Sex Determination of Carolina Wrens in the Mississippi Alluvial Valley

Daniel J. Twedt Carolina Wren, but offers little guidance in USGS-Patuxent Wildlife Research Center discriminatingbetween sexes for the nominative 2524 South Frontage Road, Suite C subspecies(T. I. ludovicianus).This subspeciesis Vicksburg, MS 39180 common to most of eastern North America. Within [email protected] a populationfrom northwest Alabama, Haggerty and Morton (1995) reported wing chords of male ABSTRACT CarolinaWrens (mean = 59.3 mm _ 1.5 SD, n = 30) were significantlylonger than wings of females I identified sexual dimorphism in wing length (mean = 55.7 mm _+1.4 SD, n = 34). While banding (unflattenedchord) of CarolinaWrens (Thryothorus Carolina Wrens in the MississippiAlluvial Valley ludovicianus)within the central MississippiAlluvial (northeastLouisiana and west-centralMississippi), Valley (northeast Louisiana and west-central I noted similar differences in wing length between Mississippi)and used this difference to assign a sexes. I developed criteria based on wing length sex to capturedwrens. Wrens were identifiedas dichotomythat can be used to identifythe sex of female when wing lengthwas less than 57.5 mm or Carolina Wrens. male when wing lengthwas greater than 58.5 min. Verification of predicted sex was obtained from METHODS recaptures of banded individualswhere sex was ascertained from the presence of a cloacal From 1994 through 2004, Carolina Wrens were protuberanceor broodpatch. Correctprediction of captured, sexed when possible, measured, and sex was 81% for adult females and 95% for adult banded during constant-effortmist-netting during males. An alternative model, which categorized summer (e.g., DeSante 1992) and duringwinter. wrens with wing lengthsof 58 and 59 mm as birds Birds were captured within two to eight forest of unknown sex, increased correct prediction of stands annually, from a total of 12 bottomland females to 93% but reduced the number of hardwood forest stands on Tensas River National •ndividualsto which sex was assigned. These Wildlife Refuge, Madison Parish, Louisiana,and simple,predictive, wing-length-based models also two intensively managed cottonwood stands on correctlyassigned sex for more than 88% of young Fitler Managed Forest, Issaquena County, (hatching-year)birds. Mississippi. All data presented are from Carolina Wrens that were identified with uniquely INTRODUCTION numbered bands. I aged birds as hatching-year (HY) young or after-hatching-year (AHY) adult Female and male Carolina Wrens (Thryothorus based on differences in plumage or skull ludovicianus) share a common plumage. Thus, ossification (Pyle 1997). Each capture (or only during the breeding season can males be recapture) of a bird was assigned categorically to •dentifiedby the presence of a cloacal protuber- one of three groups:(1) adult birdsthat were sexed ance and females by the development of a brood by presence of cloacal protuberance or brood patch. As this species is generally resident and patch,(2) adult birdsthat couldnot be sexed,or (3) frequently encountered by banders outside the young birds that could not be sexed. Because I breedingseason, the abilityto distinguishbetween sought to use wing length (unflattenedwing chord sexes based on other criteria is warranted. measured to the nearest 1 mm) to discriminate between sexes, only birds with measured wing Pyle (1997) stated that morphometric measure- lengths (mm) and no flight feather molt were used ments are useful for sexing within subspecies of for analyses.

