The Auk 124(2):444–462, 2007 © The American Ornithologists’ Union, 2007. Printed in USA.

POPULATION STRUCTURE AND MITOCHONDRIAL POLYPHYLY IN NORTH AMERICAN ( STREPERA) Jeffrey L. Peters1 and Kevin E. Omland Department of Biological Sciences, University of Maryland Baltimore County, 1000 Hilltop Circle, Baltimore, Maryland 21250, USA

Abstract.—We examined population genetic structure in Gadwalls (Anas strepera) to test the prediction that female philopatry and fi delity to migratory fl yways have contributed to the partitioning of mitochondrial variation across North America. Sequencing a 658–659 base-pair fragment of the mitochondrial DNA (mtDNA) control region from 348 Gadwalls, we found two distinct clades that were broadly intermixed across both breeding and nonbreeding populations. Clade A was abundant in North America as well as among published sequences from Eurasia. Clade B was sequenced from 5.5% of North American Gadwalls and was more similar to Asian Falcated (A. falcata) haplotypes than to clade A haplotypes. Maximum likelihood indicated that clade B haplotypes were a monophyletic group nested within Falcated Duck haplotypes, which sug- gests mtDNA introgression of clade B into Gadwalls. However, that topology was weakly supported, and we could not reject topologies that were consistent with incomplete lineage-sorting as the cause of mitochondrial polyphyly. Migratory fl yways did not contribute signifi cantly to population structure and, in general, we found a lack of genetic structure among most populations. However, Gadwalls sampled in Alaska and Washington were well diff erentiated from other popula- tions. Coalescent analyses supported a historical population expansion for clade A, and this expansion could have contributed to the high genetic similarity among some populations but the strong diff erentiation of others. Female-mediated gene fl ow, along with both historical and contemporary population and range expan- sions, has likely contributed to the overall weak mtDNA structure in North American Gadwalls. Received 27 October 2005, accepted 25 March 2006.

Key words: Anas falcata, Anas strepera, , Falcated Duck, Gadwall, mitochondrial DNA, phylogeography, polyphyly, population expansion.

Estructura Poblacional y Polifi lia Mitocondrial en Anas strepera

Resmen.—Examinamos la estructura genética poblacional en Anas strepera para poner a prueba la predicción de que la fi lopatría de las hembras y la fi delidad a los corredores de vuelo han contribuido a la partición de la variación mitocondrial a través de Norte América. Secuenciamos un fragmento de la región control del ADN mitocondrial (ADNmt) de 658–659 pares de bases para 348 individuos, y encontramos dos clados distintos que se encontraron ampliamente distribuidos a través de poblaciones reproductivas y no reproductivas. El clado A fue abundante en Norte América, y también en secuencias publicadas de Eurasia. En el clado B se ubicaron el 5.5% de los individuos de Norte América, y fue más similar a haplotipos de A. falcata que a los haplotipos del clado A. De acuerdo a análisis de

1Present address: Institute of Arctic Biology, University of Alaska Fairbanks, Fairbanks, Alaska 99775, USA. E-mail: [email protected] 444 April 2007] Population Structure in Gadwalls 445

máxima verosimilitud, los haplotipos de A. strepera presentes en el clado B forman un grupo monofi lético que se encuentra anidado dentro del clado de haplotipos de A. falcata, lo que sugiere que ha existido introgresión mitocondrial del clado B hacia A. strepera. Sin embargo, esa topología no está bien respaldada, y no pudimos rechazar topologías consistentes con la hipótesis de que la polifi lia mitocondrial sería el resultado de un proceso de separación incompleta de linajes. Los corredores de vuelos migratorios no contribuyeron signifi cativamente a la estructura poblacional y, en general, no encontramos estructura genética entre la mayoría de las poblaciones. Sin embargo, los individuos de A. strepera muestreados en Alaska y Washington estuvieron bien diferenciados de los de otras poblaciones. Análisis de coalescencia apoyaron una hipótesis de expansión poblacional, y esta expansión podría haber contribuido a la alta similitud genética entre algunas poblaciones y la marcada diferenciación de otras. El fl ujo genético mediado por las hembras, junto con expansiones poblacionales históricas y contemporáneas, probablemente han contribuido a la estructura débil del ADNmt entre las poblaciones norteamericanas de A. strepera.

Pylgegraic srveys f mitochondrial lineage sorting and interspecifi c hybridization DNA (mtDNA) have revealed diverse patt erns likely contributed to these patt erns (Avise et al. of genetic variation among species (reviewed in 1990; Rhymer et al. 1994; Omland 1997; Johnson Avise 2000). This variation is partitioned among and Sorenson 1999; McCracken et al. 2001; populations as a result of complex interactions Kulikova et al. 2004, 2005; Peters et al. 2005b). between contemporary and historical gene are well known for their ability to hybrid- fl ow as well as other demographic processes. ize and produce fertile off spring, and some spe-

Many species are strongly structured across cies regularly hybridize in sympatry (Johnsgard their ranges, and diff erent geographic regions 1960, Tubaro and Lij tmaer 2002, Mallet 2005). sometimes harbor distinct mtDNA lineages, In addition, the genus Anas underwent a rapid which suggests that gene fl ow has been histori- radiation (Johnson and Sorenson 1999), and cally restricted (e.g., Milot et al. 2000, Omland some species likely diverged too recently for et al. 2000, Drovetski et al. 2004). Other species mtDNA to have sorted to monophyly (Avise et seem to be genetically homogeneous across al. 1990, Omland 1997). Determining the reasons their ranges, which suggests regular gene fl ow. why diff erent species share mtDNA lineages can However, range expansions can also result in be diffi cult in this group, and in other recently high genetic similarity among populations. radiated groups, because species that are likely For some species, expansions from multiple to have incompletely sorted mtDNA are also source populations seem to have contributed to likely to hybridize (e.g., McCracken et al. 2001, broad intermixing of distinct mtDNA lineages Tubaro and Lij tmaer 2002, Kulikova et al. 2004). across a species’ range (e.g., Avise et al. 1992, The Gadwall (Anas strepera) has a Holarctic Quinn 1992, Lambert et al. 2002, Ruokonen et distribution extending across Europe, Asia, and al. 2004). North America. This range crosses a number of Multiple mtDNA lineages sometimes render potential barriers to dispersal, including oceans, a species’ mtDNA paraphyletic or polyphy- mountain ranges, deserts, and other uninhab- letic with respect to mtDNA of other closely ited areas. Although the Gadwall is migratory related species; that is, some haplotypes are and can likely traverse large expanses of inhos- more closely related to haplotypes of other pitable environments, several behavioral charac- species than to haplotypes of some conspecif- teristics make it likely that populations will be ics. In a survey of published phylogenetic and genetically structured, even within continents. phylogeographic studies, Funk and Omland In North America, Gadwalls migrate within (2003) observed that 17% of species had four delineated fl yways (Atlantic, Mississippi, nonmonophyletic mtDNA. Within ducks in the Central, and Pacifi c; Fig. 1), and most individu- genus Anas, several species have paraphyletic als that breed within a given fl yway migrate or polyphyletic mtDNA, and both incomplete to wintering grounds within that same fl yway 446 Peters and Omland [Auk, Vol. 124

g enerally occurs on the wintering grounds, where multiple populations intermix, and males follow their mates back to the female’s breeding site (Rohwer and Anderson 1988). Therefore, the maternally inherited mtDNA is especially likely to be structured among duck populations (Tiedemann et al. 1999, Scribner et al. 2001, Gay et al. 2004, Pearce et al. 2005). In several species of waterfowl, mtDNA is not structured among fl yways or among popu- lations (Cronin et al. 1996, Pearce et al. 2004). Historical population expansions may partially explain this lack of structure (Pearce et al. 2004, 2005). There are two reasons to predict a his- torical expansion in North American Gadwalls. First, environmental changes during the last glacial episode probably had a major eff ect on Gadwalls. More than 90% of North American Gadwalls breed in the Prairie Pothole Region of the Central Flyway (Canadian Wildlife Service Fig. 1. North American breeding and win- 2005; U.S. Fish and Wildlife Service 2005a, b; tering distribution of the Gadwall (from Fig. 1), and this region was covered by ice dur- LeSchack et al. 1997). Migratory flyways are ing the last glacial maximum. In fact, glacial ice labeled and indicated by differences in shading that completely melted only 9,000 years ago (from Baldassarre and Bolen 1994); the Prairie created the abundant wetlands that now pro-

