. ee See , serie era swans we HARVARD UNIVERSITY IR Si LIBRARY OF THE Museum of Comparative Zoology — se ry : - im) , : ’ A 7 4 i wy -, f iy ; 7 a a : > a 7 ’ ms 4 cc. _ . - > re : ' i q : o 7 i in 7 e > : 5 + _ _ - i) - m7 r o] 'Y ) : P - 1? 7 A a ! 7 - oy _ al ; 7 i : ’ ; ay hy _ 9 ' aa — & 7 ¢ “4 , , , ' - ! 7 ° - - , 7 a? i _ a5 7 es O@ oe el BS ’ 7 ’ " > ' t , , ‘ n ' ‘ Co] - - , j 7 »» 7 — ss > , - a 7 UNIVERSITY OF KANSAS PUBLICATIONS MUSEUM OF NATURAL HISTORY VeO UM be A ee Ne OniOe 1 7k EDITORS FRANK B. Cross, Chairman WiLuiAM E.. DUELLMAN HEnry S. Fircu Puiure S. HUMPHREY J. Knox Jones, Jr. RoBERT M. MENGEL MusEuM OF NATURAL HISTORY UNIVERSITY OF KANSAS LAWRENCE 1972 MusEUM OF NATURAL HISTORY UNIVERSITY OF KANSAS LAWRENCE PRINTED BY UNIVERSITY OF KANSAS PRINTING SERVICE LAWRENCE CONTENTS OF VOLUME 19 1. Geographic variation of red-winged blackbirds in central North America. By Dennis M. Power. Pp. 1-83, 32 figs. April 20, 1970. Ecological study of the worm snake Carphophis vermis ( Ken- nicott). By Donald R. Clark, Jr. Pp. 85-194, 48 figs., 4 pls. August 10, 1970. 3. Systematics and zoogeography of Middle American shrews of the genus Cryptotis. By Jerry R. Choate. Pp. 195-317, 20 figs. December 30, 1970. 4. The phylogenetic significance of vocal sac structure in hylid frogs. By Michael J. Tyler. Pp. 319-360, 10 figs. January 18, 197% 5. Mammals of northwestern South Dakota. By Kenneth W. Andersen and J. Knox Jones, Jr. Pp. 361-393, § figs. January oie alle Index, pp. 395-399. bo on | en : ‘ a Pr ee), 1 7 ic j : : aa 7 7 | 4 : : _ : ie. af 7 live § & :¢ : i‘! i? i Vala sade Ce wi a : la i oe va Bien). , ar Viet 7 kL 7 _ heit 7 ; ” : : 7 i : } , at 4 : : , 9 n - 7 a) iy ' s : p , ; 19 ve 1. = ) ehda - v7 | ie gts thats A re Ti 7 oa ’ 7 6 ry = ¢ , 7 6 v : v P 7 34 R - ; 7 i : 1 » - : : a , re _ 7 7 a . =~ Ate MUS. COMP. ZOOL. bi LIBRARY. JUN 18 1970 UNIVERSITY OF KANSAS PUBLICATIONS HARVARD UNIVERSITY; MusEUM OF NATURAL HISTORY Volume 19, No. 1, pp. 1-83, 32 figs: April 20, 1970 Geographic Variation of Red-winged Blackbirds in Central North America BY DENNIS M. POWER UNIVERSITY OF KANSAS LAWRENCE 1970 UNIVERSITY OF KANSAS PUBLICATIONS MusEUM OF NATURAL HIsToRY Volume 19, No. 1, pp. 1-83, 82 figs. April 20, 1970 Geographic Variation of Red-winged Blackbirds in Central North America BY DENNIS M. POWER UNIVERSITY OF KANSAS LAWRENCE 1970 UNIVERSITY OF KANSAS PUBLICATIONS, MUSEUM OF NATURAL HIsTORY Editors: Frank B. Cross, Philip $. Humphrey, Robert M. Mengel Volume 19, No. 1, pp. 1-83, 32 figs. Published April 20, 1970 UNIVERSITY OF KANSAS Lawrence, Kansas PRINTED BY ROBERT R. (BOB) SANDERS, STATE PRINTER TOPEKA, KANSAS 1970 Geographic Variation of Red-winged Blackbirds in Central North America BY DENNIS M. POWER CONTENTS ENTRODUGTIONA Aone ee ae ee ey, WU Ste ae eG MEREHODSE Set 8 OA oe See ee. BD te LR, bee Be Regression analyses and the use of isophanes......... oh CHRONOLOGICAT? VARIATION", 2. SOMSCP Salo 6! Gat) Pee IDIEKERENCES DUE TOVAGE 44 Of bititan.. Sh S29 ore ete AEM GeteruMmBALION tt. eee 2 SUR lA Uo.) een oe intraslocalityNecomparisoms 25%... 2). 2) oe 1 ee UNIVARIATE ANALYSIS OF GEOGRAPHIC VARIATION ...__.. PEGCECUITES ut eer AI Sek IC RN tno) RU We interpretation or the*hpures*: (Soe! BYE eine Wi4 os PII Results Shee est 4. stan e, Je ites 2). 0. Ob oe aR Correlations among the characters ...................... Correlations between males and females ................. Sexual -Gimonp nism este thy: a, Hee RD SOU el. aR DIscussION OF UNIVARIATE COMPARISONS ................... Hearsand bodyweight’? 88.208 Ogee se EE ul AG Bilésize and shape’ So 9 ei ie ee eee AaTSUS 9 ee. eee: Ree eed elm: Ot ok pone Sheed Oegyt VV anne ara eCaTl eS, oy sie 2) See Pe hn at One Meh ee @olor Bere en, FEE Bb fel ee Hiei Sa Oe Sexualedimonphisms es Pe See et, Lea Wen ees MULTIVARIATE ANALYSIS OF GEOGRAPHIC VARIATION: Pe *CanonicalV ariates ‘Analysis’... 09" 20. 3). Methods 8) oe. feo et et ie Ret ae 2 eSUiS eet see, wee eat Bie bee I , PD is per cent difference and is the difference between adult and yearling means expressed as a per cent of the adult mean. 20 UNIVERSITY OF KANSAS Pusts., Mus. Nat. Hist. cant differences (P < .05) between adults and subadults. In these four cases and in the remaining sixteen non-significant tests, adult birds had larger mean bill lengths than subadults. Because of the consistency of this trend subadult males were omitted from later analyses involving bill length. The range of per cent difference (PD) between subadults and adults was 0.1 to 6.9, and the mean PD weighted according to sample size was 2.8 per cent. The abso- lute differences ranged from 0.03 to 1.66 mm with a weighted mean of 0.67 mm. There were no systematic geographic trends of PD of male bill length over the study area, and a regression of PD on isophane was not significant (less than 2 per cent of the variation in bill length was attributable to variation in isophane). Bill lengths of adults and subadults were positively correlated in their distribu- tion over central North America (r = .74; P < .01). In females age variation of bill length does not parallel that of males. Only one sample (Loc. 31) showed a significant difference between adults and subadults. A total of sixteen of twenty-four samples tested were such that adult birds had longer bills, but many of these differences were minute. A sign test was applied to test the null hypothesis that the number of locality samples with adults averaging longer bills and the number with yearlings aver- aging longer bills occur with equal frequency. The results were non-significant, indicating no separation of age classes was nec- essary. In another icterid, the Boat-tailed Grackle (Quiscalus [Cassidix] mexicanus ), bill length averages slightly less in subadult birds than in adults (Selander, 1958). Bill height and width. Bill measurements other than length showed no meaningful age differences. There were no significant differences in bill height and lower bill width in males, and the number of samples with adult birds larger was roughly half of the total in both characters. Two samples of males (Locs. 15 and 50) had significant age differences in upper bill width, with older birds being larger. However, of all samples tested only about half showed adults that were larger. For females, bill height had one sample with a significant difference, (Loc. 19), lower bill width had two (Locs. 5 and 6), and upper bill width had no significant samples. For these characters female adults were larger than yearlings in roughly half of the samples. No separation of age classes is war- ranted for any of these characters. Tarsus length. Tarsus length appeared as a relatively variable character in these tests. Significant differences between male year- GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS DAN lings and adults were found in four samples; adults were larger in two cases (Locs. 11 and 54), while subadults were larger in the others (Locs. 5 and 15). Of the remaining sixteen non-significant samples, half had adults averaging longer tarsi. Similarly, in fe- males, significant differences occurred in two samples (Locs. 22 and 47), but only in the first did adults average longer. Of the remaining twenty-two non-significant samples adult females averaged larger than subadults in nine. The above results as well as sign tests applied to these indicate that no separation of age classes is war- ranted. Wing and tail measurements. Differences due to age were seen consistently in feather measurements (Tables 8-13). In both males and females, adults were larger than subadults in tail length, wing length to the longest primary feather (except for the female sam- ples of Locs. 52 and 53), and wing length to the first secondary feather. In males, statistically significant differences were seen in length of the tail, wing length to the primary, and wing length to the secondary in eighteen, twenty, and seventeen samples respec- tively, while in females significant differences occurred in seventeen, twelve, and ten samples, respectively. To determine if the degree of difference in feather measurements between adults and subadults exhibits meaningful patterns of geo- graphic variation, the per cent difference (PD) was calculated for each locality and plotted on maps. In addition, linear regression analyses of PD on isophane were made. Visual inspection of the map-plots showed no major trends or obvious patterns of geographic variation of PD. The distribution of high and low per cent differ- ences appeared random. The results of the regression analyses also indicated that variation of PD does not follow variation in isophane. In no case was the proportion of PD sums of squares attributable to variation in isophane greater than .012. Also calculated for feather measurements were (i) product- moment correlation coefficients (7) of adult sample means with first-year sample means and (ii) grand means of per cent difference (PD) and absolute difference (AD) of adult and first-year sample means (Table 14). In all cases the correlation coefficients were posi- tive and, except for tail length in males, all were significantly dif- ferent from zero. The degrees of correlation in feather measurements between adult males and subadult males indicate that geographic variation in the two age classes is roughly parallel, although there is certainly not a one-to-one relationship. The variation in the dif- ferences between age classes is attributable particularly to indi- 22 Untversiry OF Kansas Pusts., Mus. Nat. Hist. vidual variation in subadult birds. The date of fledging as well as nutritional factors during the time of the complete postjuvenal molt could affect the rate of growth and the ultimate length of the flight feathers. Thus, the general positive trend in correlation seems to indicate that the same selective forces are operating on the length of flight feathers in both age classes, but the moderate values of r suggest that age and growth factors or environmental factors or both may well impose an amount of local variation. In females correla- tion coefficients between adults and yearlings are higher than in TABLE 8.—Comparisons of Adults and Yearlings, Wing Length to the Longest Primary Feather in Males. Loe. no. tapas ee PD P Ae ee ly Ob G (3s) 121.6 (9) 3.2 19.01*** jee ae eels eel 2686807) 121.0 (4) 4.4 21.09*** Tee eal! A 2bu a COS) 119.8 (5) 4.7 15.19*** Be cites eese ee 125.7 (24) 119.5 (4) 4.9 30.15*** iit 8 ano 124.2 (30) 120.7 (6) 2.8 0:94*+* 11 124.0 (11) 119.8 (5) 3.4 7.95* 12 125.4 (12) 119.8 (4) 4.5 9.52** 13 121.6 (11) 118.4 (5) 27 15.88*** 15a eee PM EIZBLO UE) 119.8 (6) 4.9 13/82** 23 124.7 (24) 120.0 (9) 3.8 17.24*** 26 125.5 (11) INT 1) 6.3 7.59* 34 124.6 (15) 118.5 (4) 4.9 29.12*** Shwe eres Hl Tele onI (40) 11972 (5) 3.2 O°G8*** AY sspeq testy el op LES. Lala) 118.7 (6) 3.6 1S 740 ne 45 121.9 (29) 115.0 (3) ier 12.90*** 46 123.6 (12) 1ISey (3) 4.0 6.20* 48 126.6 (18) 119.1 .(7) 5.9 43. 84*** 49 129.2 (11) 1207-7) 6.6 a4 30%* BOSE Coe eel | 129.4 (16) 123.4 (5) 4.7 14.10*** 54 | 12120 (13) 114.7 (6) 5.2 24. 57*** GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS 23 TABLE 9.—Comparisons of Adults and Yearlings, Wing Length to the Longest Primary Feather in Females. Loe. no. Pit) eee flee F | Lorde > ee 103.8 (10) 101.3 (3) 2.4 4.625" Seles cists tee 105.0 (19) 101.3 (4) 3.6 PAB 3 oe ade. 103.4 (8) 100.0 (3) 3.3 14.09%2° OPS ae es: 103.5 (8) 101.8 (8) Dai, 5.12* pcan eee 103.3 (8) 101.4 (5) 1.8 1.91 (Ske i cee 103.3 (8) 101.8 (8) 1.5 4.34 CS oe, Eee 101.9 (11) 99.0 (5) 2.9 4.12 7 pyr 99.8 (11) 95.3 (4) 4.6 HSS 22 103.5 (4) 100.0 (3) 3.4 3.89 24 100.4 (5) 98.9 (10) 1.5 1.48 27 103.9 (14) 102.6 (7) 1.3 3.83 29 101.0 (17) 98.3 (4) 2.7 5.36* 3. ae res ars 100.3 (23) 98.3 (3) 2.0 2.96 31 101.5 (20) 98.7 (7) 2.7 9.28** 32 103.5 (11) 102.8 (5) 0.6 0.81 33 104.7 (11) 101.0 (6) 3.6 1187755 34 101.5 (10) 98.8 (6) 2.6 5.54* 37 101.2 (5) 98.0 (3) 3.2 9.00* 44 98.8 (6) 96.3 (3) 2.5 1.58 45 99.4 (25) 95.5 (6) 4.0 LO; 10s 47 100.7 (10) 99.3 (6) 1.4 2.16 BO Ren Pare Ot 10621((17) 103.0 (3) 2.9 4.66* LV ARSE eee 102.0 (4) 103.0 (3) 1.0 2.14 Bates ay Rice: 101.2 (5) 101.3 (3) O.1 0.01 males. As with males a nearly one-to-one relationship is not ob- tained, suggesting at least a partial effect of age and growth or en- vironmental factors. It is interesting that the per cent (PD) and absolute differences (AD) as averaged over several localities (and weighted according 24 University OF KAnsAS Pusts., Mus. Nat. Hist. TABLE 10.—Comparisons of Adults and Yearlings, Wing Length to the First Secondary Feather in Males. Loc. no ee Pst ED F thee | CCN 95.8 (33) 92.2 (9) 3.8 20: 20*** 5 97.4 (7) 92.8 (4) 4.8 20 1Gee= 7 95.5 (38) 91.2 (5) 4.5 14.54*** Seretoaden alts 95.9 (24) 90.0 (4) 6.1 23 .92*** 9 95.4 (31) 91.0 (6) 4.6 23 .29*** 1 94.4 (11) 90.6 (5) 4.0 6.21* 12 95.3 (12) 91.5 (4) 3.9 4.01 13 92.3 (11) 90.0 (5) 2.5 8.23** Vee Meters ool: 96.8 (4) 90.7 (6) 6.3 livswonte DBsigtenct dacs ccteu 94.8 (24) 90.1 (8) 4.9 22.05" 26 96.1 (11) 89.3 (3) 7.0 0.557% 34 96.5 (15) 89.3 (4) 7.5 40.83*** 37.. 93.7 (41) 89.2 (5) 4.8 2p:a0° 40... 93.9 (14) 88.7 (6) 5.5 SLL tte 45 93.8 (29) 87.0 (3) dee 28.37*** 46 95.4 (12) 94.3 (3) i 1.09 48 97.9 (18) 90.6 (7) 0.5 58. 84*** cA! id | Le eee 99.5 (10) 92.1 (7) 7.4 34.34*** 50 99.7 (16) 93.8 (5) 5.9 16.92 "5% a ee ae 93.5 (18) 88.2 (6) ey) 1S511**4 to sample size) for all feather characters are lower in females than in males (Table 14). The differences between males and females in the grand means of PD and AD are highly significant (P < .005). Thus, geographic variation in first-year females more closely approx- imates that of the adults than is the case in males, and first-year females are more similar to adult females at each locality than are first-year males to adult males. It is known that Red-winged Black- birds are polygynous and that females breed in their first year but that males do not (Nero, 1956; Orians, 1961). It is possible that the greater difference between age classes and the retarded maturation GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS Bb in males, coupled with postponement in time of first breeding, is a consequence of polygyny. There is a low probability of a young male obtaining mates in competition with older, more experienced males, and males which postpone first breeding until the second TABLE 11.—Comparisons of Adults and Yearlings, Wing Length to the First Secondary Feather in Females. Hoe-sn0 eas Riess ) PD F ie ieoker een 79.3 (10) 78.3 (3) 122 0.43 SECO 80.1 (19) 76.3 (4) 4.8 2229089 aa Seo eete 78.8 (8) 77.0 (3) 2.2 3.08 ero Ohara 3, 21 78.4 (7) 77.5 (8) 22 1.51 Meee phos 78.4 (8) 76.2 (5) 2.8 5.22* Beye nests 78.8 (8) 76.8 (8) 2.5 6 .05* 1 Se Ree 76.8 (11) 75.4 (5) 1.8 2.63 72.4) Ar ae 76.4 (11) 72.3 (4) 5.4 they eam ae Bo, 78.3 (4) 75.0 (3) 4.2 3.15 7a, VN eRe 77.0 (5) 75.5 (10) 129 1.56 76 (CO ae 79.0 (14) 1520 (7) 1.3 2.02 7) ee ae 1629 (17) 75.5 (4) 1.8 3.38 rigs hie ine 76.7 (23) 74.7 (3) 2.6 5.99* 3 Eee eee 78.0 (20) 75.6 (7) 3.1 157s 2 Ale oe oc: (CL) 79.2 (5) 0.3 0.12 33 79.8 (11) C20) 3.3 13 344°7> Ch ACen See 77.9 (10) 75.8 (6) 2.7 4.90* 2G eee 1136%(5) 75.0 (3) 3.4 8.30* 44.. 76.3 (6) 73.0 (3) 4.4 2.71 Jee eee 76.5 (25) 73.8 (6) 3.5 12 .43*** cl ee 77.6 (10) 76.7 (6) 1.2 2.32 SUC D Sa 81.1 (17) 79.0 (3) | 2.5 3.82 THO ae ee 80.0 (4) 79.7 (3) 0.4 0.36 52) 5) eee 78.6 (5) 77.3 (3) 1.6 0.43 26 Unrversiry OF KANsAS Pusxs., Mus. Nat. Hist. TABLE 12.—Comparisons of Adults and Yearlings, Tail Length in Males. Loe. no. ey Dee ee 93.1 (31) 5 94.3 (7) 7 92.5 (35) eu As ae Saas 92.4 (22) 9 92.6 (29) |) ee 92.2 (11) Meee stoa,'s Grane 91.8 (9) aera seeks. ns 89.3 (11) A eee ate ope 3 90.5 (4) ee er 91.8 (22) 7 | ar ee 93.1 (9) 5 eae 94.4 (14) Sie Senne 90.9 (40) 40.. 91.4 (13) AD aes Daca eey: 91.8 (25) 46 92.6 (11) 48 eae 94.6 (14) 49 97.6 (10) Oe n arrest 96.1 (16) BA, Senter 91.3 (13) Yearling means (mm.) 88.2 (9) 85.0 (3) 87.0 (5) 87.0 (3) 86.5 (6) 86.0 (5) 86.5 (4) 86.8 (4) 85.0 (6) 85.0 (9) 86.0 (3) 85.3 (4) 86.8 (5) 85.8 (6) 83.0 (3) 89.5 (2) 85.2 (6) 86.7 (6) 90.5 (4) 82.3 (6) ED bo — ou Ot. Os Co -Ovs Ov Oi cu onwnonrnrsnvoo Oo w CO Cone. Oe Olmak Hee CO NT Ni O3 O Oo WS Oo — ID F 21. 