Dispersal from the frying pan to the fire


James D. Forester, Dept. Fisheries, Wildlife, and Conservation Biology, University of Minnesota, 1980 Folwell Ave., St Paul, MN 55108, USA.
Email: jdforest@umn.edu

Worldwide, animal species are being threatened by the loss and fragmentation of their habitat. The loss of habitat area is thought to have the greatest impact on extinction risk (Fahrig, 2002). However, once a landscape is altered, the resulting land-cover configuration (and the relative habitat quality of altered areas) will clearly affect the degree of isolation experienced by the fragments of once contiguous animal populations. Where these fragmented populations fall on the gradient between non-equilibrium populations, classic metapopulations, and simply patchy populations depends largely on the dispersal ability of the species, and on individual animals' perceptions of the landscape mosaic (Harrison, 1991; Driscoll, 2007). From a conservation point of view, knowing a species' location on this gradient is critical for determining both short- and long-term management actions. Determining that location can be very difficult, however, especially when dealing with large mammals (Elmhagen & Angerbjörn, 2001).

van Oort, McLellan & Serrouya (2011) expanded and reanalyzed a long-term dataset of relocations of mountain caribou Rangifer tarandus caribou (Wittmer et al., 2005) with the specific goal of quantifying the degree of adult and juvenile dispersal within the population. Ultimately, they wanted to determine if the existing rate of dispersal was sufficient to ‘rescue’ the many small subpopulations distributed throughout the region (or at least recolonize areas after extirpation). Although their sample of 27 collared calves was likely too small to accurately measure natal dispersal (they observed none), the adult sample was much larger and left little doubt that dispersal among subpopulations is at best an infrequent event. Also troubling was their finding that within subpopulations, very few animals changed the location of their summer/fall ranges between years. This means that many groups of animals are effectively reproductively isolated despite being separated by only a few kilometers. A large part of this isolation is the result of strong range fidelity, which has also been observed in the boreal ecotype of this subspecies (Faille et al., 2010). This represents both an opportunity and a challenge for conservation. Because the animals tend to return to the same areas year after year, conservation efforts can ideally be very focused. However, if this range fidelity is maintained despite extensive habitat degradation (as seen by Faille et al., 2010), extremely intensive (and expensive) landscape-scale mitigation or animal relocation may be required.

Despite the main findings of van Oort et al. (2011), their data also show that the most immediate challenge facing mountain caribou is not one of dispersal, but of demography. With the exception of one subpopulation (for which the census area expanded, yielding a higher count), all subpopulations declined between 1990 and 2006. There is a growing consensus that these declines are the result of predator-mediated apparent competition (sensu Holt, 1977) – a process also linked to declines of woodland caribou elsewhere in Canada (Seip, 1992; James et al., 2004; Wittmer et al., 2007; Latham et al., 2011). As the area of late seral-stage coniferous forests (the preferred habitat of mountain caribou) has been reduced and fragmented by timber harvest, power transmission lines and roads, populations of other ungulate species have grown. This increased prey biomass supports a much higher density of predators that now opportunistically prey on caribou.

Small subpopulations of mountain caribou are already at risk due to demographic stochasticity. Apparent competition means that caribou living in landscapes with large amounts of early-seral-stage forest also face reduced survival rates (Wittmer et al., 2007). Thus, the rare dispersing individuals will likely find they have moved to a situation even worse than the one they left. van Oort et al. (2011) point out that managers cannot rely on a natural rescue of these subpopulations, so more intensive management is required. Population augmentation is one option and a recent population viability analysis indicates that translocation efforts may be effective at rescuing or reintroducing some populations of mountain caribou (Decesare et al., 2011). However, this approach is unlikely to be broadly successful without somehow mitigating the impact of apparent competition. Direct population management through increased harvest of other ungulates and predators may work in the short term, but will be difficult to maintain consistently over a broad spatial scale. A more effective and long-term mitigation strategy will require broad-scale forest management directed at manipulating the spatial distribution of both predators and ungulates.

One general result of the research by van Oort et al. (2011) is a reminder that not all fragmented populations are metapopulations. Without dispersal linking subpopulations and allowing for recolonization in the event of local extirpation, the remnant population fragments may simply represent an unpaid portion of the extinction debt (Tilman et al., 1994). Thus, for species of conservation concern, it is important to determine if subpopulations are linked by dispersal, and if not, what demographic rates and level of dispersal would be required to reconnect them as a population or metapopulation. If the subpopulations are indeed isolated, is the lack of dispersal a result of the species' natural history (e.g. strong range fidelity or density-dependent dispersal), or is it a constraint of landscape structure (e.g. habitat fragmentation)? Answering these questions requires an understanding of the mechanisms underlying local population and community dynamics as well as individual movement behavior (McRae et al., 2008; Morales et al., 2010; Smouse et al., 2010). Such mechanisms are beginning to be incorporated into spatially-structured population models (e.g. Revilla & Wiegand, 2008) that will ultimately provide greater insight into how animal communities interact with spatially complex and temporally dynamic landscapes.