Does biodiversity protect humans against infectious disease? Comment

Identifying how changes in biodiversity alter infectious disease dynamics is important for both basic science and policy. Biodiversity, broadly conceived, denotes “the variety of life in all its manifestations” (Loreau and Kinne 2010 ), and encompasses not only species richness, but also phylogenetic diversity, trait and functional distinctions among taxa, and indeed the complexity of community organization, such as food web interactions. Wood et al. ( 2014 ) note that an increase in biodiversity can at times amplify disease risk in a focal host species existing across a gradient in biodiversity, and some empirical examples do seem to demonstrate amplifi cation (Young et al. 2013 ). Keesing et al. ( 2006 ) outline several mechanisms that could underlie such amplifi cation, and it is certainly the case that if there is no biodiversity (the “parking lot ecosystem”) there will be no zoonotic diseases. Where matters get more interesting, complex, and relevant to conservation is when one examines how disease risk might shift across broadscale gradients in more realistic scenarios, such as when one compares largely intact natural ecosystems with systems degraded or fragmented by human activity (the focus of this comment). The recent literature contains a variety of examples of situations in which high biodiversity reduces disease risk, a phenomenon termed the “dilution effect” (Keesing et al. 2006 , Civitello et al. 2015 ). Wood et al. ( 2014 ) review a number of cases in which they hypothesize that one might fi nd that instead of a dilution effect, an increase in biodiversity is associated with increased disease risk. They postulate in particular (p. 821) that prior work has focused on the developed world, and that infectious disease in developing nations in the tropics often increases with increasing diversity. We respectfully suggest that for many of the systems discussed by Wood et al. ( 2014 ), the role of diversity is not yet clear, at least if comparisons are between intact forest and degraded habitats, including in particular in tropical biomes. Such comparisons are particularly pertinent to conservation policy, and would facilitate predictions regarding whether disease risk will increase when forests of high conservation value are degraded or fragmented by human activities. One consequence of eroding wildlife diversity in fragmented forests is that smallerbodied hosts, and hosts at lower trophic levels, most famously rodents, can become hyperabundant either due to increased food availability, or to ecological release following extirpation of competitors or predators (Adler and Levins 1994 , Nupp and Swihart 1998 , Terborgh et al. 2001 , Keesing and Young 2014 ). The disappearance of top predators (Estes et al. 2011 , Ripple et al. 2013 ), a widespread signature of anthropogenic impacts, releases their prey from topdown regulation and can also lead to an upsurge in infectious disease because of reduced mortality of infected hosts (Packer et al. 2003 ). Ecological release of prey species can occur across a gradient in body size from small rodents, which can be far more abundant in forest fragments (Nupp and Swihart 1998 , Debinski and Holt 2000) , to largebodied ungulates, which can become hyperabundant in the absence of predation or when feeding on agricultural landscapes (Ripple et al. 2013 , Wilmers and Levi 2013 ). Shifts in competition can also lead to ecological release; McCauley et al. ( 2008 ), for instance, report that removal of large herbivores in African savanna permitted surges in the abundance of smallbodied rodents, which sustain pathogencarrying fl eas at higher abundance (see also Young et al. 2014 ). Fragmentation of habitats can also lead to the ecological release of hosts through other mechanisms. For example, because of matrix subsidies and mesopredator release, generalist mesocarnivores such as raccoons and opossums often reach much higher densities in fragmented landscapes near agricultural fi elds and on the periphery of exurban development (Crooks and Soule 1999 , Ritchie and Johnson 2009 ). For example, the most heavily infected hosts for the trematode vector of salmonpoisoning disease ( Nanophytes salmincola ) in Oregon are mesopredator raccoons and skunks Ecology, 97(2), 2016, pp. 536–546 © 2016 by the Ecological Society of America