camera 25 cm off the seabed. Cameras were recovered 1 h after the final camera was deployed, and with known start and finishing times for each camera, along with their order of deployment, we were able to view the same time of day for each camera. Water visibility was measured once at each location and sampling day by holding a camera below the surface and lowering a 30 cm diameter Secchi disk attached to a rope with markings every meter. Video sampling of fish communities was performed over two separate sampling times at each location, between April 2 and 13, 2015 and again between August 4 and 11, 2015. All observations were made between 0900 and 1600. A total of 186 cameras were deployed across the two sampling times. Due to occasional camera and stand malfunctions the number of cameras deployed at each location at each time varied between 15 and 20 (Clontarf, n = 20 and 15; Hunters Bay, n = 18 and 18; Manly, n = 17 and 18; North Harbour, n = 16 and 15). The video footage was analyzed using EventMeasure software (www.seagis.com.au). One hour of footage was viewed from each of the deployed cameras with the first 5 min post deployment never analyzed to prevent boat disturbance affecting the fish community. Species

2. Methods 2.1. Study locations The study was performed in Port Jackson, the main estuary of the city of Sydney, Australia (Fig. 1). Port Jackson is a highly urbanized estuary within the Greater Sydney region, which has a population of 4.8 44

Marine Pollution Bulletin 129 (2018) 43–51

B.S. Lanham et al.

Fig. 1. The six sampling locations in Port Jackson, Sydney. Clontarf, Hunters Bay, Manly Cove and North Harbour have mooring fields while Quarantine Bay, and Rose Bay lack moorings. Zoomed boxes show the sampling sites (white dots) and mooring locations (black dots) within each location.

Table 1 Analyses of the fish community at locations with moorings. Results from univariate and multivariate generalized linear models contrasting total abundance, community composition and feeding behaviour (the likelihood of fish feeding and number of observed bites) of the demersal fish assemblage against distance to nearest moorings, distance to the shore, sampling time, and location and a linear model contrasting Shannon's diversity index (H) against the same predictor variables. Dev is the deviance from generalized linear models and significant effects (P < 0.05) are in bold. Total abundance

Diversity (H)

Composition

Likelihood

Bites

Source

df

Dev

P

F

P

Dev

P

Dev

P

Dev

P

Distance to nearest mooring Distance to shore Sampling time Location Distance to moorings x Location

1 1 1 3 3

25.20 5.53 30.42 87.23 5.76

< 0.001 0.018 < 0.001 < 0.001 0.118

1.98 0.09 23.67 17.88 3.16

0.17 0.77 < 0.001 < 0.001 0.03

21.92 25.69 142.25 261.54 50.67

0.014 < 0.001 < 0.001 < 0.001 0.002

0.46 3.47 0.04 90.10 51.99

0.50 0.06 0.85 < 0.001 < 0.001

4.26 3.29 18.39 80.39 9.89

0.035 0.071 < 0.001 < 0.001 0.017

be feeding (each observed fish was marked as ‘feeding’ or ‘passing’ while in frame) in relation to the predictor variables and was tested using a generalized linear model with a binomial distribution. As an estimate of predation pressure on the benthos, the total number of observed bites in relation to the predictor variables was tested using a generalized linear model with a negative binomial distribution. Differences in feeding behavior among fish species in response to the predictor variables were explored with a negative binomial multivariate generalized linear model as described above for relative abundance. The analyses were conducted with manyglm and manylm functions in the R package mvabund (Wang et al., 2012), with statistical inference from likelihood ratio tests using the anova function and Monte Carlo resampling with 10,000 iterations. Given that the abundance of fish observed in a video was dependent on variation in water visibility among sampling locations and days, we used Secchi disk depth in all relative abundance and diversity models as a log linear offset to standardize relative abundance (Parsons et al., 2016; Smith et al., 2017). Analyses of the likelihood of a fish to be feeding did not use water visibility as an offset given that observations were only taken from fish able to be viewed in the videos.

relative abundance was measured as MaxN, defined as the maximum number of individuals observed in a single frame of footage. This is a conservative measure of abundance commonly used for underwater video (Willis et al., 2000; Cappo et al., 2003). Feeding behaviour was recorded when individual fish were observed feeding from the benthos whilst in frame. Where individuals remained in the frame for extended periods of time, their behaviour was recorded every 5 min. The sum of MaxN values for all species observed produced an estimate of total relative abundance for each camera. This was contrasted with the predictor variables of distance to the nearest mooring, distance to shore (as a proxy for depth), location, and sampling time using a negative binomial generalized linear model. Species diversity was measured with Shannon's diversity index (H) and was contrasted to the same predictor variables using a linear model with a gamma distribution. To contrast species composition with the same predictor variables, we used a negative binomial multivariate generalized linear model. The model included the community matrix as the response variable containing the MaxN of species observed on > 20 occasions (reducing the final species count from 38 to 8, accounting for 79% of observations). Species with few individuals found throughout the study are rarely important to the multivariate analysis and may lead to zero-inflation issues during model specification (Zuur et al., 2009). Feeding behaviour was contrasted to the same predictor variables stated above using two methods. The likelihood of an observed fish to 45

