Molecular Systematics and Historical Biogeography of the Rock- Thrushes (Muscicapidae: Monticola)

Jun 10, 05:56 AM

Current Headlines: By Outlaw, Robert K Voelker, Gary; Outlaw, Diana C

ABSTRACT.- The genus Monticola consists of 13 putative species with distributions throughout Eurasia and the sub-Saharan region of Africa. As such, this genus provides an excellent model with which to explore historical intercontinental movements and forces driving speciation in southern Africa. To address these questions, we reconstructed a hypothesis of species relationships using the mitochondrial ND2 and cytochrome-b genes. Monticola forms a well- supported, monophyletic clade within the avian family Muscicapidae. Our results support previous studies suggesting that the Malagasy genus Pseudocossyphus be subsumed into Monticola, and suggest that several of the Malagasy species (notably M. bensoni and M. erythronotus) are not valid. Sequence data, along with morphological and distributional evidence, support the elevation of M. pretoriae to species status. Historical biogeographic analyses suggest an area of origin for Monticola in the arid region of northern Africa plus Saudi Peninsula or the African savanna, or both. Determination of speciation timing suggests that Monticola arose ~5.5 mya, with subsequent lineage splits occurring throughout the Pliocene and Pleistocene. We propose that climate-driven ecological vicariance as well as dispersal were important in the biogeographic history of this group and are responsible for present-day species relationships and distributions. Received 14 September 2005, accepted 11 May 2006.

Key words: Africa, Asia, biogeography, molecular systematics, Monticola, speciation.

Systematique moleculaire et biogeographie historique des monticoles (Muscicapidae: Monticola)

RESUME. - Le genre Monticola est compose de 13 especes reparties a travers l'Eurasie et l'Afrique sub-saharienne. Ce genre fourni un excellent modele pour permettre d'explorer les mouvements intercontinentaux historiques et les forces qui commandent la speciation dans le sud de l'Afrique. Pour aborder ces questions, nous avons reformule une hypothese des relations entre les especes en utilisant le gene mitochondrial ND2 et le gene du cytochrome-b. Le genre Monticola forme un clade monophyletique bien etabli dans la famille Muscicapidae. Nos resultats vont dans le meme sens que des etudes anterieures qui suggerent que le genre malgache Pseudocossyphus doit etre subsume en Monticola et que plusieurs especes malgaches (notamment M. bensoni et M. erythronotus) ne sont pas valides. Les donnees sequentielles, combinees a la morphologie et la repartition, appuient l'elevation de M. pretoriae au statut d'espece. Les analyses biogeographiques historiques suggerent que la region aride du nord de l'Afrique + la peninsule saoudienne ou la savane africaine (ou les deux) correspondraient a l'aire d'origine de Monticola. La determination de l'evolution de la speciation dans le temps suggere que Monticola soit apparu il y a environ 5,5 millions d'annees et que des separations subsequentes des lignees se sont produites pendant le Pliocene et le Pleistocene. Nous proposons que la vicariance ecologique causee par le climat ainsi que la dispersion etaient importants dans l'histoire biogeographique de ce groupe et qu'ils sont responsables des relations entre les especes et de leur repartition actuelles.

FEW STUDIES HAVE examined the biogeographic patterns of birds distributed in arid regions of Asia and Africa (but see Voelker 1999). Attention has primarily been paid to forest birds whose ranges can ostensibly be tracked through time with well-documented paleoclimatic reconstruction. However, climatic cycles are known to affect arid regions as well (see Zink et al. 2001).

The rock-thrushes of the genus Monticola provide an opportunity to evaluate biogeographic hypotheses concerning the effect of historical climate change on largely arid-adapted species. Monticola is a genus of Old World thrushes in the avian family Muscicapidae, currently considered to comprise 13 species (Sibley and Monroe 1990, Clement 2000). Of the 13 species of Monticola, 3 are found in southern and eastern Asia, 8 are distributed across sub-Saharan Africa and Madagascar, and 2 span the north African savanna and Eurasia. With the exception of these latter two species, most species of Monticola possess largely disjunct distributional ranges, with geographically overlapping species generally segregated by elevation. IntraAfrican and intra-Asian diversification, as well as African-Asian interchange, can be explored within Monticola, to add to the growing body of literature examining these processes.

Any attempt to reconstruct historical biogeographic patterns within a lineage, however, requires a robust estimate of phylogenetic relationships. We therefore present a comprehensive molecular analysis of species relationships within Monticola using mitochondrial DNA (mtDNA). Using this hypothesis of species relationships, we inferred historical movements to account for present-day distributions and approximated a maximum age of lineage splits within Monticola. These analyses allowed us to address the taxonomy of Monticola, which, as with many Old World genera, has had a turbulent history.

Taxonomic considerations.-There is some confusion within the current taxonomy of Monticola with respect to its placement in the family Muscicapidae, the placement of certain species in Monticola versus Pseudocossyphus, and the validity of particular species. Historically, Monticola was considered to be a member of the subfamily Turdinae (Muscicapidae), though recent molecular evidence supports the placement of Monticola in the Muscicapinae tribe Saxicolini (Voelker and Spellman 2004).

Considerable debate has surrounded the relationships between Monticola and the Malagasy genus Pseudocossyphus. Ripley (1952) and later Goodwin (1956) considered Pseudocossyphus a distinct genus. Subsequent treatments and opinions based on morphology and behavior, however, have suggested that Pseudocossyphus is invalid and that its member species belong in Monticola (Dowsett and Dowsett-Lemaire 1993, Morris and Hawkins 1998). A more recent study by Goodman and Weigt (2002) used mtDNA to reconstruct relationships within Pseudocossyphus; that study also suggested that Pseudocossyphus is invalid and placed its constituent species in Monticola.

Uncertainty has also been evident in the taxonomic status of these Malagasy species. Clement (2000) suggests three distinct Malagasy species: M. sharpei, M. bensoni, and M. imerinus. Goodman and Weigt (2002) suggested a taxonomic revision involving the elevation of M. erythronotus, previously considered a subspecies of sharpei, to full species. Prior to this, Morris and Hawkins (1998) had also separated M. erythronotus from M. sharpei on the basis of morphology. Monticola bensoni, previously separated from M. sharpei on the basis of morphology (Farkas 1971), was later found to show no substantial genetic differentiation and was suggested to be a pseudonym of M. sharpei (Goodman and Weigt 2002).