Potholes Region is outlined by a solid black line vide excellent waterfowl habitat in that region (from Bellrose 1976). (Pielou 1991). Therefore, Gadwalls must have colonized their primary North American breed- (Bellrose 1976). Migratory fl yways may con- ing range aft er glacial retreat. Second, using an tribute to the partitioning of genetic variation mtDNA phylogeny of the genus Anas, Johnson among populations (Anderson et al. 1992, Esler and Sorenson (1999) reconstructed the ancestral 2000). Comparing mtDNA sampled from diff er- area of Gadwalls to be in Eurasia, which sug- ent fl yways has provided support for this possi- gests that Gadwalls recently colonized North bility in Red-crested Pochards (NeĴ a rufi na; Gay America. Under this hypothesis, a population et al. 2004), Lesser White-fronted Geese (Anser expansion must have occurred to account for the erythopus; Ruokonen et al. 2004), and Wood large population size and widespread distribu- Ducks (Aix sponsa; Peters et al. 2005a). However, tion of Gadwalls in North America. migratory fl yways do not have discrete boundar- North American Gadwalls have also experi- ies, and some Gadwalls that breed in the Central enced population and range expansions during Flyway migrate to wintering areas within other the past century (LeSchack et al. 1997). Since fl yways (Bellrose 1976). Interfl yway migrants 1955, the Canadian Wildlife Service (2005) and could potentially facilitate gene fl ow between the U.S. Fish and Wildlife Service (2005a) have distantly spaced populations. counted breeding waterfowl along traditional Strong breeding philopatry, however, can transects that cover most of the Gadwall’s Central give rise to population structure, even when Flyway distribution. Within the traditional diff erent populations intermix on the wintering survey area (which also includes Alaska), the grounds (Rhodes et al. 1993, Esler 2000, Scribner Gadwall population has increased at an average et al. 2001, Kimura et al. 2002; see also Webster rate of 2.5% per year (Canadian Wildlife Service et al. 2002). In contrast to the sex-specifi c dis- 2005). Breeding populations have also increased persal patt erns of most , female ducks are in size and range in the Pacifi c Flyway (Canning philopatric to natal and previous breeding sites, and Herman 1983, U.S. Fish and Wildlife Service whereas males disperse (Rohwer and Anderson 2005b), and breeding Gadwalls only recently 1988, Anderson et al. 1992). Pair formation colonized the Atlantic Flyway (Henny and April 2007] Population Structure in Gadwalls 447

Holgersen 1974). Thus, contemporary and his- Quebec (QC), Manitoba (MB), and south-coastal torical expansions have likely contributed to the Alaska (AKsc). Nonbreeding populations partitioning of genetic variation in this species. included adult males, adult females, and juve- The primary objective here was to examine niles sampled during or aft er peak migration: population genetic structure in North American Maritime Provinces (MP; Prince Edward Island, Gadwalls. Specifi cally, we tested the hypothesis Nova Scotia, and New Brunswick), Delmarva that female breeding philopatry and fi delity Peninsula (DM; Delaware and Maryland), to migratory fl yways has contributed to the Arkansas (AR), and Kodiak Island, Alaska partitioning of mtDNA variation. In addition, (AKko). Finally, we sampled 44 individuals we tested the hypothesis that Gadwalls expe- from additional widespread locations, for a total rienced a historical population expansion in of 348 North American Gadwalls (Appendix). North America. Early in the sampling stage, Laboratory methods.—We extracted DNA from however, we found multiple Gadwall haplo- various sources (Appendix) using the DNeasy types that were more closely related to Falcated Tissue Kit (Qiagen, Valencia, California). We Duck (Anas falcata) haplotypes (designated sequenced a 658–667 base-pair fragment of the “clade B”) than to the majority of Gadwall hap- mtDNA control region, including the hyper- lotypes (“clade A”). The Gadwall and Falcated variable 5’ end, using primers L78 and H774 Duck are each other’s closest phylogenetic rela- (Sorenson and Fleischer 1996, Sorenson et al. tives (Johnson and Sorenson 1999, Peters et al. 1999). Polymerase chain reaction and sequencing 2005b), but fi nding clade B haplotypes in North followed standard protocols (see McCracken et America was surprising, because the Falcated al. 2001, Peters et al. 2005b). Eleven haplotypes Duck is restricted to eastern Asia (eastern were previously published (GenBank accession Russia, Japan, northern China, and ). numbers: AY881776–78, AY881784–91; Peters Therefore, a second goal of the present study et al. 2005b); we also included 12 published was to estimate the frequency of clade B haplo- haplotypes from Eurasian Gadwalls (AY112944,

types, and to test hypotheses regarding genea- AY881779–83, AY881792–96, AY881809; Donne- logical relationships and historical demography Goussé et al. 2002, Peters et al. 2005b). All new within clade B. sequences have been deposited in GenBank (DQ449086–422). Metds Phylogenetic methods.—We aligned control- region sequences using SEQUENCHER, ver- Sample collection.—We sampled at least 12 sion 4.1 (Gene Codes Corporation, Ann Arbor, individuals from each of 14 Gadwall popula- Michigan). To build phylogenetic trees, we used tions in North America. Each population was equally weighted maximum parsimony (MP) assigned to one of the following three catego- and maximum likelihood (ML) in PAUP*, ver- ries: “known breeding,” “putative breeding,” sion 4.0b10 (Swoff ord 1999). For tree searches, or “nonbreeding” (Appendix). Known breeding we included one representative of each unique populations included samples collected from Gadwall haplotype and one representative of nests, broods (one representative per brood), each of four Falcated Duck haplotypes (Peters adult females and juveniles trapped between et al. 2005b). We also randomly chose one hap- 22 May and 23 August, or individuals shot lotype from American (A. americana), during the hunting season but unlikely to have Chiloe Wigeon (A. sibililatrix), and Eurasian migrated (the fi rst and second primaries had Wigeon (A. penelope; Peters et al. 2005b), because not developed beyond the blood quill stage; this group is sister to the Gadwall–Falcated see Krapu 2000). We sampled seven known Duck clade (Johnson and Sorenson 1999). breeding populations: North Dakota (ND); Finally, we included haplotypes from a Eurasian Saskatchewan (SK); Alberta (AB); Utah (UT); Wigeon (Peters et al. 2005b) and an American Washington (WA); Klamath Basin, California Black Duck (A. rubripes; McCracken et al. 2001) (CAkl); and Central Valley, California (CAcv). that were identical to putatively introgressed Three populations were designated as “puta- haplotypes found in Gadwalls (see below), and tive breeding” because some individuals a randomly chosen Mallard (Anas platyrhynchos; were collected in September, aft er the onset of McCracken et al. 2001). Following the topol- migration but before peak migration, including ogy of Johnson and Sorenson (1999), we rooted 448 Peters and Omland [Auk, Vol. 124 trees using the “mallard clade” haplotypes. For richness should refl ect demographic processes MP, we conducted a heuristic search with tree- bett er than other standard measures (Petit et bisection-reconnection (TBR) branch-swapping al. 1998). Both clade A and clade B haplotypes and maximum trees set to 100,000. For ML, we were included in calculations. In addition, we used an HKY + I + Γ mutation model that was calculated r(12) and nucleotide diversity (π, selected using Akaike’s Information Criteria in the average pairwise diff erence among haplo- MODELTEST, version 3.7 (Posada and Crandall types, calculated in ARLEQUIN, version 2.0; 1998). The initial ML search was conducted using Schneider et al. 2000) among all haplotypes this model and a randomly chosen MP tree as a within each clade. starting point. We then re-estimated parameters We tested models of population struc- using the ML tree and repeated the search using ture using analyses of molecular variance those parameters. We repeated this process (AMOVA; Excoffi er et al. 1992) executed in until tree scores and parameter estimates sta- ARLEQUIN (Schneider et al. 2000). First, we bilized (Wilgenbusch and de Queiroz 2000). tested for structure between clade B Gadwall The fi nal parameters included a transition- and Falcated Duck haplotypes. Second, we to-transversion (ti:tv) ratio of 10.2, 74% invariant restricted analyses to clade A haplotypes and sites (I), and a gamma distribution (Γ) of 0.8. We used genetic distances (Φ-statistics) to test for evaluated MP bootstrap support by conducting population structure at three hierarchical lev- 1,000 pseudoreplicates and using a heuristic els: among fl yways, among populations within search with TBR branch-swapping and maxi- fl yways, and within populations. (Clade B mum trees set to 1,000. Finally, we constructed was excluded from analyses because the deep haplotype networks showing alternative evo- divergence between clades contributed dispro- lutionary pathways using the median-joining portionately to within-population variation algorithm in FLUXUS, version 4.1 (Bandelt et al. when using Φ-statistics; data not shown.) For 1999; see Acknowledgments). analyses of breeding populations, we included

