Using benthic species abundance records (2003–2009) collected by volunteer divers through ‘Seasearch’, coastal biodiversity patterns around the United Kingdom and Ireland were investigated. The first aim was to assess the utility of volunteer data as a baseline for monitoring environmental change. Separation of the influences of within- and between-surveyor variation from broadscale (year, latitude, longitude) and finer scale (month, depth, local habitat) biogoegraphical factors was achieved using a multilevel mixed effects framework. A high degree of consistency within surveyors was evident and, by modelling between-surveyor variations efficiently, expected trends in species richness and taxonomic distinctness were recovered. Moving from patterns to processes, the second aim was to test the effects of sea surface temperature on prevalence, on a species by species basis. This approach allowed identification of species likely to display range shifts in direct response to future warming trends. Within limitations, volunteer data can provide a valuable contribution to understanding of biodiversity, climate change and aquatic ecosystem health.

Introduction

Climate change is likely to be a major factor influencing biodiversity and ecosystem services over the coming decades (Montoya and Raffaelli, 2010). Effects already documented include northward or westward expansion of species ranges in the northern hemisphere and other spatial dislocations (Walther, 2010), and such disturbances extend to aquatic ecosystems (Harley et al., 2006; Woodward et al., 2010).

While not evident for all species groups, marine biodiversity patterns across the United Kingdom and Ireland are well established from exhaustive, quantitative surveys (Kuklinski et al., 2006; Hiscock and Breckels, 2007). Based upon a synthesis of data for six major phyla, latitudinal patterns of species richness, and longitudinal patterns of taxonomic distinctness, are quite marked (Figure 9 in Hiscock and Breckels, 2007). Various hypotheses have been proposed to explain latitudinal biodiversity gradients (Ricklefs and Schluter, 1993; Hawkins et al., 2003a,b). Temperature is believed to be one controlling factor, and indeed changes in northern and southern limits of many marine species occurring in the Northeast Atlantic, in response to ocean warming, have been reported (Beaugrand et al., 2002; Perry et al., 2005; Chen et al., 2006; MacLeod, 2009).

Past quantitative surveys of overall marine biodiversity, or selected species groups, provide useful ‘baseline’ data for determining shifts in species ranges, in response to global warming and/or other environmental changes. However, repeat surveys at the same locations, using the same methodologies, are extremely resource-demanding, and time-consuming. Additionally, by the time the surveys are undertaken, and the data are analyzed, significant shifts in species richness may have already occurred. Given that biological responses to climate change are likely to be used increasingly to inform national policy and environmental legislation (Leonard et al., 2006), efforts to make monitoring more efficient and cost-effective could be welcome.

In this article, the contention is that survey data from diver observations in the United Kingdom and Ireland (Seasearch) represent a vast, largely untapped resource. Expanded datasets on marine epibenthos, albeit arguably of lower ‘quality’ and resolution, can usefully complement more quantitative data derived from more robust surveys which, due to their restricted taxonomic focus, might be less representative of wider biodiversity trends (Price, 1990; Price et al., 1997; Price and Harris, 2009). Biodiversity data from Reef Check (http://www.reefcheck.org/) and the earlier Reef Watch (Roberts and Ormond, 1987; Roberts et al., 1992) are similar examples of extensive datasets originating from diver surveys in tropical settings. In the United Kingdom, ‘Shore Thing’ (http://www.marlin.ac.uk/shore_thing/about.html) gathers comparable data on intertidal biodiversity, with the aim of monitoring the impact of rising sea temperatures. More generally, the value of rapid environmental assessment (REA), to augment more quantitative data, is becoming increasingly recognized (Price, 1990, 2004; Price et al., 1998). Similarly, terrestrial volunteer-based surveys, for example the UK Butterfly Monitoring Scheme (Pollard and Yates, 1993), have been used to successfully identify patterns of biodiversity and test specific hypotheses about the likely effects of climate change on ecological processes in a rigorously quantitative manner (Oliver et al., 2009).

The overall aim of this study was to assess the utility and limitations of Seasearch data for testing hypotheses on marine biodiversity, with special reference to shifts in species ranges induced by climate change. Specific objectives include: (1) partitioning the effects of broadscale (year, latitude, longitude) and finer scale variables (month, depth, local habitat), as drivers of species richness and taxonomic distinctness (Warwick and Clarke, 1998, 2001); (2) disaggregation of these biogeographical variables from variation within and between volunteer surveyors; and (3) investigation the effects of sea surface temperature on species prevalence around the UK and Ireland. Conclusions include some recommendations for improvements to Seasearch in terms of training and recording.

Methods

Seasearch project

The Seasearch project (http://www.seasearch.co.uk) was devised in the mid-1980s, and developed through the 1990s by the Marine Conservation Society, on behalf of the UK Joint Nature Conservation Committee, as part of their Marine Nature Conservation Review. From 2003, a standardized training and recording framework has been used with the aim of gathering information on seabed habitats and associated marine wildlife in the United Kingdom and Ireland, through the participation of volunteer recreational divers.

Three levels of training exist: ‘Observer’, ‘Surveyor’ and Specialist courses focusing on specific biodiversity issues, such as threats to species identified as at risk (for example, targeted national surveys of sea fans, crawfish and skates and rays). Whilst the first training course is designed to establish a basic level of marine species identification skill amongst participants, the Seasearch Surveyor course provides additional training and experience required to produce consistent, high quality records of species distributions, suitable for biodiversity assessment. Beyond this, a number of specialist courses equip recorders to conduct more exhaustive surveys, which may in future be used to validate and extend the findings of the present study.

