SHORT COMMUNICATIONS 
843 
FIG. I. Haplotype network of all P. fulva samples in the study showing two clusters, one for each subspecies group. 
Each line represents a single mutational step with solid circles indicating unsampled haplotypes. The dashed line indicates 
the division between P. fulva subspecies groups. The size of each circle is proportional to the total number of samples with 
the corresponding haplotype and the number in parentheses indicates the number of samples carrying that haplotype trom a 
given location. 
based on the measurements presented by West 
(1995). 
Kirchman et al. (2000). using mtDNA cyt b 
data, found strong bootstrap support (79% for 
each clade in a maximum parsimony analysis) for 
'be reciprocal monophyly of P.pallida and P. J. 
fulva. The addition of the sequences trom the 
Illinois, New Jersey, and New York vagrants 
caused an unexpected breakdown of reciprocal 
monophyly (Fig. 2). However, the haplotype 
network shows clear affinities of all three vagrants 
to P. f. palluh (Fig. I). despile all three vagrants 
as well as the newly-added Texas specimen 
having nrtDNA haplotypes that differ trom those 
published by Kirchman et al. (2000). That these 
three individuals carried previously unsampled 
haplotypes suggests they may have originated 
from populations other than those sampled by 
Kirchman et al. (2000), although it is possible that 
sampling more individuals may have revealed 
these haplotypes. 
A combination of plumage, genetic distances, 
haplotype network, and the Bayesian support for 
the P. f fulva group are consistent with the 
vagrant Cave Swallow specimens having origi¬ 
nated from the P. f pallida populations of the 
southwestern USA or Mexico. This evidence 
reaffirms the putative link between rapid popula¬ 
tion expansion and the spate ot vagrancy in this 
species over the past two decades. 
ACKNOWLEDGMENTS 
We thank David Willard for providing the FMNH tissue 
sample and the Cape May Bird Observatory, Richard 