Several species designations within Monticola have been historically problematic. Sibley and Monroe (1990; see also White 1967, Farkas 1979) classified M. pretoriae as a distinct species, whereas several authors consider it merely a race of M. brevipes (Clancey 1968, Dowsett and Dowsett-Lemaire 1993, Clement 2000). Meise (1934, in Vaurie 1955) suggested that M. gularis and M. cinclorhynchus were conspecific. Vaurie (1955), however, in his assessment of the Palearctic species M. saxatilis, M. solitarius, M. rufiventris, M. gularis, and M. cinclorhynchiis, considered M. gularis and M. cinclorhynchus separate species.

Geographic distributions.-Monticola cinclorhynchus, M. rufiventris, and M. gularis occur in Asia, and all exhibit migratory behavior. Monticoln cinclorhynchus breeds from the Himalayas to Afghanistan, with a disjunct wintering range in the Western Ghats of India. Monticoln rufiventris and M. gularis occur primarily in China. An altitudinal migrant, M. rufiveiitris does not show disjunct wintering and breeding ranges, but has several populations sympatric with M. cinclorhynchus in the Himalayas, with most of its distribution in southeasern China. Monticola gularis breeds in northeastern China and Mongolia, with disjunct wintering ranges chiefly south of the range of M. rufiventris in extreme southeastern China and the Malay Peninsula.

Monticola solitarius and M. saxatilis are widespread and migratory, breeding throughout Eurasia and wintering in southeastern Asia and Africa. They show some overlap in breeding ranges, primarily in the western Palearctic and southwestern Asia. Although there is overlap in the African wintering ranges of M. solitarius and M. saxntilis, the latter does not winter in Eurasia or southeastern Asia.

The African species M. angolensis, M. rupestris, M. explorator, M. brevipes, M. pretorine, and M. nifocinereiis inhabit sub-Saharan Africa in disjunct, yet in some cases abutting, ranges (Peters 1964, Sibley and Monroe 1990, Clement 2000). Monticoln nngolensis is restricted to the Miombo woodlands of south-central Africa, and its range does not overlap with any other species of Monticoln. Monticoln rupestris and M. explorntor inhabit the mountain ranges of southern and eastern South Africa. These species exhibit almost complete range overlap, though M. rupestris extends to lower elevations than M. explorator. Monticola rupestris and M. explorator differ in habitat use, in that M. rupestris inhabits open, largely rocky terrain, whereas M. explorator is restricted to open montane grasslands. Although M. brevipes and M. pretorine seem to be separated geographically, with M. pretorine extending farther east, they are both found in the arid region of western and central South Africa, with M. brevipes extending northwest into Namibia. The ranges of M. brevipes and M. pretorine overlap in the eastern North Cape and western Northwest Provinces of South Africa. This overlap may allow for hybridization and is, thus, a possible point of contention over the taxonomic status of M. pretoriae discussed earlier. Monticola rufocinereus is restricted to the mountains of Ethiopia, Somalia, Uganda, and extreme northeast Tanzania, as well as the Asir Mountains of western Saudi Arabia. The range of M. nifocinereus overlaps with wintering populations of the Eurasian migrants M. saxatilis and M. solitarius in Africa, and M. solitarius in Saudi Arabia, but not with any of the exclusively African species.

MATERIALS AND METHODS

Sampling strategy.-We sampled 12 of the 13 putative species based on previous taxonomy (Sibley and Monroe 1990, Clement 2000); we were unable to obtain a representative of M. imeriniis. We received tissue, blood, or toepad clippings from museum collections of 29 taxa comprising the 12 representative species (see Table 1 for a complete listing of samples, museums, and collecting localities). Where possible, we acquired at least two individuals from each of the ingroup species with as disjunct collecting localities as we could obtain. Whole-genomic DNA was extracted from all tissue, blood, and toepad samples using the DNeasy tissue extraction kit (Qiagen, Valencia, California), following the manufacturer's protocols for animal tissue and whole-nucleated blood.

Sequencing.-We used polymerase chain reaction (PCR) to amplify the mitochondrial NADH dehydrogenase subunit 2 (ND2) and cytochrome- b genes. We performed whole-gene amplification and sequencing using the primers listed in Voelker and Spellman (2004). For degraded DNA, it was necessary to split the genes during amplification. To do this, we used the above primers, as well as the following internal primers designed for the cytochrome-b gene in our laboratory: H15234 (57prime;AGTGTAGGGTTG TCTACTGAGAA3') and L15372 (5'TACACGAA ACAGGATCAAACAACCCAC3'). We amplified all gene fragments in 50-[mu]L reactions using the following thermal regime: denature at 94[degrees]C for 2 min, followed by 35-40 cycles of denaturation at 94[degrees]C for 45 s, annealing at 50-54[degrees]C for 45 s, and extension at 72[degrees]C for 2 min. Following this was a final extension cycle of 72[degrees]C for 10 min and a 4[degrees]C soak. Negative controls were established for all reactions. All PCR product was purified using Qiaquick PCR purification kit (Qiagen) following manufacturer's protocols.

We performed 20-[mu]L cycle sequencing reactions using 20-40 ng of purified PCR product and Big Dye dye-labeled terminators (Applied Biosystems, Foster City, California) under the following thermal regime: 25 cycles of 95[degrees]C for 10 s, 50[degrees]C for 10 s, 60[degrees]C for 4 min. Reactions were purified using isopropanol precipitation (ABI protocols) and run out on an ABI377 automated sequencer using Long Ranger (Cambrex, East Rutherford, New Jersey) acrylamide slab gels. Sequences were generated at the Barrick Museum of Natural History at the University of Nevada, Las Vegas, and the University of Memphis.

We used SEQUENCHER, version 4.2 (Gene Codes, Ann Arbor, Michigan) to align 1,041 base pairs (bp) of ND2 and 998 bp of cytochrome b. Sequences from split-gene PCR of degraded DNA were lacking approximately 20-30 bp at the center of the gene, which was coded in SEQUENCHER as missing data. For each toepad sample, no fewer than two separate DNA extractions were performed, and each of these was sequenced separately. Sequences from these toepads were aligned and found to be identical. To ensure the accuracy of amplification of the ND2 and cytochrome-b genes, we verified that all sequence data were double-stranded and protein-coding, as referenced against the published Gallus gallus mitochondrial genome (Desjardins and Morais 1990). Also included in our data set for phylogenetic analysis are sequences of two individuals from a previous study by Goodman and Weigt (2002), obtained from GenBank. Sequences generated by the present study are deposited in GenBank under accession numbers EF434508-EF434561.

