PHYLOGEOGRAPHIC ANALYSIS OF PURPLE SANDPIPERS (CALIDRIS
MARITIMA) AS REVEALED BY MITOCHONDRIAL DNA AND
MICROSATELLITES
by
Nathalie M. LeBlanc
Thesis submitted in partial fulfillment of the requirements for the Degree of Master of Science (Biology)
Acadia University Fall Graduation 2015
© Copyright by Nathalie Marie LeBlanc, 2015 This thesis by Nathalie M. LeBlanc was defended successfully in an oral examination on September 15, 2015.
The examining committee for the thesis was:
______Dr. Michael Robertson, Chair
______Dr. David Toews, External Reader
______Dr. Marlene Snyder, Internal Reader
______Dr. Mark Mallory, Co-supervisor
______Dr. Don Stewart, Co-supervisor
______Dr. Brian Wilson, Head
This thesis is accepted in its present form by the Division of Research and Graduate Studies as satisfying the thesis requirements for the degree Master of Science (Biology).
………………………………………….
ii I, Nathalie M. LeBlanc, grant permission to the University Librarian at Acadia University to reproduce, loan or distribute copies of my thesis in microform, paper or electronic formats on a non-profit basis. I, however, retain the copyright in my thesis.
______Nathalie LeBlanc, Author
______Dr. Don Stewart, Co-supervisor
______Dr. Mark Mallory, Co-supervisor
______Date
iii Table of Contents
Title...... i
Approval of Thesis...... ii
Library release form...... iii
Table of contents...... iv
List of figures...... vi
List of tables...... vi
Abstract...... vii
Acknowledgements...... viii
Chapter 1. Literature Review
Overview...... 1
Application of biogeographic information data to the conservation biology of avian species...... 1
The use of microsatellite DNA in phylogeographic studies...... 6
Current trends in phylogeographic case studies...... 8
Natural history of the Purple Sandpiper...... 12
Chapter 2. Genetic differentiation and phylogeography of Purple Sandpipers using mitochondrial markers.
Introduction...... 18
Methods...... 23
Results...... 33
Discussion...... 41
iv Chapter 3. Genetic differentiation and phylogeography of Purple Sandpipers using nuclear microsatellite markers.
Introduction...... 62
Methods...... 66
Results...... 72
Discussion...... 77
Chapter 4. General Discussion and Conclusions...... 90
Literature Cited...... 91
v List of Figures (abbreviated captions)
Figure 1. Range map of Purple Sandpipers...... 54
Figure 2. Schematic diagram showing approximate annealing locations...... 55
Figure 3. Maps of different breeding population groupings...... 56
Figure 4. Rooted bayesian phylogenetic tree...... 57
Figure 5. Rooted maximum parsimony phylogenetic tree...... 58
Figure 6. Rooted maximum likelihood phylogenetic tree...... 59
Figure 7. Unrooted statistical parsimony haplotype network...... 60
Figure 8. Mismatch distribution...... 61
Figure 9. Mean Ln Likelihood values...... 87
Figure 10. Below: Individual assignment...... 88
List of Tables (abbreviated captions)
Table 1. List of primers...... 50
Table 2. Pairwise estimates of ΦST...... 51
Table 3. Standard diversity estimates...... 52
Table 4. Demographic statistics for breeding populations...... 53
Table 5. Average success...... 83
Table 6. Diversity estimates for 10 microsatellite loci...... 84
Table 7. Pairwise estimates of FST...... 85
Table 8. Pairwise estimates of Jost's D...... 86
Table 9. P-values from Wilcoxon and mode-shift...... 89
vi Abstract
The Purple Sandpiper (Calidris maritima) is a medium-sized shorebird that breeds in the
Arctic and winters along northern Atlantic coastlines. Due to difficulty in recapturing banded birds, migration routes and affiliations between breeding grounds and wintering grounds are poorly understood. Populations appear to be declining, and future conservation efforts for this species will benefit from a thorough understanding of their migration patterns. This study used two mitochondrial DNA markers and 10 microsatellite loci to analyze current population structure and demographic trends.
Samples were taken from breeding locations in Nunavut, Iceland and Svalbard, representing all three putative subspecies, as well as wintering locations along the coast of
Maine, Nova Scotia, New Brunswick, Newfoundland, and the United Kingdom.
Mitochondrial haplotypes displayed low diversity, a shallow phylogeny and haplotype network. F-statistics show significant differentiation between Iceland, Svalbard, and
Canada, as well as differentiation between Iceland/Svalbard and North American wintering populations.
vii Acknowledgements
This project would not have been possible without the generous contributions of Greg
Robertson, Julie Paquet, Lindsay Tudor, Glen Mittelhauser, Mark Elderkin, Snaebjörn
Pálsson, Ron Summers, Roger Bull, Michel Gosselin, and Pamela Mills, who provided the samples used in this study. Without their support we would not have been able to make use of such a large number of samples from such a wide variety of locations. I would like to thank my funding sources, NSERC and the STAGE program at
Environment Canada.
I would also like to thank the people who supported my research here at Acadia. My supervisors, Don Stewart and Mark Mallory, for funding and advice these past two years.
My labmate Brent Robicheau for feedback and moral support, and Dr. Steve Mockford and Jose Lefebvre for lending their experience with microsatellite markers. I'd like to thank Lisa Taul for always being available for advice about administration and shipping, and my parents for supporting me through this venture.
viii CHAPTER 1
Literature Review
Overview
This thesis is an analysis of the phylogeography and population genetic structure of breeding and wintering (migratory) populations of Purple Sandpipers, Calidris maritima
Brunnich 1764. In this section of the thesis, I review how historical biogeographic information has been applied to questions of conservation biology in avian taxa and how various molecular markers such as mitochondrial DNA and microsatellite alleles have been used to study population genetic structure in avian taxa. More specifically, I also review the current state of knowledge about the taxonomy and hypothesized migratory routes of Purple Sandpipers and how molecular markers can be used to help refine our knowledge about this species.
Application of biogeographic information data to the conservation biology of avian species
The evolution of distinct species and subspecies in avian taxa has been greatly affected by geological events such as the Pleistocene glaciations (Pielou 1991). During ice advances, populations retreat to refugial pockets of habitable climate (Avise et al. 1998). Small population sizes and isolation from other refugia facilitate genetic divergence in these
1 disjunct populations (Avise et al. 1998). Intra-specific divergence of some Arctic avian populations has been attributed to isolation in multiple refugia during the Pleistocene ice ages (e.g., Wenink et al. 1996; Rönkä et al. 2012), while other species with little or no divergence (often accompanied by low genetic diversity), are thought to have resided in a single refugium during the Pleistocene (e.g., Buehler and Baker 2005; Miller et al. 2013).
Phylogeography is the study of these kinds of historical processes that have contributed to the pattern of genetic variation within and among populations and species (Avise 2000).
Phylogeographic studies have become increasingly important in recent decades for identifying populations that have diverged sufficiently to be considered separate conservation units thus enabling conservation biologists and wildlife managers to maximize overall genetic diversity within a species (Haig et al. 2011). Identifying such genetic divergence is particularly important for migratory species as these populations are influenced by multiple factors across several geographic locations (Webster et al. 2002).
An analysis of genetic differentiation among populations can provide insight into current and past gene flow and connectivity between breeding and wintering sites that cannot always be obtained through other methods such as banding (Webster et al. 2002).
Identifying which specific geographic locations are most affected by population declines can help inform coordination of conservation efforts of migratory species that inevitably involve multiple locations (Webster et al. 2002).
2 As mentioned, intra-specific genetic differentiation in avian species is often attributed to isolation within separate refugia during the Pleistocene glaciations or earlier. A brief review of several papers illustrates how phylogeographic analyses have been applied to better understand evolutionary processes in several shorebird species. Species that have been isolated in separate refugia during an historical glaciation are characterized by populations with monophyletic or nearly monophyletic collections of haplotypes and/or genotypes. One strategy for assessing this is to construct haplotype networks. Haplotype networks of mtDNA sequences possess two or more clusters of distinct haplotypes that are separated by one or more nucleotide substitutions (Clement et al. 2000). In cases where secondary contact has occurred between two distinct refugial populations, the two mitochondrial lineages will remain distinct (for reasons to be discussed below) and an analysis of DNA sequence divergence, known as a mismatch distribution, will show a bimodal curve reflecting the presence of two distinct haplotype clusters (Slatkin and
Hudson 1991). In contrast, a species that survived the last glaciation in a single refugium may exhibit characteristics typical of a population experiencing multiple bottlenecks due to rapid expansion (Provan and Bennett 2008). Haplotype networks often have a 'star-like' shape, defined as a single common haplotype shared by all populations, with all others diverging from this common haplotype by a few base-pair substitutions (e.g., Buehler and
Baker 2005). Networks that are not strictly star-like still have this common haplotype, with one or more clusters separated by very few base-pair differences. Tests of neutrality of these populations often indicate recent expansion, and genetic structure found is low to
3 moderate, with population genetic statistics such as Wright’s FST, for example, often being less than 0.1 (e.g., Pearce et al. 2004; Lounsberry et al. 2013).
In one of the first extensive studies of geographic genetic variation in Arctic shorebirds,
Wenink et al. (1996) described five distinct mtDNA lineages in Dunlins (Calidris alpina), a sandpiper species that has up to 11 described subspecies. These lineages were thought to be the result of multiple isolation events in the Pleistocene era (Wenink et al. 1996). A subsequent study found evidence for multiple distinct management units within some of these lineages of Dunlins and recommended that conservation efforts be directed at each of these separate populations (Wennerberg et al. 2008). Another Arctic shorebird is
Temminck's Stint (Calidris temminckii), which is thought to have expanded from two isolated refugial populations that have since introgressed (Rönkä et al. 2007, 2012). These two refugial populations were inferred due to the presence of two divergent mtDNA lineages (1.5% sequence difference); however, current breeding populations show no genetic structuring according to geographic locations, suggesting either extensive gene flow due to dispersal of breeding adults to other populations, or incomplete lineage sorting due to recent colonization of most breeding locations (Rönkä et al. 2007, 2012). In another mtDNA-based study, Pruett and Winker (2005) found evidence for mixed genetic structure among four recognized subspecies of Rock Sandpipers (Calidris ptilocnemis).
One subspecies, C. p. quarta, was separated into its own monophyletic clade, while C. p. ptilocnemis haplotypes resided in a single, paraphyletic clade. In contrast to a
4 monophyletic named subspecies, which consists of a common ancestral sequence and all descendant sequences, a paraphyletic subspecies will be defined by a phylogenetic structure in which some sequences attributed to that subspecies are ancestral but among the descendant lineages, some sequences are identified with other subspecies (e.g.,
Remsen 2010). Based on an analysis of these mtDNA sequence data, Rock Sandpipers are thought to have been isolated in multiple refugial areas during the last glaciation (Pruett and Winker 2005).
Outside of the Calidris genus, but staying within the family Scolopacidae, a recent study of Black-Tailed Godwits (Limosa limosa) found that all three recognized subspecies were reciprocally monophyletic for the mitochondrial HVR regions, while microsatellite analysis showed strong differentiation for only one subspecies, L. l. islandica. This study determined that the deep divergence observed between the L. l. melanuroides subspecies and the other two subspecies was the result of isolation in a separate Beringian refugium, while differentiation between L. l. islandica and L. l. limosa was likely the result of random genetic drift after the last Pleistocene deglaciation (Trimbos et al. 2014).
Phylogeographic studies have also been applied to Arctic populations of birds outside the family Scolopacidae. For example, Sonsthagen et al. (2011) sampled five subspecies of
Common Eider (Somateria mollissima) and found evidence of significant genetic structure that corresponds to four ancestral refugia that existed during the Pleistocene
5 glaciations; in this case the genetic lineages identified correspond largely down subspecific ranges (Sonsthagen et al. 2011).
Many avian species do not show signs of population structuring, or they show low levels of structure with some admixture due to current gene flow or incomplete lineage sorting due to recent expansion from one or more refugia or due to recovery from a bottleneck.
For example, King Eiders (Somateria spectabili) were assessed using mitochondrial DNA and microsatellites (Pearce et al. 2004). These authors found the following: 1) no evidence of genetic structure (FST = -0.03 to 0.01), 2) a single common mtDNA haplotype was shared among all populations, and 3) haplotypes formed a network with three or fewer base-pair substitutions from the haplotype that was the most divergent from the most common haplotype. Mismatch distribution of mtDNA sequences for these populations of King Eiders was a unimodal curve that matched the expected curve under population expansion, which the authors interpreted as due to recent population growth after isolation in a single refugium (Pearce et al. 2004). Similar inferences were drawn from a study of mtDNA and microsatellite markers in Buff-Breasted Sandpipers
(Tryngites subruficolli) by Lounsberry et al. (2013) who found no differentiation between populations; these authors suggest that Buff-Breasted Sandpipers expanded from a single refugium after the last Pleistocene deglaciation followed by continued gene flow among the newly established breeding colonies.
