Geographic Variation and Evolution in the California Gull (Larus Californicus)

GEOGRAPHIC VARIATION AND EVOLUTION IN THE CALIFORNIA GULL (LARU$ CALIFORNICU$)

JOSEPHR. JEHL, JR. Sea World Research Institute/ Hubbs Marine Research Center, 1700 South Shores Road, SanDiego, California 92109 USA

ABSTRACT.--TheCalifornia Gull (Laruscalifornicus), currently regarded as a monotypicspecies, is separableinto a small, dark-mantledrace (L. c. californicus)that breedsprimarily in the Great Basin of the United States,and a larger, paler race (L. c. albertaensis)from the Great Plains of the north-central U.S. and south-centralCanada. The breeding rangesof thesetwo races,previously disjunct, have expanded recently, and a zone of secondarycontact seems to be forming eastof the RockyMountains in the northern United States.Received 13 November 1986, accepted23 January1987.

THE California Gull (Laruscalifornicus) has a distinct assemblages.I then used stepwisediscrimi- discontinuousbreeding range in the western nant function analysis(BMDP7M; Dixon 1983) to de- United Statesand Canada,extending from Great termine which variablescontributed to the separation Slave Lake, N.W.T. (62øN) south to San Fran- of groups,using locality as the grouping variable. ciscoBay and Mono Lake, California (38øN)and east to the Dakotas (A.O.U. 1983, pers. obs.). Except for a brief study by Zink and Winkler RESULTS (1983), who found only minor size and genetic Although California Gulls are well repre- differencesbetween gulls from Mono Lake and sented in museum collections, material from the Great Salt Lake, Utah, there has been no effort breeding grounds is scanty. Even so, consid- to investigate geographic variation in this erable geographic variation in size is obvious species,which is pronounced. (Appendix). A preliminary comparisonof the only two large samples,from Alberta and Mono METHODS Lake, showed that Alberta birds were signifi- Museumspecimens used in this analysiswere taken cantly larger in culmen,wing, tarsus,and mass in the breedingrange between 1 April and 15 August. (t-test, P < 0.05); they also had a paler mantle. Becausesome gulls are still migrating to inland breed- Simple inspectionof the data showed that the ing stationsin April, and becauseadults begin to size differences did not conform to a simple move coastwardby the lastdays of July, samplesfrom latitudinal gradient, as might be expectedfrom western parts of the range may have included mi- Bergmann'srule; someof the largestbirds were grants. I took standardmeasurements on length of from North Dakota (48øN) and some of the the exposedculmen, flattenedwing and tarsus,depth smallestfrom Utah (41øN) (Fig. 1). Principal of the bill at the gonys, and body mass.The large sample from Mono Lake is based mainly on birds componentsanalysis showed that 58% of the variation in females and 61% in males was ex- found freshlydead on nestingislets; a representative series was preserved as museum specimens. Some plained by a single factor and that within each California Gulls begin to show extensivewear of the sex two assemblagescould be recognized (Fig. primaries by March. Specimenswhose wing length 2). These assemblagescorresponded broadly to could not be determined accuratelywere excluded physiographic provinces in western North from the analysis.Additional size and massdata were America. Gulls from the northern Great Plains extracted from the literature. I also noted the pat- of the north-central United States,the prairie terning of the primariesand the color of the mantle. provinces of Canada, and the Northwest Ter- Although variation in the latter was evident, I could ritories were large and clustered with the Al- not measureit consistentlybecause of the generally soiled condition of most specimens. berta sample. Those from the Great Basinof the For analysisof size variation I separatedmensural western United States (Washington, Oregon, data by sex and used principal componentsanalysis Nevada, Utah, and California), as well as those (BMDP4M; Dixon 1983) to test for patterns of varia- from nearby regions in Colorado, Wyoming, tion. The analysisshowed that the gulls fell into two and Idaho, were small like those from Mono