We tested alternative topologies for clade B seven populations from the Central (ND, SK, that were generated under diff erent hypotheses and AB) and Pacifi c fl yways (UT, WA, CAcv, regarding the cause of lineage sharing. If this and CAkl). We then tested for structure among shared lineage is explained by mtDNA intro- all 14 populations and three fl yways: Eastern gression from Falcated Ducks into Gadwalls, Flyway (QC, MP, DM), Central–Mississippi we predicted that Gadwall clade B haplotypes fl yways (ND, MB, SK, AB, AR), and Pacifi c would be derived from (i.e., nested within) Flyway (UT, CAcv, CAkl, WA, AKsc, AKko). Falcated Duck haplotypes (the observed ML Third, because clade B could potentially be tree; see below). We compared that topology to informative regarding population structure, an ML tree constrained to having Falcated Duck we repeated the above analyses including haplotypes a monophyletic group nested within haplotypes from both clades, but using con- Gadwall clades, which might be expected if ventional F-statistics so that only haplotype Falcated Ducks recently diverged from a poly- frequencies were considered. Finally, we morphic Gadwall-like ancestor via peripheral performed an ad-hoc AMOVA to test for diff er- isolation (e.g., Harrison 1991). We compared ences between Alaska (AKsc and AKko) and trees using a Shimodaira-Hasegawa (SH) test the remaining 12 populations at three hierar- with resampling-estimated log likelihood and chical levels (between Alaska and “others,” 1,000 bootstrap replicates (Shimodaira and among populations within these two groups, Hasegawa 1999). and within populations). To determine which Genetic diversity and population structure.—For populations were genetically diff erentiated, each intensively sampled population, we calcu- we calculated pairwise ΦST (clade A only) and lated allelic richness, standardized to a sample FST (clade A and clade B) among all popula- size of 12 [r(12)], using RAREFAC (Petit et tions (including Eurasia) in ARLEQUIN. al. 1998). The parameter r(12) is the average Population demography.—We tested for sig- number of diff erent haplotypes expected if 12 natures of historical population expansions individuals were sampled (the smallest sample for clades A and B using the coalescent meth- size among the 14 populations). We chose this ods implemented in LAMARC, version 2.0.2 measure of genetic diversity because allelic (Kuhner et al. 1998, 2005). On the basis of results April 2007] Population Structure in Gadwalls 449 from MODELTEST (see above), we defi ned a The ML topology suggested that Gadwall mutation model that included a ti:tv ratio of 10.2 clade B haplotypes were a monophyletic group and two rates of substitution among sites (74% nested within Falcated Duck haplotypes (–lnL = of sites were invariant, and 26% were allowed 1732.1; Fig. 2A), and MP also supported this to vary). We used a Bayesian approach to simul- topology (length = 148 steps). However, alter- taneously estimate theta (θ = 2Nef µ; where Nef native MP topologies suggested that clade is the eff ective number of females and µ is the B haplotypes were intermixed in a polyphy- –gt neutral mutation rate) and growth (θt = θ0e ; letic relationship between species (Fig. 2B). where θ0 is the current eff ective population size Furthermore, a monophyletic Falcated Duck scaled to µ, θt is the eff ective population size at nested within Gadwall was not a signifi cantly time t, and g is the scaled exponential growth worse fi t (–lnL = 1735.2; SH-test, P = 0.1), and this rate). On the basis of preliminary runs, we tree was only one step longer than MP trees. defi ned uninformative priors for θ (0.00001 – 1) Genetic diversity.—Mean allelic richness that included the entire posterior distribution [r(12)] among the 14 populations was 4.14 for clade A. Because the posterior distribution (95% CI: 3.64 to 4.65; Appendix). Allelic rich- of θ for clade B had a distinct peak but a fl at tail ness was highest within the Utah and North that continued indefi nitely (data not shown), Dakota breeding populations and the Arkansas we used the same priors as used for clade A. and Delmarva nonbreeding populations. For the scaled exponential growth rate (g), we Washington, Quebec, and the two Alaskan used the maximum range of priors (–5,000 to populations had low allelic richness (below 15,000). We also randomly subsampled 19 clade the 95% confi dence interval). It is noteworthy A haplotypes (the number of clade B haplotypes that breeding and putative breeding popula- that we found; n = 20 replicates) to evaluate tions near the center of North America tended how sensitive the results were to diff erences to have higher allelic richness than those near in sample sizes. For fi nal analyses, we ran 5 the periphery (Appendix). Overall, clade A had

million steps in a single chain that followed a higher allelic richness than clade B [r(12) = 4.58 burn-in period of 100,000 steps; analyses were and 2.26, respectively], but the two clades had repeated with diff erent random-number seeds comparable nucleotide diversity (π = 0.0013 and to check that multiple runs gave similar param- 0.0010, respectively). eter estimates. Population structure.—Gadwall clade B haplotypes diff ered from Falcated Duck hap- Reslts lotypes by 1 to 8 base pairs (Fig. 3), and these haplotypes were strongly structured between

Phylogenetic analyses.—Among the 348 North species (AMOVA, ΦST = 0.67, df = 1, P < 0.00001). American Gadwalls sequenced for mtDNA, we We found clade B in 10 of the 14 intensively found 31 diff erent haplotypes that generally sampled populations, including geographically clustered into two main clades. Clade A con- disparate locations such as Alaska and the east tained 27 diff erent haplotypes and 328 (94.3%) and west coasts (Fig. 4). Within populations, the of the North American individuals sequenced, frequency of clade B ranged from 0.0% (e.g., as well as 11 of the 12 individuals sequenced Washington, Maritime Provinces) to 13.3% from Eurasia (Fig. 2A). Clade B contained (Quebec), and the mean frequency was 5.5% three diff erent haplotypes sampled from 19 (95% CI: 3.2% to 7.9%). Among the known individuals (5.5% of North American Gadwalls) breeding populations, we found clade B in both that clustered with, but were not shared with, the Central and Pacifi c fl yways (0.0–5.6% and Falcated Ducks (Figs. 2 and 3). Cytochrome-b 0.0–9.1%, respectively), and 3 of the 12 known and ND2 sequences indicated that these diver- breeding individuals (25%) from the Atlantic gent lineages were not an artifact of amplifying Flyway had clade B (Fig. 4; Appendix). a nuclear copy of mtDNA (J. L. Peters and K. E. Clade A variation was not partitioned Omland unpubl. data). In addition, one Gadwall between the Central and Pacifi c fl yway breed- from South Dakota had a haplotype that was ing populations, but 5.5% of the variation was shared with an American Black Duck, and a partitioned among populations within fl yways European Gadwall had a haplotype shared with (Table 1A). Including all populations in analy- Eurasian (Fig. 2A). ses, variation among the three fl yways did not 450 Peters and Omland [Auk, Vol. 124

Fig. 2. (A) Maximum-likelihood (ML) tree (–lnL = 1732.1) that is the same length as the maxi- mum-parsimony (MP) trees (length = 148 steps; >100,000 equally parsimonious trees; most MP trees supported a monophyletic wigeon group; see Peters et al. 2005b.) Numbers above branches indicate MP bootstrap support; the six most common haplotypes are labeled A1–A5 and B1. Note that clade B Gadwalls are nested within Falcated Ducks. (B) Alternative MP topologies for clade B that show polyphyletic relationships between Gadwall and Falcated Duck haplotypes. April 2007] Population Structure in Gadwalls 451

Fig. 3. Parsimony networks showing alternative connections among Gadwall and Falcated Duck haplotypes. The six most common haplotypes are labeled A1–A5 and B1 (see also Fig. 2). Additional labels indicate the number of individuals (if >1) and the populations from which each was sampled (see Appendix for population codes).

