Functional and structural responses to marine urbanisation

Intertidal epibenthic assemblages were sampled at 3 locations of each type of habitat or structure (pilings, seawalls and rocky shores) in Sydney Harbour, Australia in January 2014 (figure 1). Locations were between 1 km and 7 km apart. Sampling was undertaken during low tide and sampling effort focused on mid to low-shore intertidal assemblages (0.2–0.8 m above low water spring tides). At each location, six quadrats (10 × 10 cm) were photographed and all organisms within the quadrat were collected and transported to the laboratory. In the laboratory, each sample was rinsed with ethanol over a 500 μm mesh to remove excess detritus. Organisms were preserved in 80% ethanol and then counted and identified, under a dissection microscope, to the lowest possible taxonomic level. Colonial organisms and algae were recorded as present or absent in each sample. All identified species were then classified according to their feeding mode [functional groups; 53], e.g. filter-feeders, predators, scavengers, etc The biomass (ash-free dry weight (AFDW)) of each replicate quadrat at each location and type of habitat was calculated for each broad taxonomic group (e.g. polychaetes, gastropods), with the exception of the oyster Saccostrea glomerata Gould, which was measured separately due to its large size. Organisms were air dried for 24 hours in a fume cupboard with airflow and weighed. Dry samples were then transferred into a crucible and ashed in a muffle furnace for 3 hours at 500 °C. AFDW was calculated by subtracting ash weight from dry weight.

We also reviewed the relevant literature from Sydney Harbour to (1) contextualise rather than directly compare our findings (because of methodological differences), (2) investigate if general patterns were consistent regardless of sampling effort, and (3) ensure that sampling done here was representative of the assemblages present without major omissions. The review included all papers and reports that involved sampling intertidal benthic assemblages on at least one of the habitats studied here (i.e. rocky shore, seawalls or pilings). Only papers done in Sydney Harbour [see 50, 56] and that reported organisms identified to species level were included in the literature review. A list of species reported on each type of habitat in the Harbour was compiled, and these species were then classified as native, non-indigenous or of unknown origin (for comparison with our survey assessment of invasibility in each habitat, see below) according to the literature.

Filtration rates of the habitat-former Saccostrea glomerata were measured in situ, using a methodology adapted from Browne et al [54] and Cole et al [55]. Filtration rates and oysters dimensions were measured twice, following the same procedure, in January and June 2014. On the first occasion, measurements were taken at 3 locations of each type of habitat, while on the second occasion measurements were taken at 4 locations of each type of habitat. Sampling was done two times to assess whether differences among type of habitats, if any, were consistent through time. Filtration chambers (of approximately 4 l volume each) were placed, inverted, on top of clusters of oysters (minimum of 5 oysters per cluster and maximum of 40) for 30 min. Each cluster was then considered a replicate, n = 5 replicates per location. Chambers were made using polypropylene containers, with a 100 mm diameter hole cut to allow access when attached to the substrata. The edge of the chamber was lined with a strip of flexible self-adhesive foam tape (12 mm thick, 19 mm wide, open cell polyurethane foam), to form a tight seal with the rocky substratum. Oysters around the selected cluster, which could prevent a proper water-tight seal, were carefully scraped using a paint scraper and/or a chisel. A tight seal along the edges of the filtration chamber was maintained over each patch by ensuring the substrata were naturally smooth and by tightly securing the chamber using eye-bolts and cable-ties (see figure S1). On pilings, chambers were secured using a ratchet strap (figure S1). Each chamber was carefully monitored throughout the filtration period to ensure no leaks were occurring. In the few occasions that water leaks were detected, the replicate in question was discarded and another cluster of oyster was chosen. Each filtration chamber was filled with 4 l of water from the location. The water was well-mixed at regular intervals (~ every 5 min, for 15 seconds) with an electric whisk. At the beginning of the experiment and after 30 min, 100 ml of water was sampled with a syringe. The samples were preserved with glutaraldehyde and stored in a −4 °C freezer. They were later slowly defrosted in a 38 °C waterbath, and particles were counted in 10 μm size classes between 10 and 100 μm (Lepesteur, Martin et al 1993). The number of particles per sample was determined and filtration rates calculated as the reduction in the number of particles over 30 min. All oysters present within each replicate were carefully removed and preserved with ethanol 70%. Oysters were then counted, measured (height and length) and weighed (wet and dry weight of the flesh). Dry weight of oyster flesh was weighed after drying each sample at 500 °C for 24 hours.

To investigate the role of artificial and natural habitats as habitat for NIS, all collected intertidal organisms were classified as native, non-indigenous or of unknown origin according to the literature. We then calculated the proportion of NIS on each structure.