Marine Pollution Bulletin 129 (2018) 43–51

B.S. Lanham et al.

to predict the total relative abundance, diversity (H), community composition and feeding behaviour on the fish community as above (excluding the location effect). 2.4. Large-scale contrast of fish communities among locations with and without moorings To test how fish communities at locations with moorings differed from those that lacked moorings we performed a large-scale comparison among all six locations. The two locations that lacked boat moorings were sampled concurrently with the mooring locations as above. The number of cameras deployed at each new location was: Quarantine Bay (n = 17) and Rose Bay (n = 18 and 14) (Fig. 1). Quarantine Bay was only visited during the second sampling time as severe weather prevented sampling for several weeks, separating it from the other locations by an unacceptable amount of time. The response variables of total relative abundance, diversity, fish community composition and feeding behaviour were contrasted with the predictor variables of distance to shore, location, and sampling time. The analyses were generalized linear models with a negative binomial distribution (total relative abundance, community composition, and bites), a generalized linear model with binomial distribution (proportion feeding) and a linear model (diversity). The analyses were run with the manyglm and manylm functions in the R package mvabund, with inferences from likelihood ratio tests and Monte Carlo resampling with 10,000 iterations. Models were offset as above. 3. Results 3.1. Variation in fish communities and feeding behaviour with distance to moorings A total of 1882 fish (sum of MaxN) from 38 species were observed during the study across the six locations (Table S1). The five most abundant species observed were Pseudocaranx georgianus (Carangidae), Acanthopagrus australis (Sparidae), Gerres subfasciatus (Gerreidae), Girella tricuspidata (Kyphosidae), and Parupeneus spirulus (Mullidae). The total relative abundance of the fish community varied with distance to nearest mooring, distance to shore, sampling time, and among locations (Table 1). At all locations, the total relative abundance of fish increased with distance from the nearest mooring, with the strongest relationships present at Hunters Bay and North Harbour (Fig. 2a). The diversity of the fish community varied with distance to moorings at Manly Cove and North Harbour while there was no relationship at Clontarf and Hunters Bay (Fig. 2b, an interaction between distance to nearest mooring and location, Table 1), and varied between the two sampling times and among locations (Table 1). The species composition of the fish community differed significantly with distance to nearest mooring, distance to shore, sampling time, and among locations (Fig. S2, Table 1). The effect of distance to moorings varied among locations (an interaction between distance to nearest mooring and location, Table 1). The direction and magnitude of changes in relative abundance with distance to moorings differed among the most abundant species (Fig. 3). Pseudocaranx georgianus increased in relative abundance with increasing distance to moorings at Hunters Bay, while remaining relatively stable at the other locations. Girella tricuspidata showed a similar trend, although with a small decrease in relative abundance at Clontarf with increasing distance to moorings, due to high relative abundance at one camera close to a mooring (< 5 m). The relative abundance of Acanthopagrus australis varied among locations, increasing with distance to moorings at Clontarf, and decreasing in relative abundance at North Harbour. The number of observed bites varied with distance to nearest mooring, and among sampling times and location (Table 1). The strongest effect was among locations (Table 1), with Hunters Bay having the highest observed bites, increasing with increased distance to

Fig. 2. The relationships between total relative abundance (sum of species MaxN values) (a), species diversity (b), and total observed bites (c) per camera and distance to nearest mooring at the four locations containing boat moorings. The fitted lines and 95% confidence intervals are predictions from linear models for each location.

2.3. Fish associations with seagrass at Manly Cove With Manly Cove being the only location containing substantial seagrass cover, we ran fine-scale analyses again, at this location only, with the addition of seagrass cover as a predictor variable. For each video, a screenshot was taken from the footage and uploaded to Vidana software (www.marinespatialecologylab.org). The percentage cover of each seagrass species (P. australis, Z. muelleri and H. australis) was recorded and added to obtain total seagrass cover. Initially, we used linear regression to test for a relationship between seagrass cover and distance to nearest mooring. We then included seagrass cover in models 46

Marine Pollution Bulletin 129 (2018) 43–51

B.S. Lanham et al.

Fig. 3. The relationships between relative abundance per camera and distance to nearest mooring for the three most abundant fish species present at all locations across both sampling times (P. georgianus, A. australis, and G. tricuspidata). Data are MaxN values for each camera. The fitted lines and 95% confidence intervals are predictions from generalized linear models for each sampling time.

