INTRODUCTION
Coevolution —the reciprocal adaptation between two or more interacting organisms (Hembry et al. 2014)— is considered to have played a central role in shaping the diversity of life on Earth (Laine 2009). Yet, empirical evidence that coevolution is the main mechanism driving reciprocal diversification among interacting clades lack alternative explanations (Medina et al. 2022). Most examples of diversification driven by coevolution involves cases of resource competition between species during adaptive radiations (reviewed in Hembry et al. 2014). In contrast, examples of diversification arising from mutualistic and antagonistic interactions at different trophic levels are comparatively rare (e.g. Hutchinson et al. 2017; Fuzessy et al. 2022). The lack of attention to such pervasive and relevant types of interactions not only leads to a biased view of the importance of coevolution in diversification processes, but also limits our understanding on how complex relationships between species arise within communities (Thompson 1994).
A major obstacle to generate empirical evidence for coevolutionary history in antagonistic/mutualistic interactions has been ruling out alternative explanations of diversification (Althoff et al. 2014; Medina et al. 2022). Although often taken as evidence for coevolution, reciprocal adaptation can emerge through adaptation by one of the interacting species alone or through neutral processes, such as genetic drift (Yoder & Nuismer 2010). Plant-animal interactions are recognized as important drivers of one-sided evolution (Strauss & Irwin 2004; Hembry & Weber 2020; Johnson et al. 2021). The correlation in features of interacting pairs, such as floral structures that match traits of pollinators (Nilsson 1988), color vision phenotypes that shape fruit color (Valenta et al. 2018), or defenses against predators that are met by counter defenses (Brodie & Ridenhour 2003), may also arise through correlated selection (Nuismer et al. 2010). Phylogenetic congruence between the interacting lineages, another major expectation of coevolutionary processes, is neither exempt of problems. While often interpreted as evidence of co-speciation, in nature, other types of evolutionary events often act concurrently, so a perfect cophylogenetic fit is rarely observed (Hutchinson et al. 2017; Fuzessy et al. 2022). Congruence may result from a combination of events unrelated to co-speciation, such as phylogenetic tracking (when one taxon tracks the evolution of its interaction partner), shared biogeographic histories (like vicariance, where common biogeographic events cause the subdivision of the distribution range of both interacting partners) or from parallel evolution driven by climate changes and other broad selective agents (Nuismer et al. 2010).
Demonstrating that interactions shape concomitant diversification on both sides has also been challenging due to two main constraints. First, interacting species on both sides are on a continuum of generalist-specialist interactions, which strongly affects the expected selective pressure each party exerts on the other, and hence the overall selective regime. In plant-animal interactions, phylogenetic tracking by adaptive switching behaviors, the ability of generalists to adaptively increase the consumption of one resource at the expense of decreased consumption of an alternative resource (Carnicer et al. 2008), is expected to be particularly relevant in creating phylogenetic congruences that can be confounded as coevolutionary diversification. This is because switching behaviors should not only facilitate animals to change feeding habits to new plants at the expense of others, but morphology and other constraints should also increase the chances that changes will favor plants that are closely related (and thus more similar) to those previously used as a resource (see Carnicer et al. 2008). The second constrain is that mutualism and antagonism themselves are only two ends of a single continuum, where no individual acts exclusively as an antagonist or a mutualist. For example, animals that feed on flowers can carry a significant amount of pollen in their body, acting simultaneously as florivores and pollinators (Kress et al. 1994). In the same vein, when feeding from fruits, many frugivores transport the seeds far from parent trees, acting as seed dispersers (Schupp et al. 2010). However, since not all seeds are defecated intact, seed dispersing animals can also act as seed predators (Norconk 2020). Due to these constrains and the difficulties to rule out alternative explanations, the role of coevolutionary processes in antagonistic/mutualistic interactions is currently backed by limited evidence (but see Hutchinson et al. 2017; Fuzessy et al. 2022).
While generating macroevolutionary evidence for coevolutionary diversification in antagonistic/mutualistic interactions remains challenging, recent advances in analytical tools developed for testing phylogenetic congruence provides a stronger framework to rule out alternative explanations of diversification (Balbuena et al. 2013; Blasco-Costa et al. 2021). A great advantage of these cophylogenetic methods (so called pattern-based methods, Dismukes et al. 2022) is the ability to use phylogenies where interacting species may have multiple pairs, which is the case of most plant-animal interactions. This allows to accommodate variation among the interacting species along the generalist-specialist continuum while considering phylogenetic uncertainties. With this framework, it is also possible to evaluate the contribution of individual links between interacting partners to the overall congruence (Hutchinson et al. 2017), and to test the extent to which the patterns of co-diversification and/or coevolution can be explained by phenotypic traits (Blasco-Costa et al. 2021).
Here, we adopt the above cophylogenetic framework to investigate macro-coevolutionary hypotheses for phylogenetic congruence of plant-lemur interactions in Madagascar. The flora and fauna of Madagascar provides an interesting study case to test such macro-coevolutionary hypotheses. About 82% of vascular plants (Callmander et al. 2011) and 84% of vertebrates (Goodman & Benstead 2005) are endemic, and hence have evolved under similar climatic conditions within a relatively small island, and have had ample opportunity for interaction and coevolution (Génin et al. 2022). Lemurs, in particular, diverged from Haplorrihini primates about 68 Ma, having originated from an ancestor on the African mainland, and occurring only in Madagascar (Pozzi et al. 2014). They arrived in the island earlier than other mammalian orders, resulting in a variety of adaptations to diverse niches and feeding habits, which include the consumption of fruits, leaves, flowers, nectar, bark and insects (Richard & Dewar 1991; Dewar & Richard 2012).
Previous work suggests that lemur colonization and subsequent diversification may have been facilitated by the divergence of endemic Malagasy angiosperms. Based on physiognomic and chronological reconstructions, Génin et al. (2022) concluded that the generalization of mutualistic interactions (such as frugivory and nectarivory) might have boosted the coevolution of modern lemurs and at least 10 new angiosperm families. However, observations were based on physiognomic and chronological reconstructions, and empirical evidence remains elusive both for pattern and process. Also, we still lack clear evidence for the potential of mutualistic and antagonistic interactions in generating congruence in the phylogenies of lemurs and Malagasy angiosperms. Here we address these issues by testing for cophylogenetic signals across a spectrum of plant-lemur interactions, including frugivory, folivory, florivory, granivory and nectarivory. To support a role for coevolutionary processes, we first expect phylogenies of lemurs and plants to be congruent. This is to be expected if mutualistic and antagonistic interactions shape reciprocal evolution regardless of interaction type. We also expect to identify putative coevolving phenotypic traits that may have propelled diversification (Blasco-Costa et al. 2021). If both predictions met, we finally expect to find a positive correlation between phylogenetic distances and dissimilarities in sets of interacting partners for both lemurs and plants. This would help to rule out the possibility that the observed cophylogenetic patterns were driven by phylogenetic tracking or vicariance instead of coevolution (Kahnt et al. 2019).