Phylogenetic analyses. - Four outgroup species, Cercomela familiaris, Myrmecocichla formicivora, Oenanthe monticola, and Thamnolea arnotti were selected by aligning Monticola cytochrome-b sequence data with a larger data set encompassing 33 muscicapid genera (Voelker and Spellman 2004, Voelker and Bowie unpubl. data). Models of DNA sequence evolution were evaluated using MODELTEST, version 3.06 (Posada and Crandall 1998) to determine the best-fit model for our data set. With respect to combining ND2 and cytochrome- b data, the incongruence-length-difference test is inappropriate, because both genes are mitochondrial and, therefore, linked (see Barker and Lutzoni 2002). For the combined ND2 and cyt-b data, MODELTEST's likelihood-ratio test (LRT) suggested the general time- reversible (GTR) model with number of invariable sites (I) and alpha shape parameter of the gamma distribution (Gamma) as the best-fit model, yet Akaike's Information Criterion (AIC) identified Tamura- Nei (TrN93) + I + Gamma as the best-fit model. Both of these models were incorporated into maximum-likelihood (ML) analyses using the phylogenetic software PAUP*, version 4.2 (Swofford 1999). Using parameters estimated for each model by MODELTEST, we initiated full heuristic ML searches using as-is addition of taxa and the TBR (tree bisection reconnection) branch-swapping algorithm. These searches were allowed to run to completion. Because both models converged on the same topology, we used the LRT (Huelsenbeck and Rannala 1997) to determine which model, if either, was a significantly better estimate of gene evolution parameters.

Support for the ML reconstructions was obtained using three techniques: ML bootstrap analysis using TREEFINDER (see Acknowledgments), maximum-parsimony (MP) bootstrap analysis using PAUP*, and Bayesian posterior probabilities using MRBAYES (Huelsenbeck and Ronquist 2001). All analyses were performed on the complete data set. Using TREEFINDER, we initiated an ML bootstrap analysis of 2,000 replicates and incorporated the model of gene evolution suggested by MODELTEST and the LRT (the result of an ML bootstrap using PAUP* [not shown] was entirely consistent with that of TREEFINDER). For MP bootstrap analysis, we first generated an uncorrected neighbor-joining tree. From this tree we estimated, using ML and empirical base frequencies, the gene- and codon- position-specific sequence characteristics, including transition:transversion (Ti:Tv) ratio, proportion of invariable sites (I), and among-site rate variation (a) (Table 2). The codon- position-specific Ti:Tv ratios were then incorporated into an MP bootstrap analysis in PAUP* via user-defined step matrices. The MP bootstrap analysis was initiated using random taxon-addition, TBR branch-swapping, and a full heuristic search of 1,000 pseudoreplicates. From the MP bootstrap analysis, we constructed a 50% majority-rule consensus tree.

One shortcoming of traditional phylogenetic support analyses when multiple genes are used is the inability to estimate different sequence characteristics for each gene. Therefore, partitioning the data by gene, MRBAYES estimated substitution model parameters (GTR + I + Gamma) for the ND2 and cytochrome-b genes individually. Starting from random trees, we initiated three individual runs of four Markov- chain Monte Carlo (MCMC) chains and 2 million generations each, sampling every 100 generations. Each run resulted in 20,000 trees and converged on the same topology. The first 50,000 generations (500 trees) from each analysis were removed as our "burn-in." The remaining 58,500 trees were combined, and a 50% majority rule consensus tree was generated. All nodes possessing 95% or greater posterior probability were considered well supported (Huelsenbeck and Ronquist 2001).

Historical biogeography.-To reconstruct the ancestral area and historical biogeography of the genus, we constructed an area cladogram based on our ML topology. Areas were defined using the full extent of breeding and wintering ranges of extant species of Monticola, as outlined in Clement (2000). Areas for our biogeographic analyses were based on and modified from the previously published areas of Voelker (1999) using current biome data (Meadows 1996, Brown and Lomolino 1998) and are illustrated in Figure 1. These areas are classified for the present study as (A) eastern plus southeastern Asia, (B) central Asian arid, (C) Himalayas, (D) southwestern Asia plus Indian subcontinent, (E) western Palearctic, (F) North African arid plus Saudi Peninsula, (G) African savanna, (H) South African arid, and (I) Madagascar. Areas in Figure 1 in which no Monticola species occur were eliminated from the historical biogeographic analyses.

For ancestral-area reconstruction, we used dispersal-vicariance analysis (DIVA; see Acknowledgments), weighted ancestral-area analysis (WAAA; Hausdorf 1998), and an ML approach using MESQUITE, version 1.06 (see Acknowledgments) and DISCRETE (see Acknowledgments). For each node within the phylogeny, we employed DIVA with no limit on the number of areas reconstructed as ancestral. Because this resulted in all areas being ancestral, we constrained the results using the "maxareas=2" command within the software. Because DIVA assumes a vicariant explanation for lineage splits, twr, this assumption will likely underestimate dispersal or range expansions throughout the history of the lineage (Voelker 1999). Weighted ancestral-area analysis (Hausdorf 1998) is a cladistic method of determining ancestral area that weights areas found in plesiomorphic positions on the tree more highly than those in apomorphic positions, because plesiomorphic areas are more likely to be ancestral ( see Bremer 1992). The weighted gain steps (GSW) and weighted loss steps (LSW) are calculated for each area by calculating 1/(GSW) or 1/(LSW) for each node on the cladogram. Determining the probability index (PI) for each area involves calculating the ratio of (GSW)/(LSW) of that area for each node. Areas with the highest PI for each node are those that are more likely part of the ancestral area for that node. To reduce the number of areas recovered as ancestral for each node, we followed Hausdorf (1998) in establishing a minimum PI threshold value of 0.2.

Finally, we employed an ML approach to ancestral-area reconstruction. Treating each area as a separate, binary (presence- absence) character, we first determined whether a symmetrical or an asymmetrical model of character evolution resulted in significantly different likelihood scores using LRTs and MCMC simulations (100 iterations) in DISCRETE. In a symmetrical model, characters have an equal rate of gain and loss, whereas an asymmetrical model estimates unequal gain and loss rates.

We reconstructed areas as ancestral states in MESQUITE, weighting plesiomorphic areas more heavily (like WAAA) and using branch lengths to reconstruct a model of evolution for each area. With this approach, the ML analysis returns a matrix of likelihood-based proportions (and scores) of the presence or absence of each area at each node. To confirm reconstruction of areas at the base of the phylogeny, we used the rootreconstruction feature in DISCRETE. This feature calculates the likelihood of character states at the root using the "local" method (Pagel 1999) calculations that optimize transition rates at each node of the phylogeny rather than applying the same transition rates throughout the phylogeny.