Seasearch records are available from author CW and summary information may be viewed on Google Earth (http://www.seasearch.co.uk/achievements.htm).

Data collection

Species abundances were recorded on standardized Seasearch reporting forms (separate Observer and Surveyor forms are available at http://www.seasearch.co.uk/recording.htm). Reports were further validated by regional coordinators, before being entered into Marine Recorder (http://esdm.co.uk/MarineRecorder/index.html), the database used by the Joint Nature Conservation Committee, all UK government conservation agencies and most local record centres.

Here, only surveys at Surveyor level were used and, consistent with the database nomenclature, an individual, completed recording form (a single dive) is referred to as an ‘Event’. Within each Event, the recorder categorized species observations into a number of discrete habitats. Species lists from each habitat (nested within Events) are known in the database as ‘Samples’. Overall, the hierarchical survey design structure is: Surveyor (= diver) / Event (= dive)/Sample (= species list).

Six levels of categorical abundance (‘superabundant’, ‘abundant’, ‘common’, ‘frequent’, ‘occasional’, ‘rare’) were recorded for all species positively identified. In the current study, abundance factor levels were collapsed into a single ‘present’ score for each species in order to remove any inconsistency in quantification between recorders. A total of 839 Surveyors (volunteer divers) completed 2163 Events (dives), comprising 4449 Samples (habitats) with validated species lists, over the period 2003–2009.

Measures of biodiversity

Two biodiversity indices were quantified for each Seasearch Surveyor Sample:

Species richness was calculated as simply the number of species present in a given Sample. In order to treat richness as ‘count data’ in statistical modelling (integer values with a lower bound of zero), ‘richness – 1’ was used for all species richness estimates (since no reports included fewer than one species).

Taxonomic distinctness (Δ+) was calculated following Clarke and Warwick (1998). Within each Sample, pairwise distances, ωij were calculated between all pairs of species present: , where s is the number of species and, for the double summation, i and j range over these s species. Distances, based on taxonomic hierarchy were weighted: species i and j are of the same genus, ωij = 2; family (but not genus) ωij = 3; order ωij = 4; class ωij = 5; phylum ωij = 6. This metric is bounded by a minimum value of one (if only a single species was reported) and a maximum value of six. This range was converted to ‘proportion data’ (range [0, 1]) for statistical modelling, using the transformation ((Δ+−1)/5).

Statistical analysis

Patterns of species richness and taxonomic distinctness were quantified separately, using a Mixed Effects Modelling framework (Pinheiro and Bates, 2000). Initially, broadscale spatiotemporal explanatory variables were investigated, treating ‘latitude’ and ‘longitude’ as continuous fixed effects, modelling both these main terms and their statistical interaction with locally fitted smoothing splines, applied to coastal data points. Additionally, ‘year’ was modelled as a random effect. The distributions of each biodiversity index were modelled using appropriate link functions (Crawley, 2009): log in the case of richness and logit in for taxonomic distinctness; and the corresponding error distributions, allowing for observed overdispersion: ‘quasipoisson’ for species richness counts and ‘quasibinomial’ in the case of taxonomic distinctness.

Subsequently, finer scale sources of variation were quantified by modelling (normally distributed) residuals from the broadscale models. An appropriate sinusoidal function, , was fitted to the numeric calendar ‘month’ (nested within ‘years’) of each survey. In addition, the depth, in metres below chart datum (corrected for tide at recorded location/time of dive), of each species list was fitted as a fixed effect, and the statistical interaction between depth and month was quantified. Here, the variation between and within surveys was also partitioned, modelling Surveyor and Event as nested random effects. Since only one species list was collected for each habitat encountered on a given dive, variation between habitats could not be partitioned from residual error under the existing sampling design. In addition to likelihood ratio tests, confidence in parameter estimates was generated by Markov Chain Monte Carlo (MCMC) sampling with 2000 iterations.

The second aim was to test the effect of sea surface temperature (SST) on prevalence, on a species by species basis.

The HadISST1 data set is a blend of historical SST and modern SST observations from ships, buoys, drifters, etc. and sea ice observations, partly from historical ship- and air-borne and partly from satellite data. HadISST1 temperature data over the period 2003–2009 were obtained as monthly means, resolved into one degree latitude by one degree longitude cells, from the British Atmospheric Data Centre (http://badc.nerc.ac.uk/home/index.html). In line with this spatial resolution, species presence/absence data for all Samples were collated within each geographic cell. At least one species was present in a total of 53 cells containing coastline. Treating each species separately, these overdispersed binomial data (prevalence, given sampling effort in a given cell) formed the response variables for generalized linear models with mean annual SST as the explanatory variable, a logit link function and ‘quasibinomial’ error distributions.

In this analysis, species with fewer than five validated records in the database were excluded. This still allowed a species by species assessment for 638 species. Species with non-zero SST gradient parameters were identified at a significance probability threshold adjusted to reduce the expected false discovery rate to 1%, here simply referred to as p < 0.01 (Storey and Tibshirani, 2003).

All analyses were conducted using the R statistical programming platform, version 2.11.1 (http://www.r-project.org), with the additional packages: ‘mgcv’ and ‘lme4’.

