Molecular Genetic Status of Aleutian Canada Geese from Buldir and the Semidi Islands, Alaska ’

Tlze Condor 102: 172-180 0 The Cooper Ornithological Society 2000

MOLECULAR GENETIC STATUS OF ALEUTIAN CANADA GEESE FROM BULDIR AND THE SEMIDI ISLANDS, ALASKA ’

BARBARAJ. PIERSON,JOHNM. PEARCEAND SANDRA L. TALBOT United StatesGeological SurveyBiological ResourcesDivision, Alaska Biological ScienceCenter, 1011 E. Tudor Rd., Anchorage,AK 99503, e-mail: [email protected]

GERALD E SHIELDS* Institute of Arctic Biology, Universityof Alaska Fairbanks, Fairbanks, AK 99775.0180

KIM T. SCRIBNER Departmentof Fisheries and Wildlife, Michigan State University,East Lansing, MI 48824.1222

Abstract. We conductedgenetic analysesof Aleutian Canada Geese (Branta canadensis leucopareia) from Buldir Island in the western Aleutians and the Semidi Islands in the eastern portion of their breeding range. We compared data from seven microsatellite DNA loci and 143 basepairs of the control region of mitochondrialDNA from the two populations of Aleutian Canada Geese and anothersmall-bodied subspecies,the Cackling CanadaGoose (B. c. minima) which nests in western Alaska. The widely separatedisland-nesting Aleutian geese were genetically more closely related to each other than to mainland-nestingsmall- bodied geese. The populations of Aleutian geese were genetically differentiated from one another in terms of mitochondrial DNA haplotype and microsatellite allele frequencies, suggestinglimited contemporary gene flow and/or major shifts in gene frequency through genetic drift. The degree of populationgenetic differentiation suggeststhat Aleutian Canada Goose populations could be consideredseparate management units. There was someevi- dence of populationbottlenecks, although we found no significant genetic evidence of non- random mating or inbreeding. Kev words: Aleutian Canada Geese,bottlenecks, Branta canadensisleucopareia, genetics, &rosatellites, mtDNA.

INTRODUCTION been released due to the island’s remotenessand Three small-bodied subspeciesof Canada Geese difficult terrain (Byrd and Woolington 1983). (Branta canadensis) breed in Alaska: Taverner’s Restricted distribution and low remnant popu- Canada Geese (B. c. taverneri) nest predomi- lation numbers prompted the United States Fish nantly in the interior portion of the state and and Wildlife Service to list Aleutian Canada north of the Brooks Mountain range in northern Geese as endangered in 1967. Alaska; Cackling Canada Geese (B. c. minima) Hatch and Hatch (1983) subsequently de- nest predominantly in coastal regions of western scribed “Aleutian-like” Canada Geese from Kil- and southwestern Alaska; and Aleutian Canada iktagik of the Semidi Islands, and Bailey and Geese (B. c. leucopareia) nest only on a few Trapp (1984) discovered a population of “Aleu- small islands (Buldir, Nizki, Alaid, Agattu, Little tian-like” Canada Geese on Chagulak Island Kiska, Chagulak, Amukta) in the Aleutian chain (Fig. 1). Conventional morphometric studies and on the Semidi Islands (Kiliktagik and An- were inconclusive regarding the taxonomic and owik) southwestof Kodiak Island. reproductive status of these populations in rela- Aleutian Canada Geese were extirpated from tion to those on Buldir (S. Hatch, pers. comm.). most of their historical range (Fig. 1) after the However, Shields and Wilson (1987) determined introduction of non-native predatorsfor fur pro- that newly discovered “Aleutian-like” Canada duction between the 1830s and 1930s (USFWS Geese from Chagulak and Kiliktagik could not 1991). The only population known to survive be genetically differentiated from B. c. leuco- was on Buldir Island where predators had not pareia from Buldir based on 21 restriction en- zymes and examination of nearly 200 restriction fragments from mitochondrial DNA (mtDNA). I Received 13 May 1999. Accepted 20 October 1999. MtDNAs of the two other small-bodied Canada 2 Current address: Department of Biology and geese in Alaska (Cackling Canada Geese and Chemistry, Carroll College, Helena, MT 59601. Taverner’s Canada Geese) were clearly distin-