Oct - Dec 2004 NorthAmencan B•rd Bander Page 171 Often, multiplecaptures of the same bird were I found few discrepanciesamong wing lengthsof withina group.I examinedthe range of wing individualscaptured more than once; 69% differed lengthsfor each individualwithin a group.If the by 1 mm or less, 19% differed by only 2 mm. I rangeof winglengths exceeded 2 mm,I removed examinedthe wing lengthsof birdswhere range of the mostdiscrepant value. I then usedthe mean wing lengthswas 3 mm (18 individuals)or 4 mm winglength within each group for analyses.Based (four individuals) and removed captures with on differencesin winglength of female and male disparatewing lengths.Thus, in my analyses,only Carolina Wrens, I developed predictivemodels six individualshad a range of wing lengths that and appliedthese models separately to adultand exceeded 2 mm. youngCarolina Wrens of unknownsex. I assessed the predictiveability of these models by comparing Mean wing length of females was 56.2 mm +_1.9 the sexpredicted from models to the knownsex of SD (n = 308 individuals)whereas mean wing length the same individual, as determined by the of males was 60.1 mm +-1.5 SD (n = 188 presenceof cloacalprotuberance or broodpatch. individuals). The frequency distributionof wing lengths (Fig. 1) indicated 80% of female wing RESULTS lengthswere less than 58 mm, 11% were 58 mm, and 9% were greater than 58 mm. Conversely, Of 2,957 capturedCarolina Wrens, 877 had a 87% of male wing lengths were greater than 58 broodpatch and were identifiedas female, 794 had mm, 9% were 58 mm, and only 4% were less than a cloacal protuberanceand were identified as 58 mm. Based on these data, applicationof a 57.5 male, whereas 1286 wrens could not be sexed. mm cut point to discriminatebetween sexes, and However, because 582 wrens were captured more excludingbirds with 58 mm wing length, should than once, these capturesrepresented only 1,637 result in less than 10% of females and less than 4% unique individualwrens. I captured 292 wrens of males incorrectlysexed. twice, 130 capturedthrice, and 160 capturedfour or more (up to nine) times. Fig. 1. Frequency of occurrence (%) in wing length (mm) of female (n=308) and male (n=188) Carolina Wrens captured in northeastern Louisiana and west-central Mississippi.

25 Female Male

20

3 15

O 10

U. 5

51 52 53 54 55 56 57 58 59 60 61 63 64 Wing length (rnrn)

Page 172 North American Bird Bander Vol. 29 No.4 Using 57.5 mm as a cut pointwith no exclusion,I CONCLUSION assigneda sex of female to unsexed wrens with wing lengthless than 57.5 mm and male to wrens Female and male Carolina Wrens in the central with wing length greater than 57.5 mm. When I MississippiAlluvial Valley (northeastLouisiana and comparedpredicted sex with known sex (identified west-central Mississippi) are sexually dimorphic from a differentcapture of the same individual),38 This dimorphismin wing lengthcan be exploitedto of 51 females (75%) and 87 of 91 males (96%) predict the sex of Carolina Wrens. Although the were correctlysexed (Table 1). However, because range of wing lengths overlaps slightly between many females and males had wing lengthsof 58 sexes, it appears that assignment of female to mm, I applieda differentmodel that used the same wrenswith wing lengthless than 57.5 mm and male cut point, but birds with 58 mm wing length to wrens withwing length greater than 58.5 mm will remained as undetermined sex. In this model, provide 95% correct identification of males and correct prediction of males was 95%, whereas 81% correct identification of females within this correct predictionof females increased to 81% populationof Carolina Wrens. Correct identifica- (Table 1).Similarly,an alternative model in which tion of females can be improved to 93% by birds remained of undeterminedsex when wing retentionof wrenswith wing lengthsof 58 or 59 mm length was either 58 or 59 mm yielded correct as birds of undetermined sex. However, by doing predictionfor 93% of females and 93% of males so, the sex of fewer individuals will be identified. (Table 1).

Table 1. Cut point and exclusion models for wing length (ram) used to predict female (F), male (M), or undeter- mined (U) sex of Carolina Wrens in the Mississippi Alluvial Valley, the number (n) of wrens for which sex was )redicted, and the number (c) and percentage (%) of correct sex determinations. After-Hatching-Year Hatching-Year Female I Male Female I Male

