You are here: Home Our Work Global habitat suitability for reef forming cold-water corals

Global habitat suitability for reef forming cold-water corals

submitted by John Guinotte, Andrew Davies, Jeff Ardron

Figure 1. Global habitat suitability for six species of reef forming cold water scleractinian “stony” corals. Spatial resolution is 1 km x 1 km.

Reef-forming cold water corals are known to be very sensitive to anthropogenic activities, are expected to be heavily impacted by ocean acidification, and are known to have very slow recovery rates. Using known locations of the six reef-forming cold water coral species, amassed from research and cruise data bases (2732 records), we predict areas of suitable coral habitat throughout the world based on 26 environmental conditions. The fine spatial resolution of these predictions (1 km x 1 km) allows for consideration of these possible EBSAs at a scale suitable for conservation measures.

Download this report

stony-coral-illustration.pdf — PDF document, 8877Kb

Scientific background

Here we show global scale predictions for reef-forming scleractinian (or “stony”) corals. All six primary framework forming coral species were considered: Lophelia pertusa, Madrepora oculata, Goniocorella dumosa, Oculina varicosa, Enallopsammia profunda, Solenosmilia variabilis. These species form reef-like habitats and are known to be very sensitive to anthropogenic activities which make contact with the seafloor, such as bottom fisheries. In addition, these ecosystems are expected to be heavily impacted by ocean acidification (Guinotte et al 2006). Disturbance in the deep sea is usually negative and these cold water coral species are known to have very slow recovery rates, on the order of hundreds to thousands of years, if at all (Roberts et al 2006). Cold-water corals form structural habitat with a range of ecosystem functions in the deep sea, including promoting local biodiversity and supporting commercially important fisheries. Over 1,300 species have been documented on the Lophelia pertusa reefs in the NE Atlantic (Roberts et al 2006).

Predictive modeling of species’ distributions is increasingly used to direct conservation actions, research and future surveys. To date, several limitations have restricted the utility of this approach in the deep sea, such as the accuracy of species presences, the lack of reliable absence data, and the coarse resolution of environmental datasets. The habitat suitability maps presented here address earlier limitations by taking the best available data at a much finer scale, and using a novel approach to generate maps of environmental conditions on the seafloor. These improved data sets were incorporated into a “maximum entropy” model (Phillips et al. 2006; http://www.cs.princeton.edu/~schapire/maxent/) that estimates the distribution of a given species taking into consideration the known occurrences of that species in relation to a series of environmental variables likely to influence its distribution.

The analysis shows that scleractinians are predicted to occur throughout many continental shelves and slopes in the world’s oceans, with the majority of suitable habitat found in the Atlantic Ocean, and around New Zealand and Australia, whilst the Pacific Ocean appears to be less suitable. Due to the significant increases in model resolution relative to other studies, we uncovered suitable habitat on thousands of seamounts that have yet to be studied. The outputs are statistically significant, but external validation of some of these areas by field surveys is warranted and would improve the prediction and utility of the analysis. The vulnerability of these sensitive ecosystems to human activities and the cost of doing research in the deep sea make it increasingly important that these results are both applicable and reliable at a range of spatial scales.

 

How the EBSA was identified

The areas where EBSAs are likely to occur were identified by investigating the environmental conditions surrounding known coral locations. Areas with similar environmental conditions to those where documented corals occur were then calculated and mapped in a Geographic Information System (GIS). The high spatial resolution of this analysis (1 km x 1 km) allows users and managers to identify areas where cold water corals are likely to occur, at a scale meaningful to management measures such as closures or protected areas.

In the case of the high seas (areas beyond national jurisdiction), this approach represents a practical way forward in identifying sensitive, fragile and slow-recovering stony coral species and their habitats in regions of the world’s oceans that have not been well studied or surveyed. This analysis also provides a means by which proposed conservation measures can be assessed for their likely conservation value with regard to this particular criterion. For example, figure 2 shows that protective bottom trawl closures in the North East Atlantic are located in areas that have a very high probability of containing cold water coral habitat and encompass a significant amount of the high probability areas. In contrast, figures 3 and 4 show that voluntary bottom trawl closures in the Southern Indian Ocean are not positioned in areas that would protect the majority of sensitive coral habitat predicted in the region (though they may have value for other kinds of EBSAs).