Table 1. Analysis of molecular variance among known breeding populations (Central and Pacifi c fl yways). (A) Ф-statistics restricted to clade A. (B) F-statistics include clade A and clade B haplotypes.

Sum of Variance F-statistics/ Percentage of Source of variation df squares components Φ-statistics variation (%) P (A) Clade A only

Among fl yways 1 0.87 –0.001 ΦCT = –0.003 –0.34 0.4 Among populations 5 4.75 0.022 ΦSC = 0.054 5.46 <0.00001 Within populations 180 68.23 0.379 ΦST = 0.051 94.88 <0.00001 (B) Clade A and clade B

Among fl yways 1 0.51 –0.003 FCT = –0.010 –0.98 0.6 Among populations 5 3.78 0.017 FSC = 0.056 5.62 <0.00001 Within populations 189 54.72 0.290 FST = 0.046 95.36 <0.00001 452 Peters and Omland [Auk, Vol. 124

Fig. 4. Distributions of clades A and B were both broadly overlapping and geographically widespread.

account for any of the observed genetic varia- the 14 populations (61% of all North American tion, but variation among populations within Gadwalls had that haplotype), A2 and A3 fl yways accounted for 14% of the variation were the most common haplotypes in WA and (Table 2A). Including clade B haplotypes and AKsc, respectively (Fig. 5). Haplotype A2 was using conventional F-statistics did not qualita- sequenced from 47% of breeding individuals tively change these results (Tables 1B and 2B). from WA but was rare in other populations

Pairwise ΦST values indicated that only (2.4% of non-Washington individuals; A2 was Washington (WA) and south-coastal Alaska never found east of Colorado and Alberta; Fig. (AKsc) were well diff erentiated from other 5). Restricting analyses to clade A, WA dif- populations (Table 3). Whereas haplotype A1 fered signifi cantly from four of the six breeding was the most common haplotype within 12 of populations and three of the seven remaining

Table 2. Analysis of molecular variance among all populations (known breeding, putative breeding, and nonbreeding; Atlantic, Central, and Pacifi c fl yways). (A) Ф-statistics restricted to clade A. (B) F-statistics include clade A and clade B haplotypes.

Sum of Variance F-statistics/ Percentage of Source of variation df squares components Φ-statistics variation (%) P (A) Clade A only

Among fl yways 2 2.50 –0.005 ΦCT = –0.010 –1.03 0.6 Among populations 11 17.88 0.061 ΦSC = 0.137 13.79 <0.00001 Within populations 273 105.76 0.387 ΦST = 0.128 87.24 <0.00001 (B) Clade A and clade B

Among fl yways 2 1.80 0.0008 FCT = 0.003 0.26 0.2 Among populations 11 8.84 0.024 FSC = 0.076 7.60 <0.00001 Within populations 290 84.38 0.291 FST = 0.079 92.14 <0.00001 April 2007] Population Structure in Gadwalls 453

ly in

g n o are

tr rasia ns 0.120 0.200 0.164 0.212 0.166 0.133 0.207 0.129 0.180 . S o all aris w p ad m o G c c f

o ns AKko Eu ise 0.080 – –0.001 0.013 0.017 0.035 0.246 0.052 0.021 0.156 0.016 0.075 0.021 0.042 0.037 0.013 0.069 –0.006 o ns acifi sc 0.158 0.198 w o – lati air 0.313 0.270 0.206 0.235 0.280 0.326 0.250 0.230 0.319 0.288 0.229 lati 105 p g popu

r g popu o P n f

o n o reedin 0.369 am b entral n o B) rrecti n 0.168 0.126 0.178

o c and

and ni A o g 0.477

nferr C clades ; reedin Bo b a nal

tlantic 0.145 0.161 go er

ft

tative

a dia

ve 0.111 bo a erences (

ff MP QC DM MB AR AR AK DM MB QC MP ST 0.030 0.025 0.027 0.127 0.047 0.187 – 0.148 0.237 0.186 0.269 0.201 0.299 0.337 0.387 0.368 l F di

0.219 and CAk cant 0.232 0.468 )

sm A nifi c nly

Pu g 0.118 0.080 0.249 si ns –0.021 –0.007 –0.021 0.006 0.000 0.038 –0.001 0.022 0.016 0.047 0.016 0.020 0.005 –0.026 –0.006 –0.044 –0.022 –0.029 –0.012 –0.027 0.035 0.027 0.035 –0.002 –0.019 0.035 –0.024 –0.030 – – –0.007 –0.035 0.009 0.020 0.070 0.031 0.010 0.011 –0.012 0.017 0.069 0.083 0.045 –0.007 0.019 0.004 0.000 – –0.017 – –0.016 0.000 0.010 –0.023 –0.015 –0.008 0.023 –0.004 –0.021 –0.012 –0.003 –0.006 – 0.035 0.014 – –0.007 0.040 –0.052 –0.011 0.026 –0.036 –0.009 – –0.009 –0.014 0.497 0.031 0.001 A o acifi o 0.234 – lati 0.249 0.166 0.105 0.270 0.212 0.290 0.265 0.253 0.197 0.282 0.201 0.229 0.248 clade ;

0.509 < 0.05); P nal ( g popu 0.156 go ld dia bo reedin 0.372 in < 0.00048).

ow b P n el ( are 0.202

P (b ow 0.158 0.141 0.400 ST n airs Φ

p 0.207 entral ise nderlined 0.238 0.439 w u air 0.173 0.202 0.354 and

erentiated 0.011 –0.008 0.009 –0.008 0.011 0.001

ld

–0.010 –0.011 –0.004 –0.007 ff l le 3. P sc sm b di bo rasia SK –0.013 – –0.013 0.022 – – –0.013 SK –0.013 0.006 AB –0.001 UT WA –0.007 –0.013 –0.020 CA 0.005 –0.025 CAk –0.006 MP – 0.010 QC –0.021 0.005 0.002 0.025 0.008 –0.022 DM –0.012 MB –0.011 –0.015 –0.010 –0.006 AR –0.022 AK –0.021 –0.008 –0.007 AKko 0.023 0.072 0.052 –0.006 Eu 0.037 Ta –0.006 0.000 –0.006 ND – ND SK AB UT WA CA WA UT AB K SK ND C 454 Peters and Omland [Auk, Vol. 124

Fig. 5. Distributions of the five most common clade A haplotypes (i.e., >10 haplotypes sampled; see Fig. 3). Note the widespread distributions of haplotypes A1, A4, and A5, and the more restricted distributions of A2 and A3 (especially in Washington and Alaska, respectively). populations aft er a Bonferroni correction (Table clade B haplotypes in analyses and calculating