To compare the secondary productivity of assemblages (i.e. biomass accumulation) among habitats, 6 quadrats (10 × 10 cm) were cleared on each habitat (i.e. pilings, seawalls and natural reefs) at each location, using a paint scraper and a metal brush. Quadrats were marked using 6.5 mm rawl plugs drilled at two opposing corners of each quadrat. This was done in January 2015. After 6 months, a new set of quadrats was cleared, following the methodology described above. After a further 6 months, all organisms within both sets of quadrats were collected and transported to the laboratory. Samples were stored at −20 °C until processed. Total biomass (AFDW, g m⁻¹) was determined for the standing-stock.

Univariate tests for differences in response variables according to Habitat and Location were tested with generalized linear mixed models (GLMMs) using the lme4 package [57] in R v.3.0.1. Habitat was a fixed factor, and Location was random. Locations were nested within Habitat. We assumed Poisson distributions for all measurements of count, but when there was significant overdispersion, we used negative binomial. We assumed gamma distributions for measurements of size of oysters. P-values were obtained with Likelihood Ratio Test (LRT) [58]. For biomass measurements, we used the Tweedie distribution in the cplm package. All post-hoc tests were done using the package multcomp in R v.3.0.1.

For comparisons of filtration rates among habitats, we assumed Poisson distributions (i.e. number of particles consumed by the oysters). We did two different analyses to test for differences in filtration rates among habitats. In the first analysis, we accounted for influences of initial number of particles in the water and biomass of oysters on filtration rates within each chamber when comparing habitats and locations by including these variables in the GLMMs. That way, we were considering the 'net' filtration rate in each habitat. In the second analysis done, we compared filtration rates of oysters without taking into account any other variables (e.g. initial number of particles and biomass of oysters). This was done because oyster biomass and the initial number of particles in the water were correlated to habitat (see Results section). Therefore, in this analysis, we are considering the total amount of particles filtered per habitat. The abundance of oysters at each location and habitat was standardised by the total area (m²) of the container. Containers had an area of 0.025 m² for seawalls and rocky shores and 0.021 m² for pilings.

To determine any significant differences between treatments regarding relative abundance and composition of assemblages, multivariate analyses were done using PERMANOVA in PRIMER 6 [59]. Analyses were run using two different similarity matrices: Bray-Curtis on untransformed data and Jaccard dissimilarities. When run on untransformed data, Bray-Curtis gives more weight to changes in species abundances, whereas Jaccard is based on changes in species composition (e.g. presence-absence) and does not take into account changes in species relative abundances [60]. When used in combination, these two measures of similarity allow the relative importance of changes in species abundances or composition to be assessed. For all analysis, we used 9999 permutations under a reduced model [59].

A total of 16 361 specimens from 112 taxa were sampled from 54 quadrats during surveys (January 2014) to compare assemblage structure among habitats (table 1). During our surveys, we found a total of 88 taxa in rocky reefs compared with 70 taxa found on pilings and 60 on seawalls (table 1). Differences were not, however, significant (supplementary table 1 available at stacks.iop.org/ERL/13/014009/mmedia).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We found 25 papers in our literature review done in intertidal areas of Sydney Harbour, spanning from 1972−2015 (table 1). Studies varied greatly regarding their sampling effort and methodology. Furthermore, our study was the first to sample intertidal pilings within Sydney Harbour. Although total taxa numbers differed, patterns from the literature review were similar to our surveys and showed that rocky reefs are generally more diverse than artificial habitats, with 162 taxa, followed by seawalls, with 113 taxa. Differences in the total number of taxa found between our survey and the literature review are likely related to the overall greater sampling effort represented by the literature review and in differences regarding the target assemblages sampled [e.g. 30, 61]. For instance, Chapman [30] sampled a total of 800 non-destructive quadrats in seawalls and rocky reefs of Sydney Harbour, compared to 36 quadrats destructively sampled here on these habitats. In contrast, Bugnot et al [62] specifically targeted biogenic habitats (e.g. oyster and mussel beds) in their sampling efforts on rocky shores and seawalls (n ~ 27 per type of biogenic habitat). Despite these differences in the methodology used and sampling intensity, results from our surveys reflect the general patterns found in the literature review, indicating that surveys done here accurately reflect patterns in the assemblages found on artificial and natural habitats in Sydney Harbour.

In our surveys, we found that intertidal rocky shores harboured 26 unique taxa (≈ 30%) while 5 and 9 taxa were exclusive to pilings and seawalls, respectively (≈ 7% and 13%, respectively; table 1). These patterns were supported by the literature review, with 70 unique taxa (or 43%) exclusive to rocky shores compared to 33 (or 29%) on seawalls (table 1).

Pilings supported a greater total biomass (1510 g m⁻² on average) than rocky shores (676 g m⁻²) and seawalls (289 g m⁻²) (LRT Chisq = 8.45; df = 2; p < 0.05; supplementary table 1, figure 2(a)), which was mainly driven by oyster abundance. When these organisms were excluded from the analyses, there were no differences in biomass among habitats (LRT = 4.66; df = 2; p > 0.05; figure 2(b)).