11721 GENETIC STATUS OF ALEUTIAN CANADA GEESE 173

FIGURE 1. Historical distribution (polygons) and current U.S. (black circles) breeding populationsof Aleutian Canada Geese (USFWS 1991). Locations sampled in this study were Aleutian Canada Geese from Buldir Island and the Semidi Islands and Cackling Canada Geese from the Yukon-Kuskokwim Delta. guishable from those of the three island popu- geese winter in the same areas as the Buldir lations of Aleutian Canada Geese. The discovery geese (USFWS 1991). of the two previously unknown breeding popu- Previous genetic studies of Aleutian Canada lations of Aleutian Canada Geese on Chagulak Geese (allozymes, Morgan et al. 1977; mtDNA, and Kiliktagik and an increase from approxi- Shields and Wilson 1987) were based on small mately 800 birds in the mid-1970s to 32,000 sample sizes and were not sufficiently polymor- birds in 1999, led the U.S. Fish and Wildlife phic to provide data with which to establishpop- Service to propose delisting Aleutian Canada ulation relationships and comparative population Geese as threatened (DeGange 1999). levels of variation. Recently, techniques includ- Despite the lack of genetic evidence of dis- ing analysis of biparentally-inherited nuclear mi- tinctions, banding data suggestthere are at least crosatellite loci have proven to be highly infor- two demographically distinct sub-populations of mative regarding questions about dispersal, gene Aleutian Canada Geese. Buldir birds stage and flow, and relationships of isolated populations winter in northern and central California, re- (McDonald and Potts 1997). In this study we spectively, whereas Semidi birds winter in used seven recently developed biparentally-in- Oregon (USFWS 1991). The large distance be- herited microsatellite loci (Buchholz et al. 1998, tween breeding sites on Buldir and those on the Cathey et al. 1998) and DNA sequenceanalysis Semidi Islands, the time interval since the breed- of 143 base pairs of domain I of the matemally- ing distribution was more continuous (ca. 1750, inherited avian mtDNA control region to study DeGange 1999), and the two populations’ cur- Aleutian Canada Geese from Buldir and the rent use of different wintering areas suggestthe Semidi Islands. These two non-coding genetic two populations may be reproductively isolated. systems are subject to higher rates of mutation Migration patterns and wintering locations for than coding regions of the genome (Brown et al. geese of Amukta and Chagulak of the central 1982, Goldstein and Pollock 1997) and conse- portion of the Aleutian Island chain are not well quently are thought to be more informative at defined, but some banding data indicate these the population level. Our objective was to use 174 BARBARA J. PIERSON ET AL. molecular genetic markers to assessthe extent primer, and non-labeled primers at locus-specific of gene flow among populations, the potential concentrations [0.8 FM (TTUCG-4) or 0.36 uM genetic effects of population bottlenecks, and (TTUCG-5, TTUCG-1, Bcap 1, Bcap. 3) non- genetic relationships of Aleutian Canada Geese labeled forward primer, 1.2 pM (TTUCG-4) or from Buldir and the Semidi Islands. For com- 0.4 pM (TTUCG-5, ‘ITUCG-1, Bcap. 1, Bcap. parison, we used Cackling Canada Geese from 3) or 0.12 pM (Beak 9) or 0.04 PM (Beak 11) the Yukon-Kuskokwim (Y-K) Delta, which also reverse primer]. Thermocycler conditions in- have experienced reductions in population num- cluded initial denaturation at 94°C for 2 min, bers, although not as drastic as Aleutian Canada followed by 25-35 cycles of denaturing at 94°C Geese have. for 1 min, annealing at locus-specific temperatures 56°C (TTUCG-4, Beak 1, Bcap 3, Beau METHODS 9, Bca(* 11) or 60°C (TTUCG-5, TTUCG-1) for SAMPLE COLLECTION AND PREPARATION 1 min, and extending at 72°C for 1 min. During 1991, whole blood samples were col- Genotypes at each microsatellite locus were lected from geese at Buldir Island (n = 29), the assigned relative to an Ml3 sequence standard Semidi Islands (n = S), and from captive geese and internal controls (individuals of known ge- raised at the University of California, Davis notype run on each gel) after standardpolyacryl- from eggs taken at Buldir (n = 5). In addition, amide gel electrophoresis and autoradiography. tissue samples (n = 12) identified as being from Microsatellite allele frequencies, estimates of Semidi Islands birds based on morphology and observed and expected heterozygosity, and ge- their discrete wintering location were collected netic distance (chord distance, Cavalli-Sforza from carcassesfound on wintering grounds in and Edwards 1967) were derived using BIOSYS 1993 and 1994 in Tillamook County, Oregon. (version 1.7, Swofford and Selander 1981). The Samples were placed in tissue preservation buff- chord distance has been shown to be robust with er (4 M Urea, 0.2 M NaCl, 100 mM Tris-HCl respect to tree topology for microsatellite loci pH 8.0, 0.5% n-Lauryl sarkosine, 10 mM surveyed for i&a-specific populations (Takezaki EDTA) and shipped at ambient temperatures. and Nei 1996). Genetic distances were used to Blood quills from molting Cackling Canada generate a multi-locus neighbor-joining tree in Geese (n = 41) were collected in 1995 on the MEGA (Kumar et al. 1993). Bootstrap values Y-K Delta and were placed in the tissue pres- could not be generated for the tree due to an ervation buffer described above. All samples insufficient number of populations. Estimates of were stored at -20°C in the laboratory prior to Wright’s (1969) inbreeding coefficient (F) were analysis. Genomic DNA was extracted from derived and tested for statistical significance as blood and tissue using standard phenol-chloro- described by Li and Horvitz (1953). Inbreeding form extraction procedures and quantified using coefficients range from - 1.0 to 1.0, with 1.0 in- a fluorometer. dicating complete homozygosity (and thus se- MICROSATELLITE ANALYSES vere inbreeding or invariance) among the indi- Each population was characterized using six dividuals sampled. nucleotide repeat microsatellite loci (TTUCG4 Analyses of variation in allele frequency with- and TTUCG-1, Cathey et al. 1998; Bcap 1, in and among Aleutian Canada Goose popula- Bcap 3, Bcap. 9, and Bcap 11, Buchholz et al. tions were conducted with both F-statistics 1998) and one 5-bp repeat microsatellite locus (Cockerham 1969) using the program FSTAT (TTUCG-5, Cathey et al. 1998). Each locus was (F,r, Fs~, and F,s, Goudet 1994) and an analog amplified in 25 ~1 reaction volumes using lOO- of Nei’s (1973) G-statistics (RSr) using the pro- 150 ng DNA, PCR buffer (10 mM Tris-HCl pH gram RstCalc (Goodman 1997). FIT is the pro- 8.5, 1.5 mM MgCl,, 50 mM KCl, 10 p,g ml-’ portion of variance among all samplesregardless nuclease-free BSA, 0.0025% Tween-20), 1.5 U of population; FSJRSTis the proportion of vari- of AmpliTaq DNA Polymerase (Perkin Elmer, ance between populations; and F,s is the pro- Foster City, California), dNTPs at 200 FM portion of variance within populations. The an- (TTUCG4, TTUCG-5, TTUCG-I), 120 FM alog, R,, and F-statistics differ in that R,, is (Bcap 3), or 80 pM (Bcap 1, Bcap. 9, and Bcap derived from variances in mean allele size and ll), 0.04 pM of y-32P-ATP-labeled forward frequency in relation to number sampled, while GENETIC STATUS OF ALEUTIAN CANADA GEESE 175

