A considerable amount of time and money is spent searching for species presumed to be extinct or on the brink of extinction. Although predictable patterns in species decline and rediscovery almost certainly exist (Fisher & Blomberg, 2011), the spatial contexts of species declines and rediscoveries have not been quantified until recently. Two opposing hypotheses have been proposed. First, the ‘abundant centre’ or ‘range collapse’ hypothesis predicts that presumed-extinct species will be rediscovered at the centres of their historical geographical ranges (Channell & Lomolino, 2000a; Sagarin & Gaines, 2002). This hypothesis holds that species’ ranges will collapse inwardly in response to anthropogenic threats such as predation or habitat destruction along the peripheries. In contrast, the ‘range eclipse’ hypothesis predicts that anthropogenic pressures will push declining species from the centre of their historical ranges outward to more secure peripheral habitats (Channell & Lomolino, 2000a; Hemerik et al., 2006). Presumed-extinct species should, therefore, be rediscovered at the peripheries of their historical ranges. Neither of these hypotheses considers elevation, despite the critical role high elevations may play as ecological refuges for threatened species (e.g. Towns & Daugherty, 1994; Benning et al., 2002; Raxworthy et al., 2008). A third hypothesis, the ‘elevation refuge hypothesis’, explicitly considers elevation as a geographical factor in species range contractions.
In a recent study, Fisher (2011a) compiled data for 67 rediscovered mammal species, providing for the first time a systematically assembled dataset to test these alternative hypotheses. She incorporated a spectrum of possible explanatory variables (e.g. human population density, anthropogenic threat type, and the locations and elevations of both the last prior sighting and the rediscovery of each species). Using this dataset, she asked: (1) How far from last prior sightings do rediscoveries occur? (2) Are rediscoveries likely to occur in habitats different from historical habitat types? (3) Are rediscoveries more likely to occur at higher elevations than past records? (4) How does human population density affect rediscovery locations? (5) Are missing mammals more likely to be rediscovered in their prior range centres or along their prior range peripheries? Fisher’s answers to these questions are discussed below. Although previous studies have addressed some of these questions for declining species (Channell & Lomolino, 2000a,b; Sagarin & Gaines, 2002; Hemerik et al., 2006), they have never been used to understand patterns in species rediscovery. Moreover, the incorporation of elevation into such an analysis is novel.
Rediscoveries can occur at locations distant from historical locations. Of the 67 species included in this study, only three were rediscovered at their original locations. Nearly half of the rediscovered species were found within 100 km of their last recorded sighting, but four species were found > 1000 km from their last recorded locations. Species with larger historical ranges were more likely to be rediscovered further from their last recorded occurrence. Coastal species with restricted ranges were found closer to their historical locations than lowland or highland species.
Rediscoveries sometimes occur in new, potentially marginal habitats. Ten of the presumed-extinct species were rediscovered in novel habitats. These species, all historically considered to be primary or lowland forest species, experienced high human population densities and were threatened mainly by habitat loss or a combination of overkill and habitat loss. Their rediscovery in novel habitats, including marginal or degraded secondary forest and plantations, suggests that rediscovery efforts should not be strictly focused on historical or even natural habitat types.
Rediscovery locations are frequently at higher elevations than historical ones. An up-slope shift of about 200 m elevation was recorded for most rediscovered species (excluding mountain-top and coastallyrestricted species). This shift towards higher elevations is probably a response to increased anthropogenic threats at lower elevations (Towns & Daugherty, 1994; Channell & Lomolino, 2000b; Benning et al., 2002).
Human population density influences the location of rediscoveries. Human population density did not differ between rediscovery and last recorded locations for individual species. Nevertheless, human population density did differ among locations with different elevations and threat types. High elevation locations had lower human population densities than low elevation and coastal locations. Human densities were higher at locations where presumed-extinct species were most affected by habitat loss or human overkill and lower at locations where presumed-extinct species were most affected by introduced species.
Rediscovery patterns are not consistent with the expectations of either the range collapse or range eclipse hypotheses. In support of range eclipse, rediscovered mammal species whose declines were caused by habitat loss were more likely to be found at the peripheries of their historical ranges. On the other hand, species whose declines were caused by human overkill or introduced species were equally likely to be rediscovered in both the centre and the periphery of their former ranges. Coastal and lowland species were more likely to be rediscovered in the peripheries of their historical ranges, while highland species were more likely to be rediscovered in the centres.
The lack of unanimous support for either range eclipse or range collapse could be explained by discrepancies in the accuracy of historical data. Former ranges for the majority of rediscovered species are not well constrained. Mammal rediscovery results could also be affected by differences in how each threat type affects declining species. Habitat loss progressively pushes a declining species into the peripheries of its former range. Conversely, overkill and invasions reduce population density throughout a species’ range (Rodríguez, 2002).
Similar to habitat loss, increased human population density can also push declining species into the peripheries of their former ranges, or even into marginal anthropogenic or secondary vegetation. Mammal species rediscovered in these novel habitats had generally been exposed to high human population densities. Fisher (2011a) further argues that declining and presumably extinct species may be eliminated first from low elevation habitats because these areas have the densest human populations and highest levels of deforestation. Accordingly, declining species might be expected to shift their ranges to higher elevations. These high elevations are operating as refuges not because they provide ideal habitat for threatened species but because they tend to be the only localities that preserve any natural habitat or offer protection from invasive predators or diseases (e.g. Towns & Daugherty, 1994; Benning et al., 2002). Coastal species with limited geographical flexibility may be particularly vulnerable to anthropogenic pressures. Although equally comprehensive datasets do not exist for other classes of animals, the mammal rediscovery results are in agreement with observed elevational shifts for arthropod, amphibian, reptile and bird taxa (e.g. Towns & Daugherty, 1994; Benning et al., 2002; Raxworthy et al., 2008; Chen et al., 2009).
Surprisingly, the influence of human disturbance on species elevation shifts has not been considered by most authors. Instead, these shifts are usually attributed to global warming (e.g. Raxworthy et al., 2008; Chen et al., 2009). Two notable exceptions are studies by (1) Towns & Daugherty (1994), who observed that several reptile and amphibian species endemic to New Zealand are able to persist at high elevations where they are relatively isolated from introduced animals, and (2) Benning et al. (2002), who similarly determined that the remaining species of Hawaiian honeycreepers persist at high elevations because they are relatively protected from anthropogenic threats. Further studies explicitly considering the relative importance of climate versus direct human disturbance will prove illuminating.
Importantly, species rediscovery is not synonymous with species survival. Recovery efforts are also essential. The rediscovery of species in secondary or marginal habitat appears promising. However, unless recovery actions are taken, the likelihood that these species will persist is low (Fisher, 2011b). To date, large-bodied charismatic species have received the majority of species recovery funding, despite the increasing number of rediscovered small-bodied species and the lower costs associated with their recovery (Fisher, 2011b). Recognition of this bias will be crucial for the persistence of rediscovered mammals.
In summary, the findings of Fisher (2011a) contribute to our understanding of the dynamics of species rediscoveries with regard to their historical geographical and elevational ranges. Although previous research has identified range eclipse in response to anthropogenic threats (e.g. Channell & Lomolino, 2000a,b; Hemerik et al., 2006), Fisher’s explicit predictions about decline trajectories and species rediscovery in three-dimensional space are novel. Future conservation and rediscovery efforts can be more targeted by taking these geographical and elevational patterns into account.