6 Red Knots (Calidris canutus), a shorebird species with up to five recognized subspecies, showed four genetically differentiated groups that were estimated to have diverged very recently (Buehler and Baker 2005). FST values among groups were low but significantly different from 0 (i.e., 0.05 to 0.27), however, the low number of base-pair differences and high level of admixture among haplotypes, the positive tests for population expansion, and the results of models indicating that the population has not reached genetic equilibrium were interpreted as evidence that these differences arose after the last
Pleistocene deglaciation (Buehler and Baker 2005). Nevertheless, genetic differentiation was such that the authors recommended that these populations not be treated as a single conservation unit (Buehler and Baker 2005). Another example of moderate population structure within a species with populations thought to have evolved in isolation since the last Pleistocene deglaciation is found in a study of Semipalmated Sandpipers (Calidris pusilla). This species has no recognized subspecies and has displayed low levels of population structure among breeding populations for mtDNA sequences using a variation on Wright’s FST called ΦST (i.e., ΦST = 0.089 to 0.221), but no substructuring for microsatellites alleles. This is attributed, again, to expansion from a single source population, with little or no dispersal among breeding populations (Miller et al. 2013). As a final example of the way that population genetic data have been interpreted for Arctic shorebirds, we can consider the case for Dunlins (Calidris alpina). The five genetically distinct lineages found within Dunlins were hypothesized to have expanded from five separate Pleistocene refugia by Wenink et al. (1996), however, variation has been found
7 within lineages as well as between them. A subsequent study of Dunlins by Wennerberg et al. (2008) used microsatellites to investigate genetic differences between two subspecies, C. alpina alpina and C. a. schinzii, both of which are located within the
European lineage described by Wenink et al. (1996). These subspecies are characterized by differences in both morphology (e.g., body size, plumage, etc.) and behaviour (e.g., time of breeding, direction of migration). These subspecies were not reciprocally monophyletic, although the Icelandic population of schinzii did show distinct differences in control region sequences (Wennerberg et al. 2008).
The use of microsatellite DNA in phylogeographic studies
Mitochondrial DNA sequences have long been a staple of phylogeographic studies, due to their strictly maternal inheritance pattern and high mutation rate (Zink and Barrowclough
2008), however, many of the assumptions about mitochondrial DNA evolution that made it such an ideal genetic marker have come into question in recent years (e.g., Galtier et al.
2009). For example, recombination is now known to be possible within mtDNA, although rates of true recombination are difficult to differentiate from false recombination that is the result of mutational hot-spots (Galtier et al. 2009). In addition, studies have found evidence that selection has had a significant impact on mtDNA evolution, and neutrality of DNA polymorphisms cannot always be assumed (Galtier et al. 2009; Toews et al.
2013). Many population genetic studies now supplement mtDNA data with several nuclear markers to correct for possible discordance between mtDNA and nuclear DNA
8 (nuDNA) due to effects such as different selective pressures, sex-biased dispersal, hybridization or biased introgression (Ballard and Whitlock 2004; McGuire et al. 2007;
Toews et al. 2013).
Studies that use markers from both genomes often find that mtDNA agrees with nuDNA data or shows greater genetic differentiation among populations than nuDNA markers, as expected due to its lower effective population size, (which is typically 1/4 of the value for autosomal nuclear markers) and its higher mutation rate (Zink and Barrowclough 2008).
A small subset of studies, however, display either conflicting patterns of differentiation between mtDNA and nuDNA or greater differentiation in nuDNA than in mtDNA (Toews and Brelsford 2012). A review of avian species found that approximately 14% of species examined (856 total) exhibited some kind of mtDNA paraphyly (that is, a clade in which not all members are part of the same group of individuals), either due to taxonomic errors in identifying species boundaries, incomplete lineage sorting, or introgression of mtDNA from one taxonomic group into another (McKay and Zink 2010). While genuine discordance between the evolutionary history of mtDNA and nuDNA is relatively uncommon, the addition of independent nuclear loci to a phylogenetic study takes relatively little effort and provides a more thorough picture of the target species' evolutionary history (Edwards and Bensch 2009). For example, a comparison of the power of autosomal, Z-linked, and mitochondrial sequences to resolve the phylogenetic
9 tree of six shorebird species found that combining loci of multiple types greatly improved the accuracy of the resulting tree (Corl and Ellegren 2013).
Traditional downsides of using nuclear genetic markers for phylogenetic studies include a slower mutation rate on average and a larger effective population size compared to mtDNA, making them more suited to inter-specific studies than intra-specific analyses
(Avise 2009). However, there are a number of nuclear markers that sidestep these difficulties. Microsatellite markers are tandem repeats of a small number (1-6) of nucleotides at a single locus. Microsatellite alleles are defined by differences in the number of repeats rather than nucleotide point-mutations and they are notable for their very fast mutation rate that makes them well-suited for intra-specific studies (Defaveri et al. 2013). They have been successfully used to assess population genetic structure in a number of avian species (e.g., Pearce et al. 2004; Sonsthagen et al. 2011; Lounsberry et al. 2013; Miller et al. 2013), and have been shown to have high power for resolving fine- scale population structuring with few (10-300) loci (Putman and Carbone 2014). Another category of nuclear markers known as SNPs (for single nucleotide polymorphisms) are distributed throughout an organism's genome and analysis of SNPs is becoming increasingly popular in phylogeographic and population genetic studies as the cost of analysing whole genomes is decreasing (Putman and Carbone 2014). Because SNP studies make use of large numbers of independent loci, they can be used to investigate areas of adaptive selection by identifying 'outlier' loci. That is, some individual SNPs may
10 show significantly more genetic structure than other SNPs (Narum and Hess 2011; Renaut et al. 2011). In addition to providing information about areas of the genome undergoing adaptive selection, outlier SNPs have been successfully used to discriminate among populations of organisms that are not well differentiated using neutral markers (Freamo et al. 2011).
Current trends in phylogeographic case studies
In the 1940s and 1950s, Sewall Wright and Gustave Malécot explored methods of measuring genetic differentiation based on the variance in allele frequency among populations compared to the variance expected if all populations are genetically homogeneous. These became known as Wright's F-statistics (Holsinger and Weir 2009).
Because of their ease of calculation, FST and FST variants (such as GST, RST, and ΦST) have become staple descriptive statistics in population genetics (Holsinger and Weir 2009).
However, in the last decade, the validity of FST and its variations have come into question
(Jost 2008), prompting renewed interest in the question of what, exactly, we intend to measure using genetic differentiation statistics and how best to calculate those statistics
(Leng and Zhang 2011; Meirmans and Hedrick 2011; Verity and Nichols 2014).
One variant of FST is known as GST. While GST is useful in mapping out relationships between populations of organisms, it suffers from a number of documented shortfalls. GST in its unmodified form is constrained by its algorithm from having values larger than the
11 average homozygosity in that population. In highly diverse populations, therefore, GST could return a very small value even though the populations are highly differentiated
(Ryman and Leimar 2009). This is especially problematic when comparing GST values for loci with different heterozygosity values, or comparing subpopulations with different effective population sizes or levels of diversity (Meirmans and Hedrick 2011).
Jost's D is a recently proposed measure of population differentiation that measures differentiation based on effective number of alleles, and is thus unconstrained by average homozygosity (Jost 2008). Values of Jost's D are also much less affected by average homozygosity, making it easier to compare values among populations and loci with different levels of diversity (Meirmans and Hedrick 2011). Jost's D is, however, more affected by the mutation rate of a marker, making comparisons between loci with different mutation rates difficult (Ryman and Leimar 2009). Indeed, in the absence of migration, Jost's D appears to measure the mutational history of a single loci exclusively, and is relatively insensitive to recent gene flow (Meirmans and Hedrick 2011). In addition, Jost’s D approaches drift-migration equilibrium much more slowly than GST, except in cases of very low haplotype diversity, and is thus less sensitive to recent demographic changes affecting gene flow and genetic drift. Measures that estimate demographic effects such as migration rate operate on the assumption that the population is in migration-mutation-drift equilibrium, and this assumption is more likely to be violated by Jost's D at low mutation rates (~10-6) (Ryman and Leimar 2009). This,
12 combined with Jost's D's insensitivity to demographic events, make Jost's D an unsuitable marker on its own for studies that are interested in teasing apart various demographic processes.
In an effort to more fully understand the effects of various demographic processes on some of these measures of population differentiation, simulation studies have shown that while GST is a reliable estimator for population differentiation when there is no mutation,
GST on its own is not sufficient to calculate “true” levels of differentiation when markers have a high level of mutation (Verity and Nichols 2014). However, using Jost's D or a standardized GST (written as G'ST) in conjunction with GST allowed conflicting signals of mutation and differentiation to be separated (Verity and Nichols 2014). For example, a study of short-term and long-term population structure of Three-Spined Stickleback
(Gasterosteus aculeatus) found that Jost's D was a more reliable measure of historical colonization history, while GST was better at detecting contemporary demographic processes (Raeymaekers et al. 2012). To summarize, using both measures of differentiation together yields a more complete picture of patterns of recent demography and historical differentiation between populations. Yet another statistic known as ΦST, which was developed by Laurent Excoffier and colleagues (Excoffier et al. 1992), has been shown to be independent of mutation rate, reflecting only demographic processes such as drift and mutation (Meirmans and Hedrick 2011). The addition of ΦST to a
13 population genetic analysis may, therefore, provide further insights in population processes.
In addition to these tests of genetic differentiation, a wealth of techniques have been formulated to assess further aspects of microevolutionary forces acting on population structure and demographic history. Developments in genotyping technology have facilitated the ability to assess genetic structure within populations using more samples and more markers than ever before (Excoffier and Heckel 2006). Among these techniques are clustering methods such as those implemented in the software package STRUCTURE published by Jonathan Pritchard’s lab. STRUCTURE estimates the number of distinct genetic clusters that a given group of samples comes from, and attempts to estimate assignment of each individual to a cluster (Pritchard et al. 2000). These methods have been shown to be effective at detecting structure in populations with a relatively low amount of differentiation (FST < 0.1; e.g., Latch et al. 2006).
Sequence data contain a number of indicators of the evolutionary history of a genome, including whether populations deviate from the assumption of mutation-drift neutrality due to selection or demographic processes such as expansion and bottlenecks (e.g.,
Jeratthitikul et al. 2013; Lounsberry et al. 2013). However, distinguishing the underlying causes of a departure from neutrality is extremely difficult, and in many cases is not yet possible (Ramírez-Soriano et al. 2008). A large number of statistical tests of genetic
14 neutrality in sequence data have been designed to shed more light on nuances of the genetic fingerprint each of these processes leaves on a population (reviewed in Ramos- onsins and Rozas 2002; Ramírez-Soriano et al. 2008). To determine effectiveness of each of these tests in relation to each other, a number of simulation studies have been run, particularly on the more common tests. These simulations attempted to assess the power of each test to correctly detect departures from neutrality, and also to determine their sensitivity to each scenario of non-neutrality (Ramírez-Soriano et al. 2008). Ramos-
Onsins and Rozas (2002) classified these tests into three classes. Statistics in class I examine the number of observed segregating sites against the expected number assuming a neutrally evolving population. Class II statistics make use of the distribution of haplotype frequencies, and class III statistics look at the distribution of pairwise differences in sequences. Class III statistics had the lowest power to detect deviations from neutrality (Ramos-Onsins and Rozas 2002), while class II statistics had the highest power (particularly FS, ZnS and R2). This same class, however, is also very sensitive to recombination (Ramírez-Soriano et al. 2008), making other classes of tests more ideal for markers that are known or suspected to undergo recombination.
Natural history of the Purple Sandpiper
The Purple Sandpiper, Calidris maritima, is a medium-sized migratory shorebird notable for its wintering range, which extends farther north than any other shorebird (Summers et al. 1998). During summer months, these birds have nesting grounds from the high
15 Canadian Arctic to the more southern Hudson Bay in North America. Outside of the
Americas, Purple Sandpipers nest in Greenland, Iceland and northern Europe (see Chapter
2, Figure 1). In the winter, these birds move south to feeding grounds along the eastern coast of Canada and the United States, southern Greenland, Iceland, Norway and throughout Europe (Payne and Pierce 2002).