Molecular clock. - To reconstruct speciation timing, we applied a molecular clock to our cytochrome-b data. We recognize that there are many assumptions and limitations surrounding the use of a standardized avian mitochondrial molecular clock. However, clocks remain useful in establishing rough estimates of speciation timing (Garcia-Moreno 2004). Attempts to reconstruct the biogeographic history of a lineage without ascertaining speciation timing may prohibit incorporation of specific geologic and climatologie events in explaining the biogeographic history of a lineage. To determine whether Monticola lineages are evolving in a clock-like manner, we implemented the two-cluster test (TCT) of Takezaki et al. (1995). The test statistically determines whether the two daughter lineages of any node in the phylogeny are evolving in a clock-like manner and provides a confidence probability (CP = 1 - [P value]) where only CP values >95% reject clock-like behavior. Although particular nodes may not be identified as clock-like, it is reasonable to infer an approximate divergence time for the node if it is bracketed on either side by clock-like nodes (Voelker 1999).

Before being analyzed, the data set was restricted to a single representative from each species (with the exception of bensoni- sharpei; see below), because we were not concerned with determining the timing of intraspecific splits. We applied the TCT using LINTREE (Takezaki et al. 1995) with uncorrected (P) sequence distances and the K2P + Gamma correction of Fleischer et al. (1998). For the analysis using the K2P + Gamma correction, we restricted the data set to the first 675 bp of cytochrome b, because this was the amount of data used in the original study (Fleischer et al. 1998).

Although the TCT in LINTREE can also provide approximate node ages, any nucleotide site with missing data for any taxon is removed from the analysis. The elimination of nucleotide sites from the analysis results in ignoring potentially valuable information from pairwise comparisons where there are, in fact, no missing data. Because we had some missing data, owing to the degraded nature of the DNA from our toepad samples, we calculated node ages using uncorrected (2% per million years) pairwise distances on 998 bp of cytochrome b and K2P + A (1.6% per million years) corrected pairwise distances restricted to the first 675 bp. Uncorrected distances (2% per million years) have been shown to be an appropriate rate of avian mitochondrial gene evolution between relatively divergent taxa (i.e., >2% sequence divergence; Ho et al. 2005). Uncorrected values are reported here.

RESULTS AND DISCUSSION

SYSTEMATICS AND TAXONOMY

Sequence characteristics of ND2 and cytochrome b were consistent (Table 2); therefore, the two genes were combined in our analyses. Phylogenetic analyses using both the TrN93 + I + Gamma and GTR + I + A models of gene evolution resulted in the same hypothesis of species relationships, with likelihood scores of-11247.99786 and - 11252.51355, respectively. Results of an LRT were significant (LR = 9.032, CP = 7.81, P < 0.05); the TrN + I + Gamma model, with fewer parameter estimates, resulted in a higher likelihood score and, thus, we chose this model of gene evolution for our phylogenetic reconstruction.

Species of Monticola form a well-supported, monophyletic clade within the family Muscicapidae (Fig. 2). Results of our phylogenetic analyses support the placement of Pseudocossyphus species within Monticola (Fig. 2). With the exception of M. rufocinereus, Monticola is made up of two geographically distinct clades, one Asian and the other largely African. The basal placement of M. rufocinereus in our phylogeny (Fig. 2) and the degree of differentiation between our two representatives (-5% sequence divergence) presents an interesting scenario. Monticola rufocinereus is smaller than other species of Monticola and lacks the characteristic blue plumage. Its range consists of an array of small disjunct distributions restricted to upland and semimontane forests of the Asir mountains of southwest Saudi Arabia and the mountains of Eritrea, Ethiopia, Kenya, and Somalia in eastern Africa. Isolation of M. rufocinereus populations in these small disjunct ranges may have resulted in the split between M. rufocinereus and other Monticola species. The high degree of genetic differentiation between our two representatives collected from disjunct mountain ranges in Ethiopia and Kenya (Table 1) suggests that these isolated montane populations have been separated for a substantial period (-2.5 mya). Given that both M. rufocinereus samples are toepads, we acknowledge that some of this divergence may be an artifact of DNA degradation and, thus, incomplete sequence data. Although we did not have access to an Arabian sample, it has been proposed that the populations in southwestern Saudi Arabia constitute the subspecies M. sclnteri (see Clement 2000). Further analysis including representatives of disjunct populations of M. rufocinereus may reveal further intraspecific genetic structure (i.e., phylogeographic structure).

To address the historical classification of M. gularis and M. cinclorhynchiis as conspecific (Meise 1934, in Vaurie 1955), we forced a topology with M. cinclorhynchus and M. gularis as sisters, with M. rufiventris sister to them, and applied the Shimodaira- Hasegawa (S-H) test (Shimodaira and Hasegawa 1999) using RELL approximation and 1,000 bootstrap replicates. Results of this test indicated that the pairing of M. cinclorhynchus and M. gularis does not result in a significantly worse estimate of the tree (P = 0.2777). Given the short internodes between these species and the conservative nature of the test, this result was not unexpected. However, all of our phylogenetic hypotheses converged on the gularis- rufiventris sister relationship with very high bootstrap and posterior-probability support (Fig. 2). We support the assessment of Vaurie (1955) that M. cinclorhynchus and M. gularis are not conspecific.

With respect to former species of "Pseitdocossyphus," our analyses support the molecular assessment made by Goodman and Weigt (2002) that M. bensoni and M. sharpei are not specifically distinct (Fig. 2). Based on the classification of M. bennoiri as a distinct species (Farkas 1971), we forced M. bensoni and M. sharpei to be reciprocally monophyletic and compared this tree to our "best" estimate of relationships via the S-H test. This resulted in a significantly worse estimate of these relationships (P = 0.05), which suggests that M. bensoni and M. sharpei are genetically indistinct. Our results are discordant with the recognition of three distinct species, M. erythronotiis, M. imeriniis, and M. bensoni-M. shnrpei. Inclusion of Goodman and Weigt's (2002) M. erythronotus ND2 and cytochrome-/) data (Fig. 2, as M. sharpei [3]) suggests that there is no evidence to support the elevation of M. erythronotiis to full species and that it should remain a subspecies of M. sharpei. Monticola imerinus is the single species that we were unable to attain. Incorporating Goodman and Weigt's (2002) limited M. imerinus cytochrome-b data into our data set (not shown) provided no evidence that M. imerinus is genetically distinct. Although M. imerinus consistently grouped within the Malagasy clade, its specific position changed with each analysis. Given that Goodman and Weigt's (2002) data consist of only 307 bp of cytochrome-b data and we were unable to contribute any new data, we believe that their suggestion that M. imerinus is "specifically distinct" is premature. Full resolution of Malagasy species relationships in this study may not be possible because of the rate of evolution of the specific markers used here. We suggest that an examination of Malagasy species relationships using more rapidly evolving markers in population- level analyses may give more insight into their true relationships. Finally, there has been some debate over the taxonomic status of M. pretoriae, and whether it should be recognized as specifically distinct or remain a subspecies of M. brevipes (White 1967, Farkas 1979, Sibley and Monroe 1990). Some of this confusion stems from the presumed distributional extents of these subspecies. It has been suggested that these races show morphological intergradation where their ranges overlap in north-central South Africa (Clement 2000). Despite the fact that our M. brevipes samples were collected in this area of intergradation, they were clearly identified as nominate M. brevipes on the basis of plumage observations; all M. pretoriae samples were collected well to the east of the presumed intergradation zone. Judging from morphology, reciprocal monophyly, and high sequence divergence (-1.6%), our results support Sibley and Monroe's (1990) classification of M. pretoriae as a distinct species (Fig. 2). Further sampling could improve our knowledge of species range limits as well as elucidate any potential interactions across the putative zone of intergradation.