3). In AKsc, A3 was found in 57% of the samples FST did not dramatically change the observed (Fig. 5). This haplotype was also relatively com- patt erns, except that there were fewer signifi cant mon in AKko (23%) but was found in only one diff erences aft er applying a Bonferroni correc- individual outside of Alaska (i.e., ND; 0.3% of tion (Table 3). all non-Alaskan individuals). Population AKsc Population demography.—Demographic analy- diff ered signifi cantly from 10 of the 13 popula- ses indicated population growth for clade A tions aft er a Bonferroni correction (ΦST; Table (Fig. 6). The posterior distribution of g peaked at 3). Although AKko did not diff er signifi cantly 3,500, and the 95% CI did not overlap zero (95% from other populations (except WA), 29.6% of CI: 1,200 to 12,000). Therefore, we rejected the clade A variation was partitioned between the null hypothesis of a long-term stable popula- two Alaskan populations and the remaining tion size. By contrast, the posterior distribution populations (P = 0.01; Φ-statistics), whereas of g for clade B peaked at –210, which suggests only 3.8% of the variation was partitioned a population decline, and the 95% CI over- among populations within these two groups (P lapped zero growth (95% CI: –2,200 to 13,000). = 0.0001). Finally, Eurasian Gadwalls tended to The broad confi dence interval for clade B was a be diff erentiated from North American popula- result of a fl at tail in the posterior distribution tions, and diff erences were signifi cant for 5 of (Fig. 6). Regardless, the signatures of growth the 14 pairwise comparisons (Table 3). Including for clades A and B were markedly di ff erent, April 2007] Population Structure in Gadwalls 455

B was between 3.2% and 7.9% (95% CI of the mean), we estimated that between 77,000 and 190,000 North American Gadwalls had clade B haplotypes. Thus, clade B haplotypes are strongly incorporated into the Gadwall gene pool, and recent, ongoing hybridization with the Falcated Duck is not suffi cient to explain this shared lineage. Despite similar distributions in North America, clades A and B most likely have dif- ferent demographic histories. We found strong evidence that clade A underwent a historical demographic expansion, whereas clade B seems Fig. 6. Posterior distributions of population to have experienced a population decline or a growth rates (g) estimated for Gadwall. Clade A more stable population size. These contrasting has experienced historical population growth, histories suggest that the broadly overlapping whereas clade B seems to have experienced a geographic distributions resulted from recent more stable population size. admixture of lineages descended from diff er- ent source populations (see Quinn 1992, Fry which suggests that these clades most likely and Zink 1998, Lambert et al. 2002, Ruokonen experienced diff erent demographic histories. et al. 2004). This diff erence cannot be explained by unequal Overall, the observed patt ern of lineage sample sizes; subsampling 19 haplotypes from sharing seems most consistent with ancient clade A, g averaged 10,000 (n = 20 replicates; introgression of Falcated Duck mtDNA into range: 4,600–13,000), and the 95% CI did not the Gadwall gene pool (perhaps from a single

overlap the value of g estimated for clade B in hybridization event). This interpretation is espe- any of those replicates. The distinctly star-like cially supported by the topology of the ML tree, parsimony network for clade A also suggests a which shows Gadwall clade B haplotypes to be rapid population expansion compared with the a monophyletic group nested within Falcated clade B network (Fig. 3). Duck haplotypes. The diff erent demographic histories of the clades are also consistent with Discussion ancient introgression. One possible scenario is that clade B introgression occurred while Mitochondrial DNA .—We found two Gadwall populations were expanding, and this distinct mtDNA lineages in Holarctic Gadwalls expansion could have facilitated the dispersion (clades A and B). Clade A was common in both of clade B across North America. An alternative North America and Eurasia. By contrast, clade scenario is that Gadwalls recently colonized B was relatively uncommon in North America North America from Eurasia (see below), and and was not present among the 12 Eurasian one or more of the founding females carried Gadwalls sampled. Clade B was shared with the clade B haplotypes. This latt er possibility seems Asian Falcated Duck, but no haplotypes were less likely, because simultaneous colonization shared between species. This absence of shared by both clades should have resulted in similar haplotypes suggests that clade B Gadwalls signatures of population growth (although and Falcated Ducks have been diverging for note the broad overlap in growth rates; Fig. some time (“allotypy”; Omland et al. 2006). 6). Regardless, introgression of Falcated Duck Furthermore, clade B was geographically wide- mtDNA following ancient hybridization seems spread in North America (Fig. 4), and one of the suffi cient to explain clade B haplotypes in North clade B haplotypes (B1) was the fourth most American Gadwalls. common haplotype overall (Fig. 3). Assuming An alternative to introgression is that the a breeding population size of 2.4 million North Gadwall and Falcated Duck recently diverged American Gadwalls in 2005 (Central and Pacifi c from a polymorphic ancestor that contained fl yways; U.S. Fish and Wildlife Service 2005a, both mtDNA lineages. Widespread species with b), and assuming that the frequency of clade structured populations can give rise to new 456 Peters and Omland [Auk, Vol. 124

species through peripatric or parapatric specia- Flyway than in the Pacifi c Flyway (U.S. Fish and tion, and mtDNA from the parental population Wildlife Service 2005a, b), and even a small frac- will likely be paraphyletic to that of the daugh- tion of Central Flyway breeders migrating to the ter population (Harrison 1991, Omland 1997, Pacifi c Flyway could result in extensive inter- Funk and Omland 2003). Under this scenario, mixing. If the gregarious behavior of Gadwalls we would expect Falcated Duck haplotypes results in females migrating with foreign fl ocks to be nested within Gadwall haplotypes, and to new breeding areas, extensive intermixing we could not statistically reject that tree topol- during winter could contribute to high genetic ogy. Furthermore, the ancestral polymorphism similarity between breeding populations from hypothesis does not predict that shared lineages diff erent fl yways. will be found in sympatry (Hare and Avise 1998, The predominant patt ern we found was a Masta et al. 2002, Sefc et al. 2005) and, in that general lack of genetic structure among both regard, it may be a bett er explanation of para- breeding and nonbreeding populations of North phyly in this group. However, given the data American Gadwalls. Although female philopa- and analyses presented, either hybridization try is well established in waterfowl, the extent or incomplete lineage sorting is suffi cient to of short-distance and long-distance dispersal is explain the shared lineage between Gadwalls not known. Competition for breeding resources and Falcated Ducks. Increased sampling of and fl uctuations in local breeding conditions Falcated Ducks and additional loci are needed likely drive regular short-distance and even to statistically test these alternative hypotheses. occasional long-distance dispersal. For example, We also found Gadwall haplotypes that were some waterfowl “over-fl y” previous breeding shared with American Black Ducks (McCracken sites to more northern locations when local et al. 2001) and Eurasian Wigeons (Peters et al. conditions decline from one year to the next 2005b). The Gadwall is known to hybridize with (Johnson and Grier 1988). Diff erences between and produce fertile off spring with these species breeding philopatry and natal philopatry could