Table 2 Analyses of the fish community at Manly Cove, the only location with extensive seagrass beds. Results from univariate and multivariate generalized linear models contrasting total abundance, community composition and feeding behaviour (the likelihood of fish feeding and number of observed bites) of the demersal fish assemblage against distance to nearest moorings (m), distance to the shore, seagrass percentage cover, and sampling time, and a linear model contrasting Shannan's diversity index (H) against the same predictor variables. Dev is the deviance from generalized linear models and significant effects (P < 0.05) are in bold. Total abundance

Diversity (H)

Composition

Likelihood

Bites

Source

df

Dev

P

F

P

Dev

P

Dev

P

Dev

P

Distance to nearest mooring Distance to shore Seagrass Sampling time Distance to moorings x time

1 1 1 1 1

0.25 0.74 0.20 2.92 0.01

0.61 0.38 0.65 0.08 0.93

2.67 0.26 0.48 2.49 0.69

0.11 0.61 0.49 0.13 0.41

4.08 4.93 19.45 23.10 4.86

0.67 0.61 0.005 0.003 0.65

2.74 0.64 5.45 10.13 0.57

0.10 0.43 0.02 0.002 0.45

0.191 0.031 1.831 0.303 0.467

0.648 0.836 0.766 0.575 0.488

decreased with increasing seagrass cover (Fig. S5, Table 2), likely due to high but variable relative abundance of sediment feeding fish (mainly P. georgianus) at this location (Fig. 3). Seagrass loss has been extensive at Manly Cove (see Fig. S6), to the point that there is no correlation between seagrass cover and distance to nearest mooring (df = 1, P = 0.93).

moorings (interaction between distance to nearest mooring and location, Fig. 2c). The composition of fish species observed feeding varied among locations and with distance to moorings (see Fig. S3 for multivariate responses and species-specific patterns for the four species accounting for 85% of the observed bites). The likelihood of a fish feeding declined with increased distance to nearest mooring location at North Harbour (Fig. S4), with no obvious trends at other locations (an interaction between distance to nearest mooring and location).

3.3. Large-scale contrast of fish communities at locations with and without moorings

3.2. Fish associations with seagrass at Manly Cove The total relative abundance of the fish community at locations with moorings did not differ significantly from locations that lacked moorings (Fig. 4a). The variation in total relative abundance among locations was largely due to the high relative abundance of fish at Hunters Bay during the first sampling time. Other mooring locations (Clontarf, Manly Cove, and North Harbour) showed similar abundances to the unmodified locations across sampling times (Fig. 4a). Across all locations, total relative abundance varied significantly with distance to shore, among locations, and between sampling times (Table 3). The species diversity of the fish community differed among locations, with the differences among locations varying between the sampling times (Fig. 4b, an interaction between location and sampling time,

The total relative abundance of the of the fish community did not vary with distance from the nearest mooring, distance to shore, or seagrass cover (Table 2). The species diversity did not vary with distance to nearest mooring, distance to shore, and seagrass cover (Table 2). Fish community composition differed significantly with seagrass cover and sampling time but not with distance to nearest mooring (Table 2). Three of the four most abundant species varied across Manly Cove, as opposed to other locations lacking seagrass (Fig. 3). The number of bites did not differ with any of the predictor variables and proportion feeding varied among sampling times and 47

Marine Pollution Bulletin 129 (2018) 43–51

B.S. Lanham et al.

those locations that lacked moorings (Fig. 4c and Fig. S4). Quarantine Bay had a high proportion feeding in contrast to the other locations (Fig. S4) but did not differ in number of bites from Manly Cove, North Harbour, and Rose Bay (Fig. 4c). Across all locations, the likelihood of a fish feeding, and number of bites varied significantly with distance to shore, and among locations, with the differences among locations varying with sampling time (an interaction between location and sampling time, Table 3). 4. Discussion Boat moorings in Sydney Harbour were associated with changes in the relative abundance and composition of the demersal fish community over fine scales. Close to moorings, fish were less abundant, and in some locations, less diverse. These impacts were not evident on larger scales when contrasting locations with and without moorings. Although boat moorings are well known to negatively affect seagrass, our results show that they continue to affect the ecosystem even when seagrass is not present at that location. 4.1. Impacts of moorings on fish abundance and composition The strongest effects of moorings on the fish community were found at the three mooring locations in this study that had no, or very little, seagrass cover. Although relative abundance increased with distance from moorings at all locations, this relationship was least obvious at Manly Cove, where seagrass is present. In general, fewer fish were observed within 5 m of a boat mooring (i.e. within the mooring scour, Hastings et al., 1995) than at distances beyond 10 m from a mooring. Similarly, the fish community varied with distance to moorings and three of the most abundant species (P. georgianus, A. australis and G. tricuspidata) showed the strongest relationships at locations lacking seagrass, generally increasing in relative abundance further from moorings. Our study provides evidence that boat moorings act differently to other artificial structures that are commonly associated with a high abundance and diversity of fish, for example those found on artificial reefs (Champion et al., 2015; Smith et al., 2017), and associated with other recreational boating infrastructure such as marinas (Clynick, 2006; Clynick, 2008). The orientation and small size of boat moorings in contrast to other structures may contribute to the reduced relative abundance. Moorings are close to the substrate with only a rope and small buoy providing any vertically oriented habitat. Vertical structures generally house diverse and abundant fish communities (Clynick, 2008; Champion et al., 2015, Smith et al., 2017), with little previously known about fish responses to smaller, moving horizontal structures like swing moorings. Noise pollution associated with boat mooring chains could be affecting the sensitive auditory organs of fish, resulting in a range of behavioral responses (Kunc et al., 2016). In response to noise, fish can alter their position in the water column (Pearson et al., 1992; Fewtrell and McCauley, 2012), and noise may affect the ability of species that utilize sound to hunt (Kunc et al., 2016) reducing the likelihood and success rate of prey strikes (Voellmy et al., 2014). The observed behavioral responses in this study suggest that any noise associated with swing moorings could impact the fish community in close proximity to moorings. At Manly Cove, where seagrass meadows persist, there was no relationship between distance from moorings and the composition of the fish assemblage. However, the community composition was affected by seagrass cover, suggesting seagrasses are of higher importance to fish than the presence of boat moorings. The composition and configuration of habitats has a strong influence on the fish community (Staveley et al., 2016), with cover of benthic habitat being a strong predictor of fish abundance and diversity (Parsons et al., 2016). Manly Cove has very patchy seagrass meadows, with no correlation between seagrass cover and boat moorings, suggesting that long term degradation has increased the fragmentation of seagrass meadows, altering the fish community