F-statistics are derived from variances in fre- nate stopsin variable regions noted in initial se- quency alone. quence optimization experiments, each dideoxy Tests to assessthe possibility of recent bottle- termination reaction was incubated for 30 min necks in each population were conducted using at 37°C with 1 (*l of a Terminal Deoxynucleo- the program BOTTLENECK which identifies an tidy1 Transferase(TdT) mix containing 3.33 mM excess of heterozygosity over all polymorphic dNTPs, 1X sequencing reaction buffer, 2.5 U loci by performing randomizations of data under TdT (Life Technologies, Inc., Frederick, Mary- two microsatellite models of mutation theory as- land), before adding 3 ~1 of stop solution. suming mutation-drift equilibrium (Cornuet and MtDNA sequenceswere aligned manually af- Luikart 1996). Outputs of these randomizations ter standard polyacrylamide gel electrophoresis were tested for heterozygosity excess and nor- and autoradiography. Haplotype designations mal distributions (Cornuet and Luikart 1996). were based on 143 bp of sequenceand were sub- Computer simulations indicate that the Stepwise mitted to GenBank under accession numbers Mutation Model (SMM) output of the Wilcoxon AF175474, AF175483AF175490, and AF175492. one-tail test for heterozygosity excess is more Estimatesof population haplotypefrequency were appropriate for this data set because our data used in subsequentanalyses. Estimates of haplo- have fewer than 30 individuals per population type heterogeneity [specifically, the nucleon (or and fewer than 10 microsatellite loci (G. Luikart, haplotype, designated by h) and nucleotide di- pers. comm.). versity (n) indices for nonselfing populations (equations 8.1 and 10.4, respectively, of Nei MTDNA ANALYSES 1987)] were estimated for each population using Primers Cl and ClR (Sorenson and Fleischer the DA function in the program REAP (McElroy 1996) were testedusing cesium chloride-purified et al. 1991) from pairwise gamma distances be- Canada Goose mtDNA and nuclear DNA to tween haplotypes (Kimura 2-parameter model, (Y confirm amplification of mtDNA and absence of = 0.5, Kimura 1980) generated in MEGA (Ku- nuclear sequences of mitochondrial origin mar et al. 1993). The nucleon diversity index (h) (Numt) product as described by Sorenson and is approximately equivalent to the probability Quinn (1998). A hypervariable portion of the that two randomly chosen individuals will have mtDNA control region (3’ end of domain I, Bak- different haplotypes. The nucleotide diversity in- er and Marshall 1997) was sequenced from a dex (IT, Nei and Tajima 1981) measures the av- sample of geese in each population (Buldir, II = erage pairwise nucleotide difference between in- 9; Semidi, n = 13; Cackling Canada Geese, n = dividuals within samples. The measurement 7~ 17). An ~400 bp fragment of mtDNA was am- corrects h for the size of the nucleon examined plified in 50 yl reaction volumes using 50 ng of (Nei 1987). total genomic DNA, PCR buffer (67 mM Tris- Heterogeneity of haplotype distribution HCl, pH 8.0, 6.7 mM MgCl,, 0.01% Tween-20) among populations was tested for significance 1 mM of each dNTP 1 p,M of Cl primer, 1 yM using the MONTE function in the program of ClR primer, and 1 U of AmpliTaq DNA Poly- REAP (McElroy et al. 1991) using 1,000 repli- merase (Perkin Elmer, Foster City, California). cates based on Roff and Bentzen’s (1989) Monte Thermocycler parameters included a hot start at Carlo chi-square test. This test has been shown 80°C for 5 min before adding the AmpliTaq to be appropriate for data sets in which many DNA Polymerase and cycling 40 times at 94°C elements (in our case, haplotypes) occur fewer for 1 min, 60°C for 1 min, and 72°C for 1 min. than five times (Roff and Bentzen 1989). PCR products were cleaned with Ultrafree- Significance of population differences in hap- MC, 30,000 NMWL Filter Unit spin columns lotype frequency (Phi,& were estimated using (Millipore Corporation, Bedford, Massachusetts) the program AMOVA (Excoffier et al. 1992), an and then sequenced using the internal labeling mtDNA-analog of standard F-statistics (Cock- (w32P-dATPprotocol in the SequiTherm EXCEL erham 1969) which are commonly employed for DNA Sequencing Kit (Epicentre Technologies, biparentally inherited loci. Phi,, was calculated Madison, Wisconsin) and 1.5 PM of the Cl both with and without weighting the haplotype primer. Thermocycler parameters were 95°C for frequencies with pairwise Euclidean distancesof 5 min followed by 30 cycles of 95°C for 30 set, nucleotide differences between haplotypes. 56°C for 30 set, and 70°C for 1 min. To elimi- Maximum parsimony analyses of mtDNA hap- 176 BARBARA J. PIERSON ET AL.