63*** 5Q*** _AQ*** 1077 067 = A aid .08** .93 1967 BLE lidad 207° on :61** -O4*5* s6D*%2 15 SOtee* 5297 ** 109375 Bh Pac year may ultimately produce a larger number of offspring (Selander, 1965). Since geographic variation in subadults is expected to follow gen- erally that of adults a separate series of interlocality comparisons was not warranted. In addition, sample sizes of yearlings were too small for this kind of comparison. In order to reduce intrasample heterogeneity, subadult birds are not included in later comparisons involving tail length, wing length to the primary, and wing length to the secondary. Selander and Giller (1960) noted in male Red-wings from Texas GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS oi that wing length (to the primary) and tail length average shorter in subadult than in adult specimens. No significant differences were found between subadult and adult females (with a sample size of 59). Mean wing length in fifty-four subadult males was 113.7 mm versus 118.0 mm in seven adults. These give a PD of 3.6 and AD TABLE 13.—Comparisons of Adults and Yearlings, Tail Length in Females. Adult Yearling PD Loc. no means (mm.) | means (mm.) PF Ae cbacasree 76.2 (10) 72.7 (3) 4.6 4.78* Sa OE pee ae 77.5 (19) 71.3 (3) 8.0 19:60*** Ly hed rare ae 74.6 (8) 72.3 (3) 3.1 6 .23* Ghia: 74.4 (7) 74.0 (6) 0.6 0.12 Ui Sicn ra! 75.6 (7) 70.6 (5) 6.6 132622" ob SOE BAY dant Ct) (L(G) 2.8 TAS% MONS steph sire ar 75.0 (11) 71.0 (5) 5.3 Dede Peres en 73.2 (11) 67.5 (4) 7.8 a Oo ee eee 74.8 (4) (Lae (3) 4.1 3.80 DATA oh ise S 73.0 (4) 70.8 (9) 3.0 2.61 Niet ena had Ph 75.9 (10) 72.8 (5) 4.1 6.54* ORR NC a a ass: 73.7 (14) 70.3 (3) 4.6 4.95* MO ee gio gs he 73.2 (22) 68.7 (3) 6.2 O783t** Sa eee ae 72.9 (17) 70.8 (6) 2.9 3.64 Dome ee et 75.0 (8) 1123) (6) 4.3 16667" 33 76.8 (9) 72.5 (2) 5.6 (08s 34.. 74.6 (8) 70.8 (5) dae 11,560%* Ol. 3. 75.6 (5) 70.5 (2) 6.7 13.56* 44 72.8 (6) 69.7 (3) 4.3 107 45 72.8 (19) 69.0 (4) 5.2 5.32* 47 74.1 (8) 70.0 (4) 5.6 1. 102* 50 79.4 (15) 75.7 (3) 4.7 5.67* 52 76.7 (3) 75.5 (2) 1.6 0.95 53 74.0 (5) 71.7 (3) 3.2 2.00 28 UNIVERSITY OF KANSAS PuBLs., Mus. Nat. Hist. TABLE 14.—Product-moment Correlation Coefficients of Adults with Yearlings and Grand Means of Per Cent Difference (PD) and Absolute Difference (AD, in mm) between Adults and Yearlings. Males Females r 1240) AD r PD AD Marliensths. sees. .3l 6.8 6.3 om 4.8 3.6 Wing length (primary)| .76** 4.3 5.4 .82** 2.6 2.6 Wing length (secondary)...... .60** 5.2 5.0 FOO 2.6 2.0 sae Dede LOU of 4.3 mm, which are slightly less than the over-all averages I obtained. Color. As shown above, the amount of time a female Red-winged Blackbird has been housed in a museum affects its rating on a color index. Older females appear lighter due at least to oxidation of the melanin in the feathers. Although many of the older specimens were deleted from the comparisons of age classes, there remained an undetermined amount of variation which precluded really mean- ingful comparison of color differences between adult and yearling females. In ventral color one sample (Loc. 27) of the twenty-four tested showed a significant difference (P < .05) between adults and yearlings, and in this case the adults were darker. Of the re- maining twenty-three nonsignificant comparisons seventeen showed the adults to be darker. A sign test indicates this is not a significant departure from equality (.05 < P< .10). In dorsal color six sam- ples showed significant differences (Locs. 3, 24, 27, 30, 31, 33), and of these the adults averaged darker in five. Of the remaining eigh- teen nonsignificant samples the adults were darker in a total of fifteen, which, combined with the significant comparisons, is a sig- nificant departure from equality. Colors of adults and yearlings tend to be positively correlated over the study area, but this relationship is not strong. For ventral color r is .59, and for dorsal color r is .52; both are significant. Although a trend is suggested in which year- lings are paler than adults, considerable variation exists among localities, and I chose to retain yearling females in later comparisons involving color. Heart and body weights. Data were available for heart and body weights from only a few localities. No differences were found be- tween adults and subadults in these characters. GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS 29 Summary. In consequence of the above results, subadult birds were not included with adults in analyzing geographic variation for bill length in males, and tail length, wing length to the longest primary, and wing length to the first secondary feather in both sexes. UNIVARIATE ANALYSIS OF GEOGRAPHIC VARIATION Procedures Patterns of geographic variation are here investigated by tests of significance and non-significance of differences among means of lo- cality samples. Among the several statistical methods used in the comparisons of means the t-test serves well in the comparison of any two samples. If n localities are sampled, n(n—1)/2 com- parisons between pairs of means can be made for each age and sex class. Modified Dice-Leraas graphs usually allow adequate graphic comparison of samples, but for multiple comparisons the method is statistically inappropriate. In addition, Dice-Leraas graphs are perhaps best applied in comparisons along a linear gradient, such as ambient temperature, or comparisons which are one-dimensional, such as a clinal transect or along a coast line. A thorough review by Sokal (1965) is recommended for readers interested in statistical techniques in systematics. A variety of “multiple comparison tests” exists for a posteriori comparisons in analysis of variance. Generally, multiple comparison tests are used to detect significantly different subsets of samples within the total collections of samples. Tukey’s, Duncan’s multiple range, and the Student-Newman-Keuls (SNK) tests are probably the best known. Descriptions of these tests may be found in Steel and Torrie (1960). Sokal and Rinkel (1963) and Sokal and Thomas (1965) have used the SNK test in analysis of geographic variation. The multiple comparison test used here is Gabriel’s (1964) Sums of Squares Simultaneous Test Procedure (STP). The Application of this method to analysis of geographic variation is discussed in detail by Gabriel and Sokal (1969). For each character this test was ap- plied after an over-all analysis of variance revealed that significant differences existed among the means. One advantageous property of STP is that of transitivity, which provides that no samples or groups of samples can be judged hetero- geneous (significantly different) if they lie within a subset of homo- geneous (non-significant) samples. The STP method uses a single “experiment-wise” error rate. A single variance (the pooled, within- group mean square in an over-all analysis of variance, MS) serves 30 UNIVERSITY OF KANSAS PusLs., Mus. Nat. Hist. for the determination of maximal non-significant subsets. Each sub- set O of some or all of the k means is judged non-significant if Bsa MS kT) eer where SSq is the sum of squares of the subset Q, MS is defined above, and F is the critical value of the F distribution at the desired probability level a (.05 was employed in this study) with k—1 and N —k degrees of freedom where k N= > (n,x—1). i=l The sums of squares of a subset containing say m samples is given as where S; and n; are the sum and sample size, respectively, of the ith sample. In the present analysis samples were ranked in decreas- ing order of the magnitude of their means, and sums of squares are calculated by sequentially adding samples until a maximal non- significant subset is found. The process of deletion and calculation of new subsets is repeated until all samples have been included in at least one non-significant subset. In the STP analysis the probability of making any type I error, that is rejecting the null hypothesis when it is in fact true, in the entire set of comparisons among means or sets of means is a, where a is the probability level of the selected F statistic (.05, here). This probability a is called the “experiment-wise” error rate. Since the probability of any type 1 error at all is a, the probability of error for any particular test of some subset is necessarily something less than a. Thus, tests of subsets the significance level is less than the experiment-wise error rate, and the tests are not as sensitive as ex- pected to differences between individual samples or differences within small subsets. On the above grounds it seems better to use an a of .05 rather than a smaller value. Gabriel and Sokal (1969) have recommended calculating maximal non-significant subsets by considering only those samples that are from contiguous localities. The criterion of contiguity may be de- GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS ai fined by an example. Locality A is contiguous with B, and vice versa, if B is geographically closer to A than is any other locality. A third locality C may also be contiguous with A, and vice versa, even though the distance A — C is greater than A — B if C is the nearest locality that lies at least 90° of arc from a line running from A to B. The comparison of only contiguous localities and subsequent cal- culation of subsets would avoid situations such as found here with the character bill length, where by listing means in decreasing order of their magnitude the samples from localities in the northwestern part of the study area may be non-significant with samples from localities in the southeast. Clearly, no locality in the northwest is contiguous with any in the southeast. There is no doubt that the use of contiguities has advantages in simplifying the STP results and in reducing the number of comparisons to be made. However, owing to the particular arrangement and number of localities in the present study, this method was not used because there were many systemat- ically valid and interesting comparisons among localities which could not be made without violating the rules for contiguous comparisons. I believe that the method adopted here of arranging means in decreasing order of their magnitude and indicating maximal non- significant subsets on this arrangement is perhaps not the simplest method of presentation for the reader, but for this particular study it seems the most efficient method. Information on most combinations of locality samples is available. The interested reader could extract from the figures any specific information regarding tests of con- tiguous localities. For each locality the mean, standard error, standard deviation, coefficient of variation, and range have been calculated for each character. (These statistics along with sample sizes comprise an appendix to Power, 1967, doctoral dissertation having the same title as this publication and available through the University of Kansas Library, Lawrence, Kansas, or University Microfilms, Ann Arbor, Michigan. ) Interpretation of the Figures Before proceeding to the results of interlocality comparisons the interpretation of the accompanying figures should be clarified; Figure 4, a map of variation in bill length of males, will serve as an example. Beside the map in each figure the means for the character are listed by locality, along with locality number, in decreasing order of magnitude. (More specific locality information than ap- pears on the maps is given in Table 1.) The lines to the right of the means each represent a maximal non-significant subset of means, 32 UNIvERSITY OF KANSAS Pus.s., Mus. Nat. Hist. being made up of those means which are included within the range of the line. The probability level is .05. Thus, the left-most line defines a subset ranging from 25.29 mm (Loc. 2) to 23.67 mm (Loc. 12) in which no mean is significantly different from any other (i. e., the subset is homogeneous). Groups within this subset also show no statistically significant differences between themselves. For ex- ample, Locs. 1, 2, and 3 in northern Alberta and Northwestern Ter- ritories as a group are not significantly different from 10, 11, and 12 in central Manitoba. Pair-wise comparisons may also be made. Thus, bill length at Loc. 2 (25.29 mm) in northeastern Alberta is not significantly different from that at Loc. 12 (23.67 mm). The question may arise as to what the relationship between two localities really is since, for example, in the first subset Locs. 2 and 12 are shown to be not significantly different, but in the second to sixteenth they appear to be significantly different. Recall that for the second subset and beyond Loc. 2 was dropped from the comparison, thus allowing additional means into the subset at the other end of the scale. Our question as to bill lengths at Locs. 2 and 12 clearly is answered as a function of which of the other means in the study are included. If we are concerned with bill lengths just at Locs. 2 and 12 irrespective of others, we may say that the two are in fact not significantly different because they have been included together in at least one of the maximal non-significant subsets. This last decision is not based solely on the within-group variation of the two samples (as would be done in a f-test) but is based on the pooled estimate of within-group variation from all localities in the study. The fol- lowing general rule may be set: two or more means, or two or more groups of means, are judged homogeneous only if all means concerned are contained within the same maximal non-significant subset(s). Note, too, that what is significant or non-significant statistically may have different biological meanings. For example, mean bill lengths of males at Loc. 2 in Alberta and at Loc. 44 in central Missouri are the two largest in the study and are from the same statistical population. Biologically, however, we cannot say that northwestern Alberta and central Missouri Redwings represent a single “population” and that their respective gene pools have more in common with one another than, say, Loc. 2 with 8 (Saskatche- wan) and 44 with 36 (Iowa). Gene pool “similarities” cannot be based on only one character, and of course, several samples in com- parison with which Locs. 2 and 44 are judged significantly different geographically separate Locs. 2 and 44. More important in assessing GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS 33 the degree of similarity may be the fact that samples 2 and 44 both represent breeding populations found in areas adjacent to forests (coniferous forest in the north and deciduous forest in the south), whereas the central samples are from more open country. Thus, important questions about bill length might relate to the ecology of the two samples and the factors that are responsible for the pre- ponderance of long-billed birds. The variously shaded circles at each locality are an aid in esti- mating the pattern of variation. Solid circles represent the largest means, open circles the smallest, and three-quarters, half, and one- quarter solid circles represent logical intermediate values. The intervals were arbitrarily calculated by dividing the total range of the means into five equal parts. Two exceptions are heart and body weights which, because of the small number of samples, have the total ranges of the means divided into three equal parts. Occasional bars between localities indicate those adjacent localities which are judged significantly different, and may represent barriers to gene flow. Results Bill length. In bill length (Fig. 4) the non-significant subsets include a large number of means, which cover in turn a wide range of localities. The pattern of variation for both sexes involves long Fic. 4. Map of variation of bill length of males (left) and females (right) in central North America. Means are ranked in decreasing order of magnitude along with locality number; lines to the right of the means each represent a maximal non-significant subset resulting from comparisons by STP (for further explanation see text). Occasional bars separating localities indicate those adjacent localities which are significantly different from one another. 