Results

Biodiversity patterns

Overall contour maps of species richness and taxonomic distinctness, derived from fitted values, are shown in Figures 1a and b, respectively. Species richness was typically higher in the south and west of the region (UK and Ireland), whereas spatial variation in taxonomic distinctness was dominated by more local influences. Broadscale spatial variation was investigated further by separating out the effects of longitude and latitude on species richness (Figures 2a and b) and taxonomic distinctness (Figures 2c and d) separately.

Figure 1.

Contour maps showing fitted estimates of benthic biodiversity around the United Kingdom and Ireland. Open circles represent coastal dive sites, with contours fitted to these coastal data and extending over some areas of land for illustrative purposes only. Contour heights denote: (a) species richness and (b) taxonomic distinctness (Δ+).

Figure 1.

Contour maps showing fitted estimates of benthic biodiversity around the United Kingdom and Ireland. Open circles represent coastal dive sites, with contours fitted to these coastal data and extending over some areas of land for illustrative purposes only. Contour heights denote: (a) species richness and (b) taxonomic distinctness (Δ+).

Figure 2.

Fitted estimates (95% c.i.) of benthic biodiversity around the United Kingdom and Ireland regressed on spatial coordinates. (a) Species richness on latitude; (b) species richness on longitude; (c) taxonomic distinctness (Δ+) on latitude; (d) taxonomic distinctness on longitude.

Figure 2.

Fitted estimates (95% c.i.) of benthic biodiversity around the United Kingdom and Ireland regressed on spatial coordinates. (a) Species richness on latitude; (b) species richness on longitude; (c) taxonomic distinctness (Δ+) on latitude; (d) taxonomic distinctness on longitude.

There was a small but significant reduction in observed species richness with increasing depth (likelihood ratio = 237, p < 0.001). Significant seasonal variation (likelihood ratio = 20.7, p < 0.001) was captured through the relative contributions of cosine and sine functions. The interaction between depth and seasonal fluctuation did not have a significant effect on species richness (likelihood ratio = 0.894, p = 0.34). Explanatory variable parameter estimates, with confidence distributions derived from MCMC simulation, are shown in Figures 3a–d.

Figure 3.

Model parameter estimate distributions resulting from MCMC simulation (2000 iterations). (a–d) Species richness; (e–h) taxonomic distinctness.

Figure 3.

Model parameter estimate distributions resulting from MCMC simulation (2000 iterations). (a–d) Species richness; (e–h) taxonomic distinctness.

Consistent with the finding that taxonomic distinctness was largely dependent on local influences, there was a significant interaction between depth and season (likelihood ratio = 13.0, p = 0.0015). Explanatory variable parameter estimates, with confidence distributions derived from MCMC simulation, are shown in Figures 3e–h.

Variance components for nested random effects were estimated: Surveyor (= diver), σ = 0.187; Event (= dive), σ = 0.046; Sample (= species list), σ = 0.762. Whilst Surveyor level variation explained a significant amount of observed deviance (likelihood ratio = 32.3, p < 0.001), removal of Event level variation from the model did not result in a significant reduction in model fit (likelihood ratio = 0.036, p = 0.85).

Sea surface temperature

Having quantified overall biodiversity metrics, the effects of sea surface temperature (SST) on prevalence of 638 species, were investigated on a species by species basis. Applying a significance probability threshold of p < 0.01, a total of 83 species were found to be more prevalent at higher annual mean SST (here designated ‘warm water species’), 111 species were more prevalent at lower SST (‘cool water species’), and SST did not have a significant effect on prevalence of 444 species. Species names in the temperature-dependent categories are listed in Table 1.

Table 1.

All species listed showed a significant (p < 0.01) trend associated with sea surface temperature (SST) around the United Kingdom and Ireland. Here ‘warm water species’ are those found at greater prevalence with increasing SST; ‘cool water species’ are those showing decreased prevalence with increasing SST. No evidence of trends associated with SST were found in an additional 444 species recorded by Seasearch.