Fig. 4. The total relative abundance (sum of species MaxN values) (a), species diversity (b), and total observed bites (c) of fish across all locations (with and without moorings) at two sampling times. Data are mean per video and standard error. Locations sharing a letter do not differ significantly from one another in a Tukey's post hoc test contrasting each combination of location and sampling time.

Table 3). The two locations lacking moorings did not differ consistently from the locations with moorings (Fig. 4b). Species diversity was highest at Hunters Bay during sampling time 1 and Manly Cove during sampling time 2, with the other locations not differing in diversity (Fig. 4b). The species composition of the fish community varied with distance to shore, location, and sampling time, with interaction present between location and sampling time (Table 3). The likelihood of a fish feeding, and the number of observed bites did not vary consistently between the locations with moorings and 48

Marine Pollution Bulletin 129 (2018) 43–51

B.S. Lanham et al.

Table 3 Large scale analyses of the fish community at locations with and without moorings. Results from univariate and multivariate generalized linear models contrasting total abundance, community composition and feeding behaviour (the likelihood of fish feeding and number of observed bites) of the demersal fish assemblage against distance to the shore, location, and sampling time, and linear model contrasting Shannon's diversity index (H) against the same predictor variables. Dev is the deviance from generalized linear models and significant effects (P < 0.05) are in bold. Total abundance

Diversity (H)

Composition

Likelihood

Bites

Source

df

Dev

P

F

P

Dev

P

Dev

P

Dev

P

Distance to shore Location Sampling time Location x Sampling time

1 5 1 5

14.12 85.65 67.19 2.47

< 0.001 < 0.001 < 0.001 0.64

0.09 14.05 52.07 3.76

0.76 < 0.001 < 0.001 0.001

67.9 337.3 187.0 66.7

< 0.001 < 0.001 < 0.001 < 0.001

22.47 195.01 0.01 42.90

< 0.001 < 0.001 0.92 < 0.001

3.73 83.56 42.13 17.31

0.047 < 0.001 < 0.001 0.002

Predation pressure on the substrate is a product of fish abundance and fish behavior, with each of these potentially altered by prey availability. Artificial structures often act as fish attractants by providing concentration of prey. Piers and pilings have abundant sessile invertebrate communities (Dafforn et al., 2012), increasing the abundance of fish in these locations (Clynick, 2008), whereas on structures with little epibiont cover, such as mooring blocks (pers. obs), fish predation is greatly reduced (Redman and Szedlmayer, 2009). Rates of fish predation may be further affected by abiotic factors. Noise associated with moorings may affect the foraging and feeding efficiency of fish (as discussed above), and chain scour disturbs soft sediments, altering grain size and potentially resuspending heavy metals into the water column. Sampling the same sites as this study, Hedge et al. (2017) recently demonstrated that sediment grain size and metal concentrations varied with distance to moorings. Sediment grain size and metal concentrations can alter the abundance and composition of sediment dwelling species (Burton and Johnston, 2010; Hill et al., 2013), possibly leading to changes in prey availability for sediment feeding fish. However, sediment characteristics not measured by Hedge et al. (2017) could also be influencing the fish community feeding patterns. For example, sediment nitrogen loads also influence fish assemblages (Warry et al., 2016), and could potentially be altered by the sediment disturbance created by swing moorings.