421 The Auk 104:421-428. July 1987 422 JOSEPHR. JEHL,JR. [Auk,Vol. 104

Lake. Although samples from many regions were very small, the available data did not re- veal any indication of clinal variation. ', "r External differences between the Great Basin I ' I and Great Plains forms are greater than those characterizingmost currently recognizedraces of North American birds. I describe them as subspeciesbecause morphological and histori- cal (seebelow) data indicate they have had sep- arate evolutionary histories.The type specimen of the species(Amer. Mus. Nat. Hist. No. 46070) was taken near Stockton, California, and is an adult (cf. Greenway 1978) female of the Great Basin population, which becomesL. c. californicus.Its measurements are: exposed culmen, 43.7 mm; depth of bill at gonys,15.8 mm; tarsus, 55.4 mm; wing, 380 mm. For the larger form I proposethe name:

Larus caliœornicus albertaensis ssp. nov.

Holotype.--University of Michigan Museum of Zoology No. 224,653, adult male, from Frog Lake, Alberta, 53ø55'N, 110ø15'W;collected by Philip H. R. Stephey, 9 July 1985. Diagnosis.--Distinguishedfrom L. c. califor- nicusby greater size, particularly in bill dimensions and body mass(Table 1, Fig. 3). Mantle paler gray, less bluish than in L. c. californicus, approachingor matching the palenessof L. ar- gentatus(Fig. 4). Measurementsof holotype: exposedculmen, 52.5 mm; depth of bill at gonys, 18.7 ram; tarsus,62.6 ram; wing, 424 ram; mass, 809 g. Using stepwise discriminant function analysisfor the entire sample,I coulddetermine the racial identification of a bird of known sex for 87.9% of the females of L. c. albertaensis and 95.3%of the femalesof L. c. californicus,and for 87.0% of the males of L. c. albertaensis and 92;4% I of the males of L. c. californicus(Fig. 5) by the following formulas, where a score >0 is alber- taensisand a score<0 is californicus:for females, 0.36 culmen (mm) + 0.31 bill depth (mm) + 0.06 wing (mm) - 45.60; for males, 0.25 culmen

Fig. 1. Mean dimensions of (A) wing length, (B) culmen length, and (C) tarsus length of California Gulls. For each state or province, data for males are listed abovethose for females.The samplefor Wyo- ming was basedon unsexedindividuals. Sample sizes are in parentheses. July1987] Evolutionin theCalifornia Gull 423

20-

10- MALES

1;•3 2333123333 3

3 34333533 z 11• 35331233333

20-

10- FEMALES

0 312 23 3

--3 --2 --1 0 I 2 3

PC-1

Fig. 2. Plot of first principal componentof male and female California Gulls. Numbers refer to localities: ! = Northwest Territories, 2 = Saskatchewan,3 = Alberta, 4 = Montana, 5 = North Dakota, 6 = Washington, 7 = Oregon, 8 = Idaho, 9 = Mono Lake, California, 10 = other California, 11 = Nevada, 12 = Utah, and 13 = Colorado.

(ram) + 0.15 tarsus(ram) + 0.05 wing (ram) -- primaries average larger and paler, although 40.21. the latter differencesare minor. The pattern of Range.--Nests on lakes, often of slight to white markings("mirrors") on the primariesof moderate alkalinity or salinity, in North (and adults is variable (see Dwight 1925: figs. 115, South?) Dakota, Manitoba, Saskatchewan, Al- 117, 119, 120), though not geographically.I berta, and the Northwest Territories (Fig. 6). found no differencesin color or pattern in the Southern and western limits unknown, as the downy and juvenile plumagesof 10 gulls from range of the speciesis changing rapidly. Pre- Mono Lake and 10 from Beaverhill Lake that sumably winters through the range of the were hatched and reared in captivity. Among species,mainly alongthe Pacificcoast from Brit- wild-taken birds, juvenile and subadultplum- ish Columbia to BajaCalifornia. ages were too variable for analysis. Juveniles Variation.--Larusc. albertaensis averages 5-12% from Mono Lake, for example,range from pale larger than L. c. californicusin linear dimensions tan to dark brown in their general coloration, and 27% greaterin body mass.All of thesedif- which may be modified dramatically by expo- ferences are highly significant (t-test, P < sure to alkaline water and intense sunlight. 0.0001). Indeed, females of albertaensisexceed Studies of allozyme variability revealed no male californicusin mean culmen length and diagnosticgenetic differences between the two mass.In addition, albertaensishas a paler mantle, populations(Karl et al. 1987). and the gray areas on the inner vane of the Specimensexamined.--The type seriesof 9 males 424 JOSEPHR.JEHL, JR. [Auk, Vol. I04