PhylumGenus species
Warm water species 
 Annelida Bispira volutacornis Hydroides norvegica Sabella discifera 
 Arthropoda Boscia anglica Maja squinado Solidobalanus fallax 
 Bryozoa Aetea anguina Amathia lendigera Cellaria salicornioides 
 Cellepora pumicosa Crisia aculeata Escharoides coccinea 
 Pentapora foliacea   
 Chlorophyta Codium tomentosum  
 Chordata Chelon labrosus Crenilabrus bailloni Crenilabrus melops 
 Ctenolabrus rupestris Distomus variolosus Lepadogaster lepadogaster 
 Parablennius gattorugine Parablennius ruber Phallusia mammillata 
 Pycnoclavella aurilucens Raja montagui Raja undulata 
 Spondyliosoma cantharus Stolonica socialis Styela clava 
 Tripterygion delaisi Trisopterus luscus  
 Cnidaria Actinothoe sphyrodeta Aglaophenia parvula Anemonia viridis 
 Aulactinia verrucosa Balanophyllia regia Caryophyllia inornata 
 Cereus pedunculatus Eunicella verrucosa Hoplangia durotrix 
 Leptopsammia pruvoti Sarcodictyon roseum  
 Mollusca Crepidula fornicata Haliotis tuberculata Hinia reticulata 
 Loligo vulgaris Ocenebra erinacea Ostrea edulis 
 Sepia officinalis Tritonia nilsodhneri  
 Ochrophyta Dictyopteris membranacea Laminaria ochroleuca Taonia atomaria 
 Zanardinia prototypus  
 Porifera Adreus fascicularis Ciocalypta penicillus Crella rosea 
 Dercitus bucklandi Desmacidon fruticosum Dysidea fragilis 
 Esperiopsis fucorum Halichondria bowerbanki Haliclona fistulosa 
 Hemimycale columella Hymeniacidon perleve Iophon hyndmani 
 Myxilla rosacea Polymastia mamillaris Rhaphidostyla kitchingi 
 Stylostichon plumosum  
 Rhodophyta Aglaothamnion byssoides Bonnemaisonia hamifera Calliblepharis ciliata 
 Chondria dasyphylla Gastroclonium ovatum Griffithsia corallinoides 
 Gymnogongrus crenulatus Halopithys incurvus Halurus equisetifolius 
 Pikea californica Rhodymenia ardissonei Rhodymenia holmesii 
 Spyridia filamentosa  
Cool water species 
 Annelida Chone infundibuliformis Eupolymnia nebulosa Pomatoceros triqueter 
 Spirorbis spirorbis   
 Arthropoda Atelecyclus rotundatus Cancer pagurus Carcinus maenas 
 Galathea intermedia Galathea nexa Galathea strigosa 
 Inachus dorsettensis Liocarcinus depurator Liocarcinus holsatus 
 Liocarcinus marmoreus Macropodia rostrata Munida rugosa 
 Necora puber Nephrops norvegicus Pagurus bernhardus 
 Pagurus prideaux Pandalus montagui  
 Bryozoa Porella compressa Securiflustra securifrons  
 Chordata Ascidiella aspersa Chirolophis ascanii Ciona intestinalis 
 Clavelina lepadiformis Corella parallelogramma Entelurus aequoreus 
 Eutrigla gurnardus Gadus morhua Lesueurigobius friesii 
 Molva molva Myoxocephalus scorpius Nerophis lumbriciformis 
 Pholis gunnellus Phrynorhombus norvegicus Platichthys flesus 
 Pollachius virens Pomatoschistus microps  
 Cnidaria Abietinaria abietina Adamsia carciniopados Aurelia aurita 
 Cerianthus lloydii Cyanea capillata Cyanea lamarckii 
 Funiculina quadrangularis Metridium senile Obelia dichotoma 
 Pennatula phosphorea Protanthea simplex Rhizocaulus verticillatus 
 Sagartiogeton laceratus Swiftia pallida Thuiaria thuja 
 Urticina eques Virgularia mirabilis  
 Ctenophora Pleurobrachia pileus   
 Echinodermata Amphiura filiformis Antedon bifida Antedon petasus 
 Asterias rubens Astropecten irregularis Crossaster papposus 
 Echinus esculentus Henricia sanguinolenta Leptasterias muelleri 
 Leptometra celtica Ophiocomina nigra Ophiopholis aculeata 
 Ophiothrix fragilis Ophiura albida Ophiura ophiura 
 Porania pulvillus Psammechinus miliaris Solaster endeca 
 Mollusca Arctica islandica Chlamys varia Circomphalus casina 
 Dendronotus frondosus Eledone cirrhosa Ensis siliqua 
 Eubranchus farrani Gibbula cineraria Helcion pellucidum 
 Lacuna vincta Modiolus modiolus Onchidoris bilamellata 
 Pecten maximus Pododesmus patelliformis Polycera quadrilineata 
 Tapes rhomboides Tonicella marmorea Tritonia hombergii 
 Tritonia plebeia   
 Nemertea Lineus longissimus   
 Ochrophyta Chorda filum Desmarestia aculeata Desmarestia viridis 
 Laminaria hyperborea Laminaria saccharina  
 Porifera Haliclona urceolus Phakellia ventilabrum  
 Rhodophyta Bonnemaisonia asparagoides Lomentaria clavellosa Nitophyllum punctatum 
 Odonthalia dentata Phycodrys rubens Polysiphonia fucoides 
 Pterosiphonia parasitica Ptilota gunneri  