and predation pressure. Seagrass friendly moorings are also present at Manly Cove, and the presence of seagrass under some of these moorings, and less disturbed sediments may also explain the lack of trends related to relative fish abundance and distance to moorings. Differences among locations were prominent throughout the study for all response variables tested. However, these differences were not due to the presence of mooring fields within an embayment, as unmodified locations were similar to at least one location with moorings across the tested variables. Contrasting fish abundance exclusively at large scales among locations would have therefore failed to detect the effects boat moorings are having on demersal fish communities. Variation among locations on these scales is consistent with previous studies on demersal fish communities possibly due to site characteristics and site fidelity of some species. For example, differences among locations are often attributed to altered processes (Lincoln Smith et al., 1991; Clynick et al. 2008) and Clynick (2006) found similar habitats in Sydney Harbour often share similar species but vary in abundances across locations. In this study, locations at a similar distance to the estuary mouth (Hunters Bay and Manly Cove) had similar fish diversity, and relative abundance during sampling time two, consistent with previous work in Sydney Harbour that found changes in the fish assemblage with increasing distance from the estuary mouth (Clynick, 2008). High site fidelity is common in coastal fish species (Gannon et al., 2015; Lowry et al., 2017) with some species rarely found > 30 m from a reef, and others > 5 m (Champion et al., 2015; Scott et al., 2015). Site characteristics operating at larger scales than the effects of boat moorings may be attracting fish to these locations, where they remain, creating differences in the fish assemblage. The variation among locations highlights that large-scale contrasts may not be effective in assessing the impacts of structures operating at fine-scales.

4.3. Recommendations and conclusions The observed changes in relative fish abundance, community composition, and feeding behaviour adds to the growing literature of the negative effects of swing moorings on benthic habitats that extend beyond the loss of seagrass. Detecting fine-scale effects of boat moorings on the fish community may enable detailed spatial planning of mooring deployments in marine estates. Here we uncovered deterrent effects of boat moorings on demersal fish, most notably within 5 m of moorings (scour area). As degradation from mooring scours increase in size, the accumulative effects of dense mooring fields may significantly reduce the habitable area for many fish species. Reducing the effects of artificial structures on benthic communities in urbanized estuaries is essential as these ecosystems continue to decline (Stuart-Smith et al., 2015) due to increased ocean sprawl (Heery et al., 2017). Improving the design of destructive structures, such as boat moorings, is also critical. Our results from Manly Cove, where seagrass friendly moorings are present, suggest that the implementation of these structures can benefit the fish community. We encourage the implementation of improved mooring designs, in combination with well-spaced deployments and the restoration of habitat formers, to allow for the recovery of benthic ecosystems.

4.2. Feeding behaviour responses to boat moorings and seagrass cover Changes in the relative abundance and composition of the fish assemblage are expected to result in shifts in predation pressure. Our large-scale observations contrasting locations with and without moorings did not detect any effects of boat moorings on feeding behavior, in contrast to Bennett and Bellwood's (2011) findings that the proportion of fish feeding varies over large scales. On a fine-scale, however, we observed relationships between the distance to moorings and the number of bites per sampling time at all mooring locations, and the likelihood of a fish feeding at North Harbour. In general, this involved increased feeding activity at increasing distance from moorings. This contrasts to previous studies that found resident fish can use artificial habitats as a refuge and rarely forage > 5 m from the structure (Champion et al., 2015), or result in reduced sediment infauna in close proximity to artificial structures (Davis et al., 1982). High predation in bare sediment was observed at Manly Cove with fish more likely to feed in areas with low or zero seagrass cover. P. georgianus were highly abundant at Manly Cove and are commonly observed in bare sediments (Parsons et al., 2016), consuming a large volume of small benthic invertebrates (French et al., 2012).

Acknowledgements This project was partly funded by an Australian Research Centre Linkage Project Grant LP140100855 Awarded to ELJ, AV and AGBP. Further funding was supplied by Transport for NSW. We thank K. 49

Marine Pollution Bulletin 129 (2018) 43–51

B.S. Lanham et al.

Griffin, R. Garthwin, S. Macolino, J. Ledet, A. Lee, and D. Cruz for assistance in the field and laboratory, and the anonymous reviewer for comments that improved this manuscript.