34 1271 80124 I 49 I 90 30127 I 90146 I TMI 88 22149 I 86 I 44 I 44 1400

F<56.5mm;M>59.5mm30 I 27 I 90 I •61 •6 I 4øø 201 491 9s I 8 I 81400

Greater accuracyin predictions,however, came at Notably, these predictive models can also be the cost of being able to predict the sex of fewer appliedto young(HY) birdswith similar expectation individuals.For example, I predictedthe sex of 142 of correct sex identification. wrenswhen consideringall wing lengthsbut sexed only 101 individualswhen birdswith wing lengthsof ACKNOWLEDGMENTS 58 and 59 mm remained of undetermined sex (Table 1). Excludingonly 58 mm wing lengthsfrom I thank Scott Somershoe, Randy Wilson, and predictionsof sex may be a judiciouscompromise. Blaine Elliott for assistance with banding and review of this manuscript. I also thank all others Applicationof this model with a 57.5 mm cut point who assisted in netting, especially Curtis Nelms, and 58 mm exclusionto young (HY) birds resulted Michelle Hunt, and Allan Mueller. Tensas River in the correct determination of sex for 27 of 30 NationalWildlife Refuge and Filter Managed For- females (90%) and 14 of 16 males (88%; Table 1). est graciouslyprovided study sites. Fundingwas providedby U.S. GeologicalSurvey and U.S. Fish and Wildlife Service.

Oct. - Dec. 2004 NorthAmerican Bird Bander Page 173 LITERATURE CITED Haggerty,T. M. and E. S. Morton. 1995. Carolina Wren (Thryothorusludovicianus). Number DeSante, D. F. 1992. Monitoringavian productivity 188 in The birds of North America, A. Poole and survivorship(MAPS): a sharp, rather and F. Gill, eds. The Academy of Natural than blunt,tool for monitoringand Sciences, Philadelphia, and The American assessinglandbird populations. Pp. 511- Ornithologists'Union, Washington,DC. 521 in Wildlife 2001: Populations. D. R. Pyle, P. 1997. Identificationguide to North McCulloughand R. H. Barrett(eds.). American birds, Part I: Columbidae to ElsevierApplied Science, London, UK. Ploceidae. Slate Creek Press, Bolinas, CA. Swainson's Hawk Productivity in Saskatchewan, 1944 - 2004

C. Stuart Houston Dan Zazelenchuk 863 University Drive, Saskatoon Box 39 Saskatchewan S7N 0J8 Kyle, Saskatchewan SOL 1T0 [email protected] [email protected]

ABSTRACT adds the final nine years to reproductiveresults reportedin Schmutz et al. (2001). Altogether,we The numberof nestlings banded was recordedat have followedthis speciesthrough a suddencrash 2,427 successful Swainson's Hawk (Buteo in its main prey item and a very gradual,patchy and swainsonOnests in Saskatchewan between 1944 incomplete recovery in numbers of Richardson's and 2004. A suddenand prolongeddrop in ground ground squirrel in grassland areas over 16 squirrel numbersappeared to explain decreased subsequent years. Swainson'sHawk productivityin grasslandregions over ten consecutiveyears, 1987 - 1996. METHODS

INTRODUCTION Since 1969, CSH concentrated much of his Swainson's Hawk banding on and near large The world's first specimens of the Swainson's Prairie Farm RehabilitationAdministration (PFRA) Hawk (Buteo swainsonO and its main prey - pastures near Kindersley in west-central Richardson's ground squirrel (Spermophilus Saskatchewan. These pastures host beef cattle richardsoniO were collected at Carlton, without feed lots. Incubating Swainson's Hawk Saskatchewan, in 1827 and 1820, respectively pairs were located during Ferruginous Hawk (Houston 1990). In Saskatchewan, long-term (Buteoregalis) banding in June, and were revisited bandingefforts have contributedto our knowledge for banding in the latter half of July, when the of Swainson'sHawk productivityand of its annual nestlingswere between 10 days and six weeks old 20,000 km migration to and from Argentina In 1972, Jean Harris began findingnests for us, in (Schmutzet al. 1996, Houstonand Fung 1999). crop land interspersedwith small pastures near Kindersley;during a period of 33 years, she found The seniorauthor (CSH) beganincidental banding 898 nests (about 28% of our total) of which 635 of Swainson'sHawks in 1944 but did not begin were successfuland from which 1,132 youngwere intensivestudies of productivityuntil 1967. This 38- banded. From 1985 to 1995, Dean Francis found year program has benefitted from extensive 228 nests along the Saskatchewan-Alberta assistancein nest-findingand bandingfrom the boundary south of Alsask, of which 150 were juniorauthor (DZ) for the past 16 years,including successful and from which 288 young were the contributionof the 71 nestlingsbanded under banded. The study area was not searched DZ's masterbanding permit in 2004. This paper completely and had no well-defined boundaries. Page 174 North American Bird Bander Vol 29 No 4