or close allies of these species (Johnsgard 1960). also explain the general lack of population These haplotypes were sampled from wings, and structure. Whereas 33% to 49% of adult female close examination of those wings revealed no Gadwalls returned to previous breeding areas in qualitative diff erences from other Gadwall wings subsequent years, only 7% to 9% of hatch-year of the same sex and age class. There was no obvi- females returned to natal areas (reviewed in ous indication of a hybrid ancestry. Therefore, it Anderson et al. 1992). Therefore, females seem is unlikely that these individuals were F1 hybrids to disperse regularly, but more information is or even F2 backcrosses (a close examination of needed on dispersal distances for Gadwalls and hybrid wing patt erns or an analysis of nuclear other species of waterfowl. However, even regu- loci is needed to fully discount these possi- lar short-distance dispersal can accumulate into bilities). These haplotypes demonstrate mtDNA substantial gene fl ow between distant popula- introgression between highly divergent species tions over evolutionary time. (see also Cronin et al. 1996); it is possible that Although it is possible that female dispersal drift or selection could drive such introgressed homogenizes widely dispersed populations, we mtDNA to high frequencies or even fi xation. found that the Washington breeding popula- Population structure.—Migratory fl yways did tion and the south-coastal Alaskan population not contribute to the partitioning of genetic (presumably breeding) were well diff erentiated variation within North American Gadwalls, in mtDNA. The Alaskan population is widely and this result is similar to patt erns found in disjunct from the Gadwall’s core breeding other species of waterfowl (Cronin et al. 1996, range (Fig. 1), and this disjunction may inhibit Pearce et al. 2004; but see Gay et al. 2004, Peters female-mediated gene fl ow (McCracken et al. et al. 2005a). One possible explanation of this 2001, Peters et al. 2005a). In addition, Gadwalls lack of structure is that winter intermixing are present in Alaska throughout the year, facilitates female dispersal, even though only which suggests that this population may be a small percentage of Gadwalls move between sedentary or only partially migratory. Indeed, fl yways during seasonal migrations (Bellrose the putative breeding (AKsc) and nonbreed- 1976). Breeding Gadwalls are about an order ing (AKko) populations from Alaska were not of magnitude more abundant in the Central strongly di ff erentiated, which suggests the April 2007] Population Structure in Gadwalls 457 genetic autonomy of Alaskan Gadwalls, at least in North American Gadwalls (see also Pearce in regard to female dispersal. et al. 2004, 2005). Selection favoring diff erent In contrast to the Alaskan population, the haplotypes in diff erent habitats could also have Washington breeding population seems con- contributed to these patt erns. tinuous with the core breeding range, and Finally, North American and Eurasian potential geographic barriers to gene fl ow are Gadwalls tended to be well diff erentiated in not obvious. Furthermore, most Gadwalls that mtDNA, which suggests that female-mediated breed in Washington migrate to California’s gene fl ow is restricted. The Holarctic Mallard is Central Valley, where they presumably inter- also strongly structured between hemispheres, mix with other populations (Bellrose 1976). The but diff erences were more pronounced in that strong diff erentiation could have resulted from species (Kulikova et al. 2005). The more shallow a shift in haplotype frequencies associated with genealogy in Gadwalls suggests a more recent a breeding-range expansion into Washington divergence between hemispheres (see also (Hewitt 1996, Ibrahim et al. 1996). Consistent Pearce et al. 2005), which might be explained with that hypothesis, Washington Gadwalls had by a recent colonization of North America fewer haplotypes than any other breeding popu- (see Johnson and Sorenson 1999). Consistent lation sampled, but a number comparable to that with that hypothesis, the ML topology placed of the putative breeding population in Quebec, a Eurasian haplotypes near the root of clade A, recently established population (LeSchack et al. which suggests that North American haplo- 1997). Although breeding has only recently been types are derived. Furthermore, the signature documented in western Washington (Canning of population growth that we found in clade A and Herman 1983), it is not clear whether is consistent with colonization of North America Gadwalls also recently began breeding in the by a small number of individuals. However, Columbia Basin (eastern Washington), where many species from the Northern Hemisphere our samples were collected. show evidence of recent expansion from glacial

Using coalescent methods, we found strong refugia (Hewitt 2000, Lessa et al. 2003). Larger support for a historical population expansion sample sizes for Eurasia, multiple independent of clade A in North America. This historical loci, and statistical analyses that account for the expansion, coupled with recent population and randomness of genetic processes are needed to range expansions, likely had a major eff ect on the test hypotheses regarding colonization history partitioning of genetic variation among North (Rosenberg and Nordborg 2002, Knowles 2004). American Gadwalls. The likelihood of any given Conclusions.—More rigorous comparisons haplotype becoming established in a recently with Eurasian Gadwalls and Falcated Ducks, founded population will be a function of its rela- including a comparison of nuclear loci, are tive abundance in the parental population, the needed to bett er understand the patt erns of number of founders, and subsequent gene fl ow genetic variation in North American Gadwalls. (see Ibrahim et al. 1996). In the present study, we Regardless, the weak mtDNA structure in North found that 61% of all sampled Gadwalls shared American Gadwalls, especially between migra- a common haplotype (A1); therefore, this haplo- tory fl yways, probably resulted from a combi- type has a high probability of being carried by nation of female-mediated gene fl ow and recent even a small number of founders. For example, and historical expansions. However, the extent 67% of known breeding samples from recently to which each of these contributed to weak established populations in the Atlantic Flyway population structure is not clear. Additional had that haplotype. If Gadwalls recently colo- data and analyses are also needed to determine nized much of their current distribution from a whether clade B is shared with Falcated Ducks single source, fi nding A1 as the most common as a result of mtDNA introgression or incom- haplotype in most populations would not be sur- plete lineage sorting. prising, even if gene fl ow is rare. Furthermore, random sampling of haplotypes during coloni- Acnoledgments zation could explain the strong diff erentiation of Washington and Alaska. In that regard, a recent We thank the following people and organi- expansion may be a bett er explanation of the zations for their participation with this study. complex population structure (or lack thereof) Such broadscale sampling would not have 458 Peters and Omland [Auk, Vol. 124 been possible without the help of P. Padding, Bandelt, H.-J., P. Forster, and A. Rl. 1999. W. Martin, N. North, D. Mauser, D. Loughman, Median-joining networks for inferring intra- T. Hames, B. Olson, B. Emery, J. Jehl, Jr., L. specifi c phylogenies. Molecular Biology and Loos, K. Chodacheck, the Burke Museum of Evolution 16:37–48. Natural History, University of Washington, Bellrose, F. C. 1976. Ducks, Geese and Swans of and the Delaware Museum of Natural History. North America. Stackpole Books, Harrisburg, Collection, import, and export permits were Pennsylvania. issued by the U.S. Fish and Wildlife Service, Canadian Wildlife Service. 2005. Population Canadian Wildlife Service, North Dakota Game status of migratory game birds in Canada: and Fish Department, Utah Division of Wildlife November 2005. Canadian Wildlife Service Resources, California Department of Fish and Waterfowl Committ ee, CWS Migratory Game, and Washington Department of Fish Birds Regulatory Report Number 16, and Wildlife. E. M. Humphries, A. Logie, and Ott awa, Ontario. K. Doshi provided lab assistance. J. Bell and B. Canning, D. J., and S. G. Herman. l983. Gadwall Alemi in the Department of Mathematics and breeding range expansion into western Statistics at University of Maryland Baltimore Washington. Murrelet 64:27–3l. County, provided computer resources for coales- Cronin, M. A., J. B. Grand, D. Esler, D. V. cent analyses. S. Freeland and N. Keulman also Dersen, and K. T. Scribner. 1996. Breeding provided computer assistance. Two anonymous populations of Northern Pintails have simi- reviewers, B. Kondo and other members of the lar mitochondrial DNA. Canadian Journal Omland lab, provided insightful comments on of Zoology 74:992–999. this manuscript. This research was supported Donne-Gouss, C., V. Laudet, and C. Hnni. by grants from Delta Waterfowl Foundation 2002. A molecular phylogeny of anseri- and Maryland Ornithological Society to J.L.P., formes based on mitochondrial DNA analy- and K.E.O. is supported by National Science sis. Molecular Phylogenetics and Evolution