High resolution maps will be placed on the world wide web for download. The high spatial resolution of the results combined with areal extent of the high seas make it impossible to depict the full spectrum of predicted sensitive bottom habitat in this document.

The habitat suitability model was generated using Maxent software 3.31 (Phillips et al. 2006; http://www.cs.princeton.edu/~schapire/maxent/). Default model parameters were used (convergence threshold of 10-5, a maximum iteration value of 1000 and automatic regularization with a value of 10-4); these default settings have been shown to achieve good performance (Phillips & Dudik 2008). The habitat suitability map was generated by calculating a raw probability value for each grid cell, such that the total of all cell probabilities summed to one. This value was then scaled logistically, resulting in a relative habitat-suitability value ranging from zero to one. The logistic habitat suitability values can be interpreted as an estimate of the probability of presence under a similar level of sampling effort as that used to obtain the known occurrence data (Phillips & Dudik 2008). We split the presence data into 75% training and 25% test data for model validation purposes.

 

Figure2

Figure 2. Predicted coral habitat and locations of bottom trawl closures in the NE Atlantic Ocean.

 

Figure3

Figure 3. Predicted coral habitat and locations of proposed bottom trawl closures in the Southern Indian Ocean.

 

Figure 4

Figure 4. Predicted coral habitat and locations of selected bottom trawl closures in the Southern Indian Ocean. Note: the majority of high probability areas are not included in the proposed closures.

 

Sources of data

Environmental variables were created using the latest global bathymetric data, available at 30 arc second resolution (Becker et al. in press), we clipped vertically oceanographic gridded data from sources such as World Ocean Atlas to areas of available seafloor at each standardized depth interval. We assumed that conditions at these depth layers were indicative of the conditions that would be found in the area. We selected relevant environmental layers, including omega aragonite (Steinacher et al. 2009), depth (Becker et al. in press), dissolved inorganic carbon (Steinacher et al. 2009), dissolved oxygen (Garcia et al. 2006a), surface productivity (MODIS L3 Annual SMI), salinity (Boyer et al. 2005), silicate (Garcia et al. 2006b) and temperature (Boyer et al. 2005). In all, 26 geophysical, hydrographic, chemical, and biological variables were considered. In total, 2732 presence points of the six primary framework-forming Scleractinia including Lophelia pertusa, Madrepora oculata, Enallopsammia profunda, Goniocorella dumosa, Solenosmillia variabilis and Oculina varicosa were obtained from sources published in journals, cruise reports, and other sources.

 

Important considerations

There are limitations that must be considered when interpreting habitat suitability maps. Our improved approach addresses many issues with scale, resolution and extent, but a critical limitation remains. These maps show the potential for suitable cold water coral habitat. Higher values of suitability indicate the likelihood that a species may be found in a given area, but this does not mean that the species is actually present within that area. There may remain barriers to colonization, such as biotic interactions in the form of competitive exclusion or dispersal pathways that are blocked by biogeographic barriers (Guisan & Zimmermann 2000). Therefore, areas that are predicted to have a high likelihood of stony coral occurrences and are being considered as likely EBSAs, should be ground-truthed through directed surveys.

In addition to addressing the criterion vulnerability, fragility, sensitivity, or slow recovery, some areas are also likely to fulfill the following CBD EBSA criteria:

  • Biological diversity
  • Special importance for life-history stages of species
  • Importance for threatened, endangered or declining species and/or habitats (where these areas are subject to anthropogenic activities such as bottom fisheries)
  • Naturalness (where these areas are not subject to anthropogenic activities such as bottom fisheries)

 


Document Actions