3—4331 34 UNIVERSITY OF KANSAS PuBLs., Mus. Nat. Hist. bills in the northwestern and central Canadian portion of the area and in the southeast, particularly Missouri, southern Illinois, Iowa, and southeastern Nebraska. These areas include means ranging from about 23.7 to 25.3 mm for males and 20.0 to 20.8 mm for fe- males, and are roughly separated by lower mean bill lengths in the central plains and in Colorado. A regression of bill length on isophane was barely significant for males, with only 8.4 per cent of the variation in bill length being attributable to variation in isophane. A similar regression coefficient for females was not significantly different from zero. Bill height. Variation in bill height (Fig. 5) is roughly clinal; in males the highest means (12.30 to 13.08 mm) occur in the north- Fic. 5. Map of variation of bill height in males (left) and females (right). western part of the study area and elsewhere in Canada, with inter- mediate and lower values being rather scattered throughout the plains of the United States. In the U. S. there is a tendency for thick bills to occur in the western plains, and this has been partly responsible for the racial separation of the western “thick-billed Redwing” (A. p. fortis) from the nominate race in the east. Many of the western samples, however, are not significantly different in bill height than samples taken well within the supposed boundaries of A. p. phoeniceus (compare, for example, central Colorado with central Illinois). A significant linear relationship is found between bill height in males and isophane (Fig. 6), where 54.6 per cent of the variation in the character is attributable to variation in isophane. Variation in bill height in females is similar to that of males, with the highest means being found in the northwestern part of the study GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS oD BILL HEIGHT OF MALES (mm) BILL HEIGHT OF FEMALES (mm) Y = 856 * OO35X Y= 985 * 0043X ° 38.2~C«‘ia 50 54 58 62. 66 50 54 ISOPHANE ISOPHANE Fic. 6. Regression of bill height in males (left) and females (right) on iso- phane. Both regressions are highly significant (P < .005); 54.6 per cent of the variation in males, and 50.6 per cent in females, is attributable to variation in isophane. area (Fig. 5), and with 50.6 per cent of the variation in the char- acter being attributable to variation in isophane (Fig. 6). One notable difference for females is that samples from central Missouri and Illinois are relatively larger in bill height than would be ex- pected from the variation in males, with the lowest values being restricted more to the central plains. Width of the lower mandible. Variation in lower bill width of both males and females is depicted in Figure 7. The general pattern 2 z LOWER BILL WIDTH (MALES) e 5 10.50 LOWER BILL WIDTH (FEMALES) 120 Seri0 TOO 30 «10041 Fic. 7. Map of variation of the width of the lower mandible in males (left) and females (right). of variation is best summarized by the regression analyses of lower bill width on isophane (Fig. 8). Here it may be seen that the changes are clinal, with widest bills occurring in the northwest and. 36 UNIVERSITY OF KANSAS PuBLs., Mus. Nat. Hist. n n np oS p eS 8s 8 @ BILL WIDTH / BODY WEIGH LOWER BILL WIDTH OF MALES (mm) Y= 72/ + Q028X Y = 830 + 0032X f-2.325 2.300 8.0 Lo wat * 1 Ss8 62 66 38 492 46 ° 38 42 46 58 62 66 50 54 50 54 (SOPHANE !SOPHANE Fic. 8. Regression of width of the lower mandible in males (left) and females (right) on isophane (solid circles). The regressions, applying only to the solid circles, are highly significant (P < .005); 71.0 per cent of the variation in the character in males, and 81.3 per cent in females, is attributable to variation in isophane. Open circles apply to the right-hand scales, which are mean bill width as a ratio of the cube root of body weight for those localities in which weight data were available; no regression lines have been drawn for these values. narrowest in the southeast. Isophanes decrease from north to south as well as from west to east, a pattern which is almost identical to that of lower bill width. In males 71.9 per cent of the variation in the character is attributable to variation in isophane, and for females the value is 81.3 per cent. Weight data were available for ten samples of males and seven of females, and with these few samples it was of interest to see if adjusting bill width for the weight or size factor had any effect on the definite clinal pattern of geographic variation in bill width. Ratios of mean lower bill width to mean cube root of body weight were calculated and plotted as open circles on Figure 8. (It should be noted that calculating a ratio by dividing two character means is satisfactory for sample comparisons, but it is not numerically equivalent to taking the mean of the individual ratios). In both males and females the data suggest that bill widths relative to weight are larger in the northwest, and over the study area become smaller in the east and in the south. In addition, judging from the fact that bill height, the width of the upper mandible, and the width of the lower mandible are all highly correlated with one an- other (Table 15), the variation of the former two characters would be largely unaffected by adjustments for weight. Width of the upper mandible. Here, as in the last character, the width of the bill decreases from north to south and from west to east (Fig. 9). In males 70.7 per cent of the variation in upper bill width is attributable to variation in isophane, and in females 80.7 per cent of the variation is accounted for by variation in isophane (Fig. 10). GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS or UPPER BILL WIDTH (FEMALES) 120 Fic. 9. Map of variation of width of the upper mandible in males (left) and females (right). xd > N ny os) fo} a to UPPER BILL WIDTH OF MALES (mm) 2) oO T p @ ee Y= 4.90 + Q027xX 38 42 46 50 54 58 62 66 38 42 46 50 54 58 62 66 ISOPHANE ISOPHANE Fic. 10. Regression of width of the upper mandible in males (left) and females (right) on isophane. The regressions are highly significant (P < .005); 70.7 per cent of the variation in the character in males, and 80.7 per cent in females, is attributable to variation in isophane. Tarsus length. There seems little if any geographic variation in absolute tarsus length (Fig. 11). However, in relation to body weight tarsus length is relatively shorter in the northern portion of the study area. The weight factor was removed in ten samples of males and seven of females (all that were available) by dividing the mean tarsus length by the mean cube root of body weight for each available sample. When regressed on isophane a negative slope was found (Fig. 12) indicating that relative to body size tarsi are shorter in the north. This relationship is much stronger in fe- males than in males. The regression slope for males is not quite significant (.05 < P < .10). Inspection of the graph suggests this non-significance is not due to the absence of the trend but to the 38 Unrversiry OF KANsAs Pusts., Mus. Nat. Hist. TARSUS LENGTH (FEMALES) Fic. 11. Map of variation of tarsus length in males (left) and females (right). T = aaa T T 7 fez) o n @ Y= 635 - 0.00726 X OF FEMALES D N a a) 7T it o w =] a = fir 5 = Be is = w 66 io ~ Feat Sten a = »~ ke on 5 7 B ss ~ cae Pe ; 8 < =< ° . ~ 65 63h me 2 o A = a| 2 bs SS fo a =, 4 a B Pe FB 7 6.4 62 = A pene Sky A L n 1 sil ° 38 42 46 50 54 58 62 66 38 42 46 50 54 58 62 66 ISOPHANE ISOPHANE Fic. 12. Regression of tarsus length expressed as a ratio of cube root of body weight in males (left) and females (right) on isophane. The regression in males is not significant (.05 << P< .10), but see text; 34.0 per cent of the variation in the character is attributable to variation in isophane. The regres- sion in females is highly significant (P < .005); 86.3 per cent of the variation in the character is attributable to variation in isophane. fact that a curvilinear regression would provide a better fit to the data. Tail length. In males the total range of mean tail length is more than 8 mm, while most of the maximal non-significant subsets in- clude a range of about 4 mm. This, along with the map (Fig. 13), suggests that the character is consistently variable throughout cen- tral North America. A regression on isophane is significant (P < .05) but explains only a small part (10.9 per cent) of the variation. The GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS 39 dominant pattern is depicted in Figure 13 where it is seen that tails are longest in the western plains of the United States and in Alberta and Northwest Territories, Canada. Another area of moderate to long tails includes Wisconsin and Illinois. A “trough” of low values occurs through the central plains. TAIL LENGTH (MALES) ~~53 98.0 TAIL LENGTH (FEMALES) Y= 96.32 - O2045X wo o =a) a ) £ wo ow . . . . / ee . . =e TAIL LENGTH OF FEMALES (mm) = ~ uo TAIL LENGTH OF MALES (mm) 72r 90 4 | a . 7h a . es: —— = 4 SS ee iar ———e—E aol aol... i Et 10 12 22 24 10 2 4 2 8 20 22 24 14 ‘6 18 20 : AVERAGE PRECIPITATION (in) AGE PRECIPITA Fic. 14. Regression of tail length in males (left) and females (right) on precipitation (April to September, inclusive; averaged from records from the years 1899-1932, for U. S. samples only). The regressions are significant (P < .005 for males and .005 < P < .01 for females); 24.4 and 26.0 per cent of the variation in the character is attributable to variation in precipitation. In females the pattern is similar (Fig. 13). The total range of variation is 8.4 mm and the range of most subsets is about 4 mm. Longer tails are found in the western plains and in the northwest. In Illinois and in southeastern Nebraska, Iowa, and parts of Minnesota another high area is found. A regression on isophane is significant 40 UnrIvERSITY OF KANSAS PuBLs., Mus. Nat. Hist. (P < .01) but explains little (17.0 per cent) of the variation in tail length. Although the relationship is not strong, there is a tendency for samples from more arid regions of the plains to have longer tails than samples from wetter areas. An inverse relationship between tail length and inches of rainfall (April to September, inclusive ) is seen in both males and females (Fig. 14). Only specimens taken in the United States were used in these analyses. Precipitation in inches, from April to September, was averaged from records for the years 1899 to 1932 (U. S. Dept. Agric., 1941). These values are given in Figure 2 for the localities concerned. Wing length to the longest primary feather. The longest wings of males and females are found in the western plains of the United States and in Alberta and Northwest Territories, Canada (Fig. 15). so 129.4 WING LENGTH TO LONGEST PRIMARY (MALES) 51 129.4 WING LENGTH TO LONGEST PRIMARY (PEMALES) 9 129.2 91 18 124.5 41 12 ow 3 Fic. 15. Map of variation of wing length to the longest primary feather in males (left) and females (right). Through central Saskatchewan, Manitoba, and Ontario mean wing length is also relatively great. Lower values occur in the central plains, with the lowest of all in eastern Kansas, Oklahoma, and Missouri. It is also of interest to note that along the western edge of the study area the values in Canada decrease from north to south whereas in the United States the decrease is in the opposite direction, or from south to north. Regressions of wing length on isophane are given in Figure 16. Because weights are known for only a few samples, ratios of mean wing length to mean cube root body weight could be calculated. Unfortunately, weight data were lacking from the western plains GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS 4] & 8 3 6 + 8 $ uw @o WING LENGTH / {BODY WEIGHT tof “tell a ua a ~~ WING LENGTH / 3/SO0Y WEIGHT . 122 A Y«H63+* OI70X 33 o Y« 93.7 + OI73X WING LENGTH (PRIMARY) OF MALES (mm) 3 WING LENGTH (PRIMARY) OF FEMALES (mm) . 58 62 66 38 42 38 42 46 6 50 54 ISOPHANE 50 54 ISOPHANE Fic. 16. Regression of wing length to the longest primary feather in males (left) and females (right) on isophane (solid circles). The regressions, apply- ing only to solid circles, are highly significant (P < .005); in males 34.0 per cent and in females 39.4 per cent of the variation in the character is attributable to variation in isophane. Open circles apply to the right-hand scales, which are ratio of mean wing length to mean cube root of body weight for those localities for which weight data were available; no regression lines are drawn for these values. where wing length is greatest. But with the few samples available the results suggest that wing length varies as a function of, rather than independently of, body size. The relationship of the two char- acters is suggested further by the high, positive correlation shared by wing length and cube root of body weight (Table 15). Tail length and wing length to the first secondary feather show the same relationship to weight. In addition to a general increase in wing length with increase in isophane, a second relationship exists between wing length and rain- fall. As with tail length, only United States samples were considered. In males and in females (Fig. 17), there is a fair inverse correlation between wing length and inches of rainfall (April to September, is) o fo ~] Y «107.02 - 0.288 X i) o fo} o Y= 129.92 -0322X fe} a ro) B 103 oo WING LENGTH (PRIMARY) OF MALES (mm) WING LENGTH (PRIMARY) OF FEMALES (mm) 8 R . 22 24 10 12 22 24 10 12 i94 16 18 20 14 16 18 20 AVERAGE PRECIPITATION (in.) AVERAGE PRECIPITATION (in.) Fic. 17. Regression of wing length to the longest primary feather in males (left) and females (right) on precipitation (April to September, inclusive; averaged for the years 1899-1932). The regressions are highly significant (P < .005); in males 47.3 per cent and in females 39.7 per cent of the variation in the character is attributable to variation in precipitation. 42 Universiry OF KANsAS Pusts., Mus. Nat. Hist. inclusive ) at each locality. Thus, birds from the more arid western plains have longer wings than those of wetter regions in the south- eastern plains of the United States. Wing length to the first secondary feather. In both sexes wing length as measured to the first secondary feather generally follows that of wing length to the longest primary. Maps and the results of STP are given in Figure 18, and regressions on isophane are given in Figure 19. ING LENGTH TO FIRST SECONDARY MALES) ING LENGTH TO FIRST SECONDARY (FEMALES) —~ 232 iid Tr = Bub SB o AAAAAAAIAAAAAAAIAAAR Sap ssssasss! soot Rasa “ue WG | e anueet bee ms wen Ssne o LaBSIVESLSBRESRA | NL ok 8 OF me OO Nd Fic. 18. Map of variation in wing length to the first secondary feather in males (left) and females (right). (mm) S (mm) £ » ey LENGTH (SECONDARY) OF MALE WING LENGTH (SEGONDARY) OF FEMALES. Y= 717 + O1Z8X WING wo 38 <2 46 50 3¢ 58 62 66 Fic. 19. Regression of wing length to the first secondary feather in males (left) and females (right) on isophane. The regressions are highly significant (P < .005); in males 23.1 per cent and in females 38.1 per cent of the varia- tion in the character is attributable to variation in isophane. Epaulet color in males. Although there seems to be a tendency for males in the southwestern portion of the study area to have redder epaulets, there are not enough consistent differences or similarities in central North America to warrant the conclusion that a definite 43 GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS *(}4SU) sayeulaz UI IO[OO [esiop puv (19}U90) SaTeUIa; UT IO[OO [eI}JUOA BOGKFEPEIP PENI BGDIDIIMIIANINMIGAING Bese rrr rr er errs eran (SA TVWG4) YOTOD ‘Ivsuod (SHIVWGS) YHOIOD IWULNAA gt *(}Je[) sepeul url 1ojoo yajnedea jo uonerea jo dep ‘0% “OT SY ool Ol ” ” TTIFVWMMOMMMMIMNAIIIOGAIAAAUTANAIAGGAINIAIIIIOMIIINANNAN ANNAN be (SHTIVW) YO1OD LatAVaT 2 7 ° a 44 UnrIversiry OF KANSAS Pusts., Mus. Nat. Hist. pattern of geographic variation exists in this character (Fig. 20). A regression of epaulet color on isophane was not significant. The differences among localities are generally small, as the range of the means is only slightly over 1.5 points on the five-point color scale. Ventral and dorsal color of females. The ventral and dorsal color of females (Fig. 20) have similar patterns of geographic variation. Females appear slightly darker in the eastern and southeastern parts of the study area, primarily Missouri, Jllinois, Iowa, southeastern Nebraska, Wisconsin, and parts of Minnesota. The birds are palest in areas of the United States west of the localities just mentioned. In Canada the color is more toward the over-all average. There is a very strong line of demarcation separating birds of Missouri, Iowa, and southeastern Nebraska from paler birds to the west. This is the greatest divergence between adjacent localities found in the entire study. The regression analyses of ventral and dorsal color on isophane explain little (5.5 and 13.6 per cent, respectively) of the variation, and, in fact, a negative slope of the regression line for ventral color was not statistically significant from zero. ect T =——T T i D [7 ToT: terre Fo 55 x eat Yu 269 + OO771X Y= 236 + Q0950X 50+ = 2. 7 5 ° ore) te} (sc b uo COLOR IN FEMALES Db fo} DORSAL COLOR IN FEMALES (score) w a VENTRAL 22 24 4 14 16 I 20 22 24 10 12 14 16 18 20 AVERAGE PRECIPITATION (in) AVERAGE PRECIPITATION (in) 10 Fic. 21. Regression of ventral color eke) and dorsal color (right) in females on precipitation (April to September, inclusive; averaged for the years 1899- 1932). The regressions are significant (.005