Foundation grant DEB-0347083. For information 23:339–356. on FLUXUS, see www.fl uxus-engineering.com. Drovetsi, S. V., R. M. Zin, S. Roer, I. V. Fadeev, E. V. Nesterov, I. Karagodin, E. A. Literature Cited Kobli, and Y. A. Red’in. 2004. Complex biogeographic history of a Holarctic pas- Anderson, M. G., J. M. Rymer, and F. C. serine. Proceedings of the Royal Society of Roer. 1992. Philopatry, dispersal, and the London, Series B 271:545–551. genetic structure of waterfowl populations. Esler, D. 2000. Applying metapopulation Pages 365–395 in Ecology and Management theory to conservation of migratory birds. of Breeding Waterfowl (B. D. J. Batt , A. D. Conservation Biology 14:366–372. Aft on, M. G. Anderson, C. D. Ankney, D. H. Ecoffier, L., P. E. Smouse, and J. M. Quattro. Johnson, J. A. Kadlec, and G. L. Krapu, Eds.). 1992. Analysis of molecular variance inferred University of Minnesota Press, Minneapolis. from metric distances among DNA haplo- Avise, J. C. 2000. Phylogeography: The History types: Application to human mitochondrial and Formation of Species. Harvard DNA restriction data. Genetics 131:479–491. University Press, Cambridge, Massachusett s. Fry, A. J., and R. M. Zin. 1998. Geographic Avise, J. C., R. T. Alisausas, W. S. Nelson, and analysis of nucleotide diversity and Song C. D. Anney. 1992. Matriarchal popula- Sparrow (Aves: Emberizidae) population tion genetic structure in an avian species history. Molecular Ecology 7:1303–1313. with female natal philopatry. Evolution 46: Fun, D. J., and K. E. Omland. 2003. Species- 1084–1096. level paraphyly and polyphyly: Frequency, Avise, J. C., C. D. Anney, and W. S. Nelson. causes, and consequences, with insights 1990. Mitochondrial gene trees and the evo- from mitochondrial DNA. Annual lutionary relationship between Mallard and Review of Ecology, Evolution and Black Ducks. Evolution 44:1109–1119. Systematics 34:397–423. Baldassarre, G. A., and E. G. Bolen. 1994. Gay, L., P. Defos Du Rau, J.-Y. Mondain-Monval, Waterfowl Ecology and Management. John and P.-A. Crocet. 2004. Phylogeography Wiley and Sons, New York. of a game species: The Red-crested Pochard April 2007] Population Structure in Gadwalls 459

(NeĴ a rufi na) and consequences for its man- Washington, Seatt le. [Online.] Available at agement. Molecular Ecology 13:1035–1045. evolution.gs.washington.edu/lamarc/index. Hare, M. P., and J. C. Avise. 1998. Population html. structure in the American oyster as inferred Kuner, M. K., J. Yamato, and J. Felsenstein. by nuclear gene genealogies. Molecular 1998. Maximum likelihood estimation Biology and Evolution 15:119–128. of population growth rates based on the Harrison, R. G. 1991. Molecular changes at coalescent. Genetics 149:429–434. speciation. Annual Review of Ecology and Kuliova, I. V., S. V. Drovetsi, D. D. Gibson, Systematics 22:281–308. R. J. Harrigan, S. Roer, M. D. Sorenson, Henny, C. J., and N. E. Holgersen. 1974. K. Winer, Y. N. Zuravlev, and K. G. Range expansion and population increase McCracen. 2005. Phylogeography of the of the Gadwall in eastern North America. Mallard (Anas platyrhynchos): Hybridization, Wildfowl 25:95–101. dispersal, and lineage sorting contribute to Heitt, G. M. 1996. Some genetic consequences complex geographic structure. Auk 122: of ice ages, and their role in divergence and 949–965. speciation. Biological Journal of the Linnean Kuliova, I. V., Y. N. Zuravlev, and K. G. Society 58:247–276. McCracen. 2004. Asymmetric hybridiza- Heitt, G. [M.] 2000. The genetic legacy of the tion and sex-biased gene fl ow between Quaternary ice ages. Nature 405:907–913. Eastern Spot-billed Ducks (Anas zonorhyn- Ibraim, K. M., R. A. Nicols, and G. M. cha) and Mallards (A. platyrhynchos) in the Heitt. 1996. Spatial patt erns of genetic Russian Far East. Auk 121:930–949. variation generated by diff erent forms of Lambert, D. M., P. A. Ritcie, C. D. Millar, B. dispersal during range expansion. Heredity Holland, A. J. Drummond, and C. Baroni. 77:282–291. 2002. Rates of evolution in ancient DNA from Jonsgard, P. A. 1960. Hybridization in the Adélie Penguins. Science 295:2270–2273.

Anatidae and its taxonomic implications. LeScac, C. R., S. K. McKnigt, and G. R. He. Condor 62:25–33. 1997. Gadwall (Anas strepera). In The Birds Jonson, D. H., and J. W. Grier. 1988. of North America, no. 283 (A. Poole, and Determinants of breeding distributions of F. Gill, Eds.). Academy of Natural Sciences, ducks. Wildlife Monographs 100:1–37. Philadelphia, and American Ornithologists’ Jonson, K. P., and M. D. Sorenson. 1999. Union, Washington, D.C. Phylogeny and biogeography of dabbling Lessa, E. P., J. A. Coo, and J. L. Patton. 2003. ducks (Genus: Anas): A comparison of Genetic footprints of demographic expan- molecular and morphological evidence. sion in North America, but not Amazonia, Auk 116:792–805. during the Late Quaternary. Proceedings of Kimura, M., S. M. Clegg, I. J. Lovette, K. R. the National Academy of Sciences USA 100: Holder, D. J. Girman, B. Mil, P. Wade, 10331–10334. and T. B. Smit. 2002. Phylogeographical Mallet, J. 2005. Hybridization as an invasion approaches to assessing demographic con- of the genome. Trends in Ecology and nectivity between breeding and overwin- Evolution 20:229–237. tering regions in a Nearctic–Neotropical Masta, S. E., B. K. Sullivan, T. Lamb, and warbler (Wilsonia pusilla). Molecular Ecology E. J. Routman. 2002. Molecular systematics, 11:1605–1616. hybridization, and phylogeography of the Knoles, L. L. 2004. The burgeoning fi eld Bufo americanus complex in Eastern North of statistical phylogeography. Journal of America. Molecular Phylogenetics and Evolutionary Biology 17:1–10. Evolution 24:302–314. Krau, G. L. 2000. Temporal fl exibility of repro- McCracen, K. G., W. P. Jonson, and F. H. duction in temperate-breeding dabbling Seldon. 2001. Molecular population genet- ducks. Auk 117:640–650. ics, phylogeography, and conservation biol- Kuner, M. K., J. Yamato, P. Beerli, L. P. ogy of the Mott led Duck (Anas fulvigula). Smit, E. Rynes, E. Walu, C. Li, J. Conservation Genetics 2:87–102. Sloan, P. Colacurcio, and J. Felsenstein. Milot, E., H. L. Gibbs, and K. A. Hobson. 2000. 2005. LAMARC, version 2.0. University of Phylogeography and genetic structure of 460 Peters and Omland [Auk, Vol. 124

northern populations of the Yellow Warbler Snow Goose Chen caerulescens caerulescens (Dendroica petechia). Molecular Ecology 9: maternal lineages. Molecular Ecology 1: 667–681. 105–117. Omland, K. E. 1997. Examining two standard Rhodes, O. E., Jr., L. M. Smith, and R. K. assumptions of ancestral reconstructions: Chesser. 1993. Temporal components of Repeated loss of dichromatism in dabbling genetic variance in wintering and migrat- ducks (Anatini). Evolution 51:1636–1646. ing American Wigeon. Canadian Journal of Omland, K. E., J. M. Baker, and J. L. Peters. Zoology 71:2229–2235. 2006. Genetic signatures of intermediate Rhymer, J. M., M. J. Williams, and M. J. Braun. divergence: Population history of Old and 1994. Mitochondrial analysis of gene fl ow New World Holarctic Ravens (Corvus corax). between New Zealand Mallards (Anas Molecular Ecology 15:795–808. platyrhynchos) and Grey Ducks (A. supercili- Omland, K. E., C. L. Tarr, W. I. Boarman, osa). Auk 111:970–978. J. M. Marzluff, and R. C. Fleischer. 2000. Roher, F. C., and M. G. Anderson. 1988. Cryptic genetic variation and paraphyly in Female-biased philopatry, monogamy, ravens. Proceedings of the Royal Society of and the timing of pair formation in migra- London, Series B 267:2475–2482. tory waterfowl. Pages 187–221 in Current Pearce, J. M., S. L. Talbot, M. R. Petersen, and J. R. Ornithology, vol. 5 (R. F. Johnston, Ed.). Rearick. 2005. Limited genetic diff erentia- Plenum Press, New York. tion among breeding, molting, and wintering Rosenberg, N. A., and M. Nordborg. 2002. groups of the threatened Steller’s Eiders: The Genealogical trees, coalescent theory and role of historic and contemporary factors. the analysis of genetic polymorphisms. Conservation Genetics 6:743–757. Nature Reviews Genetics 3:380–390. Pearce, J. M., S. L. Talbot, B. J. Pierson, M. R. Ruokonen, M., L. Kvist, T. Aarvak, J. Petersen, K. T. Scribner, D. L. Dickson, and Markkola, V. V. Morozov, I. J. Øien,