PhylumGenus species
Warm water species 
 Annelida Bispira volutacornis Hydroides norvegica Sabella discifera 
 Arthropoda Boscia anglica Maja squinado Solidobalanus fallax 
 Bryozoa Aetea anguina Amathia lendigera Cellaria salicornioides 
 Cellepora pumicosa Crisia aculeata Escharoides coccinea 
 Pentapora foliacea   
 Chlorophyta Codium tomentosum  
 Chordata Chelon labrosus Crenilabrus bailloni Crenilabrus melops 
 Ctenolabrus rupestris Distomus variolosus Lepadogaster lepadogaster 
 Parablennius gattorugine Parablennius ruber Phallusia mammillata 
 Pycnoclavella aurilucens Raja montagui Raja undulata 
 Spondyliosoma cantharus Stolonica socialis Styela clava 
 Tripterygion delaisi Trisopterus luscus  
 Cnidaria Actinothoe sphyrodeta Aglaophenia parvula Anemonia viridis 
 Aulactinia verrucosa Balanophyllia regia Caryophyllia inornata 
 Cereus pedunculatus Eunicella verrucosa Hoplangia durotrix 
 Leptopsammia pruvoti Sarcodictyon roseum  
 Mollusca Crepidula fornicata Haliotis tuberculata Hinia reticulata 
 Loligo vulgaris Ocenebra erinacea Ostrea edulis 
 Sepia officinalis Tritonia nilsodhneri  
 Ochrophyta Dictyopteris membranacea Laminaria ochroleuca Taonia atomaria 
 Zanardinia prototypus  
 Porifera Adreus fascicularis Ciocalypta penicillus Crella rosea 
 Dercitus bucklandi Desmacidon fruticosum Dysidea fragilis 
 Esperiopsis fucorum Halichondria bowerbanki Haliclona fistulosa 
 Hemimycale columella Hymeniacidon perleve Iophon hyndmani 
 Myxilla rosacea Polymastia mamillaris Rhaphidostyla kitchingi 
 Stylostichon plumosum  
 Rhodophyta Aglaothamnion byssoides Bonnemaisonia hamifera Calliblepharis ciliata 
 Chondria dasyphylla Gastroclonium ovatum Griffithsia corallinoides 
 Gymnogongrus crenulatus Halopithys incurvus Halurus equisetifolius 
 Pikea californica Rhodymenia ardissonei Rhodymenia holmesii 
 Spyridia filamentosa  
Cool water species 
 Annelida Chone infundibuliformis Eupolymnia nebulosa Pomatoceros triqueter 
 Spirorbis spirorbis   
 Arthropoda Atelecyclus rotundatus Cancer pagurus Carcinus maenas 
 Galathea intermedia Galathea nexa Galathea strigosa 
 Inachus dorsettensis Liocarcinus depurator Liocarcinus holsatus 
 Liocarcinus marmoreus Macropodia rostrata Munida rugosa 
 Necora puber Nephrops norvegicus Pagurus bernhardus 
 Pagurus prideaux Pandalus montagui  
 Bryozoa Porella compressa Securiflustra securifrons  
 Chordata Ascidiella aspersa Chirolophis ascanii Ciona intestinalis 
 Clavelina lepadiformis Corella parallelogramma Entelurus aequoreus 
 Eutrigla gurnardus Gadus morhua Lesueurigobius friesii 
 Molva molva Myoxocephalus scorpius Nerophis lumbriciformis 
 Pholis gunnellus Phrynorhombus norvegicus Platichthys flesus 
 Pollachius virens Pomatoschistus microps  
 Cnidaria Abietinaria abietina Adamsia carciniopados Aurelia aurita 
 Cerianthus lloydii Cyanea capillata Cyanea lamarckii 
 Funiculina quadrangularis Metridium senile Obelia dichotoma 
 Pennatula phosphorea Protanthea simplex Rhizocaulus verticillatus 
 Sagartiogeton laceratus Swiftia pallida Thuiaria thuja 
 Urticina eques Virgularia mirabilis  
 Ctenophora Pleurobrachia pileus   
 Echinodermata Amphiura filiformis Antedon bifida Antedon petasus 
 Asterias rubens Astropecten irregularis Crossaster papposus 
 Echinus esculentus Henricia sanguinolenta Leptasterias muelleri 
 Leptometra celtica Ophiocomina nigra Ophiopholis aculeata 
 Ophiothrix fragilis Ophiura albida Ophiura ophiura 
 Porania pulvillus Psammechinus miliaris Solaster endeca 
 Mollusca Arctica islandica Chlamys varia Circomphalus casina 
 Dendronotus frondosus Eledone cirrhosa Ensis siliqua 
 Eubranchus farrani Gibbula cineraria Helcion pellucidum 
 Lacuna vincta Modiolus modiolus Onchidoris bilamellata 
 Pecten maximus Pododesmus patelliformis Polycera quadrilineata 
 Tapes rhomboides Tonicella marmorea Tritonia hombergii 
 Tritonia plebeia   
 Nemertea Lineus longissimus   
 Ochrophyta Chorda filum Desmarestia aculeata Desmarestia viridis 
 Laminaria hyperborea Laminaria saccharina  
 Porifera Haliclona urceolus Phakellia ventilabrum  
 Rhodophyta Bonnemaisonia asparagoides Lomentaria clavellosa Nitophyllum punctatum 
 Odonthalia dentata Phycodrys rubens Polysiphonia fucoides 
 Pterosiphonia parasitica Ptilota gunneri  