Hastings, K., Hesp, P., Kendrick, G.A., 1995. Seagrass loss associated with boat moorings at Rottnest Island, Western Australia. Ocean & Coastal Management 26, 225–246. Hedge, L.H., Dafforn, K.A., Simpson, S.L., Johnston, E.L., 2017. Uncovering hidden heterogeneity: geo-statistical models illuminate the fine scale effects of boating infrastructure on sediment characteristics and contaminants. Mar. Pollut. Bull. 119, 143–150. Heery, E.C., Bishop, M.J., Critchley, L.P., Bugnot, A.B., Airoldi, L., Mayer-Pinto, M., Sheehan, E.V., Coleman, R.A., Loke, L.H., Johnston, E.L., 2017. Identifying the consequences of ocean sprawl for sedimentary habitats. J. Exp. Mar. Biol. Ecol. 492, 31–48. Hill, N.A., Simpson, S.L., Johnston, E.L., 2013. Beyond the bed: effects of metal contamination on recruitment to bedded sediments and overlying substrata. Environ. Pollut. 173, 182–191. Johnston, E.L., Mayer-Pinto, M., Hutchings, P.A., Marzinelli, E.M., Ahyong, S.T., Birch, G., Booth, D.J., Creese, R.G., Doblin, M.A., Figueira, W., Gribben, P.E., Pritchard, T., Roughan, M., Steinberg, P.D., Hedge, L.H., 2015. Sydney Harbour: what we do and do not know about a highly diverse estuary. Mar. Freshw. Res. 66, 1073–1087. Kunc, H.P., McLaughlin, K.E., Schmidt, R., 2016. Aquatic noise pollution: implications for individuals, populations, and ecosystems. Proceedings of the Royal Society BBiological Sciences 283, 20160839. Lincoln Smith, M.P., Bell, J.D., Hair, C.A., 1991. Spatial variation in abundance of recently settled rocky reef fish in southereastern Australia: Implicaitons for detecting change. Mar. Ecol. Prog. Ser. 77, 95–103. Lowry, M., Becker, A., Folpp, H., McLeod, J., Taylor, M.D., 2017. Residency and movement patterns of yellowfin bream (Acanthopagrus australis) released at natural and artificial reef sites. Mar. Freshw. Res. 68, 1479–1488. Lynch, T., Green, M., Davies, C., 2015. Diver towed GPS to estimate densities of a critically endangered fish. Biol. Conserv. 191, 700–706. Mayer-Pinto, M., Johnston, E.L., Hutchings, P.A., Marzinelli, E.M., Ahyong, S.T., Birch, G., Booth, D.J., Creese, R.G., Doblin, M.A., Figueira, W., Gribben, P.E., Pritchard, T., Roughan, M., Steinberg, P.D., Hedge, L.H., 2015. Sydney Harbour: a review of anthropogenic impacts on the biodiversity and ecosystem function of one of the world's largest natural harbours. Mar. Freshw. Res. 66, 1088–1105. NSW, T.f, 2013. Sydney Harbour boat storage strategy. In: T.f.N.S (Ed.), Wales, . https:// www.transport.nsw.gov.au/newsroom-and-events/reports-and-publications/sydneyharbour-storage-strategy. Orth, R.J., Carruthers, T.J.B., Dennison, W.C., Duarte, C.M., Fourqurean, J.W., Heck, K.L., Hughes, A.R., Kendrick, G.A., Kenworthy, W.J., Olyarnik, S., Short, F.T., Waycott, M., Williams, S.L., 2006. A global crisis for seagrass ecosystems. Bioscience 56, 987–996. Parsons, D.F., Suthers, I.M., Cruz, D.O., Smith, J.A., 2016. Effects of habitat on fish abundance and species composition on temperate rocky reefs. Mar. Ecol. Prog. Ser. 561, 155–171. Pearson, W.H., Skalski, J.R., Malme, C.I., 1992. Effects of sounds from a geophysical survey device on behavior of captive rockfish (Sebastes spp). Can. J. Fish. Aquat. Sci. 49, 1343–1356. Perez-Ruzafa, A., Grarcia-Charton, J.A., Barcala, E., Marcos, C., 2006. Changes in benthic fish assemblages as a consequence of coastal works in a coastal lagoon: the Mar Menor (Spain, Western Mediterranean). Mar. Pollut. Bull. 53, 107–120. Perkol-Finkel, S., Ferrario, F., Nicotera, V., Airoldi, L., 2012. Conservation challenges in urban seascapes: promoting the growth of threatened species on coastal infrastructures. J. Appl. Ecol. 49, 1457–1466. Pister, B., 2009. Urban marine ecology in southern California: the ability of riprap structures to serve as rocky intertidal habitat. Mar. Biol. 156, 861–873. Redman, R.A., Szedlmayer, S.T., 2009. The effects of epibenthic communities on reef fishes in the northern Gulf of Mexico. Fish. Manag. Ecol. 16, 360–367. Rilov, G., Benayahu, Y., 2000. Fish assemblage on natural versus vertical artificial reefs: the rehabilitation perspective. Mar. Biol. 136, 931–942. Rivero, N.K., Dafforn, K.A., Coleman, M.A., Johnston, E.L., 2013. Environmental and ecological changes associated with a marina. Biofouling 29, 803–815. Scott, M.E., Smith, J.A., Lowry, M.B., Taylor, M.D., Suthers, I.M., 2015. The influence of an offshore artificial reef on the abundance of fish in the surrounding pelagic environment. Mar. Freshw. Res. 66, 429–437. Serrano, O., Ruhon, R., Lavery, P.S., Kendrick, G.A., Hickey, S., Masque, P., Arias-Ortiz, A., Steven, A., Duarte, C.M., 2016. Impact of mooring activities on carbon stocks in seagrass meadows. Sci. Rep. 6, 23193. Silberberger, M.J., Thormar, J., Fredriksen, S., 2016. Small-scale removal of seagrass (Zostera marina L.): effects on the infaunal community. Mar. Biol. Res. 12, 993–1002. Smith, T.M., Hindell, J.S., Jenkins, G.P., Connolly, R.M., Keough, M.J., 2011. Edge effects in patchy seagrass landscapes: the role of predation in determining fish distribution. J. Exp. Mar. Biol. Ecol. 399, 8–16. Smith, J.A., Cornwell, W.K., Lowry, M.B., Suthers, I.M., 2017. Modelling the distribution of fish around an artificial reef. Mar. Freshw. Res. 68, 1955–1964. Staveley, T.A., Perry, D., Lindborg, R., Gullström, M., 2016. Seascape structure and complexity influence temperate seagrass fish assemblage composition. Ecography 40, 936–946. Stevens, D.L., Olsen, A.R., 2004. Spatially balanced sampling of natural resources. J. Am. Stat. Assoc. 99, 262–278. Stuart-Smith, R.D., Edgar, G.J., Stuart-Smith, J.F., Barrett, N.S., Fowles, A.E., Hill, N.A., Cooper, A.T., Myers, A.P., Oh, E.S., Pocklington, J.B., Thomson, R.J., 2015. Loss of native rocky reef biodiversity in Australian metropolitan embayments. Mar. Pollut. Bull. 95, 324–332. Thrush, S.E., 1999. Complex role of predators in structuring soft-sediment macrobenthic communities: implications of changes in spatial scale for experimental studies. Aust. J. Ecol. 24, 344–354. Unsworth, R.K.F., Williams, B., Jones, B.L., Cullen-Unsworth, L.C., 2017. Rocking the boat: damage to eelgrass by swinging boat moorings. Front. Plant Sci. 8, 1309.

