Allochronic and geographic reproductive isolation in an African songbird

Mortega KG, van Toor M, Illera JC, Apfelbeck B, Johnson PCD, Matthiopoulos J, Dawson DA, Grant BR, Burke T, Helm B. (manuscript)

Abstract

Ecological speciation - the divergence by local adaptation to different environments – is being studied extensively with a focus on the role of geographic isolation for population divergence. In contrast, allochronic isolation - the separation of populations by timing - remains mostly unconsidered when studying speciation. Populations might exhibit temporal barriers to gene flow because seasonal activities must be accurately fitted to local conditions to avoid mismatches with the environment, which can have severe fitness consequences. The evolution of rigid reproductive schedules, via endogenously generated biological rhythms, buffers organisms from responding to misleading, unpredictable variability of environments, but may thereby restrict gene flow. The African stonechat, Saxicola torquata axillaris, is a well-studied species with robust reproductive rhythms. We studied genetic, song, and morphological divergence between synchronic and allochronic breeding populations, and populations separated by geographic barriers. We modeled the spatial connectivity between populations to quantify the relative contributions of temporal and spatial isolation to population divergence. We found that allochronic and spatially isolated populations are genetically differentiated as judged from microsatellite loci and mitochondrial DNA. Allochronic, genetically distinct populations are spatially connected with no geographic or environmental barriers to gene flow, which strongly indicates population divergence as the result of allochronic isolation. The genetic structure is strongly concordant with patterns of song and morphology divergence of Stonechats.

Notably, females preferred vocal and visual stimuli associated with synchronously breeding males, thus promoting reproductive barriers by behavioral isolation. Our results indicate that the evolutionary dynamics of allochronic isolation, reinforced by sexual selection especially for culturally transmitted song may be key drivers for population divergence and ultimately speciation.

Introduction

Ecological speciation, i.e. divergence by local adaptation to different environments, has become recognized as an important mechanism of evolutionary change (Rundle and Nosil 2005; Nosil 2008;

Elmer and Meyer 2010). Like other forms of speciation, it involves reduced gene flow and eventual reproductive isolation between populations. While contributions to isolation by spatial aspects of the environment (e.g., parapatry, micro-habitat structure in sympatry) are under extensive investigation (Huxley 1943; Coyne and Orr 2004), little attention is being paid to temporal aspects. However, the existing evidence indicates that allochrony, i.e. divergence driven by different mating time (Coyne and Orr 2004; Yamamoto and Sota 2009), can occur rapidly even without geographic isolation. Aligning

biological processes with environmental rhythms (Durant et al. 2007) is thought to be under strong selection, which can act on the innate biological time-keeping mechanisms which in most organisms regulate reproductive timing. Studies of allochronic isolation in invertebrates have indeed linked its fast evolution to changes in biological rhythms (Schwarz et al. 2009; Fuchikawa et al. 2010). In tetrapods, we are aware of only two studies that provide robust evidence for accelerated diversification of populations by temporal isolation. Madeiran storm petrels, Oceanodroma castro, which breed sympatrically at different times of the year, display genetic isolation within the same colonies (Friesen et al. 2007). The second study, involving 57 New World bird species, relates allochronic speciation to different cycles in food availability associated with variation in precipitation regimes (Quintero et al.

2014). Although enticing, this study does not explicitly disentangle the relative roles of the spatial and temporal environment for allochrony, nor to they address the mechanistic basis, such as possible involvement of biological rhythms.

We address this knowledge gap using the Stonechat, Saxicola torquata, a taxon that is emblematic of endogenous, circannual rhythmicity (i.e., endogenous rhythmicity with a period length close to one year) (Dittami and Gwinner 1985; Gwinner and Dittami 1990). Stonechats are widespread songbirds with high genetic variation and well-known ecology and behavior (Dittami and Gwinner 1985; Illera et al. 2008; Woog et al. 2008; Mortega et al. 2015). In the wild, Stonechats display clear, and geographically distinct, reproductive rhythms, i.e. growth and regression of reproductive organs (Dittami and Gwinner 1985; Gwinner et al. 1995b; Helm 2009). In captivity, their distinct reproductive schedules diminished breeding success in crossbreeding trials and were shown to have a circannual basis (Gwinner and Dittami 1990; Gwinner et al. 1995b; Helm 2009). Circannual rhythms persisted for a decade in individuals kept under constant conditions and were also expressed by offspring that had never experienced environmental fluctuations (Sorek and Levy 2014). These genetically hard-wired timing programs could thereby provide potent evolutionary substrate for allochronic diversification. This possibility can be tested in Stonechat populations in East Africa, which show high genetic sub-structuring in a region of distinctly heterogeneous environmental seasonality (Jetz and Rahbek 2002; Illera et al. 2008; Woog et al. 2008). In East Africa, rainy seasons that are suitable for breeding recur within long-term predictable times of year, but these times differ regionally on fine spatial scales. Remarkably, captive Stonechats from East Africa showed particularly robust circannual rhythms, which could gate breeding to occur during the periodic rainy seasons instead of induction by spurious environmental variability (Dittami and Gwinner 1985; Sorek and Levy 2014). Stonechats could thus enhance their fitness if they aligned their circannual rhythms to local seasonality using reliable predictive cues. Even at the equator, such cues are provided by the annual changes in sunrise and sunset times, to which Stonechats have been shown to synchronize in appropriate phase position (Goymann et al. 2012).

While rhythms can feasibly act as barriers, movements and breeding dispersal may counteract local adaptation. Given the high avian dispersal ability, additional barriers might therefore be necessary to uphold the potential isolation by differences in local seasonality. Demonstrated behavioral barriers against gene flow include discrimination against non-local recruits. Specifically, signaling in the context of mate attraction may promote reproductive isolation through assortative mating and settlement, although heightened territory defense against local birds may lower their mating prospects (Grant and Grant 2002; Edwards et al. 2005; Price and Sol 2008; Podos 2010). In many songbirds, songs are a key component of signaling and are transmitted across generations via vocal learning, which often is possible only during the breeding season (Slabbekoorn and Smith 2002). Accordingly, song dialects, i.e. the unique repertoire of shared songs within a population, combined with female preference for a local dialect, may promote reproductive divergence (Marler and Tamura 1962;

Nottebohm 1969; Baker 1975; Searcy 1992), presumably, vocal traits in conjunction with selected traits of different sensory (e.g., visual) modalities, such as plumage patterns including patches. Recent findings from European Stonechats make local divergence through fine local discrimination abilities of song and morphology a distinct possibility, and because song is expressed seasonally, we suggest that it may promote allochronic isolation (Mortega et al. 2014).

Clearly, temporal aspects of the environment that could potentially drive population divergence act in combination with spatial aspects. Disentangling the contributions of the temporal and spatial environment provides a major challenge, but the rapid development of spatially explicit modeling techniques offers new avenues to address it. We estimated generalized functional responses from spatial data (Matthiopoulos et al. 2011), and implemented these to obtain resistance distances (McRae et al. 2008) describing the spatial connectivity between populations. Many habitat features, such as open space or mountain ranges, may function as additional behavioral or ecological barriers (Harris and Reed 2002).

We accounted for these possibilities by studying twelve local populations of Stonechats from an extensive area in East Africa (Fig. 7.1; total sample size=512), in which Stonechats have previously been known to breed at high altitudes, and up to six months apart. Using newly developed molecular tools (Mortega et al. 2015), we quantified genetic diversity using 18 microsatellite loci (nDNA) to assess recent gene flow, and the mitochondrial cytochrome b (mtDNA) to reconstruct the deeper phylogenetic divergence. Combining nDNA and mtDNA markers has improved the power for testing phylogenetic and phylogeographic hypotheses. We quantified breeding time, spatial information, song, morphology, and genetic variation for the twelve study populations. Using information on song and morphology, we examined potentially concordant divergence in genetic, acoustic, and visual

traits. We also tested acoustic and visual discrimination against conspecifics by playback of song and exposure to decoys. Tests were conducted to better understand behavioral mechanisms of reproductive isolation through sexual selection (Flinks et al. 2008; Illera et al. 2008), and to compare influences of distance, habitat features and synchronicity.

We used the data to distinguish between the following hypotheses: the null hypothesis is that geographical distance (D) between populations sufficiently determines genetic diversity, so that no further genetic substructure would be identified. This is contrasted with the alternative “Spatial Environment” (Es) hypothesis of greater explanatory power by models that include resistance calculated from spatially explicit landscape features, and the “Temporal Environment” (ET) hypothesis that proposes significant contributions of allochrony to population structure. Finally, we hypothesized that song may elicit stronger responses than morphological traits in both sexes because it may have diverged more rapidly, and because song is closely tied to seasonality.

Stonechat populations revealed considerable substructure and concordance in patterns of genotypic and phenotypic traits between three regions that related to allochronic and spatial features of the environment, implying that both restrict gene flow between populations.

Results