278
R.M. link and J. Klicka
Because many bird species must have recently experienced population increases
associated with recolonization of deglaciated areas, we are interested in whether
avian species show the predicted imprints (Fig. 1), such as geographically unstructured haplotype trees (Nee et al. 1995), a northward decrease in nucleotide diversity
(Hewitt 1996), and particular shapes of mismatch distributions (Rogers and
Harpending 1992; Rogers 1996) and plots of the In number of lineages versus time
(Nee et al. 1995). Unfortunately, few North American species have been studied
thoroughly enough to test the theoretical predictions in a general way; we focus on
data from a recent phylogeographic study (Fry and Zink 1998) of the Song Sparrow
(Melospiza melodia).
2 Methods
2.1 Interspecific Analyses
We use the same mtDNA sequence data as Klicka and Zink (1997); restriction site
data were ignored because of the difficulty in making corrections for saturation
(nonetheless, correlation coefficients for sequence and restriction site distances are
often high). To account for potential saturation effects, we computed gamma-corrected Kimura (1980) two-parameter (K2P) distances in Paup*, following the protocol of Fleischer ct al. (1998). We derived a gamma value of 0.4 using the Sullivan
et al. (1995) parsimony method in Paup* (Swofford 1996) using a minimum length
tree for the 21 species for which over 1000 bp of cytochrome b existed. To be
conservative, we subtracted 0.01 from these pairwise distances to account for ancestral haplotype diversity; Moore (1995) estimated a value of 0.007. To estimate
divergence dates (i.e., "lineage sundering" as defined by Avise and Walker 1998),
we divided the distances by 0.016, the gamma-K2P rate estimated for Hawaiian
songbird cytochrome b evolution (Fleischer et al. 1998). We also considered data
(Randi 1996) from galliform birds, which were used by Arbogast and Slowinski
(1998) to derive a rate of 5%/MY. Using a calibration date of 20 MY (Randi 1996),
and an alpha value of 0.4 (from the Sullivan et al. parsimony method in Paup*) we
obtained a rate of 0.9%/MY. It has been suggested that galliforms and non-passerines evolve at a slower rate than passerines, although the validity of this claim is
uncertain. If the rate were 2%/MY, then the average distance should be closer to
40%, whereas the gamma-K2P distance is ca. 20%. Using a calibration date of 20
MY (Randi 1996), and an alpha value of 0.08 (calculated under the gamma-HKY85
model) we obtained a gamma-K2P rate of 3.5%/MY. It appears likely that there is
not a single rate for all birds. Hence, to test the LPO model as broadly as possible,
we used both the songbird rate of Fleischer et al. (1998) and the gamma-K2P
galliform rate of 3.5%.
The use of molecular dating methods involves several assumptions. First, the
sequences should be evolving in a selectively neutral manner. We have no evidence
that our sequences are under strong selection (Fry and Zink 1998). Most changes in
R.M. link and J. Klicka
Because many bird species must have recently experienced population increases
associated with recolonization of deglaciated areas, we are interested in whether
avian species show the predicted imprints (Fig. 1), such as geographically unstructured haplotype trees (Nee et al. 1995), a northward decrease in nucleotide diversity
(Hewitt 1996), and particular shapes of mismatch distributions (Rogers and
Harpending 1992; Rogers 1996) and plots of the In number of lineages versus time
(Nee et al. 1995). Unfortunately, few North American species have been studied
thoroughly enough to test the theoretical predictions in a general way; we focus on
data from a recent phylogeographic study (Fry and Zink 1998) of the Song Sparrow
(Melospiza melodia).
2 Methods
2.1 Interspecific Analyses
We use the same mtDNA sequence data as Klicka and Zink (1997); restriction site
data were ignored because of the difficulty in making corrections for saturation
(nonetheless, correlation coefficients for sequence and restriction site distances are
often high). To account for potential saturation effects, we computed gamma-corrected Kimura (1980) two-parameter (K2P) distances in Paup*, following the protocol of Fleischer ct al. (1998). We derived a gamma value of 0.4 using the Sullivan
et al. (1995) parsimony method in Paup* (Swofford 1996) using a minimum length
tree for the 21 species for which over 1000 bp of cytochrome b existed. To be
conservative, we subtracted 0.01 from these pairwise distances to account for ancestral haplotype diversity; Moore (1995) estimated a value of 0.007. To estimate
divergence dates (i.e., "lineage sundering" as defined by Avise and Walker 1998),
we divided the distances by 0.016, the gamma-K2P rate estimated for Hawaiian
songbird cytochrome b evolution (Fleischer et al. 1998). We also considered data
(Randi 1996) from galliform birds, which were used by Arbogast and Slowinski
(1998) to derive a rate of 5%/MY. Using a calibration date of 20 MY (Randi 1996),
and an alpha value of 0.4 (from the Sullivan et al. parsimony method in Paup*) we
obtained a rate of 0.9%/MY. It has been suggested that galliforms and non-passerines evolve at a slower rate than passerines, although the validity of this claim is
uncertain. If the rate were 2%/MY, then the average distance should be closer to
40%, whereas the gamma-K2P distance is ca. 20%. Using a calibration date of 20
MY (Randi 1996), and an alpha value of 0.08 (calculated under the gamma-HKY85
model) we obtained a gamma-K2P rate of 3.5%/MY. It appears likely that there is
not a single rate for all birds. Hence, to test the LPO model as broadly as possible,
we used both the songbird rate of Fleischer et al. (1998) and the gamma-K2P
galliform rate of 3.5%.
The use of molecular dating methods involves several assumptions. First, the
sequences should be evolving in a selectively neutral manner. We have no evidence
that our sequences are under strong selection (Fry and Zink 1998). Most changes in