TABLE 1. Summary of F-statistics after 500 permutations (Goudet 1994) and R-statistics after 1,000 Somid,blandr permutations(Goodman 1997) at individual loci and csekungcanada acrossall microsatellite loci, and mtDNA Phi,, values Gg= after 500 bootstrapiterations (Excoffier et al. 1992). 0.01 l-l Locus FIT FST FLY RST FIGURE 2. Neighbor-joining tree based on Cavalli- Sforza and Edwards (1967) chord distancesfrom seven TTUCG-1 -0.144 0.080 -0.243 0.078” microsatellite loci of Canada Geese. TTUCG-4 0.560 0.075 0.525 0.113a TTUCG-5 0.078 0.016 0.062 -0.012 Bca)* 1 0.107 0.064a 0.047 0.009 lotype sequences could not be completed be- Beau 3 0.096 0.03 1 0.067 0. 124a Bca; 9 0.219 0.124’ 0.108 -0.016 cause there were fewer informative sites than Beak 11 -0.042 -0.005 -0.038 0.065 haplotypes (i.e., identical site changes found in Total over more than one haplotype used to infer evolu- all micro- tionary relationships). As an alternative to a satellites 0.081 0.05@ 0.033 0.052 maximum parsimony analysis, pairwise gamma mtDNA Phi,, Without pairwise distances between haplotypes (Kimura 2-param- distance informa- eter model, ct = 0.5, Kimura 1980) were used tion 0.345” to construct a minimum spanning tree network. Including pairwise An alpha level SO.007 (i.e., 0.05 divided by distance informa- 0.280” 7 [the number of tests]) was considered signifi- tion cant for all tests or sets of multiple comparisons. d Value is rigmlicantly dtfferent from expectation. Means + SE are presented in the text. erogeneity in allele frequency as did the overall RESULTS values when all loci were considered for F,, (Ta- MICROSATELLITE DNA ble 1). Individually significant loci differed be- All seven microsatellite loci were polymorphic, tween FST (Bcap. 1, Beak 9) and R,, (TTUCG- with the number of alleles observed at individual 1, TTUCG-4, Bca(*. 3). This may be because R,, loci ranging from 2 to 21. Cackling Canada is corrected for the variance from the mean al- Geese and Semidi Islands geese were mono- lele size in each population. Although the over- morphic for one allele at locus TTUCG-4, but all R,, value was not significant after a correc- all three populations were polymorphic at each tion for multiple tests (P = 0.02), the value is of the other loci. Mean heterozygosities over all approximately the same as the overall F,, value. loci were similar across all three populations These results suggest that these populations ex- (range: 0.57-0.59), although heterozygosities at perience little genetic exchange with each other individual loci varied greatly between popula- and/or that population bottlenecks have resulted tions. A neighbor-joining tree, based on Cavalli- in large shifts in allele frequency over recent Sforza and Edwards (1967) chord distances, re- time periods. However, we failed to detect recent vealed that the Buldir and Semidi populations bottlenecks using the program BOTTLENECK were more genetically similar (0.32) than either (Cornuet and Luikart 1996) for each population Aleutian Canada Goose population was to Cack- (SMM Wilcoxon one-tail test for heterozygosity ling Canada Geese (Buldir x Cackling Canada excess Buldir P = 0.22; Semidi P = 0.42; Cack- Geese = 0.40; Semidi X Cackling Canada Geese ling Canada Geese P = 0.42). = 0.48; Fig. 2). Inbreeding coefficients (E Wright 1969) were not significant at any locus MTDNA or in any population except locus TTUCG-4 for Cackling Canada Geese and Aleutian Canada Buldir Island (x2, = 7.0, P < 0.01). Geese did not have distinct mtDNA in the 143 All loci appeared to be in Hardy-Weinberg base pairs of the mtDNA genome surveyed, al- equilibrium given the small and nonsignificant though substantial differences in haplotype fre- values of FIs (variance within populations, Table quency were observed between the two subspecies 1). When either F,, or R,, (variance between and between the Buldir and Semidi populations of populations) values were calculated, two to three Aleutian Canada Geese. Ten mtDNA haplotypes of the seven loci showed significant spatial het- with 10 variable sites were detected (seven tran- GENETIC STATUS OF ALEUTIAN CANADA GEESE 177

