Characterising and predicting benthic biodiversity for conservation planning in deepwater environments
Understanding patterns of biodiversity in deep sea systems is increasingly important because human activities are extending further into these areas. However, obtaining data is difficult, limiting the ability of science to inform management decisions. We have used three different methods of quantifying biodiversity to describe patterns of biodiversity in an area that includes two marine reserves in deep water off southern Australia. We used biological data collected during a recent survey, combined with extensive physical data to model, predict and map three different attributes of biodiversity: distributions of common species, beta diversity and rank abundance distributions (RAD). The distribution of each of eight common species was unique, although all the species respond to a depth-correlated physical gradient. Changes in composition (beta diversity) were large, even between sites with very similar environmental conditions. Composition at any one site was highly uncertain, and the suite of species changed dramatically both across and down slope. In contrast, the distributions of the RAD components of biodiversity (community abundance, richness, and evenness) were relatively smooth across the study area, suggesting that assemblage structure (i.e. the distribution of abundances of species) is limited, irrespective of species composition. Seamounts had similar biodiversity based on metrics of species presence, beta diversity, total abundance, richness and evenness to the adjacent continental slope in the same depth ranges. These analyses suggest that conservation objectives need to clearly identify which aspects of biodiversity are valued, and employ an appropriate suite of methods to address these aspects, to ensure that conservation goals are met.
Biophysical patterns in benthic assemblage composition across contrasting continental margins off New Zealand
To examine whether benthic assemblages are more diverse in a region of high topographic and oceanographic complexity by comparing benthic invertebrate assemblages across continental margins with contrasting environments. Location
Challenger Plateau and Chatham Rise, to the west and east of New Zealand, respectively. Methods Benthic faunal data were sourced from extensive seabed surveys in 2007, when both margins were sampled with an epibenthic sled and a towed video system. Three methods were used to investigate benthic assemblages in relation to environmental variables: one based on individual species distribution models (SDMs) using boosted regression trees analysis (BRT), and two community-based modelling methods using generalized dissimilarity modelling (GDM) and gradient forest analysis (GF), respectively. Each method was used to model and predict the turnover in assemblages with respect to environment – the ‘biophysical patterns’ – across the study region. Results
Across Chatham Rise, a complex oceanographic environment arising from steep gradients in productivity and temperature at the Subtropical Front produced a high diversity of assemblages associated with the sub-Antarctic water mass, the Subtropical Front, steep-sloping regions and fast tidal currents. In contrast, Challenger Plateau lies entirely beneath a single (subtropical) water mass, and assemblage diversity was lower, with a distinctive assemblage on the plateau itself and a deep-water assemblage similar to the northern deep-water assemblage of Chatham Rise. Across both regions, assemblage turnover was fastest in cold waters, at shallow depths and in deep mixed layers. Main conclusions
Benthic assemblages were more varied on Chatham Rise than on Challenger Plateau, supporting the hypothesis that environmentally heterogeneous margins have higher assemblage diversity. Differing assemblages on the northern and southern flanks of Chatham Rise suggest a biogeographical boundary for benthic taxa across the Subtropical Front. These results demonstrate that oceanographically and topographically complex margins have a diverse assemblage structure that should be considered in planning for the sustainable management of diversity.
Multi-scale marine biodiversity patterns inferred efficiently from habitat image processing
Cost-effective proxies of biodiversity and species abundance, applicable across
a range of spatial scales, are needed for setting conservation priorities and planning action. We
outline a rapid, efficient, and low-cost measure of spectral signal from digital habitat images
that, being an effective proxy for habitat complexity, correlates with species diversity and
requires little image processing or interpretation. We validated this method for coral reefs of
the Great Barrier Reef (GBR), Australia, across a range of spatial scales (1 m to 10 km), using
digital photographs of benthic communities at the transect scale and high-resolution Landsat
satellite images at the reef scale. We calculated an index of image-derived spatial
heterogeneity, the mean information gain (MIG), for each scale and related it to univariate
(species richness and total abundance summed across species) and multivariate (species
abundance matrix) measures of fish community structure, using two techniques that account
for the hierarchical structure of the data: hierarchical (mixed-effect) linear models and
distance-based partial redundancy analysis. Over the length and breadth of the GBR, MIG
alone explained up to 29% of deviance in fish species richness, 33% in total fish abundance,
and 25% in fish community structure at multiple scales, thus demonstrating the possibility of
easily and rapidly exploiting spatial information contained in digital images to complement
existing methods for inferring diversity and abundance patterns among fish communities.