A. Mosbech. 2004. Lack of spatial genetic E. E. Syroechkovsky, Jr., P. Tolvanen, structure among nesting and wintering and J. Lumme. 2004. Population genetic King Eiders. Condor 106:229–240. structure and conservation of the Lesser Peters, J. L., W. Gretes, and K. E. Omland. White-fronted Goose Anser erythropus. 2005a. Late Pleistocene divergence between Conservation Genetics 5:501–512. eastern and western populations of Wood Schneider, S., J.-M. Keuffer, D. Roessli, and L. Ducks (Aix sponsa) inferred by the “isola- Ecoffier. 2000. ARLEQUIN, version 2.000: tion with migration” coalescent method. A soft ware package for population genetics. Molecular Ecology 14:3407–3418. University of Geneva, Switzerland. Peters, J. L., K. G. McCracken, Y. N. Zhuravlev, Scribner, K. T., M. R. Petersen, R. L. Fields, Y. Lu, R. E. Wilson, K. P. Johnson, and K. E. S. L. Talbot, J. M. Pearce, and R. K. Chesser. Omland. 2005b. Phylogenetics of wigeons 2001. Sex-biased gene fl ow in Spectacled and allies (Anatidae: Anas): The importance Eiders (Anatidae): Inferences from molecu- of sampling multiple loci and multiple lar markers with contrasting modes of individuals. Molecular Phylogenetics and inheritance. Evolution 55:2105–2115. Evolution 35:209–224. Sefc, K. M., R. B. Payne, and M. D. Sorenson. Petit, R. J., A. el Mousadik, and O. Pons. 1998. 2005. Genetic continuity of brood-parasitic Identifying populations for conservation on indigobird species. Molecular Ecology 14: the basis of genetic markers. Conservation 1407–1419. Biology 12:844–855. Shimodaira, H., and M. Hasegaa. 1999. Pielou, E. C. 1991. Aft er the Ice Age: The Return of Multiple comparisons of log-likelihoods Life to Glaciated North America. University with applications to phylogenetic infer- of Chicago Press, Chicago, Illinois. ence. Molecular Biology and Evolution 16: Posada, D., and K. A. Crandall. 1998. 1114–1116. MODELTEST: Testing the model of DNA Sorenson, M. D., J. C. Ast, D. E. Dimcheff, T. substitution. Bioinformatics 14:817–818. Yuri, and D. P. Mindell. 1999. Primers for Quinn, T. W. 1992. The genetic legacy of Mother a PCR-based approach to mitochondrial Goose—Phylogeographic patt erns of Lesser genome sequencing in birds and other April 2007] Population Structure in Gadwalls 461

vertebrates. Molecular Phylogenetics and U.S. Fish and Wildlife Service. 2005a. Waterfowl Evolution 12:105–114. population status, 2005. U.S. Department of Sorenson, M. D., and R. C. Fleischer. 1996. the Interior, Washington D.C. Multiple independent transpositions of mito- U.S. Fish and Wildlife Service. 2005b. 2005 chondrial DNA control region sequences to Pacifi c Flyway databook. U.S. Department the nucleus. Proceedings of the National of the Interior, Division of Migratory Bird Academy of Sciences USA 93:15239–15243. Management, Portland, Oregon. Swofford, D. L. 1999. PAUP*: Phylogenetic Webster, M. S., P. P. Marra, S. M. Haig, S. Analysis Using Parsimony (*and Other Bensch, and R. T. Holmes. 2002. Links Methods), version 4.0. Sinauer Associates, between worlds: Unraveling migratory con- Sunderland, Massachusett s. nectivity. Trends in Ecology and Evolution Tiedemann, R., K. G. von Kistowski, and 17:76–83. H. Noer. 1999. On sex-specifi c dispersal Wilgenbusch, J., and K. de Queiroz. 2000. and mating tactics in the Common Eider Phylogenetic relationships among the Somateria mollissima as inferred from the phrynosomatid sand lizards inferred from genetic structure of breeding colonies. mitochondrial DNA sequences generated Behaviour 136:1145–1155. by heterogeneous evolutionary processes. Tubaro, P. L., and D. A. L tmaer. 2002. Systematic Biology 49:592–612. Hybridization patt erns and the evolution of reproductive isolation in ducks. Biological Journal of the Linnean Society 77:193–200. Associate Editor: K. P. Johnson

Aendi. Sampling details of North American Gadwall for each state or province.

a b

Population Code Flyway n r(12) Tissue source Known breeding populations North Dakota ND Central 36 5.28 Blood (1), web (27), wing (8) Saskatchewan SK Central 22 4.27 Wing (6), nest (16) Alberta AB Central 31 4.50 Wing Utah UT Pacifi c 22 6.00 UWBM-70833, wing (3), nest (18) Washington WA Pacifi c 19 2.99 Web (14), nest (5) California: Central Valley Cacv Pacifi c 32 3.75 Nest California: Klamath Basin CAkl Pacifi c 34 3.88 Web Putative breeding populations c Quebec QC Atlantic 10(5) 2.97 Wing Manitoba MB Central 7(6) 3.77 Wing Alaska: South-coastal AKsc Pacifi c 4(17) 3.40 Wing Nonbreeding populations Maritime Provinces MP Atlantic 12 4.00 Wing Delmarva DM Atlantic 20 4.99 Wing Arkansas AR Mississippi 14 5.29 Wing Alaska: Kodiak Island AKko Pacifi c 13 2.92 Wing Widespread samples Known and putative breeding c Delaware DE Atlantic 1 DMNH-83682 New York NY Atlantic 1 Wing South Dakota SD Central 6 Wing Alberta ABp Central 0(3) Wing Wyoming WY Central 1 Wing Montana: East MTe Central 3 Wing Montana: West MTw Pacifi c 1 Wing Colorado CO Pacifi c 3(1) Wing 462 Peters and Omland [Auk, Vol. 124

Appendix. Continued.

Population Code Flyway n r(12) a Tissue source b Nevada NV Pacifi c 1 Wing California: Mono Lake CAml Pacifi c 5 Feathers Oregon OR Pacifi c 1 Wing British Columbia BC Pacifi c 1 Wing Nonbreeding North Carolina NC Atlantic 1 Wing Maine ME Atlantic 1 Wing Louisiana LA Central 1 Wing Texas TX Central 1 Wing California: Central Valley CAnb Pacifi c 6 Wing Washington WAnb Pacifi c 1 GKD-173 Adult males: Breeding North Dakota NDam Central 2 Blood California CAam Pacifi c 3 UWBM-68930, web (1), feathers (1) a Allelic richness [r(12)] is the number of alleles sampled per population standardized to a sample size (n) of 12 (includes clade A and clade B haplotypes). b Number in parentheses indicates the number of samples of that tissue source; “wing” refers to samples collected by the U.S. and Canadian wing surveys (see text) currently held at the University of Maryland Baltimore County; web = foot-webbing from trapped birds; nest = salvaged nesting material (i.e., egg membranes and feathers); vouchered museum specimens are indicated as follows: UWBM and GKD = University of Washington Burke Museum; DMNH = Delaware Museum of Natural History. c For “putative breeding populations,” the number of samples assignable to the local breeding population is given outside the parentheses, and the number that did not meet those criteria is given inside the parentheses.