HISTORICAL BIOGEOGRAPHY

Ancestral area. - Results from our DIVA analysis suggest that the ancestral area of the genus encompasses northern arid Africa plus Saudi Peninsula and either eastern plus southeastern Asia, Himalayas, southwestern Asia plus India or the African Savanna (Fig. 3). This is not surprising, considering that Monticola is composed of an Asian clade and a primarily African clade, and that DIVA assumes a vicariant origin. We suggest that northern arid Africa plus the Saudi Peninsula in combination with eastern plus southeastern Asia or the Himalayas is unlikely, because these areas are not adjacent.

Results from WAAA suggest a North African arid plus Saudi Peninsula ancestral area for Monticola. Although WAAA cannot eliminate the possibility of the Himalayas, African savanna, southwestern Asia plus India, and eastern plus southeastern Asia, their PI values are considerably lower than that of north African arid plus Saudi Peninsula (Table 3 and Fig. 3). On the basis this result, we interpret WAAA as reconstructing north African arid plus Saudi Peninsula as the ancestral area.

In ML ancestral-area analyses, the asymmetrical model of character evolution did not have a significantly higher likelihood score than the symmetrical model for all areas, and we therefore used the symmetrical model in all reconstructions. These analyses reconstruct areas as significant (present) at terminal nodes, but not at deeper nodes within the phylogeny. Terminal reconstructions in ML are consistent with those from WAAA, and the basal uncertainty in these analyses certainly reflects the possibility of increased character change throughout long internodes within the tree (Pagel 1999). Results of root reconstruction analyses indicate the African savanna as the only significant ancestral area. Results of ML-based analysis are not incongruent with the parsimony-based analyses, because the ancestral area from the ML analysis is geographically adjacent to the ancestral area reconstructed from parsimony-based analyses.

Molecular clock.-Using LINTREE with the K2P + Gamma model of gene evolution, we were unable to generate an appropriate topology. We verified this result by generating NJ and ML phylogenies in PAUP* using both the uncorrected and K2P + Gamma models. We recovered the appropriate topology with the uncorrected data but not using the K2P + Gamma model. To confirm these results, we forced our ML topology into LINTREE and applied the two-cluster test using the K2P + Gamma model. We recovered results that were inconsistent with sequence- divergence values from PAUP*. Because of the inconsistencies with the K2P + Gamma correction, we deemed the model inappropriate for our data.

Using uncorrected (P) data, results of the two-cluster test suggest that 10 of the 12 nodes in the reduced taxon analysis possess daughter lineages or clades that are evolving in a clocklike manner (Fig. 2; wide nodes). Only the node linking the angolensis- solitarius clade and the node linking the bensoni-explorntor clade are not reconstructed as clock-like. Clock-like nodes, however, bracket each, allowing approximate dates to be assigned to these nodes. Uncorrected pairwise sequence-divergence values calculated using the 2% sequence-divergence per-million-years calibration (Tarr and Fleischer 1993) fell within the range of values obtained by the two-cluster test plus or minus the standard error estimates and are illustrated in Figure 2.

Speciation timing and movements.-Our results suggest that Monticola originated in the arid regions of northern Africa plus Saudi Peninsula or the adjacent African savanna, or both, during the late Miocene, ~5.5 mya. The initial split of M. rufocinereiis from the common ancestor of all remaining species of Monticola, combined with its restricted range, suggests long-term isolation of M. rufocinereus. The current northeast African distribution of M. rufocinereus follows closely the extent of eastern African montane forests that occurred during the late Miocene-early Pliocene (Meadows 1996), which suggests that this isolation is perhaps not recent.

Shortly after, ancestral species of Monticola may have expanded their ranges eastward into Asia as well as southward throughout Africa. Such intracontinental African movement is spatially, albeit not temporally, consistent with patterns previously observed in African lizards (Amer and Kumazawa 2005). Given that species of Monticola generally inhabit open and arid habitat, the arid conditions that existed in the greater eastern Mediterranean region in the late Miocene (Rouchy et al. 2001 and references therein) may have provided conditions suitable for species of Monticola to expand their ranges eastward. Furthermore, open, grassy conditions existed in Pakistan and eastern Africa from 9 to 5 mya (Vrba 1985), which would have allowed Monticola to gain access to the Himalayas (the area reconstructed as ancestral for the Asian clade; see Fig. 3) as well as to expand southward throughout Africa. Contemporaneous patterns of movement have been shown in other avian genera that, like Monticola, do not occupy tropical forest habitats (Voelker 1999, 2002).

The split between the African and Asian clades is estimated at - 5 mya (Fig. 2). The Miocene-Pliocene boundary, at -5.3 mya, is defined by environmental changes associated with the end of the Messian Salinity Crisis (Cita 1975), including the filling of the Mediterranean Sea. Contemporaneous cessation of open grassy conditions in Pakistan and eastern Africa (Vrba 1985) could have provided a vicariant barrier sufficient to result in the divergence between the African and Asian taxa.

Asia.-The Asian clade of Monticola consists of three species distributed from eastern Afghanistan and northern Pakistan eastward into northern China and Mongolia south to the Malay Peninsula. According to our ancestralarea analyses, the region encompassing the Himalayas and associated plateaus is reconstructed as the ancestral area for this clade. The split between M. cinclorhynchus and M. gularis + M. rufiventris is dated at -5 mya (Fig. 2). This split occurs very shortly after the initial movement into Asia, which suggests uninterrupted movement into suitable habitat in eastern China. The early divergences between the Asian species of Monticola may indicate rapid speciation following expansion into new areas, as observed in other avian lineages (Voelker 1999).