Thus, the spectral signal of unprocessed remotely sensed images provides an efficient and lowcost
way to optimize the design of surveys used in conservation planning. In data-sparse
situations, this simple approach also offers a viable method for rapid assessment of potential
local biodiversity, particularly where there is little local capacity in terms of skills or resources
for mounting in-depth biodiversity surveys.
Comparing large-scale bioregions and fine-scale community-level biodiversity predictions from subtidal rocky reefs across south-eastern Australia
In the absence of knowledge of the large-scale structure and distribution of marine biota, bioregionalisations, that is, spatial classifications of data on a range of environmental and⁄ or biological attributes, are often viewed as one of the most appropriate frameworks within which to develop networks of marine protected areas (MPAs). However, despite their potential usefulness, few studies have assessed whether bioregionalisations can be used for management of species other than those it was derived from or whether bioregionalisations capture fully fine-scale community-level biodiversity patterns.
We investigated the large-scale structure and distribution of demersal fishes and macroinvertebrates in south-eastern Australia, using rank abundance distributions (RADs). We used a recently developed community modelling method that allows their multivariate distribution to vary according to environmental gradients, assessing the congruency of mapped biogeographic patterns between the different taxa, and in the light of the Interim Marine and Coastal Regionalisation for Australia (IMCRA).
A clear pattern in our analysis based on RADs showed a large difference in assemblage structure (i.e. in abundance, richness and evenness) between South Australia, where assemblages were generally more species rich and even, and Victoria and Tasmania, where assemblages were generally more species poor and uneven. The strong longitudinal pattern in species richness and evenness was generally congruent for both demersal fishes and macroinvertebrates and related to regional differences in oceanography.
We found that the regions of highest species richness were found in the ‘core’ bioregions rather than ‘transition’ bioregions as defined in the IMCRA and for both taxa. Moreover, we found that not all assemblage structures were equally alike and that South Australia had the greatest range of unique assemblage structures.
Synthesis and applications. While bioregionalisations are typically based on data from a single taxon, our findings highlight that they can be used as a surrogate for biological patterns seen in other taxa. Bioregionalisations, however, may not capture fully fine-scale communitylevel biodiversity patterns, and this may compromise the ability of protected area networks to protect the full variability in assemblage types. We suggest that it may be necessary to validate existing regionalisations with additional data and analyses such as the RAD analyses conducted here.
Gradient forests: calculating importance gradients on physical predictors
In ecological analyses of species and community distributions there is interest in
the nature of their responses to environmental gradients and in identifying the most important
environmental variables, which may be used for predicting patterns of biodiversity. Methods
such as random forests already exist to assess predictor importance for individual species and
to indicate where along gradients abundance changes. However, there is a need to extend these
methods to whole assemblages, to establish where along the range of these gradients the
important compositional changes occur, and to identify any important thresholds or change
points. We develop such a method, called ‘‘gradient forest,’’ which is an extension of the
random forest approach. By synthesizing the cross-validated R2 and accuracy importance
measures from univariate random forest analyses across multiple species, sampling devices,
and surveys, gradient forest obtains a monotonic function of each predictor that represents the
compositional turnover along the gradient of the predictor. When applied to a synthetic data
set, the method correctly identified the important predictors and delineated where the
compositional change points occurred along these gradients. Application of gradient forest to
a real data set from part of the Great Barrier Reef identified mud fraction of the sediment as
the most important predictor, with highest compositional turnover occurring at mud fraction
values around 25%, and provided similar information for other predictors. Such refined
information allows for more accurate capturing of biodiversity patterns for the purposes of
bioregionalization, delineation of protected areas, or designing of biodiversity surveys.