TABLE1. Sizedifferences between races of CaliforniaGulls (Larus californicus). a

Wing (mm) Culmen (mm) Race Sex No. Mean Range SD No. Mean Range SD albertaensis• M 48 412.2 385-431 9.6 49 53.0 47.3-57.4 2.1 F 38 391.3 369-408 9.7 38 47.8 41.2-52.0 2.4 californicus• M 146 393.7 365-415 9.6 145 47.3 40.5-59.2 2.5 F 202 372.3 341-390 8.2 165 42.6 38.0-48.4 1.8 Pooleddata from Appendix. Northwest Territories, Saskatchewan,Alberta, and North Dakota. Washington,Oregon, Idaho, California, Utah, Nevada, and Colorado.

and 3 femalesfrom Frog Lake and Beaverhill lation (albertaensis) became isolated in the Lake, Alberta, is housedin the Universityof northern lakeson outwashplains near the mar- Michigan Museum of Zoology (2 skins), the gins of the glaciers.Until recently albertaensis Provincial Museum of Alberta (2 skins), and the bredmainly at lakesnear the edgeof the boreal SanDiego Museum of Natural History (4 skins, forest(see Fig. 6). The other population(cali- 4 skeletons).Other specimensexamined are in fornicus)became isolated around lakes in the the museumsacknowledged below. Tissuesam- steppe-desert of the western United States, plesof both forms,including the type of L. c. where the speciesis currently most abundant albertaensis,have been deposited in the Museum (Conover 1983). of Zoology,Louisiana State University. In the pleniglacial period (ca. 12,500-15,000 Etyraology.--Named for Alberta, Canada, yr B.P.) the refugium for the cold-desertflora provinceof the type locality. characteristic of the modern Great Basin, and presumablyfor L. c. californicus,was in a small DISCUSSION areaof the presentMohave Desert(Wells 1983). As the climate ameliorated in the Holocene, Evolutionand subspeciation in the California Gull: steppe-desert conditions shifted north into the ahypothesis.--Rapid speciation among large gulls Great Basin, where extensive lakes of varied in the Northern Hemisphere is presumed to hydrologyprovided breeding habitat for gulls have resulted from the splitting of ancestral (Hubbs and Miller 1948, Hubbs et al. 1974). I populationsby eventsin the Pleistocene(Mayr inferthat the absence of suitablenesting habitat 1963, Smith 1964). I postulatethat when the in the High Plainshelped isolate the GreatPlains CaliforniaGull stockwas divided, one popu- and GreatBasin gulls, which as recently as 1915 were700 km apart(Cooke 1915; Fig. 6). Shortly thereaftera majorrange expansion began, abet-

Fig. 3. Extremesof geographicand sexualvaria- tion in bill dimensionsin CaliforniaGulls. Top: L. c. albertaensis,male, Frog Lake, Alberta (UMMZ No. Fig. 4. Variation in body size and mantle color in 226,654).Bottom: L. c. californicus,female, Mono Lake, CaliforniaGulls. Top: L. c. albertaensis.Bottom: L. c. California (SDNHM No. 44123). californicus.Specimens as in Fig. 3. July 1987] Evolutionin the CaliforniaGull 425

TABLE 1. Extended.

