As well as other ecological paradigms, the theory of extinction debt (ED, sense Tilman et al., 1994) is elegant and appealing for field ecologists. Field ecologists (including us) have tried to identify ecological conditions that meet the theoretical assumptions needed to test ED (see also Wiens, 1977 for a similar case). Although the theory of ED assumes habitat homogeneity to reduce ecological complexity, field ecologists usually encounter habitat heterogeneity. Differences in site history are another obstacle to test the ED theory using a snapshot approach (or space–time substitution; Johnson & Miyanishi, 2008). Despite the difficulty of testing ED in the field, empirical evidence of the presence of ED has been accumulated over the last decade. Most studies adopt space–time substitutions, while others track sites over a long period, and these studies convincingly supported the presence of ED (e.g. Stouffer, Strong & Naka, 2008).
The importance of ED is now widely acknowledged and future research should now shift from demonstration to application (Kuussaari et al., 2009). How is the concept of ED useful for conservation? Fortunately, commentaries by Hanski (2013) and Vellend & Kharouba (2013) provided several insights. In this response, we discuss the challenges and opportunities for application of ED to real-world biodiversity conservation.
Challenges for application: importance of mapping for biodiversity conservation
The application of ED to conservation presents numerous challenges, including those provided by Vellend & Kharouba (2013). Conservation practice entails big money, and accurate predictions on the status of populations are required. One of the most relevant questions is: when will a focal population become extinct in a specific patch/landscape? Further, we may also be interested in future decreases in abundance and species richness and/or occupancy resulting from habitat fragmentation in relation to ecosystem services or functions (Gaston & Fuller, 2008).
Mapping ED is a straightforward way to enhance the application of ED in real-world situations. Generic simulation models and meta-analysis may also provide ‘gross’ estimates associated with ED. Moreover, as pointed out by Hanski (2013), use of an information-rich connectivity index would improve predictions. However, it is not until ED is mapped that we can grapple with the number of EDs or establish where ED remains. Despite of its usefulness, mapping ED was rarely conducted in previous studies (but see Cowlishaw, 1999 and Wearn, Reuman & Ewers, 2012). Although accurate estimation of ED may not be achieved immediately, mapping may enable the identification of land patches with large amounts of ED. For example, in our study area, we clarified that high amounts of unpaid ED remain in small forest patches (Soga & Koike, 2013, see also Fig. 1). Seeing is believing.
‘Debts’ or ‘Degradation'? Importance of habitat quality in fragmented landscapes
Our study did not consider the differences in habitat quality among the patches as in most of the previous ED studies (but see Vellend et al., 2006). Although we tried to search for studied patches with similar habitat structure and composition, we observed that some small patches experienced higher human disturbances (e.g. pavements) per area. Large amounts of residual variation in our study may be explained by these variations of habitat quality. Landscape modification usually entails habitat degradation (Fischer & Lindenmayer, 2007); this general phenomenon may be called the ‘norm’. Early theoretical studies on ED adopted simplified assumptions and focused on the effects of patch area and arrangement, rather than habitat quality. The importance of habitat quality in fragmented landscapes has recently been indicated (e.g. Mortelliti, Amori & Boitani, 2010). We suggest that apparent effects of ED would be partly confounded by habitat degradation. Ford et al. (2009) similarly suggest that additional extinctions of woodland birds after deforestation were likely to be caused by vegetation thickening in Australian agricultural landscapes. The relevant question is which ED or habitat degradation is more important to explain present biodiversity?
If ED is caused by habitat degradation, it may be possible to revive several living dead species (species doomed to extinction through ED) by improving habitat quality, which clearly indicates a conservation opportunity in fragmented landscapes. Huth & Possingham (2011) also suggest the importance of habitat quality in studies on species richness–patch area relationships, and that small high-quality patches would harbor many more species than prevailing degraded small patches. As Hanski (2013) points out, improving habitat quality may not be a panacea for long-term species persistence. Land acquisition for nature reserves in urban area is expensive (Naidoo et al., 2006), thus improving habitat (and matrix) quality may be a feasible conservation option, and bring high conservation returns. It is also suggested that conservation strategies through enhancement of habitat connectivity involves many uncertainties, thus improving habitat quality may be a more robust conservation intervention (Hodgson et al., 2009, 2011).
‘Crisis’ or ‘Chance'? The fate of living dead in changing landscapes
In this century, regimes of human land-use are largely shifting. In many developed countries, intensive urbanization has slowed and restoration practices using revegetation have been initiated. In Tokyo, as Vellend & Kharouba (2013) suggest, several restoration projects are ongoing (e.g. ‘the 10-Year Project for Green Tokyo’) and many green spaces have been created during the last few decades. Moreover, Tokyo's population is projected to be peak in 2020 (populations in marginal urban areas will be peak in 2015), after which the population is predicted to decrease gradually (Tokyo Metropolitan Government, 2013). Surprisingly, by 2100, Tokyo's population is expected to the same as it was at around 1950 (Tokyo Metropolitan Government Bureau of General Affairs, 2013). Expansion of the urban area to our study site began in the 1960s, and marginal urban areas such as this will begin to retreat in the near future. Yamaura et al. (2012) also present a similar story for planted forests in Japan. It is interesting to consider how quickly land-use can change.
In our very dynamic landscapes, large amounts of ED are likely to exist. Further, as Hanski (2013) points out, payment period of landscape-level ED would be much longer than those of patch-level ED. The example of Tokyo described earlier suggests that landscapes will easily change before the ED is completely paid out. We now have rigorous scientific evidence of the presence of ED and destinies of ‘living dead’ species hinge on our present and future land-uses. It is time to develop conservation schemes that take into account ED in the face of regime shifts of land-use. The early theoretical suggestions made by Hanski & Ovaskainen (2002) and Schrott, With & King (2005) are now extremely relevant: we need to restore landscapes within the payment period. The effects of restoration practices will be elicited over time. This reversed aspect of ED was coined ‘species credit’ by Hanski (2000) and ‘immigration credit’ by Jackson & Sax (2010). Restoration of habitat quality itself will also require a long time (Soga et al. unpubl. data). When is the payment period of ED longer than payback period of species credit? It depends on the range of factors from life-history traits to landscape structure. We should search for and achieve such conditions by novel conservation schemes. Previously, ED has been regarded as an ecological pessimism: a crisis for future biodiversity. Is talking about ED really pessimism? The answer will be no. In the changing world, ED can provide us with an important conservation chance. It's time to take this opportunity by the forelock.