According to the molecular-clock data, the split between the migratory M. gnlaris and the sedentary-altitudinal migrant M. rufiventris occurred ~4.2 mya. One possible mechanism for speciation within this clade is that populations of the common ancestor may have moved northward in response to available habitats in eastern Asia. These northern populations (incipient M. gnlaris) may subsequently have experienced increased seasonality in northeastern China forcing a southward migration and, over time, developed a disjunct southern wintering range as the result of competition with nonmigratory populations (incipient M. rufiventris). The current wintering range of M. gularis is located just south of the range of M. rufiventris, in southern China and the Malay Peninsula, with minimal overlap. This scenario is consistent with a long-standing hypothesis for the evolution of long-distance migration in birds (Cox 1968, 1985). Other studies have shown molecular evidence for the role of migration in speciation in Old World Motacillidae lineages (Outlaw and Voelker 2006).

Another possible scenario is that speciation between M. gularis and M. rufiventris was driven by climatic events associated with glaciation or increasing aridity of central Asian deserts. These events have been recorded by the deposition of aeolian sediments on the Loess Plateau in northern China since the early Miocene (Guo et al. 2002). The action of these events has previously been hypothesized to be a vicariant mechanism of speciation among widespread Eurasian songbird lineages (Voelker 1999). The disjunct breeding distributions of M. gularis northeast of the Loess Plateau, and M. rufiventris south of the plateau, also are consistent with the hypothesis of climate-driven vicariance and speciation in historical refugia known to have existed in eastern Eurasia (Frenzel 1968).

Africa.-Our results suggest that the African savanna is the ancestral area for this clade. Species in the largely African clade occur primarily throughout the sub-Saharan region; this includes the wintering distributions of the two Eurasian migratory species M. saxatilis and M. solitarius. No species of Monticola occur in the lowland rainforest region of Africa. After the initial divergence between the Eurasian clade and the largely African clade, diversification within the African clade began at ~4.5 mya and resulted in two subclades (Fig. 2). Subclade 1 has one species occupying South Africa, with the remainder distributed from Zimbabwe north. Subclade 2 is generally restricted to South Africa and Madagascar. Subclade 1. - During the late Miocene-early Pliocene, the eastern extent of African lowland rainforest had begun to recede from Ethiopia and Kenya westward, with a subsequent increase in cool and dry savanna habitat (Meadows 1996). This likely provided a narrow eastern avenue of dispersion from northern arid Africa plus Saudi Peninsula or African savanna, or both, southward into the increasingly arid region of central and southern Africa (Moreau 1952). The break between M. atigolensis + M. rupestris, which occur south of the lowland rainforest, and the widespread and migratory M. saxatilis + M. solitarius, inhabiting Eurasia and Africa north of the rainforest, occurred ~4.0 mya (Fig. 2). This split may be partially the result of separation by lowland rainforest. The ancestor of M. angolensis + M. rupcstris likely used the eastern arid corridor to move southward into the increasingly arid central and southern Africa, whereas the ancestor of M. saxatilis + M. solitarius possibly expanded its range northward and eastward. Given the still considerable extent of lowland rainforest at that time, there was likely little if any opportunity for subsequent contact between these populations, once separated, because the rainforest would have provided a substantial barrier to north-south movement. Instances of exchanges between Africa and Eurasia through the Saudi Peninsula have been shown in other arid or dry-habitat adapted species at approximately 3.5-4 mya (Vrba 1985; Voelker 1999, 2002). The putative secondary colonization of Eurasia from Africa by the two widespread and long-distance migrants, M. saxatilis and M. solitarius, may have occurred along the same route.

Both M. saxatilis and M. solitarius occur throughout Africa and Eurasia, with M. solitarius extending into Japan and the Southeast Asian archipelago. The breeding areas of these two species overlap in Europe and southwestern Asia, but to the east, their breeding ranges overlap minimally. Wintering ranges of M. saxatilis and M. solitarius overlap only in sub-Saharan Africa, because M. saxatilis does not winter in Eurasia. Populations of M. saxatilis that breed as far east as the Gobi Desert return to Africa to winter, one of the longest migrant passerine journeys into Africa (Clement 2000). Wintering distributions of M. solitarius, however, extend from Africa south of the Sahara eastward through the Indian subcontinent to southeastern China, Japan, the Malay Peninsula, and the Southeast Asian archipelago. Separation of ancestral populations as a result of differing migration patterns and disjunct wintering ranges may have led to a split between M. saxatilis and M. solitarins estimated at 2.5 mya.

The current extent of the range of M. nipestrift in southeastern Africa, combined with a molecular-clock date, suggests a secondary colonization of South Africa by the ancestor of this species before 0.6 mya. Because of differences in morphology and ecology, there seems to be no apparent competition between M. rupestris and M. explorator (subclade 2), even though their ranges are depicted as almost completely overlapping.

Subclade 2.-The distribution of the Malagasy lineage was the direct result of an over-water dispersal to Madagascar estimated at - 3.4 mya. The Mozambique Channel separating Madagascar from Africa appears to have been crossed repeatedly by birds and mammals (Yoder et al. 1996, 2003; Jansa et al. 1999; Cibois et al. 2001).

Monticola brevipes and M. pretoriae occur in the deserts and semideserts of southwestern Africa, whereas M. explorator inhabits South Africa's central plateau and southern and southeastern mountain ranges. Throughout the early to middle Pliocene, there was a continuing trend toward arid conditions in southwestern Africa (Meadows 1996). The ancestor of M. brevipes and M. pretoriae likely expanded westward through southern Africa with expanding arid habitat. In the absence of geographic vicariant barriers, segregation by habitat could explain the east-west separation between brevipes + M. pretoriae and the M. explorator + Malagasy clade, a split we date at -4.2 mya. Movement eastward onto the higher elevations of the central South African plateau by the ancestor of M. brevipes and M. pretoriae may have resulted in the establishment of populations in this region. It is conceivable that these populations, isolated from western populations, resulted in the divergence of incipient M. pretoriae, a split we date at -0.8 mya.

