Symposium Participants: Diana Outlaw
Examples
I and my collaborators have applied these techniques to several systems, Catharus thrushes (Outlaw et al. 2003), the avian family Motacillidae (Voelker and Outlaw submitted), Turdus robins (Voelker and Outlaw unpublished data), and Ficedula flycatchers (Outlaw unpublished data). I began asking questions about the evolution of migratory behavior within Catharus, where I specifically addressed Cox’s (1985) model. As I learned more about available techniques, the questions asked became more complicated as evidenced by the Motacillidae study.
Here, I will limit discussion to the completed studies, Catharus and Motacillidae.
Catharus
The specific objectives within the Catharus study were to:
- Examine the pattern of evolution of migratory behavior; i.e., are migratory species monophyletic? As I have already shown, the molecular data do not support this.
- Specifically test whether migratory species share ancestry with sedentary, subtropical (highland) species, a prediction of Cox’s (1985) model. Even though migration evolved independently several times in Catharus (objective 1), is there evidence of the same potential triggers for each of these events? This is supported by the data.
- Examine the age of migratory species. The data show that ages vary greatly (from five to > one million years).
While I have already presented most of the Catharus data (Figures 2, 3, and 5), I will now discuss how I provided support for Cox’s (1985) model through the integration of the techniques described above (Figure 6). The ancestral area reconstructions support a tropical origin for the genus, at about six million years ago, with several range expansions of sedentary species into the subtropics. Three independent speciation events, which are only evident because of a molecular phylogeny, at five, three and >one million years ago, led to the evolution of migratory species (“dispersal” into Eastern and Western North America from Mexico and/or Central America ). The entirely theoretical portion involves these migratory species returning to their ancestral home as temperate climates decline, where they are unable to winter probably due to competition with sedentary species. Hence, they must overstep these sedentary species, and winter in South America . This is consistent with wintering distributions (but see Outlaw et al. 2003 for a discussion of Catharus guttatus), and with ancestral areas.
Motacillidae
Within the Motacillidae, our specific objectives, using previously published phylogenies of Anthus (Voelker 1999), Motacilla (Voelker 2002), and Motacillid genera (Voelker and Edwards 1998), were to:
- evaluate ancestral character states within the family,
- evaluate the evolutionary precursor hypothesis (Levey and Stiles 1992; Chesser and Levey 1998) as applied to a cosmopolitan family, and
- examine the relationship between migratory behavior, breeding latitude and ancestral geographic areas.
A phylogeny that includes all genera and the majority of species within the family facilitated these objectives. Figure 7 presents the ancestral character state reconstruction (Maximum Likelihood) for both migration and habitat (essentially buffered and non-buffered), which provides a starting point for actual tests, both parsimony- and likelihood-based. Using these ancestral character states we asked several questions to address our second objective (and briefly report the results here; Table 1):
- Are gains of migration concentrated on branches that have open habitat as the ancestral character state (Concentrated changes test; Maddison and Maddison 1992)? Our analyses suggest no relationship (Table 1).
- Are migration and habitat evolving in an independent or dependent manner (Correlated evolution; Pagel 1999a, b)? Our results suggest that the two traits are evolving independently (Table 1).
- Is the rate at which migration state changes dependent on the habitat state at that node (Conditional evolution; Pagel 1999a, b)? There is no relationship between habitat state and the rate of migratory-state change (Table 1).
Thus, we found sedentary behavior and open habitat to be ancestral, but no relationship (regardless of algorithm) between migration and habitat. However, we found significant, positive results of the same set of questions/tests when looking at breeding latitude and migration (Table 1). In fact, Chesser (1998) and many others, have suggested the importance of latitude in affecting the distribution of migratory species. Many would suggest that these results seem intuitive, which they are, but even intuitive relationships should be rigorously tested.
Table 1. Results of evolutionary tests (From Voelker and Outlaw submitted) |
||
Independent/Dependent Variable |
Type of Test |
P Value |
|---|---|---|
Habitat/Migration |
Concentrated changes (Maximum Parsimony) |
0.455 |
Habitat/Migration |
Correlated evolution (Maximum Likelihood) |
0.18 |
Habitat/Migration |
Conditional evolution (Maximum Likelihood) |
0.40 |
Latitude/Migration |
Concentrated changes (Maximum Parsimony) |
<0.01 |
Latitude/Migration |
Correlated evolution (Maximum Likelihood) |
<0.01 |
Latitude/Migration |
Conditional evolution (Maximum Likelihood) |
<0.01 |
An additional aspect of this study is its hierarchical nature: we performed the same set of tests on each genus, groups of genera, and on the whole family. We did this particularly to examine taxonomic-level effects on the results of the analyses. We found the results to be consistent, which provides support for the approach itself at a variety of levels.
Summary of Approach
To summarize the methodology as applied to actual examples, let us revisit Motacillidae in a slightly different context (Figure 8). Figure 8 attempts to capture the integrative nature of the approach. Once again, I begin with a molecular phylogeny, which provides the foundation for evolutionary tests. I then examine both ancestral character states and trait evolution, whether I am simply interested in migration, or exploring a range of potentially related traits – those explicit factors that may drive migration in the first place.
Then, I add another layer by exploring the relationship between migration and the physical place of species origin (ancestral geographic area). In this example (Figure 8), many migratory taxa are of a southern origin (as are most taxa in the family), with dispersal (probably via the evolution of migration) into northern breeding grounds.
Let us briefly return to the utility of molecular phylogenies. With complete (all species) phylogenies, particularly of speciose genera, I can potentially quantify the role of behavior in speciation rate using various measures of tree shape (Agapow and Purvis 2002; Purvis et al. 2002). Migratory species may be less prone to speciation, particularly if there is little philopatry on the breeding grounds, but migration may lead to higher speciation rates as new habitats are colonized and genetic isolation occurs on the breeding grounds. Obviously, the roles of behavior in speciation require the biogeographical and ecological contexts of those “speciating,” but the methods I propose directly consider these contexts.
References
see link