Exploring the role of environmental variables in shaping patterns of seabed biodiversity composition in regional-scale ecosystems
Environmental variables are often used as indirect surrogates for mapping biodiversity because species survey data are scant at regional scales, especially in the marine realm. However, environmental variables are measured on arbitrary scales unlikely to have simple, direct relationships with biological patterns. Instead, biodiversity may respond nonlinearly and to interactions between environmental variables.
To investigate the role of the environment in driving patterns of biodiversity composition in large marine regions, we collated multiple biological survey and environmental data sets from tropical NE Australia, the deep Gulf of Mexico and the temperate Gulf of Maine. We then quantified the shape and magnitude of multispecies responses along >30 environmental gradients and the extent to which these variables predicted regional distributions. To do this, we applied a new statistical approach, Gradient Forest, an extension of Random Forest, capable of modelling nonlinear and threshold responses.
The regional-scale environmental variables predicted an average of 13–35% (up to 50–85% for individual species) of the variation in species abundance distributions. Important predictors differed among regions and biota and included depth, salinity, temperature, sediment composition and current stress. The shapes of responses along gradients also differed and were nonlinear, often with thresholds indicative of step changes in composition. These differing regional responses were partly due to differing environmental indicators of bioregional boundaries and, given the results to date, may indicate limited scope for extrapolating bio-physical relationships beyond the region of source data sets.
Synthesis and applications. Gradient Forest offers a new capability for exploring relationships between biodiversity and environmental gradients, generating new information on multispecies responses at a detail not available previously. Importantly, given the scarcity of data, Gradient Forest enables the combined use of information from disparate data sets. The gradient response curves provide biologically informed transformations of environmental layers to predict and map expected patterns of biodiversity composition that represent sampled composition better than uninformed variables. The approach can be applied to support marine spatial planning and management and has similar applicability in terrestrial realms.
Biological surrogacy in tropical seabed assemblages fails
Surrogate taxa are used widely to represent attributes of other taxa for which data are sparse or absent. Because surveying and monitoring marine biodiversity is resource intensive, our understanding and management of marine systems will need to rely on the availability of effective surrogates. The ability of any marine taxon to adequately represent another, however, is largely unknown because there are rarely sufficient data for multiple taxa in the same region(s). Here, we defined a taxonomic group to be a surrogate for another taxonomic group if they possessed similar assemblage patterns. We investigated effects on surrogate performance of (1) grouping species by taxon at various levels of resolution, (2) selective removal of rare species from analysis, and (3) the number of clusters used to define assemblages, using samples for 11 phyla distributed across 1189 sites sampled from the seabed of Australia's Great Barrier Reef. This spatially and taxonomically comprehensive data set provided an opportunity for extensive testing of surrogate performance in a tropical marine system using these three approaches for the first time, as resource and data constraints were previously limiting. We measured surrogate performance as to how similarly sampling sites were divided into assemblages between taxa. For each taxonomic group independently, we grouped sites into assemblages using Hellinger distances and medoid clustering. We then used a similarity index to quantify the concordance of assemblages between all pairs of taxonomic groups. Surrogates performed better when taxa were grouped at a phylum level, compared to taxa grouped at a finer taxonomic resolution, and were unaffected by the exclusion of spatially rare species. Mean surrogate performance increased as the number of clusters decreased. Moreover, no taxonomic group was a particularly good surrogate for any other, suggesting that the use of any one (or few) group(s) for mapping seabed biodiversity patterns is imprudent; sampling several taxonomic groups appears to be essential for understanding tropical/subtropical seabed communities. Consequently, where resource constraints do not allow complete surveying of biodiversity, it may be preferable to exclude rare species to allow investment in a broader range of taxonomic groups.
