The Key Biodiversity Area (KBA) approach was initially developed for terrestrial species, and is appropriate for the conservation and management of species that require conservation at the site scale (Eken et al., 2004). We can infer from Langhammer et al. (2007) that the identification of KBAs is relatively straightforward for species listed on the IUCN Red List of Threatened Species as Critically Endangered (CR) or Endangered (EN) and also have highly restricted global ranges; this remains true even in situations of data scarcity. Alliance for Zero Extinction (AZE) trigger species provide a perfect example, being CR or EN species which are restricted to a single site worldwide (Ricketts et al., 2005; http://www.zeroextinction.org). Marine turtles are wide-ranging and highly migratory, but nevertheless have site-specific conservation needs based on seasonal congregatory behavior; they clearly present a challenging example for applying the KBA approach. Testing the applicability of the KBA approach to such wide-ranging species and devising appropriate thresholds for site identification were the fundamental challenges that we faced in our analysis. Our findings highlighted the importance of scale and context in determining appropriate population thresholds for site identification for globally ranging species such as marine turtles.
As Edgar & Brooks (2011) describe, developing criteria and thresholds for KBA identification that can be consistently applied across taxa and biomes requires numerous trade-offs. Simplicity and consistency enable cross-biome and cross-taxonomic comparison. However, we run the risk of generating results that are not applicable within a country, or that are useless for a specific group of species. Additional complexities are introduced by comparing data-poor regions (such as Melanesia) with data-rich areas.
One possible solution would be to nest scales of conservation priorities. Globally significant priorities would use a single threshold applied globally, while regional priorities would use the same threshold applied at a smaller geographic scale. For instance, in the case of marine turtles, this could mean sites holding 1% or more of the global population of a species would trigger a KBA, as proposed initially by Edgar et al. (2008). At the same time, sites holding 1% or more of the regional population of the species would trigger a Regional KBA. In other words, within a Management Unit (Moritz, 1994) or Regional Management Unit (Wallace et al., 2010), a marine turtle species would trigger a Regional KBA if it held 1% or more of the total population found within that unit.
In the context of the data for Melanesia analyzed in Bass, Anderson & De Silva (2011), this would result in 11 KBAs that are globally significant and 54 Regional KBAs. As described by Wallace et al. (2011), the regional scale is likely the most appropriate for conservation and management of marine turtles. Given variability among RMUs, the identification of sites within these units might be most relevant for determining specific conservation actions. At the same time, using the global threshold will yield numbers of sites that can be used for cross-taxonomic comparisons and other global analyses and prioritization efforts. The end result would be a nested set of priority sites: a smaller set of global priorities within a larger set of regional priorities.
This parallels the development within IUCN of regional and national Red Lists to complement the global Red List (IUCN, 2003). These sub-global lists utilize the same criteria, but are scaled down to a country or region. If applied correctly, any endemic species identified as threatened on a regional or national list could also be listed as threatened on the global list. In some cases, these lists contain more detailed information or additional species that have not yet been assessed globally. Regional assessments can be pooled together to inform global assessments, but cannot themselves be scaled-up to create a global assessment. The same would be true for regional versus global KBAs.
Such a process of nested priorities would also parallel the regional assessments of extinction risk called for by many marine turtle specialists, including Seminoff & Shanker (2008). These regional assessments would fit within a larger global assessment for each species, and would take into account geographic variability in threat and population trends.
Another relevant parallel is the development of regional Alliances for Zero Extinctions that build from and expand the work undertaken through global identification of AZE sites. This trend is most advanced in Latin America, through relatively new organizations such as the Brazilian Alliance for Zero Extinction (http://www.biodiversitas.org.br/baze/intro.htm). These regional alliances allow for the identification of AZE sites for CR or EN species found in a single site that belong to taxonomic groups that have not yet been assessed comprehensively at the global level. They also potentially allow for the identification of purely regional priorities.
Overall, we believe the Bass et al. (2011) analysis highlights some useful preliminary results for marine turtle conservation and raises a number of interesting possibilities for further research. Indeed, much work remains to be done. As recommended by Wallace et al. (2011) the newly defined RMUs should be used to refine the KBA analysis for Melanesia. The entirety of Melanesia appears to fall within a large Pacific RMU for species such as the leatherback, while there region is split into two RMUs for several other marine turtle species (Wallace et al., 2010). In addition, as highlighted by Edgar & Brooks (2011) further testing of criteria and thresholds is needed both in different geographies and using other wide-ranging species. We had previously recommended that a KBA analysis be undertaken for marine turtle species in larger rookeries such as the Caribbean, using the same thresholds developed in this study. We believe comparing such a relatively data-rich region with the comparatively data-poor Melanesian context would be useful. As Wallace et al. (2011) suggest, it would be interesting to explore the feasibility of marine turtle KBA identification for other aggregations such as for feeding and ontogenetic life stages. Beyond the marine realm, testing similar absolute and percent thresholds for wide-ranging freshwater fish might be particularly interesting.