Depth (mm) Tarsus(mm) Mass(g) No. Mean Range SD No. Mean Range SD No. Mean Range SD 48 17.7 15.6-19.2 0.7 48 63.1 56.8-72.0 3.3 32 841 653-1,045 103.3 36 15.8 14.8-17.0 0.6 36 57.7 52.8-65.1 2.4 19 710 568-903 91.2 142 16.6 13.7-18.8 0.9 132 59.0 47.7-65.0 2.3 64 657 490-885 85.3 157 14.9 13.1-16.3 0.6 153 54.8 43.8-60.5 2.6 84 556 432-695 53.4

ted by the creationof nestinghabitat. By 1932 southwest.The small samplefrom Montana (4 the Great Basinbirds had bred in Washington males only), for example, includes two birds (Deckerand Bowles1932), and somehad spread that clump with the Great Basinbirds, one with acrossthe Rockies,forming coloniesin Wyo- the Great Plains birds, and one that is inter- ming and Colorado(Bailey and Niedrach1965), mediate (Fig. 2). a processthat continuestoday (Findholt 1986, The ancestryof the CaliforniaGull is uncer- C. Chasepets. comm.).The regularity of non- tain. Most authors have considered it a member breederssummering in New Mexico suggests of the Herring Gull complex(Stegmann 1934, that colonizationmay be imminent (J.P. Hub- bard pets. comm.); at least 5 of 6 specimens taken there in the summerof 1981 representL. c. californicus.Data for the Great Plains birds are scanty,but the recentestablishment of a colony in SouthDakota (Harris 1982)suggests that their range, too, is expanding. Between 1930 and 1980 the U.S. population increasedfrom 101,000gulls in 15 coloniesto 276,000 in 80 colonies, and the hiatus that ex- istedin the High Plainsbegan to fill (Conover 1983).The increasedabundance and range expansion apparently resulted in a zone of sec- ondary contacteast of the Continental Divide, with birds in that region being derived from breeding areas to the northeast and from the

I1 Ilfi --3.0 20 10 O0 10 2O 30

Fig. 6. Breedingrange of California Gull, based on Conover(1983), Findholt (1986),Godfrey (1966), Harris(1982), C. Chase,III (pers.comm.), and original -'. -t -'. '. : 2'.o 3:0 4'.o do do observations.The main range of L. c. californicusis hatched;outlying colonies are indicatedby opencir- cles.The rangeof L. c.albertaensis is stippled; a colony Fig. 5. Resultsof discriminantfunction analysis. in SouthDakota ('k) is presumedof be of this race. Stippledareas refer to samplesfrom the rangeascribed Forprecise location of U.S.colonies see Conover (1983). to L. c. albertaensis,open areasto rangeof L. c. califor- Coloniesknown to Cooke(1915) are indicatedby nicus. crosses. 426 JOSEPHR. JEHL,JR. [Auk, Vol. 104