Discussion

The key findings of this study were that through appropriate statistical modelling of a national dataset, largely comprising trained volunteer survey information, discernible trends in benthic biodiversity over a range of spatial and temporal scales were evident. A high degree of consistency from individual surveyors was found, and variation between surveyors did not overshadow systematic trends in the data. Furthermore, a large number of benthic species displayed a clear relationship between prevalence and sea temperature. This broadscale finding is novel, as well as providing a baseline by which other, more focused/quantitative studies may be set in context.

Detection of long-term trends in biodiversity or climate is inherently dependent on observations over many years. In the current, relatively short-term study, such overall trends were neither expected nor observed. However, stochastic fluctuations from year to year did form an important part of the model. In future, this type of dataset should allow detection of long-term trends in biodiversity associated with climate change.

Species richness was found to be higher in the west of the region, compared to the east. Based on geographic context, longitudinal patterns around the UK and Ireland are likely to be driven by factors associated with proximity to continental Europe vs. open ocean. These include spatial distribution of habitat types (sandy/muddy substrate is relatively more common in the east, whereas rocky reefs prevail in the west of the region), exposure to prevailing oceanic weather patterns (typically from the southwest), tidal range (greater in the south and west), pollution (likely greater close to the continent) and possibly temperature range (likely to be more stable in regions open to the deeper waters of the open ocean, but relationship to inshore benthic habitat may not be straightforward). Overall species richness decreased with increasing latitude over the study area. This is in line with previous studies on restricted taxonomic groups (synthesis in Hiscock and Breckels, 2007), where latitudinal gradients are hypothesized to be largely underpinned by climate (Currie, 1991; H-Acevedo and Currie, 2003; Hawkins, 2003a,b).

Seasonal fluctuations in biodiversity were modelled using additive cosine and sine functions. The relative magnitude of the respective parameter estimates for these components described the orientation of peaks and troughs in biodiversity, with respect to calendar month. The approximate 2:1 ratio of these parameters indicated species richness and taxonomic distinctness reach a minimum in February/March, and maximum in July/August. This finding is consistent with the hypothesis that seasonal benthic biodiversity around the UK and Ireland is associated with sea temperature, which lags calendar date by approximately two months. There was an observed trend of a slight decrease in species richness with depth, and this effect was independent of season (no statistical interaction). It is conceivable that this was due to habitat depth distribution, differential identification of species associated with depth, or that it is an artefact of divers’ recording ability at increased depth. However, no such trend was observed for taxonomic distinctness. Rather, an interaction between depth and season was found, which would be hard to explain in terms of recorder bias. It would be necessary to conduct further, more detailed studies, likely of specific taxonomic groups and habitats, in order to explore the biological basis of our findings.

Overall, the successful exposition of expected spatiotemporal trends in biodiversity suggests that volunteer data, gathered through the Seasearch programme, is of sufficient quality to provide useful information on epibenthic species distributions. One caveat is that sampling effort is confounded with location (shown by uneven coverage of dive sites in Figure 1). In this analysis, by keeping recorded species lists as separate, point estimates of species richness and taxonomic distinctness, each data point represents the same amount of sampling effort. This means that reduced sampling in a particular geographic area (for example, the east coast of England) may increase residual error but does not create a systematic spatial bias. In other analyses, where surveys are aggregated into bins (for example, the study of the association between species distribution and sea temperature, grouped by 1° × 1° geographic cell), sampling effort needed to be explicitly controlled for in order to avoid bias—the more times an area is surveyed, the more species are likely to be identified.

In this study, it was not possible to resolve differences in biodiversity associated with local habitat from unexplained (residual) sources of variation. It would certainly be expected that different species were present according to habitat, but overall biodiversity is harder to predict. However, this was an area where diver bias, in terms of choice of dive site, was also likely to influence the current dataset. Some Seasearch events target specific ‘under-represented’ habitats, such as apparently barren sediment, and the relatively high residual variance (compared to Surveyor and Event variance components) that was found is suggestive that there were marked differences in biodiversity between habitats encountered on a single dive. However, the level of effort and expertise needed to quantify such patterns is probably better suited to targeted studies by specifically trained researchers.

By modelling the hierarchical structure of the sampling design employed by Seasearch, it was possible to address the important issue of reliability of data gathered by trained volunteers. Whilst some sources of bias were identified (choice of dive site and season, as well as recording skills associated with depth), overall, a very high degree of consistency by individual Surveyors was found. Furthermore, by explicitly incorporating variation between Surveyors, recovered trends were consistent with those found in ‘expert’ studies could be recovered. The absolute numbers of species recorded in Seasearch reports was considerably lower (up to two orders of magnitude) than those reported by Hiscock and Breckels (2007), based on compilation of numerous expert studies. However, importantly, it was found that the overall pattern of species richness was very similar. Where the objective is identifying broadscale trends in species richness, Seasearch data provides a valuable source of information at considerably less cost than other studies.

In the second part of this study, the effects of sea temperature on benthic species distributions around the UK and Ireland were investigated. A substantial number of species showing a trend in prevalence with sea surface temperature was identified (Table 1). This analysis provides a useful indication of species that are likely to be affected by climate change and allows prioritisation of future monitoring efforts. Here, SST was averaged over whole years as a predictor of species prevalence. In preliminary analysis, average SST was found to be well-correlated with both maximum and minimum monthly means, but not with within-year variation (coefficient of variation of monthly means). Since global warming is predicted to result in greater variation in temperatures (Dai et al., 2001), as well as higher means, exploring more subtle relationships between temperature and species prevalence would be an important future research direction. This aim was not pursued here; rather extensions of the current work should be directed by specific mechanistic hypotheses to ensure focussed research.

A key challenge is to supplement studies of pattern with testable hypotheses on process, in order to develop mechanistic understanding and predictive frameworks for biodiversity. Many theories have been put forward to explain observed patterns of species diversity gradients, with processes underpinned by energy flow and history (Hawkins et al., 2003a,b; Cardillo et al., 2005). The consistent methodology and reporting will make Seasearch data suitable for not only more detailed pattern-based investigation, such as identifying the appropriate spatial scale to quantify β-diversity (through Jaccard or taxonomic similarity; Izsak and Price, 2001) for specific management motivated aims, but also mechanistic approaches like analysis of density feedback processes (Bjørnstad et al., 2001; Post, 2005).