Appendix A. Supplementary data Supplementary data to this article can be found online at https:// doi.org/10.1016/j.marpolbul.2018.02.010. References Able, K.W., Manderson, J.P., Studholme, A.L., 1998. The distribution of shallow water juvenile fishes in an urban estuary: the effects of manmade structures in the lower Hudson river. Estuaries 21, 731–744. ABS, 2016. http://www.censusdata.abs.gov.au/census_services/getproduct/census/ 2016/quickstat/1GSYD?opendocument, Accessed date: 15 November 2017. Airoldi, L., Turon, X., Perkol-Finkel, S., Rius, M., 2015. Corridors for aliens but not for natives: effects of marine urban sprawl at a regional scale. Divers. Distrib. 21, 755–768. Bacchiocchi, F., Airoldi, L., 2003. Distribution and dynamics of epibiota on hard structures for coastal protection. Estuarine Coastal and Shelf Science 56, 1157–1166. Bennett, S., Bellwood, D.R., 2011. Latitudinal variation in macroalgal consumption by fishes on the Great Barrier Reef. Mar. Ecol. Prog. Ser. 426, 241–252. Bohnsack, J.A., 1989. Are high-densities of fishes at artificial reefs the result of habitat limitation or behavioral preference. Bull. Mar. Sci. 44, 631–645. Bohnsack, J.A., Sutherland, D.L., 1985. Artificial reef research—a review with recommendations for future priorities. Bull. Mar. Sci. 37, 11–39. Bulleri, F., Chapman, M.G., 2010. The introduction of coastal infrastructure as a driver of change in marine environments. J. Appl. Ecol. 47, 26–35. Burton, G.A., Johnston, E.L., 2010. Assessing contaminated sediments in the context of multiple stressors. Environ. Toxicol. Chem. 29, 2625–2643. Cappo, M., Harvey, E., Malcolm, H., Speare, P., 2003. Potential of video techniques to monitor diversity, abundance and size of fish in studies of marine protected areas. In: Aquatic Protected Areas - What Works Best and How Do We Know, pp. 455–464. Champion, C., Suthers, I.M., Smith, J.A., 2015. Zooplanktivory is a key process for fish production on a coastal artificial reef. Mar. Ecol. Prog. Ser. 541, 1–14. Clynick, B.G., 2006. Assemblages of fish associated with coastal marinas in north-western Italy. J. Mar. Biol. Assoc. U. K. 86, 847–852. Clynick, B.G., 2008. Characteristics of an urban fish assemblage: distribution of fish associated with coastal marinas. Mar. Environ. Res. 65, 18–33. Clynick, B.G., Chapman, M.G., Underwood, A.J., 2008. Fish assemblages associated with urban stuctures and natural reefs in Sydney, Australia. Austral Ecology 33, 140–150. Cresson, P., Ruitton, S., Harmelin-Vivien, M., 2014. Artificial reefs do increase secondary biomass production: mechanisms evidenced by stable isotopes. Mar. Ecol. Prog. Ser. 509, 15–26. Dafforn, K., Johnston, E., Glasby, T., 2009. Shallow moving structures promote marine invader dominance. Biofouling 25, 277–287. Dafforn, K.A., Glasby, T.M., Johnston, E.L., 2012. Comparing the invasibility of experimental “reefs” with field observations of natural reefs and artificial structures. PLoS One 7, e38124. Dafforn, K.A., Glasby, T.M., Airoldi, L., Rivero, N.K., Mayer-Pinto, M., Johnston, E.L., 2015. Marine urbanization: an ecological framework for designing multifunctional artificial structures. Front. Ecol. Environ. 13, 82–90. Davis, N., Vanblaricom, G.R., Dayton, P.K., 1982. Man-made structures on marine-sediments - effects on adjacent benthic communities. Mar. Biol. 70, 295–303. Demers, M.C.A., Davis, A.R., Knott, N.A., 2013. A comparison of the impact of 'seagrassfriendly' boat mooring systems on Posidonia australis. Mar. Environ. Res. 83, 54–62. Dugan, J.E., Airoldi, L., Chapman, M.G., Walker, S.J., Schlacher, T., 2011. Estuarine and coastal structures: environmental effects, a focus on shore and nearshore structures. Treatise on Estuarine and Coastal Science, Vol 8: Human-Induced Problems (uses and abuses). pp. 17–41. Edgar, G.J., Shaw, C., 1995. The production and trophic ecology of shallow-water fish assemblages in southern Australia .3. General relationships between sediments, seagrasses, invertebrates and fishes. J. Exp. Mar. Biol. Ecol. 194, 107–131. Fewtrell, J., McCauley, R., 2012. Impact of air gun noise on the behaviour of marine fish and squid. Mar. Pollut. Bull. 64, 984–993. Folpp, H., Lowry, M., Gregson, M., Suthers, I.M., 2013. Fish assemblages on estuarine artificial reefs: natural rocky-reef mimics or discrete assemblages? PLoS One 8, e63505. French, B., Platell, M.E., Clarke, K.R., Potter, I.C., 2012. Ranking of length-class, seasonal and regional effects on dietary compositions of the co-occurring Pagrus auratus (Sparidae) and Pseudocaranx georgianus (Carangidae). Estuarine Coastal and Shelf Science 115, 309–325. Freon, P., Dagorn, L., 2000. Review of fish associative behaviour: toward a generalisation of the meeting point hypothesis. Rev. Fish Biol. Fish. 10, 183–207. Gannon, R., Payne, N.L., Suthers, I.M., Gray, C.A., van der Meulen, D.E., Taylor, M.D., 2015. Fine-scale movements, site fidelity and habitat use of an estuarine dependent sparid. Environ. Biol. Fish 98, 1599–1608. Glasby, T.M., 1999. Effects of shading on subtidal epibiotic assemblages. J. Exp. Mar. Biol. Ecol. 234, 275–290. Harasti, D., 2016. Declining seahorse populations linked to loss of essential marine habitats. Mar. Ecol. Prog. Ser. 546, 173–181.