Purple Sandpipers are difficult to track using methods such as banding (Summers 1994;
Mittelhauser et al. 2006), and they are too small to tolerate attachment of satellite GPS devices that would negate the need for recapture or re-sighting (M. Mallory pers. comm.).
Much of the research on Purple Sandpipers has been conducted in Europe (reviewed in
Payne and Pierce 2002), and several European migratory routes have been described based on band recoveries and biometric comparisons (Summers et al. 2009). Purple
Sandpipers migrate from breeding populations in Norway to the coast of Britain (Nicoll et al. 1988), from Svalbard to more southern locations in Norway and Sweden (Hake et al.
1997), and from colonies in western Russia to Norway (Strann et al. 2006). A recent study used light-based geolocators to confirm a trans-Atlantic migration route between breeding populations in Canada to wintering populations in Scotland (Summers et al. 2014). Many routes, however, are uncertain, particularly in the case of Purple Sandpipers that winter in
North America (Mittelhauser et al. 2006).
16 A study conducted by Engelmoer and Roselaar (1998) found that Purple Sandpipers from different breeding locations had differing morphological characteristics, particularly wing size, and proposed three putative subspecies based on these differences. The largest
Purple Sandpipers were found in Iceland, and were proposed as a separate subspecies,
Calidris maritima littoralis. Similarly, the smallest Purple Sandpipers in southern Hudson
Bay were proposed as the subspecies, C. m. belcheri. All other breeding populations were grouped into a third subspecies, C. m. maritima (Engelmoer and Roselaar 1998). These size differences have been used to estimate breeding origin in wintering locations across the species' range, but results are tentative because of overlap caused by size differences between sexes (Hake et al. 1997; Summers et al. 2001, 2009; Mittelhauser et al. 2006).
Based on size data, Purple Sandpipers wintering in Maine are suspected to be of mixed origin, possibly including European breeding birds (Mittelhauser et al. 2006).
Because of their high dispersal potential, gene flow among populations of migratory birds is determined largely by fidelity of individuals to their birth site. This is known as natal philopatry, which is the tendency of individuals to return to the location where they were born after returning from the wintering grounds for their first breeding attempt (Oring and
Lank 1984). A similar measure, breeding fidelity, measures the proportion of adult birds that return to their last nesting area in subsequent years. Natal philopatry and breeding fidelity in Purple Sandpipers is similar to that of other monogamous Calidris sandpipers.
Observed breeding fidelity in a breeding location in Scotland was 64.7% (Smith and
17 Summers 2005), and in a Svalbard breeding population it was 54% for males and 60% for females (Payne and Pierce 2002). In comparison, breeding fidelity of the sister species to
Purple Sandpipers, the Rock Sandpipers, was 74% (Gill et al. 2002). Values for the closely related Dunlin species were 77% for males and 71% for females, and for
Semipalmated Sandpipers the values were 58% for males and 49% for females (Oring and
Lank 1984; Gill et al. 2002). Natal philopatry of Purple Sandpipers in a breeding location in northern Scotland was 4.5% (one bird out of 22 banded chicks; Smith and Summers
2005). This is, again, comparable to measured natal philopatry of Semipalmated
Sandpipers (3.1%) and Dunlins (3.6% - 11.2%; Oring and Lank 1984). Dispersal between breeding populations of all these species is considered low, and they show a variety of levels of genetic structure, as discussed above (Wenink et al. 1996; Pruett and Winker
2005; Miller et al. 2013).
Population size estimates of Purple Sandpipers are inconsistent. Population data on breeding grounds is scarce because breeding pairs are widely dispersed in low nesting densities, and thus exceedingly difficult and expensive to survey (Wenink et al. 1996;
Pruett and Winker 2005; Miller et al. 2013). The breeding population in Canada was estimated at 25,000 (Summers 2014) and in Iceland is thought to be 90,000 including offspring (Andres et al. 2012). The Christmas Bird Count (CBC) has estimated the number of wintering Purple Sandpipers in eastern North America as 16,000 (Burton et al.
2008), 15,000 (Morrison et al. 2001), and 15,000 – 20,000 (Morrison et al. 2006), and
18 notes that a large proportion of the wintering population resides on islands off the coast outside of the CBC's range (Andres et al. 2012). The concern that CBC census counts significantly underestimate population size was supported by a later survey of the wintering population in Maine, which estimated 14,000 – 17,000 birds in that state alone
(Andres et al. 2012). Similarly, in Britain most wader monitoring programs are focused on estuarine habitats, and so do not include the primarily non-estuarine Purple Sandpiper
(Mittelhauser et al. 2013). Three non-estuarine surveys have been conducted: the Winter
Shorebird Count (WSC) in 1994/1995, the first Non-estuarine Waterbird Survey (NEWS-
I) in 1997/1998, and the second Non-estuarine Waterbird Survey (NEWS-II) in
2006/2007. Change in population size of wintering Purple Sandpipers between the WSC and NEWS I was reported at -20.5% (Rehfisch et al. 2003; Summers 2014). Preliminary comparisons between NEWS II and the WSC also found a decline in population numbers of 27%, though these have not been standardized using a method known as “paired count stretch approach” and according to Austin et al. (2008) these should be treated as rough estimates. More localized counts undertaken in Scotland, where the majority of Purple
Sandpipers are from Canadian breeding grounds, conducted since the NEWS-II survey have also shown declines (Corse and Summers 2009; Summers et al. 2012b).
Purple Sandpipers are protected under the Migratory Birds Convention Act in Canada
(Env. Canada 2012), and under the U.S. Shorebird Conservation Plan in the United States, where they are listed as moderate concern (U.S. Shorebird Conservation Plan, 2004). In
19 addition, Purple Sandpipers are one of the species affected by rockweed (Ascophyllum nodosum) harvesting in Maine (Seeley and Schlesinger 2012).
Identifying migratory pathways of this species is essential for determining the source of observed population declines and directing conservation efforts towards preventing or mitigating threats that are having the most deleterious effects on certain populations
(Webster et al. 2002). Identification of genetically differentiated populations is also important in maintaining genetic diversity in populations that are conserved (Webster et al. 2002; Deane et al. 2013). In this study, genetic data from mitochondrial DNA sequences and microsatellite markers will be assessed 1) to determine the validity of the proposed Purple Sandpiper subspecies, 2) to determine the level and distribution of genetic structure among breeding populations, and 3) to identify trends in genetic patterns within wintering populations to help identify breeding population sources.
20 CHAPTER 2
Genetic differentiation and phylogeography of Purple Sandpipers using mitochondrial DNA markers.
Introduction
Population Declines in North American Shorebirds
Range-wide surveys have shown that many shorebird populations are showing statistically significant or persistent declines in Canada and the United States (Morrison et al. 2001). In Atlantic Canada and the northeastern United States, 73% of 30 surveyed shorebird species showed a decline in population numbers (Morrison et al. 2006; Bart et al. 2007; Andres et al. 2012). Declines have been attributed to factors such as exposure to pesticides at wintering grounds and stop over sites, climate change, increase in predator populations, habitat loss and decline of habitat quality (Rehfisch et al. 2003; Bart et al.
2007; Strum et al. 2010). In Atlantic Canada and Maine, rockweed (Ascophyllum nodosum) harvesting is a concern for many organisms that make use of the rocky, tidally- inundated habitats in which this plant grows, including many bird species (Seeley and
Schlesinger 2012).
21 As an increasing number of shorebird species come under conservation concern, it is important to characterize the level of genetic variation and genetic population structure for each species as this genetic information provides a key tool for developing conservation strategies. Identification of distinct populations is essential for developing management policies aimed at maintaining maximum possible genetic diversity (Webster et al. 2002). In addition, the degree to which migratory bird populations share wintering sites, breeding sites and migratory stop over sites, can influence the extent that a particular species is affected by detrimental environmental threats. By identifying gene flow and migratory connectivity, conservation efforts are able to address concerns for the population at the locations that are most affected (Deane et al. 2013; Lounsberry et al.
2013).
One species with a documented population decline is the Purple Sandpiper, Calidris maritima. Purple sandpipers are migratory shorebirds notable for their wintering range, which extends from the coast of Newfoundland to southern Maine in North America, southern Greenland, Iceland, and Europe from the United Kingdom to northern Russia, farther north than any other shorebird (Summers et al. 1998). In summer, they move north from their wintering sites to the high Canadian Arctic and Hudson Bay, Greenland,
Iceland, Svalbard, Scandinavia, and the Severnaya Islands (Summers et al. 1998, 2014;
Payne and Pierce 2002). While many European migration routes have been determined through mark recapture data or hypothesized through biometric comparisons (Atkinson et
22 al. 1981; Rae et al. 1986; Summers et al. 1988, 1990; Strann et al. 2006) our knowledge of North American populations is limited, as few studies have centred on Purple
Sandpipers in this part of the world (Mittelhauser et al. 2006). Most migration routes are short, for example from the Svalbard Archipelago to mainland Norway and Sweden
(Hake et al. 1997) or from Russia to Norway (Strann et al. 2006). Some Purple
Sandpipers in Iceland and Greenland appear to be permanent residents (Summers et al.
1988; Payne and Pierce 2002). It has been suggested that most Purple Sandpipers that breed in Canada winter along the eastern coast of North America (Mittelhauser et al.
2006), however, it has long been suspected that a subset cross the Atlantic Ocean to winter in the United Kingdom (Payne and Pierce 2002). A recent geolocator study tracked
12 individuals wintering in the United Kingdom and confirmed this route (Summers et al.
2014), making this the longest demonstrated migration route for the species at 2000 to
3500 km (Summers 1994).
In Canada, Purple Sandpipers are protected under the Migratory Birds Convention Act
(Env. Canada 2012). A decline of 1.8% for wintering North American Purple Sandpipers has been observed during the Christmas Bird Count (CBC), although a significant portion of the wintering population occurs on offshore islands outside of the CBC's range (Andres et al. 2012). The Christmas Bird Count estimate of the number of wintering Purple
Sandpipers in all of North America is 16,000. However, a recent census in Maine places the number of wintering Purple Sandpipers in that state at 14,000-17,000, suggesting that
23 the CBC may significantly underestimate Purple Sandpiper population size in North
America (Mittelhauser et al. 2013). In Britain, abundance is poorly studied due to the fact that Purple Sandpipers that winter here do so largely in non-estuarine habitat, and most bird surveys focus on estuarine areas. However, the Non Estuarine Wader Survey of
1997/98 indicated a 21% decrease in abundance of Purple Sandpipers over all of Britain relative to the 1984–85 survey (Rehfisch et al. 2003). More recent localized counts have also shown declines in Scotland (Corse and Summers 2009; Summers et al. 2012a), where the majority of Purple Sandpipers were hypothesized to originate from Canada according to morphological characteristics. Additionally, recruitment in northern Britain appears to be correlated with summer snow fall in Baffin Island, Canada, suggesting that size of the wintering population in Britain could be affected by breeding success in Canada
(Summers 2014).
Any conservation effort directed towards Purple Sandpipers may affect populations in several countries on both sides of the Atlantic Ocean, requiring international cooperation.
These efforts will be most efficient and effective if it is possible to pinpoint which habitats and populations are involved in the overall decline and identify populations that may represent discrete evolutionary groups.
24 Identification of breeding origin and migration patterns of shorebirds
There are a number of difficulties inherent in detecting migration patterns in migratory shorebirds. Traditional methods involve banding individual birds at one location and then resighting or recapturing them at another (Webster et al. 2002). However, this method typically has relatively low success. Similarly, the use of morphological characteristics to trace migratory birds back to their source breeding populations is often confounded by overlap in the range of morphological measurements, either between sexes or populations or both (Atkinson et al. 1981). Satellite telemetry equipment is still too large to allow use on most migratory shorebirds, and geolocators require recapture of the bird to access the data they record (Clark et al. 2010). Increasingly, genomic markers are being used as a way to assess connectivity between populations, as well as to measure demographic processes such as range expansion that are currently affecting or that have recently impacted the species (Pruett and Winker 2005; Rönkä et al. 2012; Lounsberry et al. 2013;
Miller et al. 2013; Arbabi et al. 2014).