wo ne tl o b N T CUBE ROOT HEART WEIGHT OF FEMALES B @ . \ . . . | Y= 0524 + 0.00I0X Y = 0426 + 0.00/2X CUBE ROOT HEART WEIGHT OF MALES 9 J a . 38 42 46 50 54 58 62 66 38 42 46 50 54 58 62 66 ISOPHANE ISOPHANE Fic. 23. Regression of cube root of heart weight in males (left) and females (right) on isophane. The regressions are significant (P < .05); in males 39.8 per cent and in females 57.7 per cent of the variation in the character is attributable to variation in isophane. Cube root of body weight. Only ten samples with data on weights were available for males and seven samples for females. In both cases there is a trend of decreasing body weight from the north- western to the southeastern portions of the study area (Fig. 24). The variation is well represented in regression analyses (Fig. 25), where 67.5 per cent of the variation of male body weight is attributable to 46 UNIvERSITY OF Kansas Pusts., Mus. Nat. Hist. variation in isophane and a very high 92.1 per cent of the variation of female weight is attributable to variation in isophane. CUBE ROOT OF BODY WEIGHT (MALES) 120 110 100 $0 we nae jo| © | @ | Fic. 24. Map of variation of cube root of body weight in males (left) and females (right). D fo a9 | rm) 3 b L w oO a wr ow nO to} 7 \e Db on T =: b Y= 3573/ + 00078 X ea Y¥=3/0I + 00082 X 3.95 + 3.40 peu cane a 1 i n J 0 38 42 46 58 62 66 38 42 46 58 62 66 CUBE ROOT BODY WEIGHT OF MALES rs . 4 CUBE ROOT BODY WEIGHT OF FEMALES \ 50 50 ISOPHANE ISOPHANE Fic. 25. Regression of cube root of body weight in males (left) and females (right) on isophane. The regressions are highly significant (P < .005); in males 67.5 per cent and in females 92.1 per cent of the variation in the character is attributable to variation in isophane. Correlations in Variation Among the Characters From the foregoing results and the accompanying illustrations a general idea is obtainable as to the degree of correlation or similarity of variation among the various characters. A more accurate state- ment of this relationship is obtained with correlation coefficients Product-moment correlation coefficients were calculated between characters using sample means (off-diagonal elements of Table 15). It should be pointed out for this and for following sections that a 47 GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS pue sojeur usemjoq suonear09 9g" pyysiem Apog ‘[BUOSLIP-YO IOMOT OY} Ur ore veneeieerieel pp tee GL" 86° — 10c= £9" resins ate Piette age ec:= abr — [ete etfee eee ees] gge— ez: epics» ols | caps 49° oF: soeceeenee| gg 16: gz: a ee --| 39° ee: go0.— eee Bn hac 6a: opts 8 [poe es Bose ny es co: er: stro Well gyi: 26° SF: seeceeeeeel orem 1s tein |e ole ae “008 Peet at 20100 pee 60° — (ud) que] sulM 6° Tg" £0° gO" 98° 06° tL £6" AU oe q33ue| eL *IO[OO [VSIOP ST ,,g IO[OD,, Puw ‘IO[09 [vyUeA *sJOOI 9qNO 0} PaUIOJsuRI] 9B SPYBIAM ‘p ST ..[ 10[0D,, So[PUlez JOY ‘os *IO[OO Jo[Nvda st saywut IOF ..[ IO[OD,, “q ‘(co’ > d) yBOYTUsIS o1e soley UT sonfeA ‘[eUOSeIp oY} BuOTe UaATS oI so[eulay so[euloy IOF Son[BA puB UOT}OS [BUOSseIP-YoO Joddn oy} UL o1v soleUL TOF SoN[eA ‘y tL 16° jhe VL: or — 08s— or — 61° - 8a" 89° es" 89° Lis Is 83° Ls: Ig’ 06° 48 06 or £8 SS" or" Cae Sue 96° 68° qypIa Tq IOMO'T 86° aoe AS GB" a 97° 1° 63° 08° 9L° 18° er qysrey ma #1431I0m Apog oo sqgSiom 4rBezT sIelsiotels "9% OOD EAM KO OO el 10[0D (‘0a8) Y4Bue, Bury Cid): mauey uray eee s ss uqBuer [eT ++ +++ yqSue] snsiey, “UV pra [tq zaddy ** 1} pI [[Tq JOMOo'T aevsiers **-qus1eYy Td tees esa suel [ta 4939] a y SopeUla PUB Saye] UVAMjOq PUP SIoJORIvYT) UIIMJeq SJUDTIOYJIOT) UOTR[AIIOT) yusWOUl-JONpPOIG—'CT A'TAV.L 48 UNIVERSITY OF KANSAS Pusts., Mus. Nat. Hist. correlation coefficient between characters calculated with sample means is an amalgam of within-group and between-group covari- ation. In both males and females, bill height, and widths of the upper and lower mandibles are highly correlated with one another in their patterns of variation, as is each to body weight. Tail length and the two wing lengths (primary and secondary feathers) also share a high degree of intercorrelation and are highly correlated with body weight as well. In considering the correlations between certain bill measurements and body weight, and between feather measurements and body weight, it should be kept in mind that body weight is neither equally nor well represented from all regions of the study area (see Fig 24). In females, ventral and dorsal color are very highly correlated. The results also show that males and females are alike in the correlation between characters. Correlations among characters are discussed further in the sec- tion on multivariate statistical analysis which deals with the extrac- tion of factors from parts of the correlation matrices. Correlation Between Males and Females Males and females generally show similar patterns of geographic variation. The relationship between the sexes is expressed as cor- relation coefficients calculated on sample means (diagonal elements of Table 15). All correlations are significant (P < .05) except for tarsus length, which showed no geographic variation using abso- lute values. Sexual Dimorphism For each of the characters except those concerned with color, a per cent difference (PD) was calculated for each sample by ex- pressing the difference between the mean for males and the mean for females as a per cent of the mean for males. This value was used as a measure of relative sexual dimorphism. For each character the distribution of PDs was inspected over the study area and regression analysis of PD on isophane was carried out. The results show that there are no obvious trends or patterns of geographic variation in the degree of sexual dimorphism. The mean PDs, range for each character, and number of samples for mean PD are given in Table 16: GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS 49 TABLE 16.—Per Cent Differences (PD) between Males and Females. Number Character PD Range of samples BMMMen eh ee lee sss ee ee 16.5 12.8-19.0 46 OL REIGN te is.< sic: aR Ae 13.9 10.5-18.4 48 Lower bill width.............. 13.3 11.0-16.2 48 Wpper bill-wadth ss)... 0:78 sae 11.0 8.7-14.4 48 MATSUBONGEN A tryna, oe: 11.3 9.4-13.6 48 mail length 35 toni tat. ee daet 19.3 16 .4-22.2 38 Wing length (primary)......... 17.9 16.3-19.2 38 Wing length (secondary)....... 18.3 15.2-19.7 38 Heartiweight*..5.. 2 eiscnsacohst 15 11.3-16.9 10 BOGy Weight ®sr.o vincsvac Secs. 14.3 13;55-15.2 6 8, Weights are transformed to cube roots. DISCUSSION OF UNIVARIATE COMPARISONS Heart and Body Weight Owing to insufficient data on weight, an interlocality correlation coefficient between heart weight and body weight could not be calculated. However, their respective patterns of variation suggest a direct correlation, implying heart weight probably varies as a function of body weight. Generally, in the eastern and northern parts of the plains where data are available, variation in body weight is positively correlated with latitude in concordance with Bergmann’s rule. Bergmann’s rule, simply stated, is that among the forms of polytypic, endo- thermic species, body size tends to be larger in cooler parts of the range and smaller in the warmer areas. Increase in weight is gen- erally thought to indicate a decrease in the surface/volume ratio of the organism, and vice versa. Since the rate of heat flow is related to the surface area across which heat transfer can occur, intraspe- cific changes in surface/volume ratios (i. e., from change in size) may represent approaches toward thermoregulatory homeostasis, a problem facing birds that have broad geographic ranges (Mayr, 1956; Hamilton, 1961; but to the contrary, see Scholander, 1955). The winter distribution of most populations of Red-winged Black- 4—4331 50 UNIVERSITY OF KANSAS PuBLs., Mus. Nat. Hist. birds in central North America is a result of a general shift south- ward from the breeding grounds (A. O. U., 1957; Campbell, 1936). Those northern populations which experience a cooler breeding season are also expected to experience a colder winter than birds of lower latitudes. There is some evidence from banding recoveries that in winter Red-winged Blackbirds and certain other icterids may funnel into the rice belt states of Tennessee, Arkansas, Missis- sippi, Louisiana, and Texas (Meanley et al., 1966). Although in winter there may be a compression into more southerly localities, it seems highly unlikely that there is “leap-frog” migration or that there are wide differences in the spatial relationships among pop- ulations between breeding and wintering ranges. Thus, in relation to adaptations concerning thermoregulation, interpopulation dif- ferences reflecting temperature on the wintering grounds may be generally inferred from knowledge of the breeding distribution. Bill Size and Shape Variation in bill size and shape may be discussed from the stand- point of (i) Allen’s rule, which states that among the forms of poly- typic, warm-blooded species, extensions of the body tend to be longer in warmer parts of the range and shorter in the cooler regions, and (ii) functional adaptation in relation to food selection and feeding habits. From present evidence it seems more likely that variation is related primarily to variation in diet and feeding habits. The partial evidence that bill size and shape may vary independently of weight further suggests that selection is operating on the bill itself, independently of the size of the bird. A correlation among bill structure, behavior, and diet, both within and among species, is well established (Hinde, 1959; Huxley, 1942; Lack, 1944). Likewise, relationships between feeding habits and bill size and shape have been suggested for the genus Loxia (Kiri- kov, as cited in Mayr, 1963:324), Parus (Snow, 1954a, 1954b; Betts, 1955), Galapagos Finches (Bowman, 1961), Alectoris partridge (Watson, 1962), and Aphelocoma jays (Pitelka, 1951) to cite just a few examples. In a study of food-getting adaptations in the Icteridae, Beecher (1951) found varying with diet the mass and length of the bill, degree of angulation of the commissure, length of the orbital process of the quadrate, and the degree of kinetics (“kinetic” expresses the fact that the palato-pterygoid framework is movable with respect to the cranial parts of the skull). Beecher (1951: 417-418) noted: GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS Py It was found that the finch-like Cowbird [Molothrus ater] has an annual diet averaging 77.7 per cent seeds and 22.3 per cent insects. Correlated with this is a short, massive bill with a highly angulated commissure (134°), reduced kinetics (33° of arc), a quadrate with the lever-like orbital process much elongate, and palatine foramen relatively far posterior. The Red-winged Blackbird, Agelaius phoeniceus, and the Rusty Blackbird, Euphagus carolinus, are progressively more insectivorous in their annual diet (26.6 per cent for the former, and 53.0 per cent for the latter). Correlated with this, movement in the skull is progressively increased (35.0° in Agelaius, 43.0° in Euphagus), and the bill is elongated. At the same time the palatine foramen advances forward, and the bill becomes less massive and more forceps-like. Finally, the orbital process of the quadrate becomes shorter. These are, then, skull changes accompanying a more insectivorous diet. Inferring from the differences Beecher found among species it might be expected that the Red-winged Blackbirds of the western plains, with short, thick bills, are adapted to consume a greater pro- portion of seeds than the thinner-billed birds from the eastern plains and elsewhere. Dietary differences among populations may not be in the proportion of seeds and insects consumed, however, but in the kinds of seeds and insects, as Kear (1962) and Bowman (1961) have shown for other species. This matter should be investigated in Red-winged Blackbirds. Another factor, one which affects bill length only, is wear of the tip. Wear may vary with the type of substrate on which food is found or with the type of food. During the breeding season Red- wings eat mostly insects and other animal matter, while consuming mostly vegetable matter at other times of the year (Allen, 1914; Beal, 1900; Bird and Smith, 1964). A summer increase in bill length is found in Red-wings, Tricolor Blackbirds (Agelaius tricolor), Brewer's Blackbirds (Euphagus cyanocephalus), and certain other passerines which eat mostly insects in the summer and are almost entirely vegetarian in the winter (Davis, 1954). A seasonal change in bill length in Boat-tailed Grackles (Quiscalus [Cassidix] mexi- canus) has also been found by Selander (1958). It is interesting that the birds from Rolla, Missouri, (Loc. 44) have bills as long or longer than many from northern populations. There is also a similarity in the habitats of the birds from these two regions. In the north, marshes are in stands of coniferous forest, and, in the south, Rolla is in an area with much deciduous forest. Along the eastern edge of the plains in the United States much Red-wing habitat is in deciduous forest or in a grassland-deciduous forest ecotone. The significance of the apparent correlation of long bills and forest association is unknown, beyond the generality concerning o2 UNIVERSITY OF KANSAS PuBLs., Mus. Nat. Hist. proportions of insects in the diet available from Beecher’s (1951) work. An analysis of variation in the bill surface/volume ratio is given elsewhere (Power, 1970). Tarsus Tarsus length relative to body weight decreases at high latitudes and increases at low latitudes. Presumably, selective pressure for relatively shorter tarsi in the more northerly parts of the breeding and wintering ranges relates to heat retention and the maintenance of an efficient surface/volume ratio. Or, conversely, relatively longer tarsi may aid in heat dissipation in warmer areas. It is known that arterio-venous countercurrent retes found in the legs of many birds and other animals are related to heat conservation (Scholander, 1955). In the Sparrow Hawk (Falco sparverius) when ambient temperature is lower than body temperature heat loss is regulated in part through vasomotor activity in the unfeathered parts of the tarsometarsus (Bartholomew and Cade, 1957). In the California Quail (Lophortyx californicus) the unfeathered portion of the legs is known to provide an important accessory pathway for heat dis- sipation (Brush, 1965). Wing and Tail In some cases, birds from northern populations of migratory species have longer wings and tails than more southerly populations of the same species (Williamson, 1958). This is most marked in long-distance and obligatory overseas migrants such as the Wheatear (Oenanthe oenanthe). However, the association is not simple since wing and tail length may or may not fluctuate with body size (generally measured as body weight). For example, the farthest migrating populations of the plover Charadrius hiaticula have rel- atively small wings (Salomonsen, 1955). Within certain species of Parus, wing length is inversely cor- related with winter temperature (Snow, 1954a). There is also a “latitude effect” in that with similar winter temperatures (measured as the mean temperature of the coldest month) wing length tends to be greater at lower latitudes than at higher latitudes. Snow considers wing length to be directly correlated with body weight and in his discussion of the adaptive significances of patterns of variation, body size rather than wing length per se is given the attention. The degree of correlation between wing length and body weight is not known, however, since Snow (1954a:20) states only that “. . . in all except one of the species [P. major, in which British weights were GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS 53 unexpectedly high] for which weights are available, wing-length [was] found to be a satisfactory indication of general size.” It would be desirable to know what “satisfactory” means quan- titatively. Thus, using wing length to indicate body size, Snow thought that a general increase in relation to low winter temperature was in accord with Bergmann’s rule, whereas the second factor or “latitude effect,” which depresses the operation of Bergmann’s rule, relates to the short feeding day at high latitudes. It was suggested that, since a larger bird requires a larger absolute amount of food per day. “ selection will act against the attainment of the size most efficient for heat-conservation” (Snow 1954a:12). In Red-winged Blackbirds at least two factors seem to be in opera- tion. First, considering all but the western plains of the United States, wing and tail lengths increase with latitude. Since these measures are correlated with body weight (Table 15), and since geographic variation seems lacking in ratios of mean wing length to mean body weight (cube root), then wing and tail lengths may simply vary as a function of body size, the latter varying according to predictions of Bergmann’s rule. A second factor is considered in the plains of the United States, where in the west the longest wings and tail are recorded. Regression analyses of wing and tail lengths of United States samples on precipitation show a good deal of variation but a definite trend for an increase in wing and tail length with decreasing rainfall. The inverse relationship is stronger in wings than in tails. For males 47.3 per cent and for females 39.7 per cent of the variation in wing length is associated with variation in pre- cipitation. With tail length the figures are 24.4 and 26.0 per cent, respectively. An “aridity factor” has been noted by previous authors. Hamilton (1958), for example, found that in intraspecific comparisons of vireos, wings were shorter in humid regions and longer in arid regions of North America. Hamilton designated this observation an “aridity effect” stating (1958:321) that “. . . within wide-rang- ing species of temperate and tropical distribution there may be a tendency for individuals of populations occurring in hot, arid regions to have greater wing lengths than individuals in hot, humid regions.” Similar variation in relation to aridity is found in Downy Wood- peckers (Dendrocopos pubescens), White-breasted Nuthatches (Sitta carolinensis), and Cardinals (Richmondena cardinalis) (Hamilton, 1958). In the Loggerhead Shrike (Lanius ludovicianus) populations with the longest wings are found in the arid lowlands of northwestern Mexico, while those with the shortest wings occur 54 UNIVERSITY OF KANSAS Pusts., Mus. Nat. Hist. in the more humid southeastern United States (Miller, 1931). Other examples are given by Salomonsen (1955) and elsewhere. In some cases shorter wings are characteristic of sedentary populations and longer wings of migratory populations of the same species. But for some kinds of birds the sedentary forms are also residents of more southerly and more humid habitats. Thus, it is difficult in some cases to determine which, if any, of these factors has primary selective influence on wing and tail lengths. Red- winged Blackbirds from the southeastern United States, which are both relatively sedentary and in a warm, humid environment, have short wings (about 113.5 mm in males, and 93 mm in females) and tails (about 88 mm in males, and 70.5 mm in females), according to Howell and van Rossem (1928). Clearly a variety of factors in the environment may directly or indirectly exert a selective influence on wing and tail length (Hamil- ton, 1961). If variation in wing and tail length is a direct function of variation in body size then with increases in these measures with increase in latitude, or more directly ambient temperature, selection may concern surface/volume ratios in relation to thermoregulation and heat loss. Variation in body size may also relate to efficiency in migration (Salomonsen, 1955). In addition, in arid regions larger individuals may have a metabolic advantage, or conversely, in more humid regions a smaller body may have greater survival value. Hamilton (1961:184), for example, speculated that for vireos “the increase in wing length in warm, arid regions is an indication of size increases that facilitate conservation of metabolic water, and that the decrease of wing length in warm, humid regions is indicative of size decreases that facilitate effective heat dissipation.” In certain desert birds, larger bodied individuals and species lose relatively less res- piratory moisture than birds of smaller size (Bartholomew and Dawson, 1953; also of interest in this regard is Dawson and Schmidt- Nielsen, 1964). Several factors may have a direct effect on the structure of the wing. Relative intraspecific increases in wing length and wing area may be correlated with distance of migration, and prove to be advantageous for sustained flight. Also important is the shape of the wing. Rensch (1934), for example, noted that migratory individuals have more pointed wings (longer and narrower) than more seden- tary members of the same species. A second factor may be altitude, and although no correlation between wing and tail lengths and altitude was detected in Red-winged Blackbirds, a positive relation has been found with other birds. For example, Moreau (1957, 1960) GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS 55 found that in Zosterops complex of Africa, wing length is cor- related with altitude independently of temperature. According to Hamilton (1961: 185), Stresemann has suggested that a relatively larger surface and length of wing may be an adaptation for flight in the thin air of high-altitudes. A third factor directly affecting wing and tail lengths may be the density of the vegetation (Hamilton, 1961) or the distance between patches of suitable habitat. The longest winged and longest tailed Red-winged Blackbirds occur along the western edge of the Great Plains in the United States. In this region the density of the vege- tation in and immediately adjacent to breeding sites does not appear markedly different from that of other areas. However, the country is more open in areas where the birds wander to feed. Also, rel- atively vast expanses of dry grassland, and in some cases desert, separate the marshy areas and irrigated farmland in which Redwings are found—a situation that does not occur in eastern and northern portions of the study area. Since birds in the western plains may have to fly greater distances during normal diurnal activity longer wings and tails may be advantageous. Clearly the environment presents a multiplicity of factors and interactions which may have evolutionary impact on variation. In addition nutritional factors may affect body size and wing and tail lengths. At this stage the cause-and-effect relationship between the environment and wing and tail lengths is confusing for at least two reasons. First, it is not know in all cases if wing and tail lengths merely reflect body size. And second, changes in factors such as latitude, altitude, precipitation, humidity and aridity, migration distance, density of the vegetation, patchiness of the habitat, and the like are often correlated with one another. In cases where such factors are not highly correlated the singling out of one factor often imposes an oversimplification. Elsewhere, I have attempted to sort out some of these factors by the use of multiple regression analysis (Power, 1969). Color There is no trend or pattern of variation in male epaulet color. Since the red epaulets of males probably serve a signal function in courtship and defense of territories, it may be significant that there is no marked variation in color. A similar situation exists in Purple Martins (Progne subis), where females and first-year males are brighter and paler in Sonora, Mexico, than in Kansas, but adult males are of uniform color throughout the range (Johnston, 1966a). Color of the body feathers in females and first-year males is thought 56 UNIVERSITY OF KANSAS PuBLs., Mus. Nat. Hist. to vary due to differential selection relating to crypticity and, per- haps, thermoregulation. In adult male martins, uniform color may reflect selection for display plumages. However, one problem with Johnston’s interpretation of the adaptive significance of variation in females and first year males is in breast coloration; it would seem that selection relating both to crypticity and efficiency in thermoreg- ulation would be maximal for dorsal rather than ventral coloration. In Red-winged Blackbirds, variation in female color tends to follow Gloger’s rule. For samples taken in the United States the in- tensity of color is roughly related to amount of rainfall and presum- ably to humidity as well. Thus, more heavily pigmented females are found in the southeastern portion of the plains and paler birds tend to occur in the central and western plains. An increase in pigmenta- tion with increases in humidity or rainfall has been found in many other bird species (Hamilton, 1958; Johnston and Selander, 1964; Packard, 1967; and Snow, 1954a). Bowers (1959, 1960), working on Wrentits (Chamaea fasciata) in the San Francisco Bay region, found a strong correlation among the degree of pigmentation, temperature, humidity, and substrate color, which suggested a causative mech- anism underlying Gloger’s rule. Bowers (1960:118) states “ in temperate regions with relatively high humidity, low temperature, and frequent cloudiness, the plants respond by growing luxuriantly; this in turn allows the production of dark, moist soils with abundant humus that is more nearly screened from the bleaching action of the sun than in other situations. Through the agency of natural selec- tion, then, the darker colors of Wrentits in time would be favored and the birds come to match the surroundings in this dark habitat.” Johnston (1966b, and pers. comm.) has shown that interpopulation variation in breast color of female House Sparrows is correlated with soil color. Packard (1967) has also discussed color variation in House Sparrows. The marshes in which female Red-winged Blackbirds nest prob- ably do not differ much among localities in background color. However, the substrate of fields in which Red-wings feed may be paler in more arid regions of the plains. Also possibly significant, but unknown, is the substrate color of the wintering quarters. A correlation of color with substrate suggests crypticity and the ad- vantage of protective coloration against visually oriented predators as at least one selective factor. Nothing more definite can be said regarding evolution of color at this time since the physiological basis of color variation is not clear, and the selective advantages of genes responsible for differences are not always evident (Mayr, 1963:324). GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS 57 A very strong line of demarcation separates samples from south- eastern Nebraska, Iowa, and Missouri (Locs. 35, 36, and 44, re- spectively) from samples in southeastern South Dakota, north- central Nebraska, and northern and central Kansas and eastern Colorado (Locs. 31, 34, 45-47, and 48, respectively ). This line sep- arates adjacent localities which are significantly different. The trend of variation in this region probably has biological significance, but the magnitude of the difference may be somewhat exaggerated. Of the samples characterized by dark females, the Missouri locality is in deciduous forest and the localities in southeastern Nebraska and Iowa fall into deciduous forest and the deciduous forest-grassland ecotone which extends into the classically recognized grassland biotic community. Thus, at first glance there seems a correlation between the occurence of dark females in dense vegetation and pale females in dry, open grassland. However, the eastern samples characterized by darker birds were collected in 1966, and the paler birds to the west were taken in 1949, and it is known that a de- tectable amount of post-mortem color change occurs within this time span (see Chronological Variation). Differences in the date of collecting may contribute to the significant difference between Loc. 52 (largely 1965-1966) in southwestern Kansas and Locs. 47 and 48 (largely 1949) in northeastern Kansas and Colorado seen in com- parisons for dorsal color. I think the trend of variation seen in color of females in the southern plains is real and has biological signifi- cance, probably along lines discussed above, but it is also likely that the magnitude of difference—and the fact that several adjacent localities tested significantly different while variation elsewhere is gradual—is artificially imposed by the differences in the time of col- lecting. Time differences probably also account for the fact that the westernmost samples of Montana, Wyoming, and Colorado (Locs. 27, 33, and 51), all collected in 1966, rate slightly darker than specimens taken just to the east in eastern North and South Dakota, eastern Nebraska, and northeastern Kansas, for the most part collected in 1949. Sexual Dimorphism Red-winged Blackbirds fit in the general trend for the family Icteridae of increased sexual dimorphism in polygamous or promis- cuous species (Orians, 1961; Selander, 1958). Presumably, selec- tion affecting color in females is related to the advantages of cryp- ticity and the role of females in caring for eggs and young. Males do not care for or feed the young. Factors governing larger size and more striking color in males are most likely related to sexual selec- 58 UNIVERSITY OF KANSAS PuB3s., Mus. Nar. Hist. tion, where a premium is placed on a bird’s ability to obtain and hold a territory and one or more mates. Experiments have shown that adult males (females were not tested) respond differently to the sex of another Red-wing when only visual cues are available to them (Noble and Vogt, 1935). The fact that males are more boldly colored than females and that owing to their epigamic and agonistic behavior they are more frequently exposed to predators possibly results in an increased mortality among males as suggested by the unbalanced tertiary sex ratios in favor of females found at breeding colonies (Meanley and Webb, 1963; Orians, 1961). Vari- ous aspects of mating systems and sexual selection, as these pertain to Red-winged Blackbirds, have been treated by Selander (1965). The evolutionary significance of sexual dimorphism also relates to differences in food and feeding habits. If the abundance of food of a certain size or kind, say, were a factor limiting the number of birds in a population (in this regard see Lack, 1966), then the probability of survival of any individual would increase if it fed on a more readily available food source. The availability of food of a certain kind depends not only on its natural abundance, but on the number of organisms which also use the particular food source. Thus, given the evolutionary opportunity, a population may evolve behaviorally and/or morphologically such that its individuals ob- tain the greatest degree of success in feeding, specifically, and sur- vival in general. Such evolution may increase behavioral and mor- phological variability, within populations. As discussed previously, differences in bill size and shape within and among species often reflect differences in the nature of their food. Differences in food and feeding habits may be reflected by other characters as well. For example, in some hawks and owls, it is body size which often reflects food diversity, and in ducks and fla- mingos more specific modifications of the bill, such as the degree of serration, are related to food selection. The point is that the degree of intrapopulation variation in char- acters such as body size and bill size and shape is very probably related to factors such as the degree of food diversity of a popula- tion. Intrapopulation variation may be continuous or discontinuous (a case of the latter is polymorphism). For example, Van Valen (1965) compared island and mainland populations of six passerine species in which one population (usually the island birds) occurred in a wider variety of local habitats (“wider niche”) than the other, and he found that variances of bill measurements