Predictions of beta diversity for reef macroalgae across southeastern Australia
We analyzed and predicted spatial patterns of turnover in macroalgal community composition (beta diversity) that accounted for broad-scale environmental gradients using two contrasting community modelling methods, Generalised Dissimilarity Modelling (GDM) and Gradient Forest Modelling (GFM). Percentage cover data from underwater macroalgal surveys of subtidal rocky reefs along the southeastern coastline of continental Australia and northern coastline of Tasmania were combined with 0.01°-resolution gridded environmental variables, to develop statistical models of beta diversity. GDM, a statistical approach based on a matrix regression, and GFM, a machine learning approach based on ensemble tree based methods, were used to fit models and generate predictions of beta diversity within unsurveyed areas across the region of interest. Patterns of macroalgal beta diversity predicted by both methods were remarkably congruent and showed a similar and striking change in community composition from eastern South Australia to western Victoria and northern Tasmania. Macroalgal communities differed markedly in predicted composition between the open coast and inshore locations. A distinct algal community was predicted for the enclosed Port Philip Bay in Victoria. Sea surface temperature standard deviation and average contributed most to changes in beta diversity for both the GDM and GFM models; changes in wave exposure and oxygen also influenced beta diversity in the GDM model, while salinity and exposure contributed substantially to the GFM model. The GDM and GFM analyses allowed us to model and predict spatial patterns of beta diversity in macroalgal communities comprising >180 species over 6600 km of coastline. These outputs advance regional-scale conservation management by allowing planners to interpolate from point source ecological data to assess the distribution of biodiversity across their full domain of interest. The congruence between methods suggests that strong environmental gradients related to temperature and exposure are the common drivers of community change in this region.
Model based grouping of species across environmental gradients
We present a novel approach to the statistical analysis and prediction of multispecies data. The approach allows the simultaneous grouping and quantification of multiple species’ responses to environmental gradients. The underlying statistical model is a finite mixture model, where mixing is performed over the individual species’ responses to environmental gradients. Species with similar responses are grouped with minimal information loss. We term these groups species archetypes. Each species archetype has an associated GLM that can be used to predict distributions with appropriate measures of uncertainty. Initially, we illustrate the concept and method using artificial data and then with application to real data comprising 200 species from the Great Barrier Reef (GBR) lagoon on 13 oceanographic and geological gradients from 12°S to 24°S. The 200 species from the GBR are well represented by 15 species archetypes. The model is interpreted through maps of the probability of presence for a fine scale set of locations throughout the study area. Maps of uncertainty are also produced to provide statistical context. The presence of each species archetype was strongly influenced by oceanographic gradients, principally temperature, oxygen and salinity. The number of species in each group ranged from 4 to 34. The method has potential application to the analysis of multispecies distribution patterns and for multispecies management.
RAD biodiversity: prediction of rank abundance distributions from deep water benthic assemblages
Rank abundance distributions (RADs) are a description of community structure common to every ecological sample where counts are recorded and are useful for managing and understanding biodiversity. We use RADs to describe patterns of biodiversity in samples with high numbers of unique species. We use a novel statistical method to analyse RADs and demonstrate prediction methods for attributes of biodiversity. The RAD is defined by the total abundance (Ni), species richness (Si) and the vector of relative abundances (nij) and the joint probability distribution of these quantities is modelled. Models were fitted to benthic biological data sampled on the Western Australian coast from depths of 100 to 1500 m and a latitudinal range of 22 to 35oS, using topographic and oceanographic data as covariates. Predictions from fitted models give attributes of biodiversity derived from RADs at a regular grid over the sampled area. The Leeuwin current and Leeuwin undercurrent appear to be key structuring forces for the predicted biodiversity attributes. The predictions show that benthic biodiversity is complex and varies with a number of different covariates. The predictions are unique, as they characterise important aspects of biodiversity and how it varies with large spatial scales. The predictions enable the complete reconstruction of the expected RAD at any point where covariates are available with estimates of uncertainty.