Variation in mitochondrial DNA (mtDNA) is regularly used for these studies, as it possesses several characteristics that in combination make it uniquely suited to fine-scale population analysis (Avise 2000). These characteristics include maternal inheritance in most organisms, as well as little to no measurable recombination (Zink and Barrowclough
2008). Avian mtDNA also lacks confounding features such as transposable elements and introns (Avise et al. 1987). As a haploid genome, mtDNA has a smaller effective
25 population size than that of nuclear DNA, and also seems to accumulate mutations at a faster rate than most nuclear DNA (Wilson et al. 1985), resulting in an approximately four-fold increase in the rate of evolution. Isolated populations will develop distinct lineages for mtDNA before they will for nuclear DNA, making it a more sensitive indicator of genetic drift, especially over shorter time-scales (Zink and Barrowclough
2008).
Pruett and Winker (2005) found genetic structure among some Rock Sandpiper (Calidris ptilocnemis) subspecies in the Beringian strait, and incomplete lineage sorting among others. Mitochondrial DNA has also been used to investigate genetic structure in the
Black-Tailed Godwit (Limosa limosa; Höglund et al. 2009), and a variety of Eurasian species (Zink et al. 2008). In the highly structured Dunlin (Calidris alpina) species, mtDNA variation has successfully been used to determine breeding origin and migration patterns of a number of wintering populations (Wennerberg 2001). In migrant birds, detection of genetically distinct populations is important, as distinct conservation units cannot always be predicted by the presence of physical barriers or, indeed, morphological characteristics (Phillimore and Owens 2006; Avise 2009). A recent meta-study of traditionally recognized avian subspecies found that only 36% had corresponding genetic structure, while the rest were not genetically distinct from other subspecific groups
(Phillimore and Owens 2006).
26 By examining variable regions of the mtDNA genomes, we can determine the nature and extent of genetic structuring across the entire range of the Purple Sandpiper range, and identify recent demographic influences on contemporary Purple Sandpiper populations.
Based on external morphology, Engelmoer and Roselaar (1998) proposed three subspecies of Purple Sandpiper. In this chapter, I use mitochondrial DNA sequence data to test the prediction that genetic divergence patterns would match morphological differences in these three putative subspecies. Further, I use sequence data to examine more general patterns of divergence among Purple Sandpiper breeding populations, to determine whether possible differences in sequences could be used to assign migratory birds to specific populations or regions.
Methods
Samples
The present study is based on an analysis of a total of 1534 bp from both control-region and cytochrome b, obtained for 279 individuals representing four breeding locations and five wintering locations. All singleton haplotypes were re-amplified from total DNA to test for PCR-based amplification errors, and error rate was 10%. Two sets of samples were analyzed and pooled for the present study. The first set was from a previous study
(Leblanc 2013), and included 25 samples of Purple Sandpiper DNA collected from Maine in 2002 and 2003, one blood sample and two museum tissue samples from Nova Scotia, taken from the Acadia University museum, and 20 tissue samples donated by the
27 Canadian Museum of Nature. These 20 samples are of varying ages from 37 – 62 years old, and represented birds collected in northern Nunavut, defined as the area north of
Hudson Bay (Prince of Wales Island, Bathurst Island, Cornwallis Island); southern
Nunavut, defined as the islands located on the southern (North Twin Island) and eastern side (Long Island) of Hudson Bay; and islands in between these two areas that have not been included in previous morphometric analyses (Coats Island, West Foxe Islands; Fig
1; Engelmoer and Roselaar 1998).
The second set consists of newly obtained samples. Of these, 92 are blood samples collected from Maine in 2013. Eighty-two of these were stored in Queen's Lysis Buffer at room temperature and nine were stored in FTA cards. Twenty-one blood samples were collected from Grand Manan Island, New Brunswick. Blood samples were stored at room temperature in Queen's Lysis Buffer (Seutin et al. 1991).
Twenty-five samples of liver tissue were obtained from the Nova Scotia Department of
Natural Resources, taken from carcasses collected in Nova Scotia in winter of 2013, from
Bon Portage Island (Shelburne Co.), Grey Island (Shelburne Co.), Little Hope Island
(Queen's Co.), Bowens Ledge (Halifax Co.), and Flat Ledge (Halifax Co.). Liver tissue was stored at -20 C in 100% ethanol.
28 A total of 16 blood samples were collected in Point LaHaye and Dildo Island in eastern
Newfoundland, as well as Broom Point in western Newfoundland, in 2013 and 2014. All samples were stored at room temperature in Queens Lysis Buffer. Forty-two feathers sampled from birds from the United Kingdom were donated by Ron Summers from the
Royal Society for the Protection of Birds. Twenty DNA samples each from Iceland and
Svalbard were donated by Dr. Snæbjörn Pálsson at University of Iceland. Three blood samples from Rock Sandpipers were obtained from Dr. Dan Ruthrauff and used as an outgroup.
All but five of the Canadian breeding samples were obtained from areas covered in the comparison of morphological characteristics published by Engelmoer and Roselaar
(1998). The 10 samples collected from breeding populations on Bathurst Island,
Cornwallis Island, and Prince of Wales Island in northern Nunavut, were grouped with most European breeding populations in the subspecies Calidris maritima maritima
(Engelmoer and Roselaar 1998). The five samples taken from Long Island and North
Twin island in the southern tip of Hudson Bay were placed in the putative subspecies
Calidris maritima belcheri, whose studied range extends up the eastern coast of Hudson
Bay (Engelmoer and Roselaar 1998). Five samples were taken from Coats Island and the
West Foxe Islands in northern Hudson Bay, which fall between the two areas sampled by
Engelmoer and Roselaar (1998), and the morphological characteristics of the birds that
29 breed there are unknown. Because of this uncertainty, I considered two possible groupings of Purple Sandpipers breeding in Canada (Figure 3):
1) Scenario A. The northern Hudson Bay birds may resemble those found farther
north, and fall within the putative Calidris maritima maritima subspecies. This
theory is supported by measurements of two birds taken from nearby
Southampton Island (Engelmoer and Roselaar 1998), which most closely
resembled those of birds on Baffin Island, and the proximity of West Foxe
Islands to the southern part of Baffin Island. Scenario A therefore groups these
five samples with the northern Canadian birds.
2) Scenario B. The northern Hudson Bay birds may, alternatively, resemble
Purple Sandpipers from southern and eastern Hudson Bay, due to proximity to
the eastern coast. Scenario B therefore groups these five samples with the
birds taken from southern Hudson Bay.
Molecular Analysis
Blood and Liver samples
DNA was isolated from blood samples using a DNeasy Blood & Tissue kit (Qiagen), spin column protocol. For blood samples, I prepared phosphate buffered saline (PBS) according to Bush et al. (2005), whereas for liver samples, I cut 25 mg from the main tissue using utensils that had first been sterilized in flame.
30 Feathers
DNA was isolated from feathers using a DNeasy Blood & Tissue kit (Qiagen), spin column protocol, modified as follows: feathers were cut into ~1 cm pieces, soaked in a
10% bleach solution for 30 minutes to remove surface contaminants, then rinsed 3 times to remove residual bleach. One feather was used for each sample. To maximize DNA yield the entire feather was incubated overnight at 56 °C in 180 μL buffer ATL (Qiagen),
20 μL 1M DTT, 20 μL 1M proteinase K (Taberlet and Bouvet 1991). The Qiagen protocol was then followed normally up to the final elution step in which the DNA was allowed to sit on the membrane for 5 min before it was eluted. After bleach incubation, samples were transported to a separate laboratory for DNA isolation. All utensils were cleaned with ethanol, bleach and UV light.
All sampled DNA was amplified using a thermal cycler (Peltier PTC-0200 or Biometra T-
Gradient) to produce double stranded control region or cytochrome b mtDNA fragments.
DNA was amplified in a reaction mixture containing 44 μL Platinum® Blue PCR
SuperMix (Cat. Number 12580-015), 2 μL forward primer (10 μM), 2 μL reverse primer
(10 μM) and 2 μL template. Specifics of the various forward and reverse primers are described below. Samples isolated from feathers were amplified with 5 μL template as needed.
31 Depending on DNA quality, the control region was amplified in one to three fragments, and the cytochrome b fragment was amplified in one or two fragments. Control region reactions were heated to 94 °C for 3 min for denaturing, followed by 30 cycles of 1 min at
94 °C for denaturing, 1 min at 50 °C for annealing, and 1 min at 72 °C for extension.
Finally, the temperature was brought to 72 °C for 5 min for a final extension. Cytochrome b reactions were heated to 94 °C for 3 min for denaturing, followed by 30 cycles of 45 sec at 94 °C for denaturing, 45 sec at 50 °C for annealing, 1 min at 72 °C for extension, and the final extension step of 72 °C for 5 min. Samples were sent to McGill University and
Génome Québec Innovation Centre for sequencing.
For high quality DNA samples obtained from fresh material (i.e., blood or liver), the control region was amplified in one fragment using two primers (Table 1). Primer L98 was taken from a previous study of Dunlins, a close relative of Purple Sandpipers
(Wenink et al. 1993), while primer H1018 was taken from a previous study of Purple
Sandpipers (Leblanc 2013). These primers amplify an 841 bp fragment of DNA. Samples isolated from feathers were amplified using two or three fragments to improve amplification success and sequence quality. Primers were again taken from either Wenink et al. (1993) or LeBlanc (2013) and used as described in LeBlanc (2013). Rock sandpipers were amplified using an alternate primer also designed by LeBlanc (2013; Table 1).
32 A cytochrome b fragment of 712 bp was amplified as one fragment initially, using primers designed for Rock Sandpipers (Pruett and Winker 2005). Samples that failed to amplify using these primers were amplified in two fragments, using internal primers developed in a previous study (Leblanc 2013). Primers all retain their original labels.
Control region and cytochrome b sequences were analysed as a single locus, as the unique inheritance pattern of mitochondrial DNA means they share the same evolutionary history
(Zink and Barrowclough 2008). Two samples from Iceland could not be sequenced fully and were excluded from the study. To ensure all haplotypes were analysed unambiguously, sequence was trimmed to remove nucleotide positions with missing data, including a section of cytochrome b that did not overlap fully when amplified and sequenced in the forward and reverse directions with the internal primers. Nineteen base pairs in total were removed from cytochrome b sequence. A singleton United Kingdom haplotype was lost as a result.
Data Analysis
Sequences were aligned using the web application Clustal Omega (Sievers et al. 2011), edited in Jalview (Waterhouse et al. 2009), and saved in fasta format. Sequences were translated into amino acid sequences in MEGA6 (Tamura et al. 2013) to check for premature stop codons, frameshifts or other evidence of nuclear pseudogenes (Rodríguez et al. 2007). The web program FaBox (Villesen 2007) was used to concatenate sequences
33 into a single mtDNA loci, convert the fasta files into the format required for Arlequin
(Excoffier and Lischer 2010) and TCS (Clement et al. 2000). The program PGDSpider
(Lischer and Excoffier 2012) was used to convert between input formats for all other software described.
Modeltest (Guindon and Gascuel 2003; Darriba et al. 2012) was used to estimate the nucleotide evolution model that best fit the sequence used in this study. The best model was found to be the Hasegawa, Kishino and Yano (HKY) model with invariant sites (+I).
The HKY+I model was used in subsequent analyses unless not supported, in which case the Jukes-Cantor model was used.
Overall levels of differentiation among breeding populations was measured using a modern version of Wright’s classic Fixation index (ΦST) that incorporates genetic distances measured from sequence data, using the software package Arlequin (Excoffier and Lischer 2010). Pairwise ΦST values were also calculated with Arlequin, for both breeding and wintering populations. P-values were generated using 10,000 non- parametric permutations to obtain a null distribution of ΦST values. Unlike the allele frequency based statistics that ΦST is derived from, ΦST incorporates information about the mutational differences between haplotypes and as a result is independent of mutation rate
(Meirmans and Hedrick 2011).
34 Standard genetic diversity indices were calculated in DnaSP (Librado and Rozas 2009) for each population as well as all populations pooled. Indices included sample size, number of haplotypes, number of private haplotypes, number of segregating sites, nucleotide diversity (average pairwise nucleotide difference between individuals, Nei
1987), and haplotype diversity (probability of any two individuals randomly selected from the population having different haplotypes, Nei 1987).
Distinct sequence haplotypes were identified using FaBox and imported into MEGA 5.10
(Tamura et al. 2011). A maximum likelihood tree was constructed using the HKY model of nucleotide substitution with invariant sites, pairwise deletion and the Nearest-
Neighbor-Interchange Heuristic Method, and bootstrapped 1000 times. As well, a maximum parsimony tree was constructed using the Subtree-Pruning-Regrafting (SPR) search method and pairwise deletion, and bootstrapped 1000 times. A Bayesian tree was similarly constructed using the program Mr Bayes (Ronquist and Huelsenbeck 2003), using the HKY+I model of nucleotide evolution. Two simultaneous analyses were run to calculate convergence values. For both analyses, a run length of 1,000,000 and a burn-in fraction of 25% was used. A statistical parsimony haplotype network was constructed in
TCS (Clement et al. 2000) for all mtDNA fragments combined as well as for just the protein-coding cytochrome b fragments, drawn in Inkscape (http://www.inkscape.org/) with the NiceCharts extension, and compared to a minimum-spanning tree constructed in
Arlequin and drawn by hand.