averaged larger for populations using a wide range of local conditions, and smaller GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS 59 in populations occurring in a more narrow range of conditions. More to the point here is discontinuous variation as found in cases of sexual dimorphism. Selander (1966) demonstrated a presumably adaptive relationship between the degrees of sexual dimorphism and sexual differences in foraging behavior in two melanerpine wood peckers (other species are also discussed). The insular His- paniolan Woodpecker (Centurus striatus) is strongly dimorphic (e.g., PD of bill length = 21.3) and shows sexual differences in foraging behavior in directions expected on the basis of differences in the bill and tongue. Conversely, the continental Golden-fronted Woodpecker (C. aurifrons) is moderately dimorphic (e. g., PD of bill length = 9.1) and shows less sexual difference in foraging. The divergence of the sexes in the insular species has allowed greater latitude in food and feeding behavior for populations as a whole, reduced intersexual competition for food, and may have allowed population densities greater than those maintained by continental woodpeckers. Selander thinks the divergence of island forms is related to reduced interspecific competition due to the absence or rarity on the island of other woodpeckers and birds with similar food requirements. In Red-wings, sexual differences in bill measurements ranging from about 11 to 16.5 per cent (upper bill width and bill length, re- spectively; see Table 21) would theoretically allow sexual differ- ences in diet. There are hints, but to my knowledge no conclusive evidence, suggesting such differences. Large flocks of females have been observed feeding regularly on beds of millet but males only rarely (Meanley, 1961). Males rather than females, however, have been observed in the heavier work of removing seeds from cones of loblolly pine (Meanley, 1962) and seeds from the fruit of white ash (Nero, 1950). It is also known that wintering flocks of Red-winged Blackbirds in Texas are largely unisexual (Mengel, 1965:443, and R. K. Selander, pers. comm. ). In summary, at this stage of our knowlege the evolutionary sig- nificance of sexual dimorphism may be accounted for in two ways: (1) Sexual selection occurs with selection pressure on males being partly related to the advantages of large size and bold color in epigamic and agonistic behavior and territorial defense, while selection pressure on female morphology may be partly related to the role of the female in parental care and the consequent ad- vantage of being cryptically colored; and (2) sexual dimorphism in over-all size and shape of bill may allow partitioning of food by size, permitting reduced intersexual competition for food and in effect increasing the variety of food available to the population. 60 UNIVERSITY OF KANSAS PuBLs., Mus. Nat. Hist. MULTIVARIATE ANALYSIS OF GEOGRAPHIC VARIATION Studies in systematic biology often compare groups of organisms by considering multiple characters. If the emphasis or interest is in detailing variation in the characters themselves then univariate com- parisons are applicable. If, however, interest lies primarily in the groups of organisms and their relationships, then the most efficient means of comparison is to consider as many characters as possible simultaneously through multivariate statistical techniques such as canonical variates analysis. Multivariate techniques may also simplify descriptions of characters. Factor analysis has been applied in this regard in the following section. With the availability of high- speed, electronic computers the application of such techniques to biological data has become practical. I. Canonical Variates Analysis Methods In this analysis only adult birds were used, and since canonical analysis considers all characters jointly, specimens with data missing for one or more characters were necessarily deleted from the study. In males this did not reduce sample sizes drastically, but did require that localities 15, 16, and 53 be excluded from the analysis. In females the effect was greater, requiring the dele- tion of twenty samples and re- sulting in the use. of only those localities shown in Figure 26. The characters body weight and heart weight could not be in- cluded because data were insuf- ficient. Thus, one color and eight linear characters were used for males and two color and eight linear characters were used for females. These are of course the same characters used in the uni- variate analysis. If only two characters are mea- sured on specimens of a sample then each specimen may be represented by a point in a simple two- Fic. 26. Samples of females used in the canonical variates analysis. GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS 61 dimensional space in which the dimensions correspond to the two characters. This 2-D phenetic space may be represented by a simple bivariate scatterdiagram on which points describing the specimens are plotted. If many localities are to be compared then the means of each sample may likewise be plotted in the 2-D space. With three characters a 3-D model might be construed with specimens or sample means plotted therein, and, as in the 2-D case, each char- acter represents an axis which is at right angles to every other axis. Beyond three dimensions the system is similar and here we may locate a specimen or a sample in an n-dimensional phenetic hyper- space, although a model may not be constructed. In the present study a single male specimen may be represented by a point in nine-dimensional space, in which the dimensions cor- respond to the nine characters used in describing the specimen. A single sample would be a small cluster of points in the 9-D space, the best representation of which is the point described by the means of all nine characters. For females the same sort of visualization is valid, but here we are concerned with 10-D space, corresponding to the ten characters used for females. Multivariate canonical analysis allows us to generate one new axis for each character and rotate these axes through hyperspace in such a way as to minimize within group variation and maximize among group variation. These axes are the so-called discriminant functions. Usually a few of the dis- criminant functions are all that are required to describe most of the variation among groups, and by projecting the group means onto these discriminant functions we may obtain the “best” views of rela- tionships among groups. “Relationship” here is of course in terms of the characters used. This is what has been done in Figures 27 and 28, where the bivariate scatter diagrams represent two-dimensional planes through the phenetic hyperspaces on which group means are projected. Other views may be obtained using other discriminant functions (Fig. 29). Further, the axes of the original characters may be projected onto the plane in order to indicate which of the char- actors contribute to the separation of locality samples. With a few bivariate and trivariate scatterdiagrams one may place the emphasis on among-group relationships and neatly summarize most of the information given in univariate comparisons as well as provide new information by considering characters jointly. Calculations involve obtaining variance-covariance matrices for each sample and pooling these matrices over all samples to obtain a within-group variance-covariance matrix W. A second matrix, the between-group variance-covariance matrix B, expresses the disper- sion of the sample means around their grand mean. The discrim- 62 UNIVERSITY OF KANSAS Pusts., Mus. Nat. Hist. inant functions are the principal components of matrix B after standardization by matrix W(BW-). Standardization adjusts the measurement of each character by the variance of every other char- acter and serves in maximizing among-group variation relative to within-group variation. The principal components of the standard- ized matrix B describe axes such that the first represents the greatest amount of among-group variation, the second the next greatest amount, the third the next greatest, and so on. For each axis there are coefficients (K) of the discriminant function such that K,Y, + KY, + ...-+ KY, = P, where K,,..., K, are eigenvectors or direction cosines, Y,, . . ., Y, are sample means for the n characters, and P is the point on the axis for a sample. Individuals may be plotted instead of sample means simply by substituting the indi- vidual character values (Y,, . . ., Y:) in place of means, assuming non-significant differences among within-locality variance-covari- ance matrices. Each discriminant axis has an eigenvalue or char- acteristic root which measures the dispersion of group means along the axis. An estimate of the per cent of the variation described by each discriminant function is obtained by the ratio of each eigenvalue to the sum of all the eigenvalues. The eigenvectors and eigenvalues for both males and females are given in Tables 17 to 20. In the present study some of the sample variance-covariance matrices were significantly different from others, and, thus, the pooled matrix W is an approximation. Inspection of sample matrices suggested that it was the covariances (reflecting intrasample cor- relations between characters) rather than the variances which con- tributed most to the significant differences among groups. Differ- ences among variances may be corrected by transformations of data but differences among covariances cannot generally be resolved in this way. A particularly helpful discussion of the application of multivariate discriminant analysis to a geographic variation study will be found in the introductory material of Jolicoeur (1959: 284-287); Dupraw (1965) and Jameson et al. (1966) have also carried out systematic studies using these techniques. More detail on the theory and pro- cedures of canonical variate analysis are available in Seal (1966: chap. 7) and in the references cited by the above authors. Results In Figure 27 where male sample means have been plotted on the first and second canonical axes it can be seen that northern popula- tions differ from more southern ones due to increased height and width of the bill and decreased tarsus length. Western populations 63 GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS ¢$¢0°0— S7L0= 0&2 0— L10°0— 009;0— LAR Selah O&1'0 600°0— 1000 qopnedq LOO SOF’ 0 0S2°0— $62 0— 660° 0— 608°0— 040°0 £960 910°0 (988) yy sue] SUI 090°0— 0gc'0— LTT°O— 80¢'0 6c0°0 9970 820°0 6E1 0 9€1°0 (‘tad) 43 3U9y BUI, 460°0— S¥0°O0— 9g¢°0 -00°0 F230 Scr '0— OLL O-— 20° 0— 620°0— Yysue] LEAF 290°T WIGsO = VALUE) 988° 0— cge'0 €242°0— LZ O— £90°0 Z0E°0— Yasue] sNnsiB J, 978 T C6P'G 606° T Sc0°T 967 = S6L°0 L¥S'O— 280°0— 968" T qypra M4 reddy 8£0°I— 80% I- S69°0 110°%— 160°1— 910°0- 082'0 O11’ 0— 929°0 uy pias IEG IaMO'T 17) OO oss "0— 798 0— PEL I PPS 0 090° T — 660°0 ser O- STE'T WsIey Wa 0&1 O— Tee'0 L6G 0— 9LE°0— IPe°0 T9€°0O oo O— FOL’ O— 1L0°0 q93ue] Wad worzoUuny queuruniiosiq ‘so[BJ JO SISATeUY SozeLIVA [ROTUOULD OF s10JPVVAUASIY— JT ATAVL UNIVERSITY OF KANSAS PuBLs., Mus. Nat. Hist. r0]09 [esi0q SiO = ge'0 LIZ°0—- 83E'0— I88°o— 006°0 $0E°0 929'°0 930 68Z'0— 26S O— €81°0 $e8 1 600 °% 006 °0 OFZ O— FL1°0 rau) a) 281°0— 266°0 882°0 162° 6— o&L'0 O2Ts0'= $82 0— 8Le°0 €Z1°0— OZ 0 Tee 0— o18 0—- 88Z°0 org 0— T61°0- (0) Gs oe Tge°0 g8s°0 008° 2— €80°T ceo'T 9€2°O— 9st '0— 6520 oor’ O— 0€8°0 LLE°0 €89°T $c8° I— 8LF°0 cIF'O L1G:0= 6c1°0 201 0—- 898°0— sZe°1= $520 0€6'°0 Stt-0o— 666° O— 8£0°0— ltr 0 F681 T6€°0— $F6'0 0€9°0 9st 0=— LET'O 8t0°0- €&6°0— 979 °0— ¥90°0 608 °0 890°0 FL1°0 £02°0 800°0 683° 0— £00°% €0¢°0 hy a | bir O— eS ae 4yBu9] 49309] ce fig q4s18q yyue] Sul SUM TeL SnereL syeddy JOMO'T ra Wa 10]09 [e1} 0904, oS omoot eee A uorouny qUBUIUOTIOSI(T ‘so[eUla,J JO SISA[TeUY soyeIIeA [POTUOURD OF SIOPOAUASI—'ST ATAV.L GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS 65 TABLE 19.—Eigenvalues for Canonical Variates Analysis of Males. Dicer ant Eigenvalue Per cent pees | Speen ue ract urd se 888 .953 37.9 PAs lees Se ey cusios acs 593.290 25.3 63.2 ] 00 beep pceaen 2 risa ere RC 5 210.053 8.9 72.1 | oor eR re Pat 181.693 Gall 79.8 \ AEE ads Ce eee 170.338 7.3 87.1 WA setae Drceidtereee oe oe 128.675 5.4 92.5 WEE Peer eee css pets 70.457 3.0 95.5 V6.0 0 Cae er Ae eee 59.037 2.5 98.0 LEX Gece artes: eke eh as 46.077 2.0 100.0 TABLE 20.—Eigenvalues for Canonical Variates Analysis of Females. Bean Eigenvalue Per cent ae Ns De he ek aes A ee 451.127 35.0 TSE geo tend en ane st 337 .286 26.2 61.2 SET Pay he Bld a GS oe 175.615 13.6 74.8 LVEE eS: See se 126.123 9.8 84.6 Wikee ce, ents e Sn nS 58.782 4.6 89.2 NG Ogee oP scarier eames a gee 45.278 3.5 92.7 Wilts Dette tettrer ae ge ete oe 44.439 3.4 96.1 VAIL ehh mon oth Steere at, 23.119 1.8 97.9 1 RIS SAS eed at fay Seal 20.456 LG 99.5 DG Re te Pec ts Cyn 7.064 0.5 100.0 of males on the other hand are characterized by longer wings and a decreased bill length. Also, western samples have slightly wider and thicker bills than eastern populations but are smaller in these characters than northern populations. The projection of group means on the discriminant functions reveals shape differences in the bill. Individuals of western populations tend to have short, thick 5—4331 66 UNIVERSITY OF KANsAs PuBLs., Mus. Nat. Hist. (3i7,9:%) FIRST DISCRIMINANT FUNCTION 16 \7 18 19 20 2\ SECOND DISCRIMINANT FUNCTION (25.3%) Fig. 27. Canonical variates analysis of 51 samples of male Red-winged Black- birds. The mean for each sample is plotted on the first and second canonical axes, describing a total of 63.2 per cent of the variation among groups. Samples are plotted with locality number and a symbol according to their present sub- specific status; circles, A. p. arctolegus; triangles, A. p. fortis; inverted triangles, A. p. phoeniceus. The directions for north and west give the approximate distribution of localities on the graph. The original coordinates of the char- acters are represented by vectors which may be imagined to be emanating from the grand mean (cross). The length of each vector corresponds to two standard deviations of the measurement (based on the pooled, within-group variances). The characters are: BL, bill length; BH, bill height; LBW, lower bill width; UBW, upper bill width; TR, tarsus length; TL, tail length; WI, wing length to the longest primary feather; W2, wing length to the first secondary feather; EC, epaulet color. bills, northern populations have relatively long, thick bills (generally thicker than western samples in the extreme northwest), and south- eastern populations are characterized by long, thin bills. It is in- teresting to note that tarsus length when considered jointly with other characters aids in discriminating groups whereas it was necessary to express tarsus length as a ratio to body weight in order to find the same pattern in a univariate analysis. Since about 37 per cent of the among-group variation is not shown in the scatterdiagram it is not surprising that details in the patterns of variation of some characters are lost. Generally speaking, the best separation of northern and southern popula- tions of males relies most heavily on differences in the thickness of the bill and to a lesser degree on tarsus length. The western samples on the other hand are separated on the basis of bill and GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS 67 wing length. An example of loss of detail is seen in the char- acter wing length, which, although is was rather variable among localities, showed in the univariate comparisons a trend of increase to the north as well as to the west. The increase in wing length to the north is not evident in the plane of the first and second dis- criminant functions. This does not reduce the utility of canonical variates analysis in geographic variation studies, but it does empha- size that the primary concern is in the relationship among groups, not in detailing trends of variation in single characters. Scatterdiagrams of canonical variates show additionally that variation among localities is largely gradual throughout the study area. This relationship is indicated by the fact that localities that are closer to one another on the diagrams are also generally closer to one another geographically. The subspecies A. p. phoeniceus, A. p. fortis, and A. p. arctolegus, rather than distinct entities, seem to be continuations of broad trends of variation representing adaptations to local conditions. The fact that the first and second canonical variates represent only 63.2 per cent of the among-group variation suggests that other views might reveal unique relationships among the localities. Several bivariate scatterdiagrams were made using combinations of canonical axes one through four; however, no additional information was obtained nor were the among-locality relationships further elucidated. These graphs are therefore not given here. In females (Fig. 28) southeastern and eastern U. S. populations are fairly well separated on the basis of being dark colored, short winged, and narrow billed. Western samples are not as distinctly separable from north-central U. S. and Canadian samples as in males, but the western females do lie in the direction of pale color. Tarsi appear relatively shorter in northern samples. From the univariate comparisons bill length is known to be greatest in northwestern and southeastern populations, and lowest in the central and western plains. The decrease in bill length in western and central popula- tions and the increase in southeastern populations is indicated in Figure 28, but the particular configuration of localities in this view does not allow the trend of increase in northwestern populations to be evident. The fact that 13.6 per cent of the among-locality variation is described by points along the third canonical axis suggests that a trivariate scatterdiagram of the first three axes would do somewhat better in assessing relationships among groups than the simple two- dimensional plot. In Figure 29 it may be seen that there is further 68 UNIVERSITY OF KANSAS PuBLs., Mus. Nar. Hist. as a . $ on (o>) N @ i 66 | —.3l .70 | —.23 43 Bilheighiawexss ee ieee .92 .06 .89 .09 25 Lower bill width.............. .90 12 oh l 27 | —.14 Wiper billswidth.!. 