Abundances of dominant taxonomic groups also differed in surveys of different habitats. Specifically, intertidal mobile species such as gastropods and polychaetes were significantly more abundant on rocky shores than on pilings or seawalls (LRT Chisq = 10.33 and 8.64, respectively; df = 2; p < 0.01; supplementary table 1, figure S2). Similar patterns were found for the biomass of these animals (figure S3); with rocky shores supporting more gastropod and polychaete biomass (69 g m⁻² and 7.5 g m⁻² on average, respectively) than seawalls (15 g m⁻² and 3.2 g m⁻², respectively), and pilings (8 g m⁻² and 0.5 g m⁻²). In contrast, barnacles were most abundant and accounted for more biomass on seawalls, followed by pilings and rocky shores (LRT Chisq = 8.47; df = 2; p < 0.01; supplementary table 1, figures S2 and S3). We found a total of 7 algal species in natural rocky shores and 5 species in seawalls and pilings (table 1). There were no differences in the algal biomass found in each habitat (supplementary table 1; figure S3).

Abundances of surveyed functional groups also differed among habitats (supplementary table 1; figure 3). Grazers were, on average, at least 40% more abundant on rocky shores than on artificial habitats (figure 3(a)). Pilings had the least amount of grazers (LRT Chisq = 30.41; df = 2; p < 0.001; figure 3(a)), while scavengers, e.g. isopods, were most abundant on seawalls (LRT Chisq = 11.71; df = 2; p < 0.01; supplementary table 1; figure 3(c)). This reflected results from our literature review (table 1). Grazers, for example, were at least 50% more diverse, in terms of number of species, on rocky shores than on artificial habitats (table 1). There were no differences in the abundance (LRT Chisq = 0.42; df = 2; p > 0.05; supplementary table 1, figure 3(b)) or number of species of filter-feeders among habitats found in our survey—with 19, 20 and 17 species found on pilings, rocky shores and seawalls, respectively (table 1).

There were major differences in species assemblages among habitats, regardless of differences in locations. The relative abundance of species (Bray-Curtis; pseudo-F_2,6 = 2.5; p < 0.01) and composition (presence/absence; Jaccard; pseudo-F_2,6 = 2.1; p < 0.05) of intertidal seawall assemblages differed significantly from pilings or reefs (figure 4). Dispersion analyses (PERMDISP) showed that abundances of species on seawalls were also more homogeneous than on rocky shores or pilings (F_{2, 51} = 25, p < 0.01).

Oysters were significantly more abundant on pilings (952 ind m⁻², on average) than on rocky shores (271 ind m⁻²) or seawalls (512 ind m⁻²) (supplementary table 2; figures 5(a) and (b)). However, these animals were significantly smaller on pilings than on the other habitats (LTR Chisq = 23; df = 2; p < 0.001; figures 5(e) and (f); figure S4). Dry weight of oyster flesh on seawalls was approximately 30% less than on rocky shores and pilings at the first time of sampling (although no significant differences were found), while rocky shores had significantly less dry weight of oysters during the second time of sampling (LTR Chisq = 6.4; df = 2; p < 0.05; figures 5(c) and (d)).

Oyster filtration rates increased in relation to particulate matter in the water column (F_{1, 37} = 59.2; p < 0.05; R² = 0.6, figure S5). The number of particles in the water varied with time and type of habitat, with more particles in the water around seawalls when compared to other habitats at the first time of sampling (LRT Chisq = 10.8; df = 2; p < 0.01). No differences in the number of particles among the different habitats were found for the second sampling time (LTR Chisq = 3.5; df = 2; p > 0.05).

The total amount of particles filtered by oysters did not differ among habitats (supplementary table 2; figure 6). However, when analyses were done after accounting for initial number of particles in the water and oyster biomass, filtration rates varied in time and per type of habitat. Filtration rates were greater on seawalls than on pilings or rocky shores at the first time of sampling (supplementary table 2; figure 6). In contrast, at the second time of sampling, rocky shores had significantly higher filtration rates than seawalls or pilings (supplementary table 2; figure 6).

In our survey, we found 9 non-indigenous species (NIS) in total (≈ 8% of the total number of species found). Results of the literature review found the proportion of total NIS was slightly lower, ≈ 4%, with 9 NIS, out of 209 taxa (table 1). Results from the literature review revealed that seawalls had almost twice the proportion of NIS (8%), compared to natural rocky reefs (4.9%) and pilings were intermediate (5.7% of species richness). Most NIS were colonial bryozoans (Watersipora subtorquata, Conopeum seurati, and Cryptosula pallasiana), but also included a polychaete (Hydroides elegans), the isopod Cirolana harfordi, the gastropod Zeacumantus subcarinatus, and the algae Colpomenia sinuosa and Ulva lactuca (table 1).

The secondary productivity of pilings, represented here by total biomass accumulated (as AFDW), was significantly greater than rocky shores or seawalls, but only after 12 months (LTR Chisq = 34.35; df = 2; p > 0.001; figure 7). This is probably due to the high recruitment of oysters to these habitats (figure S4).