ORNITHOLOGICAL LITERATURE 
651 
have a score of 4, 5, 6. or 7-8 is noted for each 
character scored on a 1-3 basis, so approximately 
half of the two-page table is composed of zeros. 
One line separating the groups of five species is 
wrongly placed, and the bottom line is omitted in 
the first page. The latter portends a problem with 
the tables throughout the book, which apparently 
were composed in a larger format and pasted into 
a book with smaller pages, where they don't 
always fit and often result in very small print. 
An analysis of pairwise comparisons of these 
color characters shows they are virtually indepen¬ 
dent with few correlations between variables. 
Where there is some correlation in large geese, 
there is a similar correlation in small geese. 
Correlations between measurements are much 
greater than between color characters. There 
may be correlations between measurement and 
color characters (page 27), but the presumably 
definitive statement on that is an incomplete, 
indefinite, sentence. The point is made that both 
color/pattem and measurement characters should 
be considered when evaluating geographic varia¬ 
tion in W-c G. and “the potential for geographic 
variation is large." 
An analysis of age and sex variation in W-c G 
shows that (page 40) "colors and patterns are 
virtually useless in separating the age and sex 
classes.” 1 do not understand the following 
statement: “However, as additional age or sex 
classes are added the range of variation is 
somewhat increased within a taxon." Within a 
taxon, males are larger than females and adults are 
larger than immatures, although there is much 
overlap. 
Chapter 3 is a study of the effect of sample size 
on analytical techniques, specifically discriminant 
function analysis (DFA). This is important because 
many of the taxa Hanson described were repre¬ 
sented by very few specimens. From a large sample 
of a single taxon, Anderson compared two random 
samples of five, then two samples of 10. two of 15. 
and on to two samples of 70. This allowed 14 
pairwise comparisons. This test was repeated lor 
various size samples of several supposed subspe¬ 
cies. This exercise shows (page 62) that with small 
samples “there is a high probability that putative 
taxa could be determined to differ when in tact they 
are made up of individuals from a single taxon. 
Comparisons should be based on samples of at least 
15 of each taxon. A conclusion, not specifically 
slated, is that this casts extreme doubt on many, if 
not most, of the subspecies named by Hanson in 
2006 and 2007, where samples were typically quite 
small. Further, it casts doubt on the taxonomic 
purity of the large subspccific samples assembled 
by Anderson. 
The repeatability of plumage scores, by the 
same observer at different times and by different 
workers, is tested in Chapter 4. Recall there are 16 
characters with scores of 1-3 or 1-8. For 180 
birds having been scored <1 or >10 years 
previously, there were changes in the score of 
~5 characters, ~30% per bird (Table 4.1). There 
were more changes for birds that had been scored 
earlier. Some characters are more susceptible to 
change in score than others. Some scores were 
higher, some lower: the net sura of the score 
differences was 5-6 points, but there is no 
indication of what proportion this is of the total 
score for any bird, or whether that difference 
would affect the subspecific identification, but 
some of the birds were identified to different taxa 
in different scoring sessions. “The chances of 
making the same identification for a given 
specimen were 80-85%" (page 83). which does 
not strike me as very encouraging. 
The subspecies concept in birds is intimately 
related to breeding areas of populations; subspe¬ 
cies are defined by morphological characters that 
differ between or among populations Lhat breed in 
different geographic areas. Individuals of migra¬ 
tory species will not always be in breeding areas, 
they will occur elsewhere as migrants or non- 
breeders. Chapter 5 is dedicated to showing that 
migrant and wintering geese have taxonomic 
value, and that they can be used to define 
subspecies. This is an important justification for 
the work of both Hanson and Anderson, who 
named and accept many subspecies on the basis of 
only wintering or migrant, i.e., non-breeding, 
birds. This result is accomplished by establishing 
large samples of six ‘core’ taxa of Hanson, which 
are analyzed within themselves and compared to 
four geographically adjacent taxa, a total of 21 
taxa. The samples of core taxa are composed of 
birds with varying degrees of certainty of 
relationship to breeding areas, from birds banded 
as goslings (and later collected as adults) to (page 
88f “banded birds wintering at Cibola NWR that 
resembled birds banded as goslings ... or as molt 
migrants...." Table 5.1 shows the makeup of all 
samples. Of the 1,019 birds, fewer than 300, 
mostly of the core taxa, had fairly firm evidence 
of representing breeding populations. Most of the 
rest were molt migrants or wintering birds. Some 