Speciation in southern Africa. -In the absence of major vicariant barriers in southern Africa, what can we propose to explain speciation in this region? Several studies suggest that allopatric speciation of arid-adapted species in southern Africa results from climatic cycling (Matthee and Flemming 2002, Herron et al. 2005). Although not a traditional vicariant barrier, ecological change (i.e., habitat) via climatic cycling may provide a significant barrier to relatively specialized taxa (Price et al. 1997, Smith et al. 1997). Herron et al. (2005) suggest that fragmentation of the arid-adapted African ground-squirrel genus Xerus is the result of isolation of populations in warm refugia during cold periods and in dry refugia during wet periods. A similar hypothesis of climate- based ecological separation was put forth by Cracraft (1986) to potentially explain speciation in Australian avifaunas. Species of Monticola are found throughout South Africa, but there is a distinct break in range between M. brevipes + M. pretorine and M. explorator at the western margin of the Drakensburg range. Furthermore, there is a putative west-east range division between M. brevipes and M. pretorine at ~23[degrees] east, although, judging from current knowledge, there is some range overlap. These two breaks are displaced in time, and though we know of no physical geographic barrier to explain these splits, they may result from climatic cycling and increased desertification throughout western and central South Africa. Thus, we can postulate cyclical habitat change to account for the separation between M. brevipes + M. pretorine and M. explorator, and more recently between M. brevipes and M. pretorine. Although the studies of Matthee and Flemming (2002) and Herron et al. (2005) examined fragmentation of populations within a species, patterns seen here between M. brevipes + M. pretorine and M. explorator may be the end result of a process still operating in those other studies.

ACKNOWLEDGMENTS

This project was completed as an independent undergraduate research project at the University of Memphis. We thank the collective staffs of the Field Museum of Natural History, the Academy of Natural Sciences in Philadelphia, the American Museum of Natural History, University of Washington's Burke Museum, Marjorie Barrick Museum UNLV, Moscow Museum, United States National Museum, and the University of Pretoria and Witwatersrand Bird Club for their generosity in supplying samples, without which this study would not have been possible. We thank J. Klicka and two anonymous reviewers, whose comments greatly improved the quality of this manuscript. The authors would also like to thank T. P. Gnoske and C. Anderson for their expertise in the field. This research was supported in part by a National Science Foundation grant to G.V. (DEB-9903544). TREEFINDER is available from its author, G. Jobb, at www.treefinder.de. MESQUITE, by W. P. and D. R. Maddison, is available at mesquiteproject.org. DISCRETE, by M. Pagel, is available at www.ams.reading.ac.uk/ zoology/pagel/. DIVA, by F. Ronquist, is available by FTP server from Uppsala University (ftp.uu.se or ftp.systbot.uu.se).

LITERATURE CITED

AMER, S. A. M., AND Y. KUMAZAWA. 2005. Mitochondrial DNA sequences of the Afro-Arabian spiny-tailed lizards (genus Urotnastyx; family Agamidae): Phylogenetic analyses and evolution of gene arrangements. Biological Journal of the Linnean Society 85:247- 260.

BARKER, F. K., AND F. M. LUTZONI. 2002. The utility of the incongruence length difference test. Systematic Biology 51:625-637.

BREMER, K. 1992. Ancestral areas: A cladistic reinterpretation of the center of origin concept. Systematic Biology 41:436-445.

BROWN, J. H., AND M. V. LOMOLINO. 1998. Biogeography, 2nd ed. Sinauer Associates, Sunderland, Massachusetts.

CIBOIS, A., B. SLIKAS, T. S. SCHULENBERG, AND E. PASQUI-T. 2001. An endemic radiation of Malagasy songbirds is revealed by mitochondrial DNA sequence data. Evolution 55:1198-1206.

CITA, M. B. 1975. The Miocene/Pliocene boundary: History and definition. Pages 1-30 in Late Neogene Epoch Boundaries (T. Saito and L. H. Burckle, Eds.). Micropaleontology Press, American Museum of Natural History, New York.

CLANCEY, P. A. 1968. The status of Monticola pretorine Gunning and Roberts 1911. Bulletin of the British Ornithologists' Club 88:126-128.

CLEMENT, P. 2000. Thrushes. Princeton University Press, Princeton, New Jersey.

COX, G. W. 1968. The role of competition in the evolution of migration. Evolution 22: 180-192.

COX, G. W. 1985. The evolution of avian migration systems between temperate and tropical regions of the New World. American Naturalist 126:451-474.

CRACRAFT, J. 1986. Origin and evolution of continental biotas: Speciation and historical congruence with the Australian avifauna. Evolution 40:977-996.

CROWE, T. M., AND A. A. CROWE. 1982. Patterns of distribution, diversity and endemism in Afrotropical birds. Journal of Zoology (London) 198:417-442.

DESJARDINS, P., AND R. MORAIS. 1990. Sequence and gene organization of the chicken mitochondrial genome, a novel gene order in higher vertebrates. Journal of ar Biology 212:599-634. DOWSETT, R. J., AND F. DOWSETT-LEMAIRE. 1993. A contribution to the distribution and taxonomy of Afrotropical and Malagasy birds. Tauraco Research Report 5:323-389.

FARKAS, T. 1971. Monticola bensoni, a new species from south- western Madagascar. Ostrich 9 (supplement):83-90.

FARKAS, T. 1979. A further note on the status of Monticola prctoriac Gunning and Roberts. Bulletin of the British Ornithologists' Club 99:20-21.

FLEISCHER, R. C., C. E. MCINTOSH, AND C. L. TARR. 1998. Evolution on a volcanic conveyor belt: Using phylogeographic reconstructions and K-Ar-based ages of the Hawaiian Islands to estimate molecular evolutionary rates. Molecular Ecology 7:533-545.

FRENZEL, B. 1968. The Pleistocene vegetation of northern Eurasia. Science 161:637-649.

GARCIA-MORENO, J. 2004. Is there a universal mtDNA clock for birds? Journal of Avian Biology 35:465-468.

GOODMAN, S. M., AND L. A. WEICT. 2002. The generic and species relationships of the reputed endemic Malagasy genus Pseudocossyphus (family Turdidae). Ostrich 73:26-35.

GOODWIN, D. 1956. Note on the genus Pseiidocossyphus Sharpe. Bulletin of the British Ornithologists' Club 76:143-144

GUO, Z. T., W. F. RUDDIMAN, Q. Z. HAO, H. B. WU, Y. S. QIAO, R. X. ZHU, S. Z. PENG, J. J. WEI, B. Y. YUAN, AND T. S. LIU. 2002. Onset of Asian desertification by 22 Myr ago inferred from loess deposits in China. Nature 416: 159-163.

HAUSDORF, B. 1998. Weighted ancestral area analysis and a solution of the redundant distribution problem. Systematic Biology 47:445-456.

HERRON, M. D., J. M. WATERMAN, AND C. L. PARKINSON. 2005. Phylogeny and historical biogeography of African ground squirrels: The role of climate change in the evolution of Xerus. Molecular Ecology 14:2773-2788.

