18. Glaciation and Avian Evolution
277
cent species' origins did not differ from a "random" sample. We (Klicka et al.
1999) also discovered a single species pair, the Timberline Sparrow (SpizeUa
tavernerz) and Brewer's Sparrow (S. brewerz), whose mtDNA characteristics were
consistent with speciation occurring within the last 100,000 yr. In particular, this
species pair exhibited a very low molecular distance, as well as a haplotype tree that
did not exhibit reciprocal monophyly (that is the structure of the tree does not
reflect species boundaries). None of the 35 species pairs exhibited these characteristics, which are genetic signatures of very recent speciation events, such as those
predicted by the LPO.
Our conclusion stemming from the distribution of mtDNA distances was recently challenged by Arbogast and Slowinski (1998). In particular, these authors
concluded that genetic distances derived from mtDNA sequence comparisons are
biased (underestimated) because of "saturation" resulting from multiple base substitutions at single nucleotide positions. According to Arbogast and Slowinski, the
songbird sister species originated much more recently than Klicka and Zink (1997)
suggested, once corrections for saturation are made. Arbogast and Slowinski's preferred method of estimating the true evolutionary distance between species employed the gamma-HKY85 maximum likelihood model, which is becoming widely
used (Swofford et al. 1996). Klicka and Zink (1998) suggested that this model
likely overestimates the degree ~f saturation (most likely via inaccurate estimation
of the alpha or shape parameter, not the model per se). For example, Arbogast and
Slowinski (1998) derived a rate of 5%/million years (MY) for galliform birds from
the cytochrome b data of Randi (1996). We analyzed the same data, finding, for
example, that the uncorrected cytochrome b distance between Gallus gallus and
Alectoris magna, 0.14, has a gamma-HKY85 distance of 1.02, suggesting a significant underestimate of the "true" evolutionary distance. Although we believe that
these galliform sequences are saturated, this corrected distance is likely biologically meaningless. Thus, Arbogast and Slowinski's (1998) calibration based on
these data (as well as one they derived from higher primates) is dubious. Nonetheless, even accepting these recomputed distances and the new calibration (5%/MY),
90% of their gamma-HKY85 divergence dates exceed one million years, a decisive
refutation of the LPO model (Klicka and Zink 1998).
In this chapter we use a different maximum likelihood model to estimate the
degree to which saturation (multiple hits) affects mtDNA distance estimates and
we employ two different calibrations on the resultant distances. The results of this
new analysis do not change the conclusion of Klicka and Zink (1997, 1998) that
most species pairs of North American songbirds originated earlier than most previous authors believed.
Nonetheless, major effects of glaciers on the genetics of birds and other organisms seem unarguable. It is clear that major biomes were displaced and often greatly
reduced in size (Pielou 1991). Such massive environmental perturbations must have
left genetic "imprints" at some taxonomic level. If these imprints were not the
speciation events themselves, the search for such imprints should proceed to lower
taxonomic levels, namely that within species (termed phylogeography; Avise 1994).
277
cent species' origins did not differ from a "random" sample. We (Klicka et al.
1999) also discovered a single species pair, the Timberline Sparrow (SpizeUa
tavernerz) and Brewer's Sparrow (S. brewerz), whose mtDNA characteristics were
consistent with speciation occurring within the last 100,000 yr. In particular, this
species pair exhibited a very low molecular distance, as well as a haplotype tree that
did not exhibit reciprocal monophyly (that is the structure of the tree does not
reflect species boundaries). None of the 35 species pairs exhibited these characteristics, which are genetic signatures of very recent speciation events, such as those
predicted by the LPO.
Our conclusion stemming from the distribution of mtDNA distances was recently challenged by Arbogast and Slowinski (1998). In particular, these authors
concluded that genetic distances derived from mtDNA sequence comparisons are
biased (underestimated) because of "saturation" resulting from multiple base substitutions at single nucleotide positions. According to Arbogast and Slowinski, the
songbird sister species originated much more recently than Klicka and Zink (1997)
suggested, once corrections for saturation are made. Arbogast and Slowinski's preferred method of estimating the true evolutionary distance between species employed the gamma-HKY85 maximum likelihood model, which is becoming widely
used (Swofford et al. 1996). Klicka and Zink (1998) suggested that this model
likely overestimates the degree ~f saturation (most likely via inaccurate estimation
of the alpha or shape parameter, not the model per se). For example, Arbogast and
Slowinski (1998) derived a rate of 5%/million years (MY) for galliform birds from
the cytochrome b data of Randi (1996). We analyzed the same data, finding, for
example, that the uncorrected cytochrome b distance between Gallus gallus and
Alectoris magna, 0.14, has a gamma-HKY85 distance of 1.02, suggesting a significant underestimate of the "true" evolutionary distance. Although we believe that
these galliform sequences are saturated, this corrected distance is likely biologically meaningless. Thus, Arbogast and Slowinski's (1998) calibration based on
these data (as well as one they derived from higher primates) is dubious. Nonetheless, even accepting these recomputed distances and the new calibration (5%/MY),
90% of their gamma-HKY85 divergence dates exceed one million years, a decisive
refutation of the LPO model (Klicka and Zink 1998).
In this chapter we use a different maximum likelihood model to estimate the
degree to which saturation (multiple hits) affects mtDNA distance estimates and
we employ two different calibrations on the resultant distances. The results of this
new analysis do not change the conclusion of Klicka and Zink (1997, 1998) that
most species pairs of North American songbirds originated earlier than most previous authors believed.
Nonetheless, major effects of glaciers on the genetics of birds and other organisms seem unarguable. It is clear that major biomes were displaced and often greatly
reduced in size (Pielou 1991). Such massive environmental perturbations must have
left genetic "imprints" at some taxonomic level. If these imprints were not the
speciation events themselves, the search for such imprints should proceed to lower
taxonomic levels, namely that within species (termed phylogeography; Avise 1994).