F QYKDelta 11.8%

Semidis 84.6% YK Delta 58.8%

YK Delta 11.8%

FIGURE 3. Minimum spanning tree indicating relationships among 10 mtDNA haplotypes identified in this study and haplotype frequencies in each population of Canada Geese. Each branch represents a single nucleotide difference with the exception of haplotype N which differed from haplotype L by two base pairs. The tree is drawn to scalebased on pairwisegamma distances between haplotypes (Kimura 2-parametermodel, CY = 0.5, Kimura 1980).

sitions, two transversions,and one insertionldele- The analysis of mtDNA haplotype variance tion; GenBank accession numbers AF175474, and resulting F-statistic, Phi,, were significant AF175483-AF175490, and AF175492) and whether or not the Euclidean squared distances aligned with published mtDNA sequencesfrom between haplotypes were used (Table 1). The Snow Geese (Chen caerulescens,Quinn 1992). minimum spanning tree network central position Haplotype L was the most common haplotype is occupied by haplotype L which is geograph- overall, and geographicallyubiquitous. Many hap- ically ubiquitous. All other haplotypes are no lotypes were found in only one subspeciesor one more than two mutations from haplotype L (Fig. population (e.g., haplotypesE S, T, and U were 3). observed only in Cackling Canada Geese, haplotypes N, P, and Q were observed only in Buldir DISCUSSION Island geese, whereashaplotype 0 was observed GENETIC INFERENCES only in Semidi Islands geese).Haplotype diversi- ties (h) ranged from 0.28 ? 0.11 (Semidi Islands The seven microsatellite loci and mtDNA se- geese)to 0.76 2 0.08 (Buldir Island) with the hap- quence data indicate significant genetic differ- lotype diversity of Cackling Canada Geese (h = entiation between the two Aleutian Canada 0.63 + 0.08) being intermediate. Nucleotide di- Goose populations, but a closer relationship to versity (IT) values were much lower becausethey each other than either population is to Cackling are correctedfor the number of base pairs exam- Canada Geese. A neighbor-joining tree based on ined, but followed the same trends as haplotype microsatellite data (Fig. 2) and an analysis of diversity (data not shown). Monte Carlo simula- molecular variance from microsatellite data (FsT tions testing for haplotype heterogeneity across and R,,), as well as the mtDNA analog (Phi,,, populationswere significant whether or not Cack- Table 1) supports this relationship. The Phi,, ling Canada Geese were included (P = 0.001 + value is approximately four times the value of 0.001). the nuclear loci, as expected,because mtDNA is 178 BARBARA J. PIERSON ET AL. maternally inherited and has an effective size l/4 alleles with the second allele occurring in only of nuclear markers (Avise 1994). Significant dif- a few individuals from Buldir) and not an indi- ferences in haplotype heterogeneity (P = 0.001 cation of inbreeding. F,s values are not signifi- ? 0.001) also indicate a high degree of repro- cant. The absence of additional significant in- ductive isolation between the two Aleutian Can- breeding coefficients or F. values suggeststhat ada Goose populations. The statistically signifi- all populations surveyed are in Hardy-Weinberg cant inter-population variance analyses of both equilibrium for all loci and do not appear to be mtDNA and microsatellite allele frequencies ar- inbred. gue for consideration of these populations as Both microsatellite and mtDNA frequency separatemanagement units (Moritz 1994). distributions are consistent with the current tax- Historically, there is evidence these popula- onomic separation of Aleutian Canada Geese tions were much smaller than they are today from Cackling Canada Geese (Bellrose 1978). (USFWS 1991). Thus, an alternative explanation Shields and Wilson (1987) were able to differ- for the degree of differentiation between these entiate Aleutian Canada Geese from Cackling two populations of Aleutian Canada Geese could Canada Geese based on restriction enzyme di- be that population bottlenecks have causedlarge gests of the entire mtDNA for two individuals shifts in allele and haplotype frequency between per population (Buldir, Chagulak, Kiliktagik, the two populations. Indeed, the data show that and Y-K