Introduction
Land-use change is hypothesized to be a key driver of the emergence of infectious disease (Foley et al., 2005; Lambin et al., 2010; Norris, 2004; Patz et al., 2000). Deforestation and the expansion of human activity into forests can alter ecological communities and interactions (Aguirrea & Taborb, 2008; Roque & Jansen, 2014) and thus potentially increase the risk of infectious disease emergence from wildlife reservoirs and vectors (Lambin et al., 2010; T. Lima et al., 2017; McCauley et al., 2015; Vanwambeke et al., 2007). However, there is debate on whether there is a generalizable effect of land-use patterns and consequently changes in biodiversity on increased disease risk (Levi et al., 2016; Ostfeld, 2013; Ostfeld & Keesing, 2013; Randolph & Dobson, 2012; Wood et al., 2014; Wood & Lafferty, 2013). The question of whether dilution, amplification, or neutralizing effects predominate for particular infectious diseases (or entire suites of pathogens) across gradients of disturbance and biodiversity is a challenging empirical problem as the mechanisms are rarely clear and often non-linear. This is particularly the case for vector-borne pathogens, which land-use change can influence by modifying the composition, density, and transmission traits of the hosts, vectors, and pathogens (Burkett-Cadena & Vittor, 2018; Kocher et al., 2022).
Improved mechanistic understanding of how land-use change influences host, vector, and pathogen networks is particularly critical in tropical forests where anthropogenic impacts are rapidly altering landscapes and the burden of disease is disproportionately high. Large-scale deforestation and/or forest fragmentation can influence biodiversity by several mechanisms that likely influence pathogen dynamics and ultimately disease risk. For example, vertebrates at lower trophic levels can become hyperabundant in fragmented forest due to energy-rich subsidies from forest edges and the agricultural matrix (Luskin et al., 2017; Marczak et al., 2007; Wilcox & Gubler, 2005), and due to ecological release when habitat loss and/or fragmentation leads to the decline of larger-bodied competitors and predators (Debinski & Holt, 2000; Nupp & Swihart, 1998; Ripple et al., 2014; Wilmers & Levi, 2013). These mechanisms are supported by field surveys in the southern Amazon showing a strong negative association between the abundance of common reservoir species and forest patch size, and the extirpation of apex predators and other large-bodied taxa in the smallest forest fragments (Michalski & Peres, 2007). However, as the context of tropical deforestation changes from agricultural expansion by many smallholders to large-scale agribusiness monocultures, the impacts of this change on vertebrate community composition may disrupt the patterns of forest loss and/or fragmentation on biodiversity witnessed thus far.
While disturbance-tolerant vertebrate reservoir hosts are likely to play a key role in influencing disease prevalence and emergence as tropical forest systems are fragmented (Johnson et al., 2013), this mechanism alone is not sufficient to predict how disease risk changes as tropical forests are degraded given how the intricate ecological network between hosts, vectors, and pathogens can respond to forest loss in complex ways. For vector-borne pathogens, even if forest edge increases the abundance of reservoir hosts in tropical forests, pathogen reproduction will be stymied if vector populations decline in edge habitats, if there is spatial mismatch between hosts and vectors, or if vectors feed disproportionately on species that are not competent reservoirs. While studies have demonstrated some of the effects of land-use change on host species (Guo et al., 2019; LoGiudice et al., 2003), little work has been done to simultaneously examine the effects of deforestation on vectors, hosts, and pathogens.
To tease apart the effects of large-scale land-use change on host and vector communities, and consequently disease risk, we implemented a multifaceted, landscape epidemiology approach using field surveys and DNA metabarcoding of sandfly vectors and their bloodmeals to disentangle the complex relationships between Leishmania parasites, the known sandfly vectors, and the potential wildlife hosts (see Fig. 1) in response to rapid deforestation across the Amazonian ‘Arc of Deforestation’. Numerous Leishmania species cause cutaneous and visceral leishmaniasis (Akhoundi et al., 2016), neglected tropical diseases that are associated with both intact tropical forest (Jones et al., 1987; Lainson, 1983; Travi et al., 1998) and forest fragments (De Luca et al., 2003; Jones et al., 1987). The natural mammalian hosts are diverse, but small mammals and armadillos (Dasypus spp.) are thought to be strongly associated with Leishmania transmission (Lainson et al., 1979; Lainson & Shaw, 1989; B. S. Lima et al., 2013; Travi et al., 1994, 1998, 2002). Domesticated species, particularly dogs (Canis lupus familiaris ), are also important hosts (Courtenay et al., 2002; Gramiccia & Gradoni, 2005; Quinnell & Courtenay, 2009), and may play a key role as conduits of disease transmission if forest fragmentation leads to increased interactions between sylvatic and domesticated species. Given that many Leishmania species that cause leishmaniasis in humans are multi-host parasites (Roque & Jansen, 2014), their prevalent transmission pathways in deforested landscapes remain an open question. Kocher et al. (2022) recently found that mammal diversity, which declined with greater human footprint, was correlated with lower reservoir host abundance, lower prevalence ofLeishmania spp. in sandflies, but high sandfly abundance. The context for this work was the intrusion of small landholders into otherwise vast, continuous tropical forest across 19 forest sites in French Guiana. Here, we use a similar landscape epidemiology approach but, in contrast to Kocher et al. (2022), we sampled a gradient of forest loss and fragmentation that represents the most typical deforestation pattern witnessed across much of the Amazon and other tropical forest ecosystems worldwide, where >70% of all remaining forests is now within 1 km of a forest edge (Haddad et al., 2015) primarily due to agricultural expansion (Geist & Lambin, 2002). Within this context, we ask how large-scale deforestation influencesLeishmania hosts, vectors, their interactions, and pathogen prevalence. Specifically, we hypothesized that deforestation would be associated with increased bloodmeals derived from competent hosts that proliferate due to matrix subsidies and relaxed top-down control, and that sandfly vector abundance may increase as a result of higher host density in degraded forests.