Discussion concludes with some general guidance in using Seasearch data for quantifying epibenthic biodiversity patterns around the United Kingdom and Ireland, as well as making recommendations for improvement to training and recording. In a large study such as this, many sources of variance in data cannot be controlled. However, such variation can be effectively managed through careful selection of hypotheses and appropriate statistical analysis. Principally, these include variation in: (1) the experience and behaviour of surveyors; (2) natural factors biasing observations, such as water turbidity; (3) sampling effort across the range of survey sites (Figure 1); and (4) natural variation in habitat type.

Conclusions

Here, it has been shown that with the current level of training available through Seasearch, surveyor standards are sufficient to recover expected spatiotemporal trends, consistent with more focused studies, despite the variance due to (1) and (2), above. However, patterns of species richness should be considered as a relative, rather than absolute measure, as expert studies report substantially greater numbers of species (Hiscock and Breckels, 2007). Further, surveyors record species sightings on a semi-quantitative, categorical scale, ranging from ‘rare’ to ‘superabundant’. In this study, this additional information was redundant due to lack of standardization in these measures, both between surveyors and between species – one surveyor might consider 20 pollack (Pollachius pollachius) sighted on a dive as ‘frequent’, where another surveyor would score this as ‘abundant’, and most surveyors would likely record 20 basking sharks (Cetorhinus maximus) as ‘superabundant’. Therefore, the standardization of an appropriate scale should be considered as a priority for future Seasearch development.

As well as variation in surveyor expertise, substantial variation in sampling effort across the United Kingdom and Ireland was evident. Measures such as taxonomic distinctness are relatively insensitive to sampling effort (Clarke and Warwick, 1998; Warwick and Clarke, 1998, 2001) but, where cumulative measures of species are estimated (for example, within geographic cells in our study), it is important to explicitly incorporate sampling effort into any analysis (as binomial regression does here).

Currently, variance due to the natural range of benthic habitats encountered by surveyors was also dealt with as ‘error’. Whilst this has not obscured the trends in biodiversity evident in this study, the role of habitat type in explaining such patterns is another area where Seasearch could be improved. An attempt has been made to match volunteer descriptions of habitats with the formal concept of biotopes (Wood, 2007). However, this proved to largely unsuccessful, both due to the qualitative nature of habitat descriptions within Seasearch and lack of standardization between those tasked with linking habitats with biotopes. Although some further categorization of habitat features was subsequently incorporated into Seasearch Surveyor forms, habitat recording remains largely descriptive and diagrammatic. Although a difficult task, the formalization of habitat recording should also be considered as a key priority for Seasearch.

If used appropriately, Seasearch records may prove to be an invaluable, low-cost yet robust dataset. It has the potential to fulfill several ecological and conservation objectives, ranging from use in analysis of national and regional biodiversity patterns, for example in the context of global warming, to informing the identification of new marine protected areas based on benthic species composition.

Acknowledgements

We are grateful to Rohan Holt (Countryside Council for Wales) and James Perrins (exeGesIS) for advice on extracting taxonomic information from Marine Recorder. We would also like to thank the many Seasearch volunteers and regional coordinators who provided the data for this study. Finally, we would like to acknowledge some important preliminary work on patterns evident in Seasearch data, which was undertaken by undergraduate students Oliver Cooper, Rachel George and Adam East, as part of final year research projects at Warwick. Whilst not directly contributing to the findings of this manuscript, their hard work provided much of the motivation for this study.