35 A large number of statistical tests have been developed over the years to measure deviations from a neutral model of evolution. Such deviations can be caused by a number of demographic events, such as population expansion, contraction, bottlenecks, selective sweeps, or background selective pressure in the population's history. Because all of these events leave a similar “molecular footprint” on an organism's DNA, distinguishing among these factors can be difficult. Ramos-Onsins and Rozas (2002) organized these tests into three categories, according to the type of sequence information used. Class 1 tests were based on segregating site frequency, class 2 tests were based on haplotype distribution, and class 3 tests were based on distribution of pairwise sequence differences. Ramírez-
Soriano et al. (2008) investigated the statistical power of a variety of common class 1 and
2 neutrality tests under different demographic processes, with and without recombination.
Class 2 tests, especially Fu's Fs (Fu 1997), were more powerful overall at detecting deviations from neutrality but were much more sensitive to recombination than class 1 tests. In this study, mitochondrial deviations from neutrality were measured using all three categories of tests. In class one, Tajima's D and Ramos-Onsin's R2 were used. Fu's Fs was chosen as the class two test, and mismatch distribution was measured for class 3 tests.
Mismatch Distribution analysis measures the distribution of pairwise sequence divergence frequencies; that is, how often two randomly selected sequences differ by zero, one, two, etc. base-pairs. Population expansion leaves a characteristic unimodal distribution of
36 pairwise divergence values that can be used to test for the presence of such an expansion in a population being studied (Harpending 1994). The observed distribution is compared to an expected distribution given a model for either a constant, stable population or a population experiencing growth. If population expansion is identified, the time since expansion can be estimated using τ (τ =2ut, where t is the time in generations and u the mutation rate per sequence per generation). A raggedness statistic was used to test the smoothness of the distribution. Lower values of τ indicate population growth (Slatkin and
Hudson 1991; Rogers and Harpending 1992).
Perhaps the most widely used test of neutrality, Tajima's D compares the number of nucleotide differences between pairs of sequences with the number of segregating sites in those sequences. The observed ratio is compared to the expected ratio under a neutral model of evolution (Tajima 1989) . Similarly, Ramos-Onsin's R2 statistic examines the number of singleton mutations on a branch in a genealogy, compared to the average number of nucleotide differences. Values significantly close to zero indicate recent population growth. This test has been shown to have more statistical power given smaller sample sizes, compared to Fu's Fs and Tajima's D (Ramos-Onsins and Rozas 2002).
Fu's Fs is based on the infinite sites model, and estimates the probability of having a certain number of alleles (i.e. the observed number), given the level of diversity (θ) observed. A significant negative value indicates an excess of rare mutations and suggests
37 population expansion or selection. Fu's Fs is commonly considered a more sensitive indicator of population expansion at p-value < 0.02 (Fu 1997), and multiple comparisons of common neutrality tests give it the greatest statistical power overall, in the absence of recombination (Ramos-Onsins and Rozas 2002; Ramírez-Soriano et al. 2008).
Results
Data Analysis
Of the 1535 nucleotide sites sequenced from a total of 279 birds, forty-one sites were polymorphic (Table 3). Forty-two haplotypes were obtained in total, 23 of which occurred in at least two individuals. Of these, 22 haplotypes were only found in wintering locations, and nine were not found in any wintering locations. The most common haplotype comprised 33% of samples, from all breeding and wintering locations. The second most common haplotype comprised 17% of samples, and was found in all wintering locations but only one breeding location (southern Hudson Bay).
Analyses of population structure
Overall ΦST
Overall ΦST values were calculated for breeding populations under both scenarios. For
Scenario A, overall ΦST was 0.089 (p = 0.004). Scenario B resulted in a slightly lower overall ΦST of 0.077 (p < 0.006). Both values were significantly higher than expected under panmixia.
38 Pairwise ΦST
In both scenarios A and B, pairwise ΦSTvalues (Table 2) were significantly higher than expected under panmixia between Iceland and Norway (ΦST= 0.12; p = 0.006) and between Norway and the southern Nunavut group (ΦST = 0.18 and 0.06; p = 0.018 and
0.048), although ΦSTvalues for the latter were three times as high in Scenario A (where northern Hudson Bay samples are included in the northern Nunavut group). Pairwise ΦST values were relatively high in both scenarios between Iceland and the northern Nunavut group (ΦST = 0.10 and 0.09; p = 0.007 and 0.061), and Iceland and the southern Nunavut group (ΦST = 0.12 and 0.17; p = 0.057 and 0.017), but the Iceland/northern Nunavut comparison was only significantly so in Scenario A, and the Iceland/southern Nunavut comparison was only significant in Scenario B. In both cases, the non-significant ΦST value had a p-value very close to 0.05. In both scenarios, there was no significant population differentiation observed between the two Nunavut groups (p = 0.490 and
0.675), or between Norway and the northern Nunavut group (p = 0.210 and 0.100).
Pairwise ΦST values of sampled wintering locations show relatively large and significant differentiation between Iceland and all wintering locations, and a similar pattern between
Norway and North American wintering locations (Table 2). Due to patterns seen in
STRUCTURE clustering results for microsatellite loci (see chapter 3 of this thesis), the four samples from western Newfoundland (Broom Point) were treated as a separate
39 population for pairwise ΦST. Differentiation between Iceland and western Newfoundland was much higher (ΦST = 0.292; p = 0.034) than with other North American wintering locations. The Norway and the United Kingdom comparison has a non-significant ΦST value (p = 0.097), and North American breeding locations were not significantly different from any wintering ground, including the United Kingdom (p = 0.086 to 0.999), with the exception of northern Nunavut in scenario A and Maine (ΦST= 0.093; p = 0.019).
Standard genetic diversity estimates
Overall nucleotide diversity, π, was 0.00157 (Table 3). Nucleotide diversity among breeding populations was low, with π ranging from 0.00063 in Iceland to 0.00169 in southern Nunavut. Nucleotide diversity in wintering populations was generally higher than in breeding populations, ranging from 0.00156 in Maine to 0.00187 in New
Brunswick.
Haplotype diversity, h, was highest in the southern Nunavut populations in scenario B
(0.911), and lowest for the northern Nunavut population in scenario B (0.533). As was observed for nucleotide diversity, haplotype diversity in wintering populations was higher than for breeding populations (Table 2).
Phylogenetic analysis
40 The Bayesian (Figure 4), maximum likelihood (Figure 5), and maximum parsimony trees
(Figure 6) all differed from each other in some branching patterns, reflecting the weak support found for certain clades within each tree. Clades in the Bayesian tree were assigned the letters A-E in order of decreasing posterior probability, and two novel clades in the maximum likelihood tree were assigned the letters F and G. Branch lengths were short throughout the tree, indicating recent divergence of haplotypes, no clade was monophyletic for a single breeding population or proposed subspecies, and no clade had posterior probabilities and bootstrap values above 95% and 60%, respectively, for all trees. All three trees displayed a 'comb-like' topology, and the lack of agreement between them reflects the general lack of resolution seen among haplotypes.
Clade B, the largest with nine haplotypes, was found in all three trees. Posterior probability (PP) for this clade was 79%, and bootstrap values were 59% and 44% for maximum likelihood (MLB) and maximum parsimony trees (MPB), respectively. This clade was defined by a single substitution in the control-region and two substitutions in the cytochrome b region; one of the cytochrome b substitutions was present in all samples within the clade. This clade was monophyletic for North American breeding samples, which may indicate a North American breeding origin for wintering birds that possess these haplotypes. In the Maximum Likelihood tree, two additional haplotypes were grouped, albeit weakly, with clade B (Figure 5, haplotypes 3 and 38). The most common of the two haplotypes, 3, is possessed by 50% of analyzed Icelandic breeders and 10% of
41 Norwegian breeders. These two haplotypes are not associated with clade B in the other two trees.
Clade A was the second largest clade, with five haplotypes, and was present in both the
Bayesian and maximum likelihood trees. This clade has a posterior probability of 96% and a bootstrap value of 52%. It was differentiated by a single transversion mutation in the control region. Samples from both Nunavut and European breeding locations were present in this clade.
The remaining clades seen in these trees each contained only two haplotype groupings.
Clade E was grouped on its own in all three trees, with a posterior probability of 62% and bootstrap values of 60% and 31% for maximum likelihood and maximum parsimony, respectively. Clades C (PP 74%, MLB 65%) and D (PP 70%, MLB 62%) were found in the Bayesian and maximum likelihood trees. Clades F (MLB 35%) and G (MLB 31%) were found only in the maximum likelihood tree.
The statistical parsimony haplotype network (SPHN) of concatenated (i.e. joined together) mtDNA fragments (Figure 7A) and minimum-spanning network (not shown) agreed with each other. Clade B, as identified in the phylogenetic trees, was also clearly present in the SPHN, coming off of an otherwise star-like cluster of haplotypes. Also haplotypes 3 and 38 were seen in the maximum likelihood phylogeny, connecting the
42 clade to the rest of the network. Haplotypes from all breeding locations were dispersed throughout most of the network, indicating limited geographic structure. Branches were also quite short, with many singleton or low frequency haplotypes differing from a common haplotype by a single substitution.
Three circular connections in the network corresponded to sets of four sequences that failed what is known as the four gamete test (Martin et al. 2011). This could be due to recombination, which is a rare but not unknown event in mitochondrial DNA (Rokas et al.
2003). Alternatively, the presence of circular connections could indicate that one to three nucleotide sites have mutated more than once (Galtier et al. 2009). Given the high mutation rate of mitochondrial DNA and the low recombination rate, these sequences may have been produced via back mutation. These ambiguities were not present in a network constructed of only cytochrome B fragments (Figure 7b). A fragment of the ND2 mtDNA gene was amplified for one sample from each node involved in the network circles in an attempt to obtain additional sequence data to resolve the circular connections, however, all ND2 sequences examined were identical.
Demographic history
Mismatch distribution analysis of all samples, either pooled or for each population analyzed separately, showed distributions that more closely matched that of an expanding population (Figure 8), although peaks were bimodal rather than unimodal. When breeding
43 populations were examined separately, Iceland and Norway samples both showed a unimodal distribution that more closely matched the expected curve of an expanding population. In both scenarios, the northern Nunavut group showed a distinct bimodal distribution, which was more pronounced in scenario B. The southern Nunavut group in scenario B was similar to the distribution of all pooled samples, while the five southern
Nunavut samples in Scenario A gave a very ragged distribution. The raggedness statistic was significantly close to zero for northern Nunavut in scenario A and southern Nunavut in scenario B, and was also significant for the pooled sample group.
Tajima's D was significantly lower than zero for the population as a whole (Table 4), as well as for Norway, indicating an excess of rare nucleotide substitutions compared to what is expected under a neutral model of evolution. Fu's Fs was significantly lower than zero for the population as a whole, Norway, and northern Nunavut in scenario A, indicating an excess of rare haplotypes. Ramos-Onsin's R2 was significantly close to zero in the population as a whole, Norway, and northern Nunavut in scenario A and northern
Nunavut in scenario B, indicating an excess of singleton mutations compared to overall number of nucleotide substitutions. Iceland and southern Nunavut in both scenarios showed no significant deviations from neutrality. Given the relative power of Fu's Fs over
Tajima's D (Fu 1997; Ramos-Onsins and Rozas 2002) and R2's relative power over both at small samples sizes, the hypothesis of neutral evolution was strongly and significantly rejected for Purple Sandpipers as a whole, as well as the Norway breeding population in particular. Less strongly, but still significantly, the northern Nunavut breeding population
44 also showed evidence of departure from neutral evolution in both scenarios. For scenario
B, the τ value indicated a very recent expansion; the τ value for the same group in scenario A was three to four times as large, while the τ value for Norway was half again as large as for scenario A northern Nunavut.
Discussion
Population structure of Purple Sandpipers, as revealed by control region and cytochrome b sequences, indicated low genetic diversity throughout the sampled breeding range, low
(but significant) genetic structure, and a largely star-like haplotype network with few substitutions between haplotypes. Neutrality indices of the population as a whole also indicated recent expansion, which together suggest recovery from a recent, severe bottleneck (Avise 2000; Ramírez-Soriano et al. 2008). Northern and southern groups of
Nunavut breeding samples showed no genetic differentiation for either scenario examined, indicating there may be mixing despite observed morphological differences.