HO, S. Y. W., M. J. PHILLIPS, A. COOPER, AND A. J. DRUMMOND. 2005. Time dependency of molecular rate estimates and systematic overestimation of recent divergence times. Molecular Biology and Evolution 22: 1561-1568.

HUELSENBECK, J. P., AND B. RANNALA. 1997. Phylogenetic methods come of age: Testing hypotheses in an evolutionary context. Science 276:227-232.

HUELSENBECK, J. P., AND F. RONQUIST. 2001. MRBAYES: Bayesian inference of phylogenetic trees. Bioinformatics 17:754-755.

JANSA, S. A., S. M. GOODMAN, AND P. K. TUCKER. 1999. Molecular phylogeny and biogeography of the native rodents of Madagascar (Muridae: Nesomyinae): A test of the single-origin hypothesis. Cladistics 15: 253-270.

MATTHEE, C. A., AND A. F. FLEMMING. 2002. Population fragmentation in the southern rock agama, Agama atra: More evidence for vicariance in southern Africa. Molecular Ecology 11:465-471.

MEADOWS, M. E. 1996. Biogeography. Pages 161-172 in The Physical Geography of Africa (W. M. Adams, A. S. Goudie, and A. R. Orme, Eds.). Oxford University Press, New York.

MOREAU, R. E. 1952. Africa since the Mesozoic: With particular reference to certain biological problems. Proceedings of the Zoological Society of London 121:869-913.

MORRIS, P., AND F. HAWKINS. 1998. Birds of Madagascar: A Photographic Guide. Pica Press, Robertsbridge, United Kingdom.

OUTLAW, D. C., AND G. VOELKER. 2006. Phylogenetic tests of hypotheses for the evolution of avian migration: A case study using the Motacillidae. Auk 123:455-466.

PAGEL, M. 1999. The maximum likelihood approach to reconstructing ancestral character states of discrete characters on phylogenies. Systematic Biology 48:612-622.

PETERS, J. L. 1964. Check-list of Birds of the World, vol. 10. Museum of Comparative Zoology, Harvard University, Cambridge, Massachusetts.

POSADA, D., AND K. A. CRANDALL. 1998. MODELTEST: Testing the model of DNA substitution. Bioinformatics 14:817-818.

PRICE, T. D., A. J. HELBIG, AND A. D. RICHMAN. 1997. Evolution of breeding distributions in the Old World leaf warblers (genus Phylloscopus). Evolution 51:552-561.

RIPLEY, S. D. 1952. The thrushes: A taxonomic study. Postula, no. 13.

RONQUIST, F. 1994. Ancestral areas and parsimony. Systematic Biology 43:267-274.

RONQUIST, F. 1997. Dispersal-vicariance analysis: A new approach to the quantification of historical biogeography. Systematic Biology 46:195-203.

ROUCHY, J. M., F. ORSZAG-SPERBER, M.-M. BLANC-VALLERON, C. PIERRE, M. RIVIERE, N. COMBOURIEU-NEBOUT, AND I. PANAYIDES. 2001. Paleoenvironmental changes at the Messinian-Pliocene boundary in the eastern Mediterranean (southern Cyprus basins): Significance of the Messinian Lago-Mare. Sedimentary Geology 145:93-117.

SHIMODAIRA, H., AND M. HASEGAWA. 1999. Multiple comparisons of log-likelihoods with applications to phylogenetic inference. Molecular Biology and Evolution 16: 1114-1116.

SIBLEY, C. G., AND B. L. MONROE, JR. 1990. Distribution and Taxonomy of Birds of the World. Yale University Press, New Haven, Connecticut.

SMITH, T. B., R. K. WAYNE, D. J. GIRMAN, AND M. W. BRUFORD. 1997. A role for ecotones in generating rainforest biodiversity. Science 276:1855-1857.

SWOFFORD, D. L. 1999. PAUP*: Phylogenetic Analysis Using Parsimony (*and Other Methods), version 4.0. Sinauer Associates, Sunderland, Massachusetts.

TAKEZAKI, N., A. RZHETSKY, AND M. NEI. 1995. Phylogenetic test of the molecular clock and linearized trees. Molecular Biology and Evolution 12:823-833.

TARR, C. L., AND R. C. FLEISCHER. 1993. Mitochondrial-DNA variation and evolutionary relationships in the Amakihi complex. Auk 110:825-831.

VAURIE, C. 1955. Systematic notes on Palearctic birds, no. 14. Turdinae: The genera Erithacus, Litscinia, Tarsiger, Phoenicians, Monticola, Erythropygia, and Oenanthe. American Museum Novitates 1731:1-30.

VOELKER, G. 1999. Dispersal, vicariance, and clocks: Historical biogeography and speciation in a cosmopolitan passerine genus (Anthus: Motacillidae). Evolution 53: 1536-1552.

VOELKER, G. 2002. Systematics and historical biogeography of wagtails: Dispersal versus vicariance revisited. Condor 104:725- 739.

VOELKER, G., AND G. M. SPELLMAN. 2004. Nuclear and mitochondrial DNA evidence of polyphyly in the avian superfamily Muscicapoidea. Molecular Phylogenetics and Evolution 30:386-394.

VRBA, E. S. 1985. African Bovidae: Evolutionary events since the Miocene. South African Journal of Science 81:263-266.

WHITE, C. M. N. 1967. Taxonomic notes on African Turdinae. Bulletin of the British Ornithologists' Club 87:150-152.

YODER, A. D., M. M. BURNS, S. ZEHR, T. DELEFOSSE, G. VERON, S. M. GOODMAN, AND J. J. FLYNN. 2003. Single origin of Malagasy Carnivore from an African ancestor. Nature 421:734-737.

YODER, A. D., M. CARTMILL, M. RUVOLO, K. SMITH, AND R. VILGALYS. 1996. Ancient single origin of Malagasy primates. Proceedings of the National Academy of Sciences USA 93: 5122-5126.

ZINK, R. M., A. E. KESSEN, T. V. LINE, AND R. C. BLACKWELL-RAGO. 2001. Comparative phylogeography of some aridland bird species. Condor 103:1-10.

Associate Editor: J. Klicka

ROBERT K. OUTLAW,1 GARY VOELKER, AND DIANA C. OUTLAW

Department of Biology, University of Memphis, 3700 Walker Avenue Memphis, Tennessee 38252, USA

1 E-mail: routlaw@memphis.edu

Copyright American Ornithologists' Union Apr 2007

(c) 2007 Auk, The. Provided by ProQuest Information and Learning. All rights Reserved.