Delta). However, these small sample Semidi Islands geese have fewer alleles overall sizes may not have been sufficient to establish and more heterozygotes than expected for five taxonomic divergence (Shields and Wilson of seven microsatellite loci, as well as the lowest 1987). With our larger sample size, we were un- haplotype and nucleotide diversity for mtDNA, able to designate individuals to a particular sub- three potential indicators of a recent bottleneck species based on this short portion of mtDNA (Nei et al. 1975). However, statistical tests of sequencealone. Shared microsatellite alleles and these data fail to indicate inbreeding or a recent mtDNA haplotypes between both subspecies, bottleneck, indicating that if a bottleneck oc- and between the Aleutian Canada Goose popu- curred, it was not of long duration. Our results lations, indicate either a common ancestry suggest the differentiation observed is more among all three populations, or recent, but lim- likely due to limited contemporary gene flow ited gene flow, or both. The minimum spanning than due to recent bottlenecks. tree network (Fig. 3) provides a graphical sum- Although low sample size in the Semidi Is- mary of haplotype relationships with some geo- lands population (n = 18) may explain the low graphic relevance. The lack of phylogenetically number of alleles in comparison with our other informative sites may be due to the small num- two populations, it does not explain the in- ber of nucleotides sequenced (too few nucleo- creased levels of heterozygosity. Furthermore, a tides sequencedfor phylogenetic analysis) or be- comparison of statistics for a Canada Goose cause the subspecies have only recently di- population with identical sample size revealed verged. greater allele abundance and lower than expected levels of heterozygosity (unpubl. data for CONSERVATION IMPLICATIONS Canada Geese from Washington state). There- Our objective was to assessthe extent of gene fore, although microsatellite allele and mtDNA flow and effects of population bottlenecks on haplotype frequencies would likely change with Aleutian Canada Geese from Buldir and the increased sample sizes from the Semidi Islands, Semidi Islands. The occurrence of shared alleles we do not believe observed locus characteristics and haplotypes indicates common ancestry, but are an artifact of low sample size. may suggest some level of historical gene flow A locus-by-locus evaluation of Wright’s as well. Statistical analyses of our data indicate (1969) inbreeding coefficient (F) and statistical there is little, if any, current gene flow between significance as described by Li and Horvitz these populations; this is based on significant (1953) shows inbreeding is insignificant for all variances between populations for both micro- populations at all loci with the exception of the satellites and mtDNA (Table 1). Although the Buldir Island population at locus TTUCG-4. The populations on Buldir Island and the Semidi Is- statistical significance of this locus is most likely lands have experienced different rates of popu- an artifact of locus characteristics(i.e., only two lation growth since listing, we found no statis- GENETIC STATUS OF ALEUTIAN CANADA GEESE 179 tically significant genetic evidence that either breeding population of the Aleutian Canada has gone through a recent population bottleneck, Goose. Am. Birds 38:284-286. BAKER,A. J., AND H. D. MARSHALL.1997. Mitochon- although increased sample size and number of drial control region sequencesas tools for under- loci would increase the power of our tests. Aleu- standing evolution, p. 51-82. In D. l? Mindell tian Canada Geese from the central Aleutian Is- [ED.], Avian molecular evolution and systematics. lands (Chagulak) were not available for analysis Academic Press, San Diego, CA. BELLROSE,E C. 1978. Ducks, geese, and swans of in this study. Because little is known about this North America. StackpoleBooks, Harrisburg,PA. population’s migration corridors, wintering are- BROWN,W. M., E. M. PRAGER,A. WANG, AND A. C. as, current population numbers, or genetic sta- WILSON.1982. Mitochondrial DNA sequencesof tus, it is possible that genetic analyses of this primates: tempo and mode of evolution. J. Mol. geographically intermediate population would Evol. 18:225-239. BUCHHOLZ,W. G., J. M. PEARCE,B. J. PIERSON,AND reveal gene flow not detected in our study. We K. T. SCRIBNER.1998. Dinucleotide repeat poly- distinguish populations of Aleutian Canada morphisms in waterfowl (family Anatidae): char- Geese as genetically differentiated from each acterizationof a sex-linked (Z-specific) and 14 bi- other. Furthermore, the lower number of alleles parentally inherited loci. J. Anim. Genet. 