References

Beaugrand, G., Reid, P. C., Ibañez, F., Lindley, J. A. and Edwards, M.
2002
.
Reorganization of north Atlantic marine copepod biodiversity and climate
.
Science
,
296
:
1692
1694
.
Bjørnstad, O. N., Sait, S. M., Stenseth, N. C., Thompson, D. J. and Begon, M.
2001
.
The impact of specialized enemies on the dimensionality of host dynamics
.
Nature
,
409
:
1001
1006
.
Cardillo, M., Orme, D. and Owens, I. P.F.
2005
.
Testing for latitudinal bias in diversification rates: An example using new world birds
.
Ecology
,
86
:
2278
2287
.
Chen, C. S., Pierce, G. J., Wang, J., Robin, J. P., Poulard, J. C., Pereira, J., Zuur, A. F., Boyle, P. R., Bailey, N., Beare, D. J., Jereb, P., Ragonese, S., Mannini, A. and Orsi-Relini, L.
2006
.
The apparent disappearance of Loligo forbesi from the south of its range in the 1990s: Trends in Loligo spp. abundance in the northeast Atlantic and possible environmental influences
.
Fisheries Res.
,
78
:
44
54
.
Clarke, K. R. and Warwick, R. M.
1998
.
A taxonomic distinctness index and its statistical properties
.
J. Appl. Ecol.
,
35
:
523
531
.
Crawley, M. J.
2002
.
Statistical computing: An introduction to data analysis using S-plus
,
John Wiley
:
New York
.
Currie, D. J.
1991
.
Energy and large-scale patterns of animal and plant species richness
.
Am. Nat.
,
137
:
27
49
.
Dai, A., Wigley, T. M.L., Boville, B. A., Kiehl, J. T. and Buja, L. E.
2001
.
Climates of the 20th and 21st centuries
.
J. Climatol.
,
14
:
485
519
.
H-Acevedo, D. and Currie, D. J.
2003
.
Does climate determine broad-scale patterns of species richness? A test of the causal link by natural experiment
.
Global Ecol. Biogeogr.
,
12
:
461
473
.
Harley, C. D.G., Randall Hughes, A., Hultgren, K. M., Miner, B. G., Sorte, C. J.B., Thornber, C. S., Rodriguez, L. F., Tomanek, L. and Williams, S. L.
2006
.
The impacts of climate change in coastal marine systems
.
Ecol. Letts.
,
9
:
228
241
.
Hawkins, B. A., Porter, E. E. and Diniz-Filho, J. A.F.
2003a
.
Productivity and history as predictors of the latitudinal diversity gradient for terrestrial birds
.
Ecology
,
84
:
1608
1623
.
Hawkins, B. A., Field, R., Cornell, H. V., Currie, D. J., Guegan, J.-F., Kaufman, D. M., Kerr, J. T., Mittelbach, G. G., Oberdorff, T., O’Brien, E. M., Porter, E. E. and Turner, J. R.G.
2003b
.
Energy, water, and broad-scale geographic patterns of species richness
.
Ecology
,
84
:
3105
3117
.
Hiscock, K. and Breckels, M.
2007
.
Marine biodiversity hotspots in the UK: their identification and protection
Izsak, C. and Price, A. R.G.
2001
.
Measuring ß-diversity using a taxonomic similarity index, and its relation to spatial scale
.
Mar. Ecol. Progr. Ser.
,
215
:
69
77
.
Kuklinski, P., Barnes, D. K.A. and Taylor, P. D.
2006
.
Latitudinal patterns of diversity and abundance in North Atlantic intertidal boulder-fields
.
Mar. Biol.
,
149
:
1577
1583
.
Leonard, D. R.P., Clarke, K. R., Somerfield, P. J. and Warwick, R. M.
2006
.
The application of an indicator based on taxonomic distinctness for UK marine biodiversity assessments
.
J. Env. Manag.
,
78
:
52
62
.
MacLeod, C. D.
2009
.
Global climate change, range changes and potential implications for the conservation of marine cetaceans: a review and synthesis
.
Endangered Species Res.
,
7
:
125
136
.
Montoya, J. M. and Raffaelli, D.
2010
.
Climate change, biotic interactions and ecosystem services
.
Phil. Trans. Roy. Soc. B.
,
365
:
2013
2018
.
Oliver, T., Hill, J. K., Thomas, C. D., Brereton, T. and Roy, D. B.
2009
.
Changes in habitat specificity of species at their climatic range boundaries
.
Ecol. Letts.
,
12
:
1091
1102
.
Perry, A. L., Low, P. J., Ellis, J. R. and Reynolds, J. D.
2005
.
Climate change and distribution shifts in marine fishes
.
Science
,
308
:
1912
1915
.
Pinheiro, J. C. and Bates, D. M.
2000
.
Mixed-effects models in S and S-Plus
,
Springer-Verlag Inc., New York
.
Pollard, E. and Yates, T. J.
1993
.
Monitoring butterflies for ecology and conservation
,
Chapman and Hall, London
.
Post, E.
2005
.
Large-scale spatial gradients in herbivore population dynamics
.
Ecology
,
86
:
2320
2328
.
Price, A. R.G.
1990
.
Rapid assessment of coastal zone management requirements: case study in the Arabian Gulf
.
Ocean & Shoreline Manag.
,
13
:
1
19
.
Price, A. R.G.
2004
. “
Rapid Coastal Environmental Assessment
”. In
Standard Survey Methods for the Red Sea and Gulf of Aden
1
26
.
PERSGA/GEF. PERSGA Technical Series No. 10. PERSGA, Jeddah
Price, A. R.G., Jobbins, G., Dawson Shepherd, A. R. and Ormond, R. F.G.
1998
.
An integrated environmental assessment of the Red Sea coast of Saudi Arabia
.
Environ. Cons.
,
25
:
65
76
.
Ricklefs, R. E. and Schluter, D.
1993
.
Species diversity in ecological communities
,
University of Chicago Press, Chicago
.
Roberts, C. M. and Ormond, R. F.G.
1987
.
Habitat complexity and coral reef fish diversity and abundance on Red Sea fringing reefs
.
Mar. Ecol. Progr. Ser.
,
41
:
1
8
.
Roberts, C. M., Dawson Shepherd, A. R. and Ormond, R. F. G.
1992
.
Large-scale variation in assemblage structure of Red Sea butterflyfishes and angelfishes
.
J. Biogeogr.
,
19
:
239
250
.
Storey, J. D. and Tibshirani, R.
2003
.
Statistical significance for genomewide studies
.
PNAS
,
100
:
9440
9445
.
Walther, G.-R.
2010
.
Community and ecosystem responses to recent climate change
.
Phil. Trans. R. Soc. B.
,
365
:
219
224
.
Warwick, R. M. and Clarke, K. R.
1998
.
Taxonomic distinctness and environmental assessment
.
J. Appl. Ecol.
,
35
:
532
543
.
Warwick, R. M. and Clarke, K. R.
2001
.
Practical measures of marine biodiversity based on relatedness of species
.
Oceanogr. Mar. Biol.
,
39
:
207
231
.
Wood, C.
2007
.
Assigning biotopes to Seasearch data. JNCC report no. 418. Peterborough: Joint Nature Conservation Committee
Woodward, G., Perkins, D. M. and Brown, L. E.
2010
.
Climate change and freshwater ecosystems: impacts across multiple levels of organization
.
Phil. Trans. Roy. Soc, Series B
,
365
:
2093
2106
.