736 
THE WILSON JOURNAL OF ORNITHOLOGY . Vol. 123. No. 4. December 2011 
TABLE 1. Descriptive statistics of morphometries of Kriiper’s Nuthatch, Anatolian Peninsula, Turkey. 
Character 
Body mass, g 
Wing , mm 
Tail, mm 
BL, mm 
BW, mm 
BH, mm 
LBaHb, mm 
LNBa, mm 
Alula, mm 
Tarsus, mm 
Back nail, mm 
Left nail, mm 
Mid. nail, mm 
Inner nail, mm 
LbF, mm 
Mean ± SD 
72 
13.11 ± 0.88 
77 
74.79 ± 2.35 
75 
37.45 ± 2.06 
77 
17.65 ± 0.76 
76 
4.55 ± 0.34 
76 
3.87 ± 0.24 
66 
36.18 ± 0.90 
77 
12.64 ± 0.68 
66 
18.55 ± 1.22 
76 
19.10 ± 0.93 
70 
6.50 ± 0.34 
62 
4.04 ± 0.36 
63 
4.63 ± 0.39 
63 
3.78 ± 0.36 
58 
14.25 ± 1.98 
Min Max 
11.17 
14.80 
65.32 
80.00 
31.04 
42.00 
15.47 
19.23 
3.78 
5.51 
3.32 
4.68 
34.09 
38.36 
11.14 
14.87 
13.62 
22.00 
16.59 
21.50 
5.56 
7.34 
3.46 
4.93 
3.95 
6.36 
3.10 
9.86 
4.71 
19.66 
surement data, and were excluded from the 
analysis. The discriminant analysis was performed 
on tour populations totaling 60 individuals. 
Our data set indiscriminately incorporated adult 
males and females, and possibly some juveniles’ 
we suspected lha. size could be a confusing 
m the discriminant analysis. We thus focused on 
Shape analy sis (morphology inslead of morphom- 
ty) 1° remove Ihc potential cffecl of size and 
avoid misinterpretation of the results. Size 
tramfom 5 ,' Were T"™* Usi " e a Centering 
transformation on the log of the measurements 
Two d du,, 1 [his mansformation 
,He , wr- mU '"^ constant and became 
identical (Gower 1076). This approach is classi 
cally used lo study morphology instead of 
morphometry (Yoecoz 1993). Bicemering wi 
bceneT w d “ch g ‘T Ade4 ,fuiKti ™ 
bicenter.wt) (Chessel el al. 21104) and R , |() , 
■ development Core Team 2009). 
The data set was subjected to stepwise 
discriminant analysis to ascertain which morpho¬ 
logic traits best discriminated among the fom 
populations. All lrails that remained in ihe mode 
when the stepwise selection process stopped wem 
^gnificantly discriminating among populations 
uaing^r^r' a " alySiS WUS P0,for '^' d 
(oaq , no/ S1 AT s °ftware (PROC STKPDISC) 
used ti 9 ^ P T DISCR,M «AS 1985) was 
function bla,n , the d,scnminan ' function. The 
^ronT^rThViir^"^ 
validation procedure Z cross- 
tion several times, based on all birds except (he 
one which is classified using the estimated 
1 unction. Each classified bird in this procedure 
was allocated lo the group with the nearest 
centroid, and the proportion of correctly allocate 
individuals was calculated (Ripley 1996). A 
graphical representation of the classification 
was also obtained using library Ade4 in R 2.10.1 
(R Development Core Team 2009). 
RESULTS 
Morphometric Characterization .—We found J 
small statistically significant differentiation be¬ 
tween individuals without a brood patch andwidi 
a brood patch for bill length, bill width, and P 
(/-test; P < 0.05). We recorded no measurement 
for brood patch individuals due to the low sample 
size (n = 5. probably females). Mean (± SDi. 
range, and sample size for the 15 morphoroetiH 
measurements used in the discriminant analysis 
(Table !), and measurements of primaries, sec¬ 
ondaries. and tail feathers varied (Table 2). The 
longest primary was P 6 and lon°est secondary 
was S 2 (Table 2). 
Population Relationships .—The stepwise div 
ciiminant analysis retained only seven measure 
ments allowing discrimination among populations 
based on body mass, wing length, length of P $• 
alula, BH, back nail, and left nail (Table 3). The 
plot of the first two canonical variables indicate" 
small differences among populations as the 
overlaps are great (Fig. 2). The discriminant 
analysis correctly classified 50 of the 60 birds, 
i-e., 83% (Table 4). However, the cross validation 