Haplotypes unique to wintering birds largely followed the pattern of differentiation found in breeding birds, suggesting that unsampled breeding locations, such as Russia, mainland
Norway and the migratory population in Greenland, likely do not contain a highly differentiated lineage. It is worth noting, however, that there is likely a permanent population of Purple Sandpipers in southern Greenland that does not migrate out of the country (Payne and Pierce 2002). Purple Sandpipers from this population would not have been among the wintering birds sampled in this study (Payne and Pierce 2002).
45 Diversity
While morphological variation can be an indicator of underlying genetic differentiation, many studies have found that subspecies defined by non-molecular traits such as morphology and plumage often do not show corresponding genetic differentiation (e.g.,
(Ball and Avise 1992; Zink et al. 2000; Zink 2004; Miller et al. 2013), likely due to selective pressures on phenotypic traits that have caused them to diverge faster than the observed neutral genetic markers (Haig et al. 2011). In other species, however, some or all subspecies that were originally proposed based on morphology were also confirmed genetically. A well known study on Dunlins uncovered the presence of five distinct genetic lineages that corresponded to all but three accepted subspecies that were recognized at the time (Wenink et al. 1996), though up to 11 subspecies have been described for this species at various points in time and these are not as well-defined genetically (Marthinsen et al. 2007). A similar study on Rock Sandpipers found that patterns of mtDNA differentiation corresponded to one subspecies, but the other three were not as well defined (Pruett and Winker 2005). A study that used AFLP markers in
Redshanks (Tringa totanus) found genetically distinct clusters that corresponded to three subspecies, while mtDNA control region sequences differentiated only one (Ottvall et al.
2005). Geographic structure that corresponds to known subspecies has also been found in species with relatively low diversity, such as Black-Tailed Godwits (Höglund et al. 2009).
46 This lack of genetic diversity within subspecies is especially common in species that breed in the high arctic, which typically display lower overall genetic variability, often attributed to recent bottlenecks in the Pleistocene glaciations (Piersma 2003). Populations that have expanded after a population decline, as happens when recovering from a bottleneck, tend to display high haplotype diversity and low nucleotide diversity
(Lounsberry et al. 2013). The very low nucleotide diversity values seen throughout the
Purple Sandpiper range, but especially in breeding locations, do not seem to coincide with very low haplotype diversity, consistent with an interpretation that Purple Sandpipers populations have recently expanded.
Nucleotide diversity values for breeding locations roughly coincided with values obtained from the ND2 gene (Barisas et al. 2015). Overall, values for individual locations were less than a quarter of the observed nucleotide diversity observed in individual Dunlin lineages.
Nucleotide diversity of all pooled Purple Sandpiper samples (0.00157) was two thirds of the lowest-diversity Dunlin lineage (Alaskan; 0.0024) and three quarters of the average nucleotide diversity of Red Knots (Buehler et al. 2006). The latter's diversity levels are theorized to have come from a single recent expansion refugium after the Wisconsinan glaciation in North America.
Neutrality indices showed mixed results for individual populations. Svalbard showed strong evidence of either recent population expansion, recovery after a strong bottleneck
47 or genetic hitchhiking, as Tajima's D and Fu's Fs was significantly negative and R2 was significantly close to zero. The northern Nunavut group also showed some signs of the same, and the southern Nunavut group in scenario B fit a model of expansion via mismatch distribution. Iceland did not deviate from a neutral population model in any of the statistics examined, and neither did southern Nunavut in scenario A. When all samples, including those taken from wintering grounds, were pooled, all statistics showed significant deviation from neutrality. Positive raggedness indices for expansion in
Canadian breeding populations and in samples as a whole were obtained despite a bimodal distribution not often attributed to expanding populations.
Interpreting deviations from neutrality can be a difficult undertaking, as many demographic and selective processes leave similar marks on an organism's genome. The neutrality tests in this study that deviated from neutrality all showed a genetic signature that could coincide with expansion, recovery from a strong bottleneck, or genetic hitchhiking from a selective sweep (Ramírez-Soriano et al. 2008). A common way of determining whether the latter is the case is to look at multiple recombining loci.
Demographic events such as bottlenecks and expansions should affect the entire genome, while genetic hitchhiking will generally only affect a small area (Galtier et al. 2000). A recent study by Barisas et al. (2015) looked at multiple markers in both mtDNA and nuclear DNA of Purple Sandpipers. The mtDNA ND2 gene examined in that study showed a similar pattern to the combined control region and cytochrome b fragments
48 here, with signs of expansion in Svalbard and the northern Canadian samples. In the nuclear markers, the northern Canada population showed a significantly negative Tajima's
D in the nuclear HMG-2 marker, however, all other markers and populations, and when all populations were considered, did not show signs of deviation from neutrality (Barisas et al. 2015), indicating that Purple Sandpiper mtDNA may have undergone some sort of purifying selection. While mtDNA is often considered a neutral genetic marker, it has recently become clear that this is not always the case (Dowling et al. 2008), particularly for migratory birds with intense metabolic demands (Toews et al. 2014).
Phylogenetic Trees and Haplotype Network
Phylogenetic trees showed poor resolution, as only one clade was present in all trees and bootstrap/PP values were relatively low. Branch lengths were quite short, indicating recent population divergence. While no breeding population grouped into its own clade, haplotypes were rarely shared between populations, except for the most common haplotype present in every location.
The haplotype network constructed with both mtDNA fragments had a similar structure to that of Red Knots, a long-distance migrant shorebird that breeds in the high arctic
(Buehler and Baker 2005). The haplotype network was largely star-like in appearance, with a single common, widespread haplotype in the centre and all other haplotypes differing by only a few base-pairs, which is characteristic of an expanding population or
49 recent selective sweep. Networks that were recently constructed from other mtDNA and nuclear markers of Purple Sandpipers (Barisas et al. 2015) also show a marked star-like shape.
The restriction of haplotypes from clade B to Canadian breeding populations is worth noting and may imply that if there is gene flow between Canadian breeding populations and those in Europe, it is limited. Supporting this, most Iceland and Svalbard haplotypes were not found in wintering birds, and were private to their respective breeding location.
Population differentiation as measured by ΦST
In contrast to the largely star-like haplotype network, overall population differentiation was significantly greater than zero in both scenarios, and pairwise ΦST values indicated
Iceland was either significantly different than all other breeding populations or showed a trend towards differentiation (p-value <= 0.06). Results of the ΦST analysis are consistent with differentiation between Norway and the southern Nunavut group in both scenarios, and no signs of differentiation between the two Canadian breeding groups. This agrees with a recent examination of Purple Sandpiper mitochondrial and nuclear sequences that found evidence for Icelandic Purple Sandpipers, which make up the proposed Calidris maritima littoralis subspecies, being differentiated from other population groups when all loci were analyzed together. Most individual loci, however, did not show this differentiation on their own (Barisas et al. 2015). That study also found greater
50 differentiation between Iceland and Norway than between Iceland and northern Nunavut, a pattern that I observed with mtDNA loci in my study. Unlike the Icelandic subspecies, which has limited genetic agreement, the proposed subspecies Calidris maritima belcheri, found in south and southeastern Hudson Bay, showed no evidence of differentiation from the northern Canada breeding populations. Despite this, the southern Canadian population was often more differentiated from European breeding locations than the northern population, especially in regards to Svalbard. Further sampling of birds in southern
Hudson Bay may help clarify this apparent inconsistency.
The large ΦST values found between sampled Iceland and Norway populations and the
North American wintering locations indicate that few or no birds from these populations winter in these areas. Particularly notable was the high differentiation seen between western Newfoundland and Iceland, as band recoveries have previously linked these two locations (Hallgrimsson et al. 2012). The low ΦST values seen between the United
Kingdom and all sampled North American locations lend genetic support to the recently confirmed connection between Canadian breeding grounds and the United Kingdom
(Summers et al. 2014).
Low FST values between Purple Sandpipers from the United Kingdom and
Canada/Svalbard support evidence of a migration route between Canada and Scotland
(Summers et al. 2014) and Svalbard and Scotland (Summers et al. 2010), though wintering Purple Sandpipers in Scotland are also known to come from breeding
51 populations in southern Norway (Hake et al. 1997). Genetic similarity of southern
Norway breeders to Svalbard breeders has not been examined. If they are similar, FST values between the United Kingdom and Svalbard could reflect the presence of southern
Norwegian breeders rather than Svalbard breeders. Barisas et al. (2015) suggested the possibility of genetic differentiation between Canadian breeders that winter in the United
Kingdom, and those that winter along the North American coast. While the United
Kingdom samples did not have morphological data to determine breeding origin and genetic differentiation between Canada and Svalbard was not sufficient to identify the
Canadian breeders, the major clades found within Canadian breeding populations did not correspond to any wintering area in particular, which suggests that this is not the case.
Summary
Low genetic diversity, short branches, a shallow haplotype network, and lack of reciprocal monophyly together from analysis of Purple Sandpiper mtDNA and nuDNA suggest recent divergence from a common ancestor, possibly from a common refugium.
In contrast, mismatch distributions in Canadian breeding populations and in samples as a whole exhibited a bimodal distribution, reflecting a relatively large, slightly divergent cluster of haplotypes (clade B). Neutrality tests showed evidence of expansion or purifying selection in Svalbard and northern Canadian Purple Sandpiper populations, and the lack of corroboration of this in a recent study indicates selection may be involved.
Despite the low level of observed diversity, ΦST values indicated significant genetic
52 structure between breeding populations. High ΦSTvalues suggested that gene flow between populations was not significant, and the lack of well-supported clades may be the result of low diversity levels and incomplete lineage sorting from a very recent divergence. While there was some evidence for differentiation of Icelandic Purple
Sandpipers, the presence of shared ancestral haplotypes makes it less credible that it should be considered as a potential subspecies and prevents the diagnosis of an Icelandic breeding origin from unknown wintering samples. There was no support for recognition of birds from southern Hudson Bay as a distinct subspecies.
53 Tables
Table 1. List of primers used to amplify and sequence mtDNA fragments of the control-region and cytochrome b of Purple Sandpipers (Calidris maritima). Primers are numbered according to their approximate location in the mtDNA genome. Primers marked with an asterisk were used to amplify Rock Sandpiper (Calidris ptilocnemis) samples.
______Primer Name Sequence (5' – 3') Source Control Region Primers L98* GCATGTAATTTGGGCATTTTTTG Wenink et al. 1993 H381 AACCTGGTACGACTGGTGTG LeBlanc 2013 L335 ACAGCTCGGAAACTCTCGAA In Lab H772* AAACACTTGAAACCGTCTCAT Wenink et al. 1993 L725* GCCCTCAGGCGTTACTGA LeBlanc 2013 H1018 GTTCATCTATTCGTTTATGGTT LeBlanc 2013 H1030* CGAATAGATGAACGCAAACG LeBlanc 2013 Cytochrome B Primers L15350 TTACAAACCTATTCTCAGG Pruett and Winker 2005 H16064 CTTCAGTTTTTGGTTTACAAGACC Pruett and Winker 2005 H15713 TGGGGAGGTGTGACTAGAGG LeBlanc 2013 L15641 ACCCCAGCAAACCCTCTAGT LeBlanc 2013
54
5 5
B A S. Nunavut N. Nunavut S. Nunavut N. Nunavut Svalbard Iceland United Kingdom Newfoundland-W Newfoundland Nova Scotia BrunswickNew Maine ______areindicated bold. Significant (P< in pvalues 0.05) diagonal arebelow. values and Table 2. Pairwise estimatesof Φ Pairwise 0.086 0.157 0.937 0.019 0.000 0.002 0.001 0.063 0.171 0.108 0.475 * Maine
ST for breeding and wintering populations of wintering forpopulations Purple Sandpipers and ( breeding 0.383 0.318 0.999 0.098 0.002 0.007 0.058 0.252 0.766 0.562 * -0.007 NB
0.721 0.899 0.904 0.564 0.020 0.026 0.425 0.262 0.426 * -0.012 0.020 NS
0.602 0.237 0.887 0.138 0.001 0.009 0.154 0.529 * 0.005 -0.033 Nfl 0.026
0.517 0.129 0.364 0.313 0.116 0.034 0.469 * -0.022 0.040 0.044 0.159 West Nfl
0.766 0.719 0.763 0.720 0.097 0.014 * -0.012 0.024 -0.002 UK 0.037 0.087
0.017 0.061 0.057 0.007 0.006 * 0.055 0.292 0.135 0.069 0.122 0.136 Iceland Calidris maritima 0.048 0.100 0.018 0.210 * 0.102 0.069 0.149 0.180 Nun. Svalbard N. 0.123 0.020 0.149
0.675 0.490 * 0.017 0.101 -0.020 0.008 0.038 -0.012 0.093
0.048
Scenario A Scenario ). ValuesΦ ). of
0.018 0.001 0.107 -0.102 -0.076 -0.122 -0.088 S. Nun. * 0.192 0.173
ST
are above the areabove Scenario B Scenario -0.019 0.044 -0.020 0.097 0.016 -0.041 0.038 N. Nun. 0.094 0.004
-0.022 -0.032 -0.023 -0.030 -0.003 0.055 S. Nun. 0.064 0.121 Table 3. Standard diversity estimates for breeding and wintering populations of Purple
Sandpipers (Calidris maritima), as well as all samples pooled together. Breeding populations are
Iceland, Norway, northern Nunavut, and southern Nunavut. Five samples from northern Hudson bay are grouped with northern Nunavut in scenario A, and with southern Nunavut in scenario B. P values indicated in brackets.