5... ¢:..-..< 22. .90 .09 .87 .09 —.17 Marsuslengtheset at. eeccse oe ot 18 .30 .02 — .38 Mailiengthe joc ee 04 .83 | —.06 .98 13 Wing length (primary)........ .23 .88 sll! .95 .08 Wing length (secondary)...... .10 .96 Bai hss .83 —.15 IB paulet: Color. ooce sciepac <'sc o s s — .27 Bae reat ceteatheets | sueahe eee cleo aka Wentralcolorsacmn rire fot 5 corse aeons tie ieee am eel: 10 .05 97 DMorsalicolonas wesc eterseis Ss Sheet esis... |. PSS oe —.01 .06 1.03 Correlations among 1.00 .16 1.00 yf —.14 Primary axes .16 1.00 aD Yh 1.00 —.17 have the highest loading of any of the characters. Thus, factor I appears to be a general “bill thickness factor.” Variation of factor scores (Fig. 30) shows a gradual decrease from north to south and from west to east, which correlates exactly with the patterns of variation of each of the three characters taken separately under univariate comparisons. Factor II for both males and females appears as a “flight feather factor” as indicated by the high loadings of tail length, wing length to the longest primary feather, and wing length to the first secondary feather. Factor scores are highest in the western plains, decrease gradually through Canada and the northern plains, and are lowest in the central, eastern, and southeastern plains of the U. S. (Fig. 31). 72 UNIVERSITY OF KANSAS PuBLs., Mus. Nat. Hist. MALE FACTOR 1 = FEMALE FACTOR I r = Fic. 30. Map of variation in locality scores for factor I (bill thickness factor) of males (left) and females (right). Factor scores were calculated as indi- cated in the text. In this and subsequent figures 10.0 was added to each score in order to remove negative numbers. MALE FACTOR II —a49 17.2 FEMALE FACTOR 11 Fic. 31. Map of variation in locality scores for factor II (feather factor) of males (left) and females (right). eUeatIsseesmmoooeoe ON eR Ue DOPE EUR NUE WO MAAAVIIIS IISc omemmrowown SS Ieee 2 2 a ree rorer eer Tor ane ary GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS de A third factor for females shows high loadings by ventral and dorsal color. Factor III, or the “female color factor,” tends to vary as expected from the univariate comparisons, with highest scores occurring in the southeastern plains, lowest values in the cen- tral-western plains, and _inter- mediate values in the north and through Canada (Fig. 32). The only character that varied independently of the factors was bill length, which had moderate loadings on almost all factors. From univariate comparisons it is known that epaulet color in males does not exhibit a mean- ingful pattern of variation and that in males and females varia- Fic. 32. Map of variation in locality tion in absolute tarsus length is “°° sae eter ad color nonexistent. Factor analytical procedure has reduced interlocality variation in nine and ten characters for males and females, respectively, to three independent patterns of variation. Three major evolutionary modes are revealed as well if this variation reflects genetic variation rather than differential phenotypic modification due to environmental variables. Although the possible adaptive significance of variation has been discussed in detail under the discussion of univariate com- parisons, it may serve well to briefly mention some of these points again. First, variation in bill thickness likely reflects morphological adaptation to interlocality differences in diet and feeding habits. Second, flight feather variation may reflect interlocality size variation or independent selection through environmental factors, such as the density of the vegetation, which would place locally different demands on the flight apparatus. And, third, female color variation may be related in part to variation in background color and the demands for crypticity. COMMENTS ON TAXONOMY The area covered by my study includes parts of the ranges of three recognized subspecies: Agelaius p. phoeniceus in the eastern plains of the United States, A. p. fortis in the western plains of the U. S., and A. p. arctolegus in the north-central plains of the U. S. 74 UNIVERSITY OF KANSAS PuBLs., Mus. Nat. Hist. and through Canada (Fig. 1). Agelaius phoeniceus fortis Ridgway (1901) was separated from A. p. phoeniceus (Linnaeus) on the basis of being larger in size and having a shorter, thicker bill. Agelaius p. arctolegus Oberholser (1907) was believed separable from A. p. fortis on the basis that females were darker and that males had shorter wings and tails, larger bills, and paler wing coverts (epaulets). The subspecies arctolegus was supposedly separable from the nominate race by having longer wing and tail, and longer and thicker bill. Taverner (1939) expressed the opinion that racial size differences between arctolegus and phoeniceus were not consistent enough to allow recognition of an individual out of its geographic range, and that arctolegus should not be accorded formal taxonomic recognition. Bent (1958: 159) noted problems in separat- ing fortis and arctolegus. In Ontario, Canada, fairly good mor- phometric separation of arctolegus in the north and west from phoeniceus in the southeast correlates with spatial isolation along the Pre-Cambriam Shield in central Ontario, where little suitable breeding habitat exists (Snyder and Lapworth, 1953). From the results of my analysis of geographic variation in Red- winged Blackbirds of central North America I think it is biologically erroneous to sustain the arbitrary boundaries and phenetic over- simplifications implicit in subspecific nomenclature for these pop- ulations. The results of univariate and multivariate analyses have verified that variation is gradual throughout the area. This is probably due to the absence of major geographic barriers to dispersal, the general mobility of individuals, and the existence of widespread and abun- dant habitat. Males and females are dimorphic but show the same general trends as follows: bill long and thick in northwestern Alberta and through much of central Canada, remaining as long or becom- ing slightly shorter but decidedly thinner in the southeastern and eastern plains of the U. S. and shortest and of moderate thick- ness in the western plains of the U. S.; a definite clinal decrease in widths of upper and lower mandibles from north to south and from west to east; wing and tail longest in the western plains region of Colorado, Wyoming, and Montana, becoming slightly shorter in western Canada, decreasing gradually through central Canada and the eastern plains, and shortest in the south- eastern plains region of eastern Kansas, Oklahoma, and Missouri; weight not well documented but existing data show clinal de- crease from northern Alberta to Oklahoma (data are lacking from the western plains); absolute tarsus length invariant but GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS ao relative to body weight (data are lacking from the western plains) is shorter in northern Alberta and increases to Oklahoma; male epaulet color invariant; female dorsal and ventral color darkest (widest bands or greatest concentration of melanin, or both) in areas of high rainfall in the southeastern plains, becoming lighter through Canada, and palest in the arid regions of the central and western plains. Females are much smaller than males in all dimentional characters; per cent differences range from about 11 to 19 per cent depending on the character. The above is a general survey of trends that does not take into account numerous local divergences and wide individual variation. In a model check-list of the birds of central North America I would like to see a paragraph such as the preceding one in place of subspecies accounts for the Red-winged Blackbird. Since the currently recognized subspecies do not accurately reflect the details of variation, and the boundaries between such are arbitrary, com- paratively little systematic information is conveyed by descriptive accounts that imply the existence of three forms. The erection of new subspecies would likewise be futile. There is no way apparent to me that the subspecies concept can logically describe or account for the variation as it exists. Owen (1963) suggested a descriptive paragraph in lieu of an account of subspecies following his study on geographic variation in the Screech Owl (Otus asio). Although the foregoing paragraph is simplified as regards intraspecific variation, as any summary must be, it is not as oversimplified as the current intra- specific nomenclature would have it, nor does it require the arbitrary establishment of boundaries. 76 UNIVERSITY OF KANSAS PuBLs., Mus. Nat. Hist. ACKNOWLEDGMENTS I am grateful to the curators of the following collections for the loan of specimens in their care: Museum of Natural History, Uni- versity of Kansas; Museum of Vertebrate Zoology, University of California at Berkeley; California Academy of Sciences, San Fran- cisco; Dickey Collection, University of California at Los Angeles; Museum of Zoology, University of Alberta; Denver Museum of Nat- ural History, Denver; Zoology Collection, Colorado State University; American Museum of Natural History, New York; Royal Ontario Museum, Toronto; Peabody Museum, Yale University; Field Museum of Natural History, Chicago; Museum of Comparative Zoology, Harvard University; Carmegie Museum, Pittsburgh; Mu- seum of Zoology, University of Michigan; National Museum of Canada, Ottawa; Academy of Natural Sciences, Philadelphia; De- partment of Conservation, Cornell University; Museum of Natural History, University of Minnesota; Museum of Zoology, University of Wisconsin; United States National Museum, Washington, D. C. I thank Dr. G. R. Smith for critically reading the manuscript, Dr. R. R. Sokal for guidance during the study and for criticism of the thesis, and Dr. R. F. Johnston who initially suggested Red-winged Blackbirds for a study in geographic variation and who has pro- vided assistance and encouragement throughout the study. Dr. Pierre Jolicoeur introduced me to the method of canonical variates analysis, and Dr. F. J. Rohlf kindly provided a Fortran IV program for the IBM 7040 to carry out these calculations, as well as assistance with the section on factor analysis. To Dr. J. C. Barlow who assisted in copy reading and to Mrs. K. Dugdale who typed the final manu- script I am also grateful. Finally, I thank my wife, Kristine, who assisted me in field work, specimen preparation, and in numerous other ways throughout the course of this research. The study was supported in part by funds from the Graduate School at the University of Kansas, a National Science Foundation Teaching Assistant Summer Fellowship, and a Traineeship from the Committee on Systematics and Evolutionary Biology at the Univer- sity of Kansas (NSF Grant GB-4446X; Dr. W. A. Clemens, princi- pal investigator). Computer time was made available through the University Computation Center. GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS 77 SUMMARY This study details geographic variation in the Red-winged Black- bird (Agelaius phoeniceus) in central Canada and the central prairie region of the United States. The results are based on fifty- four samples (“localities”) of males and forty-eight of females, with a total of about 1,850 specimens. Fourteen male and ten female samples were chronologically heterogeneous and allowed comparisons of adult birds for chrono- logical variation within localities. The results suggested exclusion of certain specimens owing in part to post-mortem color changes, in part to inadequate specimen preparation, and possibly in part to a recent phenotypic shift. Several samples allowed comparison of first-year with adult birds, with the results of excluding first-year males in analyzing bill length and epaulet color, and excluding first-year males and females in analyzing tail length, wing length to the longest primary feather, and wing length to the first secondary feather. No geographic variation was found in degree of difference between yearlings and adults, and in fact the two age classes are correlated in their patterns of variation. Univariate comparisons were made using Gabriel’s “Sums of Squares Simultaneous Test Procedure,” a method of multiple com- parison analysis of variance. Males and females are dimorphic but show the same general trends of variation. Bills tend to be long and thick in northwestern Alberta and through much of central Canada, remain as long or become slightly shorter but decidedly thinner in the southeastern and eastern plains of the U. S., and are shortest and of moderate thickness in the western plains of the U. S. A definite clinal decrease in widths of the upper and lower man- dibles occurs from north to south and from west to east, as evi- denced by a strong positive relationship between isophane and bill width. Although no direct evidence is available for Red-winged Blackbirds, it seems most likely that variation in bill size and shape is due to differences in food and feeding habits. At localities from which weights of birds were recorded, it was found that bill width varies independently of body size. Wings and tails are longest in the western plains region, become slightly shorter in western Canada, decrease gradually through central Canada and the eastern plains, and are shortest in the southeastern plains. In the U. S. there is a negative relationship between wing and tail lengths and the amount of rainfall in the breeding season. Long winged and long tailed birds are found in the drier, more open country of the western plains, and short winged and short tailed birds are more common 78 UNIVERSITY OF KANSAS PuBLs., Mus. Nat. Hist. to humid, more forested areas. Weight is not well documented but my existing data show a clinal decrease from northern Alberta to Oklahoma. Absolute tarsus length is invariant, but in relation to weight is relatively shorter at higher latitudes and longer in more southerly localities. This may relate to the maintenance of an efficient surface/volume ratio, as predicted by Bergmann’s and Allen’s rules. Male epaulet color is invariant—due, perhaps, to sexual selection. Slight variations in female color roughly conform to Gloger’s rule. In the U. S. color is positively related to the amount of rainfall. Females are darkest in areas of high rainfall in the southeastern plains, become lighter through Canada, and are palest in the arid western plains. This may relate to selection for crypticity and the role of females in care and feeding of young. Bill height and widths of the upper and lower mandibles are highly correlated with one another in their patterns of variation, as is each to body weight. Tail length and primary and secondary wing lengths also share a high degree of intercorrelation and are highly correlated with body weight as well. There are no patterns or trends of geo- graphic variation in the degree of sexual dimorphism for the several characters. A multivariate analysis using canonical analysis considered all characters simultaneously and assessed the among-group relation- ships. Factor analytical procedure reduced highly correlated vari- ables to single “factors” and summarized the major trends of vari- ation. The gradual nature of variation in all characters suggests considerable interpopulation gene flow in central North America. Finally, taxonomic conclusions are discussed. GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS 79 LITERATURE CITED ALLEN, A. A. 1914. The Red-winged Blackbird. A study in the ecology of a cattail marsh. Proc. Linn. Soc. New York, Nos. 24-25:43-128. AMERICAN ORNITHOLOGISTS’ UNION. 