29:323- 325. observed in the Semidi Islands population are BYRD,G. V., AND D. W. WOOLINGTON.1983. Ecology consistent with the lower population numbers of Aleutian Canada Geese at Buldir Island, Alas- and slower growth rate relative to the Buldir Is- ka. U.S. Fish and Wild]. Serv., Special Scientific land population. Continued poor recruitment in Rep., Wildlife No. 253, Washington,DC. the Semidi Islands population may jeopardize an CATHEY,J. C., J. A. DEWOODY,AND L. M. SMITH. 1998. Microsatellite markers in Canada Geese irreplaceable component of the overall diversity (Bruntu can&en&). J. Hered. 89: 173-175. of the subspecies. We recommend that future CAVALLI-SFORZA,L. L., AND A. W. E EDWARDS.1967. conservation measures for Aleutian Canada Phylogeneticanalysis: models and estimationpro- Goose populations include increasing genetic cedures.Am. J. Human Genet. 19:233-257. COCKERHAM,C. C. 1969. Variance of gene frequencies. sample size in the Buldir and Semidi Islands, Evolution 23:72-84. genetic sampling and analyses of geese from CORNUET,J. M., AND G. LUIKART.1996. Description Chagulak Island in the central Aleutian Island and power analysis of two tests for detecting re- chain to ascertain the role of that population in cent population bottlenecksfrom allele frequency Aleutian Canada Goose genetic structure, as data. Genetics 144:2001-2014. DEGANGE,A. 1999. Proposal to remove the Aleutian well as careful monitoring of the Semidi Islands Canada Goose from the list of endangered and population. threatenedwildlife. Fed. Reg. 64:42058-42068. EXCOFFIER,L., P. E. SMOUSE, AND J. M. QUATTRO. ACKNOWLEDGMENTS 1992. Analysis of molecular variance inferred from metric distances among DNA haplotypes: Ann Rappaport and Tony DeGange, USFWS Region application to human mitochondrial DNA restric- 7 gave guidance.USFWS Region 7, Anchorage, Alas- tion data. Genetics 131:479-491. ka, provided funding. Aleutian Canada Goose blood GOLDSTEIN,D. B., AND D. D. POLLOCK.1997. Launch- samples were collected by Vern Byrd and Jeff Wil- ing microsatellites:a review of mutationprocesses liams, USFWS, while on Buldir Island, by Roy Lowe and methods of phylogenetic inference. J. Hered. on the Semidi Islands, and from captive birds at the 88:335-342. University of California-Davis by Scott McWilliams. GOODMAN,S. J. 1997. Rst Calc: a collection of com- Dave Pitkin, Oregon Coastal Refuges, USF’WS, New- puter programsfor calculatingestimates of genetic port, Oregon collected tissue samplesfrom Semidi Is- differentiation from microsatellite data and deter- lands carcasses. Ada Fowler, LJSGSlBRD provided mining their significance.Mol. Ecol. 6:881-885. Cackling Canada Goose samples. Gordon Luikart, GOUDET,J. 1994. FSTAT, a program for IBM PC com- Universitt JosephFourier, Grenoble, France, provided patibles to calculateWeir and Cockerham’s (1984) assistancewith statistical interpretation of BOTTLE- estimatorsof F-statistics, version 1.2. Distributed NECK analyses. Comments on the draft manuscript by the author. Inst. Zoo]. Ecol. An., Univ. Lau- were provided by and appreciatedfrom Brian Ander- sanne, Lausanne,Switzerland. son, Vernon Byrd, Dirk Derksen, Dan Esler, Walter HATCH,S. A., AND M. A. HATCH. 1983. An isolated Koenig, and two anonymousreviewers. population of small Canada Geese on Kiliktagik Island, Alaska. Wildfowl 34:130-136. LITERATURE CITED KIMURA, M. 1980. A simple method for estimating evolutionary rate of base substitutions through AVISE, J. C. 1994. Molecular markers, natural history, comparative studies of nucleotide sequences.J. and evolution. Chapman and Hall, New York. Mol. Evol. 16:111-120. BAILEY,E. I?, AND J. L. TRAPP.1984. A second wild KUMAR, S., K. TAMURA, AND M. NEI. 1993. MEGA: 180 BARBARA .I. PIERSON ET AL.