______
N H Pr S π ĥ
Iceland 18 4 2 4 0.00063 (0.00014) 0.660 (0.078) Norway 20 9 5 9 0.00081 (0.00016) 0.789 (0.086) Scenario A N. Nunavut 15 8 2 11 0.00133 (0.00038) 0.733 (0.124) S. Nunavut 5 3 0 5 0.00169 (0.00043) 0.800 (0.164) Scenario B N. Nunavut 10 4 1 6 0.00109 (0.00044) 0.533 (0.180) S. Nunavut 10 7 1 9 0.00178 (0.00034) 0.911 (0.077) Breeding Total 58 20 - 20 0.00102 (0.00014) 0.782 (0.049) Maine 115 19 5 21 0.00156 (0.00008) 0.797 (0.025) New Brunswick 21 11 0 13 0.00187 (0.00016) 0.914 (0.038) Nova Scotia 28 13 0 17 0.00160 (0.00023) 0.812 (0.072) Newfoundland 16 10 2 11 0.00173 (0.00028) 0.905 (0.054) United Kingdom 42 23 7 24 0.00170 (0.00020) 0.920 (0.032) Wintering Total 221 33 - 33 0.00167 (0.00007) 0.854 (0.016) Grand total 279 42 - 41 0.00157 (0.00006) 0.853 (0.016)
N = sample size, H = number of haplotypes, Pr = number of private haplotypes, S = number of segregating sites. Nucleotide diversity (π), and haplotype diversity (ĥ) are shown with standard deviations in parentheses.
56 Table 4. Demographic statistics for breeding populations of Purple Sandpipers (Calidris maritima), as well as all samples (breeding and wintering) pooled together. P-values indicated in brackets, obtained through coalescent simulations, 10,000 replicates. Significant values (p < 0.05) indicated in bold.
______Tajima's Fu's D Fs R2 r τ
Iceland -0.521 (0.340) -0.215 (0.411) 0.147 (0.434) 0.096 (0.147) 0.961 Norway -1.748 (0.013) -5.531 (0.000) 0.072 (0.000) 0.115 (0.273) 1.249 Scenario A N. Nunavut -1.532 (0.051) -3.010 (0.020) 0.072 (0.000) 0.039 (0.041) 0.849 S. Nunavut 0.562 (0.695) 1.090 (0.724) 0.241 (0.376) 0.36 (0.646) 2.023 Scenario B N. Nunavut -0.886 (0.219) 0.195 (0.525) 0.1172 (0.013) 0.285 (0.727) 0.265 S. Nunavut -0.617 (0.293) -2.314 (0.053) 0.129 (0.077) 0.0415 (0.038) 2.7 Total -1.804 (0.008) -32.837 (0.00) 0.0288 (0.036) 0.0181 (0.016) 1.829
Tajima’s D values, Fu’s Fs, Ramos-Onsins and Rozas’s R2, raggedness index (r) and τ are represented with their statistical significance.
57 Figures
Figure 1. Range map of Purple Sandpipers. Shaded areas represent breeding (yellow), wintering
(blue) and year-round (green) populations. Squares represent sampling locations. Dotted lines represent the sampled north American locations of putative subspecies Calidris maritima maritima (in the north) and Calidris maritima belcheri (in the south), as proposed by Engelmoer and Roselaar (1998). The area in between contains breeding populations whose morphometric characteristics have not been examined.
58 Figure 2. Schematic diagram showing approximate annealing locations of primers used to amplify fragments of the control-region (top) and cytochrome b (bottom) genes of Purple
Sandpipers. Locations are mapped onto an entire-gene template of a Turnstone taken from
Genbank. Primers are numbered and labelled with F or R to indicate direction, as well as their approximate locations within each gene. Primer 3-1R was only used on Rock Sandpiper samples.
Primer pairs were as follows: control-region: 1F-1R, 2F-2R, 3F-3R, 3F-3-1R, 1F-2R, 2F-3R, 1F-
3R, 4F-4R, 5F-5R, 4F-5R.
59 Figure 3. Maps of different breeding population groupings.
60 Figure 4. Rooted bayesian phylogenetic tree (left, root not shown), variable site matrix (centre) and frequency of haplotypes across populations and in total (right). Tree was constructed using
HKY model of nucleotide evolution, with invariant sites, and a run length of 1,000,000. Site matrix shows variable positions relative to haplotype 1. Breeding populations and European wintering population frequencies are highlighted in grey.
61 Figure 5. Rooted maximum parsimony phylogenetic tree (left, root not shown), variable site matrix (centre) and frequency of haplotypes across populations and in total (right). Tree was constructed using the Subtree-Pruning-Regrafting (SPR) search method and pairwise deletion, and bootstrapped 1000 times. Breeding populations and European wintering population frequencies are highlighted in grey.
62 Figure 6. Rooted maximum likelihood phylogenetic tree (left, root not shown), variable site matrix (centre) and frequency of haplotypes across populations and in total (right). Tree was constructed using the Hasegawa-Kishino-Yano model of nucleotide substitution with invariant sites, pairwise deletion and the Nearest-Neighbor-Interchange Heuristic Method, and bootstrapped
1000 times. Breeding populations and European wintering population frequencies are highlighted in grey.
63 Figure 7. Unrooted statistical parsimony haplotype network of a) concatenated mitochondrial control region and cytochrome b fragments, totalling 1534 base-pairs and b) cytochrome b fragments only, taken from 279 Purple Sandpipers. Nodes indicate individual haplotypes, and lines indicate 1 base-pair difference.
Black squares indicate missing intermediate haplotypes. The size of a node roughly corresponds to the number of individuals with that haplotype, and the population origin of those individuals are marked by colour, according to the legend. Wintering locations are shown in white and designated with numbers as follows: 1) Newfoundland, 2) New Brunswick, 3) Nova Scotia, 4) Maine, 5) United Kingdom. Breeding locations are shown in colour.
64 Figure 8. Mismatch distribution of concatenated mitochondrial control region and cytochrome b fragments, totalling 1534 base pairs, of 279 Purple Sandpipers (Calidris maritima). Observed frequencies of pairwise distances were plotted against expected frequencies given a neutral model of evolution and either population expansion or stable population size.
65 66 CHAPTER 3
Genetic differentiation and phylogeography of Purple Sandpipers using nuclear-encoded microsatellite markers.
Introduction
The status of shorebird populations in North America is of some concern, with many species showing evidence of consistent decline in population size in recent years (Andres et al. 2012). This is particularly evident in Atlantic Canada and the northeastern United
States, where a recent survey showed that 22 of 30 surveyed species of shorebirds had declining populations (Bart et al. 2007). Rockweed (Ascophyllum nodosum) harvest has led to concern over the status of Purple Sandpipers, Calidris maritima, a migratory shorebird that winters in Rockweed habitat in Maine and Atlantic Canada (Seeley and
Schlesinger 2012) and has shown declines in several wintering populations (Summers et al. 2001; Burton et al. 2008). Purple Sandpipers breed from the high Canadian Arctic to the southern tip of Hudson Bay, as well as in Greenland, Iceland, Norway, Scandinavia, and the Severnaya Islands in Russia (Summers et al. 2014). During the winter they migrate south along the northeastern coast of North America, southern Greenland,
Iceland, and along the coast of central and northern Europe; their wintering range is farther north than that of any other shorebird (Payne and Pierce 2002). Migratory routes
67 between North American breeding and wintering sites of Purple Sandpipers are largely unknown, though a recent study confirmed a trans-Atlantic route from the Canadian
Arctic in the summer to the United Kingdom in the winter (Summers et al. 2014).
Conservation efforts on behalf of Purple Sandpipers in North America would benefit from more thorough knowledge of migratory routes and population connectivity throughout the species' range, allowing conservation efforts to be focused on locations and populations that are contributing the most individuals to help bolster a species in decline (Webster et al. 2002; Deane et al. 2013; Lounsberry et al. 2013).
Efforts to identify migration routes of Purple Sandpipers via banding studies have met with limited success (Mittelhauser et al. 2006; Summers et al. 2010). Purple Sandpipers exhibit morphological differences among some breeding locations (Engelmoer and
Roselaar 1998), which have also been used to estimate breeding origin of wintering populations (Mittelhauser et al. 2006; Hallgrimsson et al. 2012). However, this approach is complicated by overlap of these measurements among several populations, and the presence of sexual dimorphism of the same magnitude as that found between some populations (Burton and Evans 2001; Barisas et al. 2015).
68 Identification of breeding origin and migration patterns of shorebirds
An additional way to assess connectivity and recent demographic events among populations is by using genetic markers. While mtDNA sequences have long been a staple of phylogeographic studies, due to the maternal inheritance pattern and high mutation rate of this haploid genome (Galtier et al. 2009), it has become increasingly standard to supplement mtDNA data with several nuclear markers, to correct for possible discordance between mtDNA and nuclear DNA due to the effects of selection, sex-biased dispersal, hybridization or biased introgression (Ballard and Whitlock 2004; McGuire et al. 2007;
Toews et al. 2013). Fast-evolving and sensitive markers such as microsatellites and single nucleotide polymorphisms (SNPs) can reveal subtle genetic structure among populations
(Defaveri et al. 2013) and detect possible discordance between nuclear and mitochondrial genomes (Toews and Brelsford 2012). Microsatellites have been used for this purpose in a number of studies of migratory birds (Pearce et al. 2004; Sonsthagen et al. 2011;
Lounsberry et al. 2013; Miller et al. 2013).
Previous studies have found strong genetic structure in the closely related Dunlin
(Calidris alpina) species (Wenink et al. 1996) and some genetic structure in Rock
Sandpipers (Calidris ptilocnemis; Pruett and Winker 2005). Notably, both mtDNA and microsatellites have been used to determine breeding origin of wintering populations of
Dunlins on a fine scale (Wennerberg 2001). In contrast, Semipalmated Sandpipers
(Calidris pusilla) show very little genetic structure in both mtDNA and neutral nuclear
69 markers (Miller et al. 2013). Purple Sandpipers show high fidelity to wintering locations
(Dierschke 1998; Mittelhauser et al. 2012; Summers et al. 2012c). Fidelity of adult birds to breeding locations is less well known, however, the return rate of adult birds in a northern Scotland population was found to be 64.7% (Smith and Summers 2005), and in
Svalbard, Norway, return rate was 54% for males and 60% for females (Payne and Pierce
2002). These rates are fairly typical of monogamous Calidris sandpipers, and they lie between the return rates of Semipalmated Sandpipers (58% and 49% breeding fidelity return rate), and the higher-fidelity Rock Sandpipers (75% breeding fidelity) and Dunlins
(77% and 71% breeding fidelity; Oring and Lank 1984; Gill et al. 2002). Natal philopatry, or proportion of newly hatched birds that return to their birth area after their first migration, was measured once in a small population of breeders in northern Scotland and was found to be 4.5% (Smith and Summers 2005), similar to observed values of
Semipalmated Sandpipers (3.1%) and Dunlins (3.6% - 11.2%; Oring and Lank 1984).
These species are all considered to have low dispersal among breeding populations
(Wenink et al. 1996; Pruett and Winker 2005; Miller et al. 2013), and it is likely that
Purple Sandpipers follow a similar pattern. In this chapter, I investigated genetic diversity and structure of a subsection of the Purple Sandpiper range using 10 microsatellite markers chosen for polymorphism between breeding populations. I found low levels of structure among Iceland, Svalbard, and Canadian breeding locations, particularly Iceland, and no signs of a recent bottleneck.