50

Marine Pollution Bulletin 129 (2018) 43–51

B.S. Lanham et al.

12377–12381. Wetzel, M.A., Scholle, J., Teschke, K., 2014. Artificial structures in sediment-dominated estuaries and their possible influences on the ecosystem. Mar. Environ. Res. 99, 125–135. Whitfield, A.K., Becker, A., 2014. Impacts of recreational motorboats on fishes: a review. Mar. Pollut. Bull. 83, 24–31. Wickham, D.A., Watson Jr, J.W., Ogren, L.H., 1973. The efficacy of midwater artificial structures for attracting pelagic sport fish. Trans. Am. Fish. Soc. 102, 563–572. Willis, T.J., Millar, R.B., Babcock, R.C., 2000. Detection of spatial variability in relative density of fishes: comparison of visual census, angling, and baited underwater video. Mar. Ecol. Prog. Ser. 198, 249–260. Zanuttigh, B., Martinelli, L., Lamberti, A., Moschella, P., Hawkins, S., Marzetti, S., Ceccherelli, V.U., 2005. Environmental design of coastal defence in lido di Dante, Italy. Coast. Eng. 52, 1089–1125. Zuur, A.F., Ieno, E.N., Saveliev, A.A., 2009. Mixed Effects Models and Extensions in Ecology with R. Springer.

Vaselli, S., Bulleri, F., Benedetti-Cecchi, L., 2008. Hard coastal-defence structures as habitats for native and exotic rocky-bottom species. Mar. Environ. Res. 66, 395–403. Voellmy, I.K., Purser, J., Flynn, D., Kennedy, P., Simpson, S.D., Radford, A.N., 2014. Acoustic noise reduces foraging success in two sympatric fish species via different mechanisms. Anim. Behav. 89, 191–198. Walker, D.I., Lukatelich, R.J., Bastyan, G., McComb, A.J., 1989. Effect of boat moorings on seagrass beds near Perth, Western Australia. Aquat. Bot. 36, 69–77. Wang, Y., Naumann, U., Wright, S.T., Warton, D.I., 2012. Mvabund - an R package for model-based analysis of multivariate abundance data. Methods Ecol. Evol. 3, 471–474. Warry, F.Y., Reich, P., Cook, P.L.M., Mac Nally, R., Thomson, J.R., Woodland, R.J., 2016. Nitrogen loads influence trophic organization of estuarine fish assemblages. Funct. Ecol. 30, 1723–1733. Waycott, M., Duarte, C.M., Carruthers, T.J.B., Orth, R.J., Dennison, W.C., Olyarnik, S., Calladine, A., Fourqurean, J.W., Heck, K.L., Hughes, A.R., Kendrick, G.A., Kenworthy, W.J., Short, F.T., Williams, S.L., 2009. Accelerating loss of seagrasses across the globe threatens coastal ecosystems. Proc. Natl. Acad. Sci. U. S. A. 106,

51