Fisher and Lockley 1954, A.O.U. 1983), and ACKNOWLEDGMENTS among extant forms it is reminiscent of Thayer's I examined collections in the Museum of Vertebrate Gull (L. thayeri)in its small size, dark iris, rel- Zoology,University of California, Berkeley;Califor- atively dark mantle, juvenile and subadult nia Academy of Sciences;University of California, plumages,east-west migration route, and west LosAngeles; Los Angeles County Museum of Natural coast wintering range. Schnell (1970, pers. History; U.S. National Museum of Natural History, comm.), on the other hand, using only speci- Field Museumof Natural History;and SanDiego Mu- mens that I would classifyas L. c. californicus, seum of Natural History. Information on other col- found that the speciesclustered more closely lectionswas provided by C. M. White (Brigham Young with the smallerRing-billed (L. delawarensis)and University, University of Utah), R. M. Mengel (Uni- Mew (L. canus)gulls than with the Herring Gull. versity of KansasMuseum of Zoology),K. C. Parkes While I suspecta reanalysisthat included large (CarnegieMuseum of Natural History), R. W. Starer (The Universityof Michigan Museumof Zoology),L. specimensfrom the Great Plains might lead to Cancadeand P. Stepney (Provincial Museum of Al- a different result, I doubt any conclusionswill berta), M. Gillman (CaliforniaAcademy of Sciences), be fully satisfactoryuntil the possiblerelation- G. Schnell (University of Oklahoma),M. R. Browning ship of L. californicusto two small Pleistocene (U.S. National Museum of Natural History), C. Chase, species,L. oregonusand L. robustus,is considered. III (Denver Museum of Natural History), T. Miller Those speciesare known only from inland lo- (British Columbia Provincial Museum), and L. Oring calities in the Great Basin (Howard 1946, Jef- (University of North Dakota). G. McCaskie, P. Step- ferson 1985), the area now inhabited by L. c. ney, S. Findholt, and C. Chasehelped in many ways. californicus. Special thanks are due S. I. Bond for continual assis- Samplingprocedures.--The paucity of pre- tance in all phasesof this and complementary re- served material from many parts of the gulls' search.Computer programs were run at the SanDiego State University Computer Center. The manuscript range is deplorablebut easily remedied. Col- was improved by the commentsof T. R. Howell, S. lecting east of the Continental Divide, espe- A. Karl, R. W. Starer, R. W. Zink, G. D. Schnell, A. cially, is needed to document current morpho- H. Brush, D. M. Power, M. Conover, and B. L. Monroe, logical variation more precisely and to Jr. This researchwas supportedby grants from the determine whether the differences described Los Angeles Department of Water and Power. above will be swamped or reinforced as the speciescontinues its increaseand range expansion. LITERATURE CITED Considerable sampling bias in currently availablematerial is manifestedby the sexratios AMERICAN ORNITHOLOGISTS' UNION. 1983. Check-list of specimensused in this study, which vary of North American birds, 6th ed. Washington, among localities (see Appendix); indeed, fe- D.C., Amer. Ornithol. Union. maleswere unrepresentedin somesamples. The BAILEY,g. M., & R. J. NIEDRACH. 1965. Birds of Col- orado. Denver, Colorado, Denver Museum of high incidence of males (109 males vs. 69 fe- Natural History. males) from localities other than Mono Lake BEHLE,W. H. 1958. The bird life of Great Salt Lake. probably stemsfrom their greater aggressive- Salt Lake City, Univ. Utah Press. nessand likelihood to approachwithin shotgun BURGER,J. 1983. Determining sex ratios from col- range (Butler and Janes-Butler 1982, Burger lected specimens.Condor 85: 503. 1983).By contrast, the high incidenceof females BUTLER,g. G., & S. JANES-BUTLER.1982. Territoriality in the Mono Lake (72 males vs. 124 females) and behavioral correlatesof reproductive success sample, derived largely from birds dying of of the Great Black-backed Gulls. Auk 99: 59-66. trauma or predation in the colony, reflectsthe CONOVER,M.R. 1983. Recentchanges in Ring-billed females' greater risk of mortality early in the and California gull populations in the western United States. Wilson Bull. 95: 362-383. nesting season(Jehl and Chase1987, Jehl MS). --, & G. L. HUNT, JR. 1984. Female-female pair- Thesecontradictory ratios illustrate further the ing and sex ratiosin gulls: an historicalperspec- inherent risks in drawing inferencesabout so- tive. Wilson Bull. 96: 619-625. cial structure from the sex ratios of preserved COOKE,W. W. 1915. Distribution and migration of specimens,when bias in collection or preser- North American gulls and their allies. U.S. Dept. vation is not, or cannot be, known (Burger 1983; Agr. Bull. 292. cf. Conover and Hunt 1984). DECKER,F. R., & J. H. BOWLES.1932. Two new breed- July1987] Evolutionin theCalifornia Gull 427