1957. Check-list of North American Birds. Fifth edition. A. O. U., Balti- more, Maryland. BARTHOLOMEW, G. A., AND T. J. CADE. 1957. The body temperature of the American Kestrel, Falco sparverius. Wilson Bull., 69:149-154. BARTHOLOMEW, G. A., AND W. R. Dawson. 1953. Respiratory water loss in some birds of southwestern United States. Physiol. Zool., 26:162-166. BEATS We Ha is 1900. Food of the Bobolink, blackbirds, and grackles. U. S. Dept. Agric., Div. Biol. Survey, Bull. no. 13, 77 pp. BEECHER, W. J. 1951. Adaptations for food getting in the American blackbirds. Auk, 68:411-440. BEHLE, W. H. 1940. Distribution and characters of the Utah Red-wing. Wilson Bull., 52:234-240. Bent, A. C. 1958. Life histories of North American blackbirds, orioles, tanagers, and allies. U.S. Nat. Mus., Bull. 211, 549 pp. Betts, M. M. 1955. The food of titmice in oak woodland. J. Anim. Ecol., 24:282-323. Birp, R. D., anp L. B. Smrrn. 1964. The food habits of the Red-winged Blackbird, Agelaius phoeniceus, in Manitoba. Canadian Field Nat., 78:179-186. Bowers, D. E. 1959. A study of variation in feather pigments of the Wrentit. Condor, 61:38-45. 1960. Correlation of variation in the Wrentit with environmental gradients. Condor, 62:91-120. Bowman, R. I. 1961. Morphological differentiation and adaptation in the Galapagos finches. Univ. California Publs. Zool., 58:1-302. Brusu, A. H. 1965. Energetics, temperature regulation and circulation in resting, active and defeathered California Quail, Lophortyx californicus. Comp Biochem. Physiol., 15:399-421. CAMPBELL, L. W. 1936. The subspecies of Red-winged Blackbirds wintering near Toledo, Ohio. Wilson Bull., 48:311-312. Davis, J. 1954. Seasonal changes in bill length of certain passerine birds. Condor, 56: 142-149. Dawson, W. R. AnD K. Scumipt-NIELSEN. 1964. Terrestrial animals in dry heat: desert birds. Pp. 481-492 in Handbook of physiology; section 4: adaptation to the environment. Amer. Physiol. Soc., Washington, D. C. 80 UNIVERSITY OF KANSAS Pusts., Mus. Nat. Hist. Dupraw, E. J. 1965. Non-Linnean taxonomy and the systematics of honeybees. Syst. Zool., 14:1-24. Dwicnrt, J., JR. 1900. The sequence of plumages and moults of the passerine birds of New York. Annals New York Acad. Sci., 13:73-360. Dyck, J. 1966. Determination of plumage colours, feather pigments and-struc- tures by means of reflection spectrophotometry. Dansk Ornith. Forenings Tidsskrift, 60:49-76. EIsSENMANN, E. 1955. The species of Middle American birds. Trans. Linnean Soc., 7:1-128. GABRIEL, K. R. 1964. A procedure for testing the homogeneity of all sets of means in analysis of variance. Biometrics, 20:459-477. GaBRIEL, K. R., and R. R. SOKAL. 1969. A new statistical approach to geographic variation analysis. Syst. Zool., 18:259-278. GouLp, S. L. 1967. Evolutionary patterns in pelycosaurian reptiles: A factor analytic study. Evolution, 21:385-401. Gururie, R. D. 1965. Variability in characters undergoing rapid evolution, an analysis of Microtus molars. Evolution, 19:214-233. Hamitton, T. H. 1958. Adaptive variation in the genus Vireo. Wilson Bull., 70:307-346. 1961. The adaptive significances of intraspecific trends of variation in wing length and body size among bird species. Evolution, 15:180- 195. HarMan, H. H. 1960. Modern factor analysis. Univ. Chicago Press, Chicago. Hinpe, R. A. 1959. Behaviour and speciation in birds and lower vertebrates. Biol. Revs., 34:85-128. Hopkins, A. D. 1938. Bioclimatics. A science of life and climate relations. U. S. Dept. Agric., Misc. Publ. no. 280. HowE .t, A. H., and A. J. vAN RossEM. 1928. A study of the Red-winged Blackbirds of southeastern United States. Auk, 45:155-163. Hvux.ey, J. 1942, pyolouen: the modern synthesis. Allen and Unwin, London. pp. JAMESON, D. L., J. P. Mackey, and R. C. RicHMonp. 1966. The systematics of the Pacific tree frog, Hyla regilla. Proc. Cali- fornia Acad. Sci., 33:551-620. Jounston, R. F. 1964. The breeding birds of Kansas. Univ. Kansas Publs. Mus. Nat. Hist., 12:575-655. 1966a. The adaptive basis of geographic variation in color of the Purple Martin. Condor, 68:219-228. GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS 81 Jounston, R. F. 1966b. Colorimetric studies of soil color-matching by feathers of House Sparrows from the central United States. Kansas Ornith. Soc. Bull., 17:19-23. Jounston, R. F., and R. K. SELANDER. 1964. House Sparrows: rapid evolution of races in North America. Science, 144:548-550. JoLIcoEvurR, P. 1959. Multivariate geographic variation in the wolf Canis lupus L. Evo- lution, 13:283-299. KEAR, J. 1962. Food selection in finches with special reference to interspecific dif- ferences. Proc. Zool. Soc. London, 138:163-204. ACK D: 1944. Ecological aspects of species formation in passerine birds. Ibis, 86:260-286. 1966. Population studies of birds. Clarendon Press, Oxfv.d. Mayr, E. 1956. Geographical character gradients and climatic adaptation. Evolu- tion, 10:105-108. 1963. Animal species and evolution. Belknap Press of Harvard Univ. Press, Cambridge, Mass. 797 pp. MEANLEY, B. 1961. Late-summer food of Red-winged Blackbirds in a fresh tidal-river marsh. Wilson Bull., 73:36-40. 1962. Feeding behavior of the Red-winged Blackbird in the Dismal Swamp region of Virginia. Wilson Bull., 74:91-93. MEANLEY, B., AnD J. S. WEBB. 1963. Nesting ecology and reproductive rate of the Red-winged Black- bird in tidal marshes of the Upper Chesapeake Bay region. Chesa- peake Sci., 4:90-100. MEANLEY, B., J. S. WEBB, AND D. P. FANKHAUSER. 1966. Migration and movements of blackbirds and starlings. U. S. Dept. Interior; 12 pp., mimeographed. MENGEL, R. M. 1965. The birds of Kentucky. A.O.U. Ornith. Monogrs., no. 3, 581 pp. Miter, A. H. 1931. Systematic revision and natural history of the American shrikes (Lanius). Univ. California Publs. Zool., 38:11-212. 1956. Ecologic factors that accelerate formation of races and species of terrestrial vertebrates. Evolution, 10:262-277. Mutter, A. H., H. FRIEDMANN, L. GrRISCOM, AND R. T. MOORE. 1957. Distributional check-list of the birds of Mexico. Part II. Pacific Coast Avifauna, no. 33. Cooper Ornith. Soc., Berkeley, California. 436 pp. Moreau, R. E. 1957. Variation in the western Zosteropidae (Aves). Bull. British Mus. (Nat. Hist.), Zool., 4:311-433. 1960. Climatic correlations of size in Zosterops. Ibis, 102:137-138. Nero, R. 1950. Red-wings feeding on white ash. Wilson Bull., 62:39-40. Nero, R. W. 1956. A behavior study of the Red-winged Blackbird. Wilson Bull., 68:5-37, 129-150. 6—4331 82 UNIVERSITY OF KANSAS PuBts., Mus. Nat. Hist. NosEL, G. K., and W. Vocrt. 1935. An experimental study of sex recognition in birds. Auk, 52:278-286. OBERHOLSER, H. C. 1907. A new Agelaius from Canada. Auk, 24:332-336. Orrans, G. H. 1961. The ecology of blackbird (Agelaius) social systems. Ecol. Monogr., 31:285-312. Owen, D. F. 1963. Variation in North American Screech Owls and the subspecies con- cept. Syst. Zool., 12:8-14. PacKARD, G. C. 1967. House Sparrows: evolution of populations from the Great Plains and Colorado Rockies. Syst. Zool., 16:73-89. Proeckas, He A. 1951. Speciation and ecologic distribution in American jays of the genus Aphelocoma. Univ. California Publs., Zool., 50:195-464. Power, D. M. 1969. Evolutionary implications of wing and size variation in the Red- winged Blackbird in relation to geographic and climatic factors: a multiple regression analysis. Syst. Zool., in press. 1970. Geographic variation in the surface/volume ratio of the bill of Red-winged Blackbirds in relation to certain geographic and climatic factors. Condor, in press. RENSCcH, B. 1934. Einwirkung des Klimas bei der Auspragung von Vogelrassen, mit besonderer Beriicksichtigung der Fliigelform und der Ejizahl. pp. 285-311 in Proc. Eighth Internat. Ornith. Congr. (Oxford). Riweway, R. 1901. New birds of the families Tanagridae and Icteridae. Proc. Washing- ton Acad. Sci., 3:149-155. Rouxr, F. J., AND R. R. SOKAL. 1962. The description of taxonomic relationships by factor analysis. Syst. Zool., 11:1-16. SALOMONSEN, F. 1955. The evolutionary significance of bird migration. Dansk Biol. Medd., 22:1-62. SCHOLANDER, P. F. 1955. Evolution of climatic adaptation in homeotherms. Evolution, 9:15-26. Sea, H. L. 1966. Multivariate statistical analysis for biologists. Methuen and Co. Ltd., London. 209 pp. SELANDER, R. K. 1958. Age determination and molt in the Boat-tailed Grackle. Condor, 60:355-376. 1965. On mating systems and sexual selection. Amer. Nat., 99:129-141. 1966. Sexual dimorphism and differential niche utilization in birds. Con- dor, 68:113-151. SELANDER, R. K., and D. R. GrLuer. 1960. First-year plumages of the Brownheaded Cowbird and Redwinged Blackbrid. Condor, 62:202-214. SELANDER, R. K., R. F. JoHNsTON, and T. H. HAMILTON. 1964. Colorimetric methods in ornithology. Condor, 66:491-495. GEOGRAPHIC VARIATION OF RED-WINGED BLACKBIRDS 83 Snow, D. W. 1954a. Trends in geographical variation in palaearctic members of the genus Parus. Evolution, 8:19-28. 1954b. The habitats of Eurasian tits (Parus ssp.). Ibis, 96:565-585. SnybeER, L. L., and E. D. Lapwortu. 1953. A "comparative study of adults of two Canadian races of Red-wings. Canadian Field Nat., 67:143-147. Soka, R. R. 1958. Thurstone’s analytical method for simple structure and a mass modification thereof. Psychometrika, 23:237-257. 1962. Variation and covariation of characters of alate Pemphigus populi- transversus in eastern North America. Evolution, 16:227-245. 1965. Statistical methods in systematics. Biol. Revs., 40:337-391. SoKaL, R. R., and H. V. Daty. 1961. An application of factor analysis to insect behavior. Univ. Kansas Sci. Bull., 42:1067-1097. SoKAL, R. R., H. V. Daty, and F. J. Rouur. 1961. Factor analytical procedure in a biological model. Univ. Kansas Sci. Bull., 42:1099-1121. SoxaL, R. R., and R. C. RINKEL. 1963. Geographic variation of alate Pemphigus populi-transversus in eastern North America. Univ. Kansas Sci. Bull., 44:467-507. SoKAL, R. R., and P. A. THoMas. 1965. Geographic variation of Pemphigus populi-transversus in eastern North America: stem mothers and new data on alates. Univ. Kan- sas Sci. Bull., 46:201-252. STEEL, R. G. D., and J. H. Torrie. 1960. Principles and procedures of statistics. McGraw-Hill, New York. 481 pp StopaRrtT, E. 1965. A study of the biology of the wild rabbit in climatically different regions in eastern Australia. III. Some data on the evolution of coat colour. CSIRO Wildlife Research, 10:73-82. TAVERNER, P. A. 1939. The Red-winged Blackbirds of the Canadian prairie provinces. Condor, 41:244-246. Tuomas, P. A. 1968. Geographic variation of the rabbit tick, Haemaphysalis leporispalus- tris, in North America. Univ. Kansas Sci. Bull., 47:787-828. UNITED STATES DEPARTMENT OF AGRICULTURE. 1941. Climate and man. Washington, D. C. 1248 pp. VAN VALEN, L. 1965. Morphological variation and width of ecological niche. Amer. Nat., 99:377-390. Watson, G. E. 1962. Three sibling species of Alectoris partridge. Ibis, 104:353-367. WiuiamMson, K. 1958. Bergmann’s rule and obligatory overseas migration. British Birds, 51:209-232. Wricut, P. L., and M. H. Wricut. 1944, ane. pypeoductive cycle of the male Red-winged Blackbird. Condor, ~ MUS, COMP. ZOOLL LIBRARY, UNIVERSITY OF KANSAS Pusat o "1970 Museum oF NATURAL History HARVARD UNIVERSITY Vol. 19,,No. 2, pp. 85-194, 48 figs., 4 pls. August 10, 1970 Ecological Study of the Worm Snake Carphophis vermis (Kennicott) BY DONALD R. CLARK, JR. UNIVERSITY OF KANSAS LAWRENCE. 1970 WU: ON PAO ROB an ray os ans i a5) fe ant of Rea: eh | Ae yey ale * a & We P|. She yy Po alaB Nita ce i a Ho Pri ios ia ASE } f A i Lae ips Ne RY? aca. = A ie When " “vf * ‘ pach ; ae ; NP te Ve 4 poe, she ng hae 2 és aa Acer Lee MS i) Noi ae ‘ fede ae Cee nl pied uf Ion i Vain Ulan ps a i, ieee SAS ce aS Es ae ee LoS , $e = Soe eS —— ee et a aE a = = a ies =) Se Re cos e ats = pe ¢ ge =a z = A aor — Vea. ay, ip ae 5 oe = ay Ae is. 2 =} oF ot ys $=. a. yy Te SL ee <2 tr a a a ees Cpe a Se oo ake Ry ES > tah ss af = FEA 2 SA Sas i te Fp ES ine Se Soe, we ee = 5 gk “se /S \ : ASS =e Rene CSS ee SS ao See) ; Bieta Se ie Te = Naa4 as eS ‘oa eas =: To =e wea oR a RP ea pe eS a ee 2 oF ryt ee = a ope wo ee een . = ean tn SD he NZ ei a ee a wa Sar oy tes or OSE = = : « Se aig we a oe a a; - y ae Mins oo >] AG a a ea Pd 7 ue lee oey DAA oe Naat hea i Naf if. LG. ALAS Sue ONE Ge yO te ye LEW want fo Ph meet sy Valle ae of WN te plc cdt Vip us Men gM ae Oe es (be ‘ pute ; Ae 4 ! i : } , fo Se aN chs a | We 3 4 ie rae at Sy UNIVERSITY OF KANSAS PUBLICATIONS MusEUM OF NATURAL HISTORY Vol. 19, No. 2, pp. 85-194, 48 figs., 4 pls. August 10, 1970 Ecological Study of the Worm Snake Carphophis vermis (Kennicott) BY DONALD R. CLARK, JR. UNIVERSITY OF KANSAS LAWRENCE 1970 UNIVERSITY OF KANSAS PUBLICATIONS, MUSEUM OF NATURAL HisTORY Editors of this number: Frank B. Cross, Henry S. Fitch Volume 19, No. 2, pp. 85-194, 48 figs., 4 pls. Published August 10, 1970 UNIVERSITY OF KANSAS Lawrence, Kansas PRINTED BY ROBERT R. (BOB) SANDERS, STATE PRINTER TOPEKA, KANSAS TEASE 2D 32-9080 Ecological Study of the Worm Snake Carphophis vermis (Kennicott) BY DONALD R. CLARK, JR.* CONTENTS PAGE PNTRODUCTIONS peta sie h oe at eh eee Pas eee oe Ae ha ag tis on ae 89 PYCKMOWICOCMENS: 3256240695 eS ub snl as Stone h Ooo dene. 89 Materials:tand, Methods: 24.0. S23 elie 5. notions «soe odd 90 SLUG Vee aT Co Sly any bee Ronee, | aay iy aD eed 91 BORE HOUOG Nga: ee hin stats eee ath coeur i OES aay oes Bisbee 94 BCULE I ATOM eae sa ee oth ate ett he eee ogre feel a ae Ne gh tie Bas 94 Siculland WOemtbOMN a. igo sas oe Ae no aici ud hee 95 @oloratiom erga cc ee eRe ee re erie Gy ee ee 96 | 5 Lion ofa) 7 RT ER rece ar ge eee ee a eee Dh a, AVNET IG TAN EEN POUL Ce acca tt ys eerie ata Scatdass BP 98 EO DOTHONS fee Ser Wi niet ar a or Pasion al cee oso cave eee 99 SextalsiDinaorp ligne su tees oer A on 8) hss. Syuces ac ae 100 Adaptation for Fossorial Habits among Worm Snakes ..... . 101 WIG EES ULI Nae eee ae rt) oe eee, oS Sk rh ra be bets 104 BSE ASST A ae tre cre re ocean ce UN ffir ka leads AO ac 106 Cerie tal ere Met ea see tee tacts We taney eels aR 106 Amphibian and: Reptile Associates.) )s..... 6.....2:.2.2-. 106 Size of Rock versus Snout-Vent Length .................. 107 iReycisvot Soil’ Moisture: See a2 eo. fs C8e eee 109 'S(CILMAL IS jo Leh pene in hy ek ee ae Oana Sars Se co eee eo i SRENTPERATURE: INELATIONSHIPS' 228..0.0.0 00) 0.0 v ke 6. 5) tne 1 1s SANA GOL Ob {01 0 ae ae eee are ee ye eee a 113 Scasonalem in, pt, Uae ao BO i> a) eM OO 113 LDIDIIN? Sa ns Rte AOR AR de Rte. Ce CURE RPE AN eee Fin 114 IBURROWING? DEHAVIOR «05 0). 222i oe hn Soe 115 POLONAGNOEINGE SG) wee ey ete g ehaAceca eage © yo ins. reatke es MSERN Vp te aes 115 PEPRODUCTION ae. G078 GHG 8 one eh Jon RS ed Gree 122 Sexual Maturity and Testicular Enlargement in Males ...... 122 Sexual Maturity and the Ovarian Cycle in Females ........ 128 Mate and WMertilization <4) cu. asics. + sis age ss ee 132 Oviposition*‘and the Clutch of Eggs ...2..4.......2..-25; 136 TUE Herre ren serge ABN cet oe! cts. te ira ae ee CR 150 Plapchingeandsbatchlngs:. 5.4... fei cn ondewlst wae ae eae 155 * Present address Department of Wildlife Science, Texas A. and M. Uni- versity, College Station, Texas 77843. (87) 88 UNIVERSITY OF KANSAS PuBLs., Mus. Nat. Hist. GRowin sor). | {o/ Xe) Xe] ‘a1 10! ee: 0! (0) (0) 1/8), 6, 10 le jere) Fe) ,e)\\e) "0,0. 0) (8) tu)