molecular evolutionary genetics analysis. Version QUINN, T. W. 1992. The genetic legacy of Mother 1.0. Pennsylvania State Univ., University Park, Goose-phylogeographic patternsof Lesser Snow PA. Goose Chen caerulescens caerulescens maternal LI, C. C., AND D. G. HORVITZ.1953. Some methodsof lineages. Mol. Ecol. 1:105-l 17. estimating the inbreeding coefficient. Am. J. Hu- ROFF. D. A.. AND P. BENTZEN.1989. The statistical man Gen. 5:107-117. analysis of mitochondrial DNA polymorphisms: MCDONALD,D. B., AND W. K. POTTS.1997. DNA mi- x2 and the problem of small samples. Mol. Biol. crosatellites as genetic markers at several scales, Evol. 6:539-545. p. 29-49. In D. P Mindell [ED.], Avian molecular SHIELDS,G. E, AND A. C. WILSON.1987. Subspecies evolution and systematics.Academic Press, San of the Canada Goose (Brunta canadensis) have Diego, CA. distinct mitochondrial DNAs. Evolution 41:662- MCELROY, D., P. MORAN, E. BERMINGHAM,AND I. 666. KORNFIELD.199 1. REAP: the restriction enzyme SORENSON, M. D., AND R. C. FLEISCHER.1996. Multiple analysis package, version 4.0. Dept. Zool., Univ. independenttranspositions of mitochondrialDNA Maine, Orono, ME. control region sequencesto the nucleus.Proc. Nat. MORGAN,R. F!, S. T. SULKIN,AND C. .I. HENNY.1977. Acad. Sci. 93:15239-15243. Serum proteins of Canada Goose (Brunta cnna- SORENSON, M. D., AND T. W. QUINN. 1998. Numts: a densis) subspecies.Condor 79:275-278. challenge for avian systematicsand populationbi- MORITZ,C. 1994. Defining “evolutionarily significant ology. Auk 115:214-221. SWOFFORD,D., AND R. B. SELANDER.1981. BIOSYS- units” for conservation.Trends Ecol. Evol. 9:373- 1: a FORTRAN program for the comprehensive 375. analysis of electrophoreticdata in population ge- NEI, M. 1973. Analysis of gene diversity in subdivided netics and systematics.J. Hered. 72:281-283. populations.Proc. Nat. Acad. Sci. 70:3321-3323. TAKEZAKI,N., AND M. NEI. 1996. Genetic distanceand NEI, M. 1987. Molecular evolutionary genetics. Co- reconstructionof phylogenetic trees from micro- lumbia Univ. Press, New York. satellite DNA. Genetics 144:389-399. NEI, M., T MARUYAMA,AND R. CHAKRABORTY.1975. UNITEDSTATES FISH AND WILDLIFESERVICE (USFWS). The bottleneck effect and genetic variability in 1991. Aleutian Canada Goose recovery plan. U.S. populations.Evolution 29:1-10. Fish Wildl. Serv., Anchorage, AK. NEI, M., ANDE TAJIMA.1981. DNA polymorphismde- WRIGHT,S. 1969. Evolution and the genetics of pop- tectable by restrictionendonucleases. Genetics 97: ulations. Vol. 2. The theory of gene frequencies. 145-163. Univ. Chicago Press, Chicago.