Data Accessibility:
No new data were generated. Code needed to reproduce results of the manuscript are available at https://github.com/em-bellis/StrigaMacroecologyMS

Abstract:

Fitness responses to environment can shape species distributions, though opposing eco-evolutionary processes can obscure environmental effects. For example, host specificity influences parasite dynamics, but is unclear how specialization of individual parasites or populations scales up to continental distributions. Here, we develop a macroecological framework to determine how host community structure affects continent-scale specialization in Striga hermonthica, an African parasitic plant of cereal crops. We find regional abundance of hosts in cultivated cereal communities is associated with parasite specialization observed in experiments. Moreover, abiotic environment at location of origin predicts parasite performance on pearl millet and sorghum but not maize, possibly due to the shorter coevolutionary history for maize andStriga . Our study demonstrates that patterns of parasite local adaptation to host communities can emerge at continental scales and that differential environmental tolerances of hosts indirectly shape the distribution of specialist and generalist parasites.

Introduction:

Two central questions in biology are what maintains diversity in environmental responses within species and what determines the distribution of organisms across environmental gradients? These two questions are intimately related but often studied separately. A common hypothesis is that adaptation by populations to different local environments is a major source of diversity in environmental responses within species (Clausen et al. 1940; Kawecki & Ebert 2004). Local adaptation is driven by genetic variation in fitness responses to environment (“environmental preferences”) and thus local adaptation may shape abundance patterns across environments. It is often hypothesized that a species’ relative abundance in different habitats reflects environmental preferences, but there are many empirical studies where additional processes (e.g. dispersal limitation, Pulliam 1988) obscure or counteract environmental effects.
Local adaptation is defined as a genotype-by-environment interaction for fitness such that home genotypes have higher fitness than foreign genotypes (Kawecki & Ebert 2004). Local adaptation is common but not ubiquitous, and often not detected in experiments (Leimu & Fischer 2008; Hereford 2009; Koskella 2014). The scale of environmental heterogeneity relative to gene flow is one factor determining whether populations adapt to local conditions, evolve generalist strategies (Slatkin 1973; Penczykowski et al. 2016), or specialize on a single environment (Brown & Pavlovic 1992). For example, local adaptation is favored when the size of habitat patches is larger than the characteristic scale of gene flow, while smaller habitat patches may favor adaptation to the average of environments encountered, due to the unpredictability of environments inhabited by offspring (Slatkin 1973; Fig. 1). This pattern is readily apparent from empirical studies of host-parasite metapopulations, where the scale of parasite adaptation to local host genotypes depends strongly on habitat configuration and relative scales of gene flow (Thrall & Burdon 1997; Thrall et al. 2002; Laine 2005; Koskella et al. 2011; Tack et al.2014). However, it is unclear how local adaptation in metapopulations scales up to determine species distribution and abundance (Petersonet al. 2019).
A simple hypothesis about distributions is that population size is greatest where environmental conditions are most favorable (Brown 1984; Fig. 1). Yet organisms are often excluded from ideal environments by negative biotic interactions such as competition or predation (Buckley & Roughgarden 2005). Organisms can also persistently occur in poor environments due to immigration (i.e. source-sink dynamics, Pulliam 1988). Similarly, high extinction rates of parasite populations followed by recolonization, or high adaptability of hosts, can also limit parasite adaptation to their local hosts (Kaltz & Shykoff 1998). As a result, recent studies have questioned the assumption underlying many macroecological studies that patterns of environmental distribution reflect individual fitness responses to environment (Osorio-Olveraet al. 2019; Holt 2020). Relatedly, parasite fitness may not always be highest in environments where they are most abundant, though several studies suggest parasite abundance is greater on preferred hosts (Krasnov et al. 2004; Poulin 2005). Higher abundance may translate to higher frequencies of parasite occurrence across sampled localities (Poulin et al. 2012), but this trend is rarely linked to patterns of local adaptation.
To address macroecological hypotheses of resource specialization, parasite systems are perhaps some of the most promising (Stephenset al. 2016). For example, the growing availability of host-parasite occurrence data has enabled continent-scale investigations of host specificity (Fecchio et al. 2019; Wells et al.2019), defined as the number or diversity of hosts a parasite can infect (Wells & Clark 2019). In contrast, host specialization (in the Grinnellian sense) refers to variance in species’ performance across a range of environments (Futuyma & Moreno 1988; Devictor et al.2010). Compared to host specificity (Fecchio et al. 2019), macroecological studies of host specialization are scarce, perhaps due to the need for quantitative measures of parasite performance which are generally more difficult to obtain than occurrence data.
Here, we investigate continent-scale patterns of host specialization in the parasitic plant Striga hermonthica. S. hermonthica is widespread across diverse abiotic environments in East and West Africa where it parasitizes grasses including the staple crops maize, pearl millet, and sorghum and is a major constraint to food security (Runo & Kuria 2018). Compared to pearl millet (Pennisetum glaucum ) and sorghum (Sorghum bicolor ), which both have centers of domestication in Africa, maize is a relatively recent host and has few natural resistances to Striga (Rich & Ejeta 2008; Timko et al. 2012). Hosts are characterized by mating systems from highly outcrossing (pearl millet and maize) to predominately selfing (sorghum) and a diversity of abiotic requirements. Unlike many microbial pathogens and endoparasites, parasite generation times are similar to their hosts (~1 year) leading to more balanced coevolutionary dynamics. Parasitic plants are also large and conspicuous, so excellent occurrence data are available from natural history collections. Host specialized as well as generalist populations are known (Parker & Reid 1979). However, the eco-evolutionary determinants of specialization are poorly characterized, despite detailed knowledge at the molecular level on parasitic plant response to different hosts (Honaas et al.2013; Johnson et al. 2019; Lopez et al. 2019).
We synthesize previous experiments on host specialization, combined with continental scale host and parasite distribution data, to ask three questions: Q1) Does regional abundance of a host crop lead to local specialization by parasites?, Q2) Is host specialization associated with patterns of parasite occurrence on different hosts?, and Q3) Do abiotic environmental gradients shape the distribution of specialist and generalist parasites, suggesting that future dynamics of specialization vs. generalization may be predicted by abiotic change?

Materials and Methods: