To study how interactions between two closely related species of nightingales affect their phenotypic evolution, we analyzed patterns of morphological variation in allopatric and sympatric populations of both species. Our analysis revealed two main patterns of morphological change. First, nightingales have converged in overall body size (as reflected in PC1) in sympatry. Second, nightingales diverged in relative bill size when compared to body size (as reflected in PC3) in sympatry. This divergence was asymmetric and was caused mainly by increased bill size in L. megarhynchos. Closer analysis of morphological variation along geographical gradients revealed that the convergence in overall body size was mainly caused by increasing body size with increasing latitude (Figure 3), a phenomenon known as Bergmann's rule [59–61]. Interspecific interactions did not have a significant effect on the convergence in body size (Table 3). On the other hand, interspecific interactions contributed significantly to the divergence in relative bill size even after controlling for the effects of geographical gradients (Table 3). Below, we argue that the observed divergence in relative bill size is most likely caused by interspecific competition for food resources and discuss how this ecological character displacement might facilitate speciation in nightingales.
Evidence for ecological character displacement in nightingales
Schluter & McPhail  summarized six criteria for demonstrating the occurrence of ecological character displacement. (1) The pattern should not occur by chance. (2) Sites of sympatry and allopatry should not differ greatly in food, climate, or other environmental features affecting the phenotype. (3) Morphological differences should reflect differences in resource use. (4) There must be independent evidence for competition. (5) Enhanced differences should result from actual evolutionary shifts, not from the biased colonization and extinction of similar-sized individuals. (6) Phenotypic differences should have a genetic basis. Meeting all of these criteria is usually quite difficult, and there are surprisingly few studies where alternative explanations for sympatric divergence have been ruled out and interspecific competition for food resources has been proven as a causal mechanism [12, 40, 62, 63]. In this study, we assemble evidence that at least partially satisfies four of these criteria (2-5).
Character displacement is typically demonstrated as a greater between-species morphological difference in sympatry than in allopatry. Such a pattern can, however, also arise if sites of sympatry and allopatry differ in environmental features affecting the phenotype (criterion 2). This is often caused by environmental gradients across species ranges. Our study area is located within central Europe, for which southwest-to-northeast climatic gradients are characteristic . We sought to disentangle the effects of environmental gradients and interspecific interactions by incorporating geographical variables (latitude and longitude) in statistical models. This approach removes the effects of geographical gradients in all directions. We found that the interspecific interactions contribute significantly to the enhanced differences in bill morphology in sympatry even when the effects of environmental gradients were controlled for. It is thus unlikely that the observed changes in bill morphology are caused by different environmental features or by different food supplies in sympatric and allopatric regions.
If phenotypic differences in sympatric populations are caused by ecological character displacement, they should reflect differences in resource use (criterion 3). Bill morphology is generally closely linked to resource use in birds  and determines the type and size of the food as well as feeding strategies [11, 12]. This is very likely also true for nightingales. Although both species of nightingales have a similar diet in general - they feed on small invertebrates on the ground in dense shrubby vegetation  - it is possible that minor differences in diet have evolved between the species in sympatry. These differences could be caused by either a separation of their feeding niches in sites where both species co-occur or by different food supplies in different microhabitats . Territories of L. luscinia tend to occur in wetter sites in the region of sympatry, while L. megarhynchos is more frequent in drier places, probably due to displacement by interference competition [43, 48]. In addition, Stadie  observed slightly different feeding strategies of the two species in sympatry. Whereas L. megarhynchos fed almost exclusively on the ground, L. luscinia was able to catch flying insects and was observed more frequently foliage gleaning . These observations suggest that both niche separation within the same habitat and habitat segregation occur in sympatric populations of nightingales and might contribute to bill size divergence in sympatry.
Bill morphology in passerines could also be affected by song characteristics, such as frequency, harmonic content and temporal patterning [67–70]. Thus, an alternative explanation for bill size differentiation in nightingales would be a change of song in sympatry. Such change could be driven for example by selection against maladaptive hybridization, a phenomenon known as reproductive character displacement or reinforcement. This has been documented in African Tinker birds . However, such an explanation is unlikely in this study system since song convergence rather then divergence occurs in sympatric populations of nightingales [43, 47, 71]. Moreover, song convergence in nightingales is caused by song change in L. luscinia, but not in L. megarhynchos, in contrast to the pattern that we observed in bill morphology.
Independent evidence for competition between species needs to be demonstrated to make a compelling case for ecological character displacement (criterion 4). Both species of nightingales have very similar habitat requirements  and show interspecific territoriality in sympatry . Moreover, playback experiments have demonstrated that males of both species respond aggressively to heterospecific songs in sympatry . This suggests that interspecific competition for resources is present in the two nightingale species. The role of interspecific competition in bill size divergence in nightingales is also supported by the observed asymmetry in morphological change in our data. Asymmetrical character displacement is expected if one species suffers higher costs during interspecific interactions; this species should diverge more than the other species . In this study system, L. megarhynchos is the weaker competitor as is suggested by four observations. First, L. megarhynchos has a smaller body size compared to L. luscinia . Second, L. megarhynchos shows partial habitat shift in sympatry . Third, L. megarhynchos responds aggressively to the heterospecific song less often than L. luscinia . Fourth, the zone of sympatry is slowly moving in the south-west direction towards the area of L. megarhynchos, which could be the result of dominance of L. luscinia in interspecific competition . In accordance with the lower competitiveness of L. megarhynchos, this species shows significant shift in bill size between sympatry and allopatry, while the bill size of L. luscinia does not differ between the regions. This result is consistent with the idea that interspecific competition drives the bill size differentiation in nightingales.
Finally, character displacement should result from a true evolutionary shift, not from biased colonization and extinction of similarly sized individuals (criterion 5). For example, it is possible that sympatric regions were colonized preferentially by L. megarhynchos with large bills or that L. megarhynchos with small bills went extinct in sympatry due to reasons other than competitively mediated selection. We addressed this question by comparing the distributions of bill size values (as reflected in PC3 residuals after removing the effects of latitude and longitude) in sympatric and allopatric populations within the same species. We found that the non-outlier range of bill size values for sympatric population of L. megarhynchos reaches beyond the non-outlier range observed for allopatric population of the same species (Figure 5). This suggests that the observed increase in bill size in sympatric population of L. megarhynchos is caused by in situ evolution of a novel phenotype and thus represents a real evolutionary shift rather than biased colonization and extinction of individuals with certain phenotypes.
These observations suggest that ecological character displacement is likely to be the causal mechanism underlying morphological differences between sympatric populations of the two nightingale species. Nonetheless, several issues still need to be addressed to provide more direct evidence for ecological character displacement. First, the observed divergence in bill morphology should be demonstrated on additional independent populations to rule out the possibility that the pattern is caused by chance (criterion 1). Second, a direct link between bill size and food preferences should be established (criterion 3). Third, the relationship between bill size and the level of interspecific competition in sympatric populations should be demonstrated (criterion 4). Finally, it remains to be shown that the divergence of sympatric populations is genetically based (criterion 6), although some non-genetic changes may also reflect ecological character displacement as discussed below.
Ecological character displacement in the face of gene flow: result of natural selection or phenotypic plasticity?
Ecological character displacement has often been regarded as a post-speciation event that occurs after the completion of reproductive isolation between incipient species [1–3, 13, 38]. In species where hybridization is common, interspecific gene flow can hinder ecological differentiation. In this study, we found that at least 3% of sympatric nightingales represent hybrids. In addition, gene flow between the species has been documented at multiple loci . This raises the question of how morphological divergence in sympatric populations of nightingales is maintained and why it is not erased by interspecific gene flow.
One possible explanation is simply that natural selection has a stronger effect on allele frequencies at loci that are responsible for bill size variability than does the rate of interspecific gene flow. This can be thought of in the context of models of migration-selection balance. Alleles at loci controlling bill size will be introduced due to gene flow from the sister species, and will be removed due to selection. Under a number of simplifying assumptions, the equilibrium frequency (q) for a dominant deleterious allele introduced by migration at rate m and removed by selection of magnitude s is given by q = m/s . In nightingales, most hybridization may not lead to gene flow since F1 females are sterile . In our study, only one bird was a later-generation hybrid. If we take this as a very rough upper estimate of the level of gene flow (m = 0.01), then a 10% selective cost (s = 0.1) would be sufficient to keep introduced alleles at a relative low frequency (i.e., 10%). This very rough calculation is only meant to illustrate that the degree of gene flow is sufficiently low that strong selection could still maintain different allele frequencies for traits of ecological importance.
In fact, previous work suggested that this sort of selection regularly acts against alleles introduced by migration between these species . That study showed that introgression between the nightingale species is significantly lower on the Z chromosome than on the autosomes, suggesting that selection acts against mis-matched Z-linked loci . Indirect estimates of the overall migration rate from patterns of DNA sequence variation analyzed under an isolation-with-migration model  were on the order of 10-7 , many orders of magnitude lower than the proportion of later-generation hybrids (10-2) observed in this study. This large difference suggests that many hybrid individuals may not contribute substantially to gene flow, perhaps because they have lower fitness. If so, then even weak selection might be sufficient to drive the evolution of bill shape differences in sympatry.
Why do alleles that increase bill size in sympatric L. megarhynchos not spread into allopatric populations? One possible explanation is that a larger bill is less optimal then the ancestral pre-displacement phenotype in allopatry. Indeed, character displacement might represent a "best-of-a-bad-situation", sensu , in that it lessens interspecific competition, but at a cost of reduction in other fitness parameters . Such fitness trade-offs can generate a selective barrier to gene flow between sympatric and allopatric populations because individuals from either population will be disadvantaged in the alternate population .
Phenotypic plasticity could provide an alternative explanation for the maintenance of phenotypic differences in the presence of gene flow. Phenotypic plasticity is the ability of an organism to change its phenotype in response to environmental stimuli [79, 80]. Since this is a non-genetic response, interspecific gene flow should not affect morphological differences caused by phenotypic plasticity. Recent evidence suggests that phenotypic plasticity in bill morphology can occur in response to poor conditions during development. Gil et al.  have demonstrated that nestlings of the Spotless Starling (Sturnus unicolor) develop larger bills and smaller body size in poor feeding conditions. Bill size of nestlings (especially the gape width) is an important determinant of food distribution among nestlings . Gil et al.  thus suggested that growth of the bill could be favored over growth of overall body size when feeding conditions worsened. In nightingales, L. megarhynchos, which is the weaker competitor, might be forced to low-quality territories with reduced food availability in sites where both species co-occur. Nutritional stress could then cause increased bill size in nestlings and, at maturity, of the adult birds.
Morphological differences caused by phenotypic plasticity are traditionally not considered character displacement, since they do not have a genetic basis (although the tendency to express different phenotypes in varying environmental conditions can be determined genetically) and thus cannot be inherited (criterion 6). Interesting exceptions represent cases where plastic phenotypic change is transmitted to the next generation through maternal effect [83, 84]. Recent studies, however, indicate that if plastic phenotypic change is adaptive (for example if it leads to change in food choice, which in turn reduces interspecific competition), it can eventually be stabilized by the evolution of genetic differences through a process known as genetic assimilation . Phenotypic plasticity could thus facilitate the evolution of character displacement in the presence of gene flow . Further developmental studies of nightingales raised on low and high quality food should provide more insight into the proximate mechanisms responsible for the divergence in bill size in these species.