ing recordsfor the stateof Washington.Murrelet Desert, California. Contrib. Sci., Los Angeles 13: 53. County Mus. No. 362. DIXON, W. J. (Ed.). 1983. BMDP statistical software. JEHL,J. R., JR., & C. CHASE,III. 1987. Foraging pat- Los Angeles, Univ. California Press. terns and prey selectionby avian predators:a DWIGHT,J. 1925. The gulls (Laridae) of the world: comparativestudy in two coloniesof California their plumage,moults, variations,relationships Gulls. Stud. Avian Biol. 10. and distribution. Bull. Amer. Mus. Nat. Hist. 52: KARL,S. A., R. M. ZINK, & J. R. JEHL,JR. 1987. A1- 63-408. lozyme analysisof the California Gull (Laruscal- FINDHOLT, S. 1986. Status and distribution of Cali- ifornicus).Auk 104: in press. fornia Gull nestingcolonies in Wyoming. Great MAYR,E. 1963. Animal speciesand evolution. Cam- Basin Natur. 46: 128-133. bridge, Massachusetts,Belknap Press, Harvard FISHER,J., & R. M. LOCKLEY. 1954. Sea-birds. An in- Univ. troductionto the natural historyof the sea-birds SCHNELL,G. D. 1970. A phenetic study of the sub- of the North Atlantic. Boston,Houghton Mifflin order Lari (Aves) II. Phenograms,discussion and Co. conclusions.Syst. Zool. 19: 264-302. GODFREY,W. E. 1966. The birds of Canada. Ottawa, SMITH,J. E., & K. L. DIEM. 1972. Growth and devel- Queen's Printer. opment of young California Gulls (Laruscalifor- GREENWAY,J. 1978. Type specimensof birds in the nicus).Condor 74: 462-470. AmericanMuseum of NaturalHistory. Bull. Amer. SMITH,N. G. 1964. Evolution of some arctic gulls Mus. Nat. Hist. 161: 1-305. (Larus):an experimentalstudy of isolatingmech- HARRIS, B. 1982. First nest record for California Gulls anisms.Ornithol. Monogr. No. 4. in South Dakota. South Dakota Bird Notes 34: 42. STEGMANN,g. 1934. Ueber die Formen der grossen HOWARD, H. 1946. A review of the Pleistocene birds Mowen ("subgenusLarus") und ihre gegensei- from FossilLake, Oregon. Carnegie Inst. Wash- tigen Beziehungen.J. Ornithol. 82: 340-380. ington Publ. 551: 141-195. VERMEER,K. 1970. Breeding biology of California HUBBS,C. L., & R. R. MILLER. 1948. The zoological and Ring-billed gulls: a study of ecologicalad- evidence: correlation between fish distribution aptation to the inland habitat. Can. Wildl. Serv. and hydrographichistory in the desertbasins of Rept. Ser. No. 12. western United States.Pp. 17-166 in The Great WELLS,P.V. 1983. Paleobiogeographyof montane Basin,with emphasison Glacial and Postglacial islands in the Great Basin since the last glacio- times. Bull. Univ. Utah 38 (Biol. Ser. 10). pluvial. Ecol. Monogr. 53: 341-382. , --, & L. C. HvB•s. 1974. Hydrographic ZINK, R. g., & D. W. WINKLER. 1983. Genetic and historyand relict fishesof the north-centralGreat morphologicalsimilarity of two California Gull Basin. Mem. California Acad. Sci. 7: 1-259. populationswith different life history traits. Blo- JEFFERSON,G. T. 1985. Review of the Late Pleisto- chem. Syst. Ecol. 11: 397-403. cene avifauna from Lake Manix, central Mojave 428 JOSEPHR. JEHL,JR. [Auk,Vol. 104