70 Methods
Samples
Purple Sandpiper blood, tissue, and feather samples were collected from Maine (n=117),
New Brunswick (n=21), Nova Scotia (n=25), Newfoundland (n=16), United Kingdom
(n=42), Nunavut (n=20), Iceland (n=18) and Norway (n=20). Three Rock Sandpiper samples were used as an outgroup, as described in chapter two of this thesis. Details of
DNA extraction are also outlined in the previous chapter.
As described in Chapter 2, samples were analyzed under two scenarios regarding the placement of five Canadian breeding samples. In Scenario A, these five samples were grouped with the northern Canadian samples. In Scenario B, these five samples are grouped with the southern Hudson Bay samples.
Molecular Techniques
Primers were developed for 10 microsatellite loci through Ecogenics GmbH (Schlieren,
Switzerland). Magnetic streptavidin beads and biotin-labeled CT and GT repeat oligonucleotides were used to create a library of size selected fragments of genomic
DNA, enriched for single sequence repeat (SSR) content. These fragments were then analyzed using a Roche 454 platform using the GS FLX Titanium reagents. Of the 3,627 fragments read, 304 were found to have a dinucleotide of at least 10 repeat units or a tetra- or trinucleotide repeat of at least six repeat units. A number of primers were tested
71 and 10 were selected based on high level of polymorphism across samples from a subset of breeding locations in North America, Iceland and Svalbard. Primers were designed with a universal 18 base pair M13 tail, and amplified with an M13 primer labelled with
FAM fluorescent dye.
The 10 microsatellite loci were amplified in the following duplex combinations: 2668 and
2988, 705 and 997, 296 and 3007, 1669 and 2198. Two singleplex reactions were also performed for loci 1422 and 3547. All PCR reactions were run in a Biometra T-Gradient thermocycler using the following protocol: 94°C for 15 min; 26 cycles of 95°C for 30 sec,
56°C for 45 sec, 72°C for 45 sec; eight cycles of 95°C for 30 sec, 53°C for 45 sec, 72°C for 45 sec; 72°C for 5 min.
Both singleplex and duplex reactions were performed in 20 μL reactions using Platinum®
Taq polymerase (Invitrogen, Carlsbad, CA) and associated reagents (2 μL 10x buffer, 0.4
μL 10 mM dNTPs, 0.1 μL Taq, 0.8 μL 50 mM MgCl2, 2 μL template DNA). Singleplex reactions used 2μM primer concentrations in the following volumes: 0.4 μL forward primer, 1.6 μL reverse primer, 1.6 μL M13 primer, with 11.1 μL of water to bring the total volume to 20 μL. Duplex reactions used 2 μM primer concentrations in the following volumes: 0.26 μL of the smaller fragment's forward primer, 0.2 μL of the larger fragment's forward primer, 1.0 μL of the smaller fragment's forward primer, 0.8 μL of the
72 larger fragment's forward primer, 1.5 μL of the M13 primer, and 7.94 μL water to bring the volume up to 20 μL.
Due to limited quantity of DNA, the Nunavut samples were pre-amplified (Arandjelovic et al. 2009) in a 10 μL reaction with all ten primer pairs, using Platinum® Taq polymerase and associated reagents (2 μL 10x buffer, 0.22 μL 10 mM dNTPs, 0.3 μL each of 10 μM forward primers, 0.3 μL each of 10 μM reverse primers, 0.1 μL Taq, 0.93
μL MgCl2, 5 μL template DNA, and 5.75 μL water). The product was then diluted 1:100 in water and 5 μL was used as template for 10 singleplex reactions (2 μL 10x buffer, 0.22
μL 10mM dNTPs, 0.63 μL of 2 μM forward primer, 2.5 μL of 2 μM reverse primer, 2.5
μL of 2 μM M13 primer, 0.07 μL Taq, 0.47 μL MgCl2, 5 μL template DNA, and 6.61 μL water). This technique has been shown to be as effective (DE Barba and Waits 2010) or more effective (Hedmark and Ellegren 2005) than traditional one-step amplification for low quality samples and significantly reduces the amount of template DNA required to genotype multiple loci.
Error Rates
Amplification success and error rates of the tissue types used to isolate DNA were compared. Amplification success was calculated as the proportion of scoreable genotypes obtained on the initial PCR performed for each sample at each locus, and was lowest for
DNA isolated from feathers and highest for DNA isolated from blood (Table 1). A subset
73 of 21% of samples were reamplified to check accuracy of genotypes (see Table 1 for percentages). Error rates per loci averaged 6% and ranged from 0% to 20% (Table 1) and were highest for DNA isolated from feathers.
Size shifting
Microsatellites were amplified and genotyped over a period of 12 months. A subsection of samples was re-amplified and compared to previous genotypes to calculate percent error rate. Five of the 10 loci (296, 2668, 705, 997, 3547) showed signs of 'size shifting' in one or more independent run, that is, all genotypes for that loci during that run were 'shifted' one or two base-pairs to the left or right compared to other runs, as revealed by re- amplified samples. For one base-pair shifts, direction of a shift was determined by selecting the majority of runs that agreed with each other and comparing re-amplified samples to these runs. For two base-pair shifts, samples that had been reamplified three times were examined and the two runs that agreed with each other were used as reference to determine the direction of shifts. Microsatellites were binned using tandem v. 1.09
(Matschiner and Salzburger 2009) and all file conversions from one input format to another were done using PGDSpider v. 2.0.8.0 (Lischer and Excoffier 2012).
Microsatellite loci are known for often having relatively high frequencies of null alleles, which are alleles that fail to amplify due to mutations in the primer annealing sites
(Oddou-Muratorio et al. 2009). The presence of null alleles falsely inflates the proportion
74 of homozygous individuals, as individuals who are heterozygous for a null allele would be genotyped as homozygotes. An overrepresentation of homozygous genotypes can lead to significant biases in population differentiation tests when null allele frequencies are
>20%. (Oddou-Muratorio et al. 2009). Null alleles are commonly detected by identifying apparent departures from Hardy-Weinberg equilibrium due to heterozygote deficit. This method can be confounded by demographic processes that result in a similar pattern, such as the Wahlund Effect (i.e. a collection of samples that come from a subdivided population will show an excess of homozygotes, corresponding with the level of differentiation between subdivisions; Dharmarajan et al. 2013). Null alleles were detected using samples from breeding locations due to the higher likelihood of mixed wintering locations deviating from Hardy-Weinberg expected values due to demographic processes and subdivision.
The presence of null alleles was tested for indirectly using two programs, ML-NullFreq
(Kalinowski and Taper 2006) and Micro-checker v. 2.2.3 (Van Oosterhout et al. 2004), as recommended in Dąbrowski et al. (2014). Indirect tests of null alleles have a high probability of giving false positives, however, false positive results tend to be inconsistent between programs and datasets (Dąbrowski et al. 2014). By using only null alleles detected by both programs, the probability of false positives is significantly decreased while still retaining power to detect true null alleles (Dąbrowski et al. 2014).
75 Pairwise linkage disequilibrium values were calculated in FSTAT v. 2.9.3.2 (Goudet
1995). Observed and expected heterozygosities were calculated in Arlequin v. 3.5.1.3
(Excoffier and Lischer 2010), and allelic richness was calculated using FSTAT to determine relative diversity levels of all populations.
There has been considerable debate over the utility of various descriptive statistics (e.g.,
FST, Jost's D, RST, etc) for determining population structure using microsatellites, as described in chapter 1 of this thesis. FST, one of the most widely used measures of population structure, was calculated using Arlequin. Pairwise FST was calculated using all sampled populations, and overall FST was calculated using only breeding populations.
Jost's D (Jost 2008) values were calculated using GenoDive v. 2.0b23 (Meirmans and Van
Tienderen 2004) and used for comparison purposes. As with FST, Pairwise Jost's D was calculated using all sampled populations, and overall Jost's D was calculated using only breeding populations.
A Bayesian clustering method was performed using STRUCTURE v. 2.3.3 (Pritchard et al. 2000). This analysis was performed on all samples without defining putative populations, as well as on only the breeding samples, both with and without putative populations used as a prior. Analyses were done using the admixture method for K = 1 to
5, where K is the putative number of clusters, and run for 2,000,000 cycles with a burn-in of 200,000 cycles. Wintering locations were broken down to more localized areas to
76 detect fine-scale population structure within the Purple Sandpiper wintering distribution.
Individuals in Maine were taken from a large number of islands off the coast; islands that had only one or two samples were grouped with their closest neighbouring location.
The program BOTTLENECK (Piry et al. 1999) tests for evidence of a recent population bottleneck by comparing heterozygosity levels to overall genetic diversity and searching for an excess or deficiency in that measure. A heterozygosity excess is interpreted as a recent population bottleneck, while heterozygote deficiency indicates recent population growth (Cornuet and Luikart 1996). I ran multiple analyses using several different models: i) a strict stepwise mutational model, ii) the infinite alleles model, and iii) a more complex two-phase model that assumes mutation via both the stepwise and the infinite alleles model (Di Rienzo et al. 1994). Studies in avian microsatellite loci have found that on average a range of 60% to 80% of mutations are acquired according to the step-wise mutation model (Miller et al. 2012), and so the two-phase model was implemented using each end of this range as a parameter. I specified two-phase model variances as 4, 9, 16,
25 or 36, according to the observed allele sizes across all loci (Di Rienzo et al. 1994).
Results over all loci were analyzed with the Wilcoxon Signed-Rank test (Cornuet and
Luikart 1996).
77 Results
Genotype Quality
Deviations from linkage equilibrium were well below the expected number of false positives (22 expected, 10 observed), and none of the significant p-values fell below the adjusted cutoff using sequential Bonferroni correction (Rice 1989). Comparisons of expected and observed heterozygosity values and deviations from Hardy-Weinberg equilibrium were performed in FSTAT using only samples from breeding populations, as wintering locations are expected to hold Purple Sandpipers from multiple breeding populations and such mixing often results in deviations from Hardy-Weinberg proportions
(Dharmarajan et al. 2013). In Iceland, locus 2668 deviated significantly from Hardy-
Weinberg equilibrium, but this was the only significant result and it is within the expected number of false positives for multiple calculations of p-values at a 0.05 cutoff (two). A null allele frequency of 15% was detected in the Iceland population for locus 2668. Null allele frequencies below 20% have limited effect on general population structure analyses
(Chapuis and Estoup 2007), and when analyses were run with or without locus 2668 results did not differ. Results are therefore reported here for all 10 loci.
Allelic diversity among the 10 microsatellite DNA loci ranged from 5 to 13 alleles, with a mean of 8.5 alleles/locus (Table 1). Samples from Svalbard showed higher observed heterozygosity than other breeding and wintering locations (Table 1). Allelic richness and expected heterozygosity were similar across all locations (Table 1).
78 None of the Wilcoxon tests for heterozygosity excess were significant, and the mode shift test was significant only for southern Nunavut under scenario A. However, this population was below the minimum recommended number of samples (n=10) and thus this result should be interpreted with caution.
Analysis of population structure
FST
Overall FST values were calculated across breeding populations for both scenarios, and were similar to those found for mtDNA sequences in chapter 2 of this thesis. FST was
0.080 (p < 0.001) for scenario A and 0.074 (p < 0.001) for scenario B, indicating moderate genetic structure.
Pairwise FST values (Table 2) were significantly higher than expected under panmixia between Iceland and North American breeding populations in both scenarios (FST = 0.126 to 0.156; p < 0.001). The values between Iceland and Svalbard was lower but still significant (FST = 0.062; p < 0.001). Similarly, the FST value between Svalbard and North
American breeding populations was significantly different from zero for both scenarios
(FST = 0.034 to 0.049; p < 0.05). In both scenarios, there was no significant difference in allele frequencies between northern and southern Canadian groups (p > 0.7).
79 There were consistently significant FST values between Iceland and all wintering locations
(FST = 0.061-0.156; p < 0.001 to p = 0.022), as well as lower but significant values between Svalbard and North American wintering locations (FST=0.043-0.76; p < 0.001 to p = 0.037). FST was calculated for western and eastern Newfoundland separately to explore clustering observed in STRUCTURE results. FST values between Iceland and western Newfoundland (FST = 0.061; p = 0.022) were much lower, though still significant, than those between Iceland and eastern Newfoundland (FST = 0.146; p < 0.001). FST values were also lower between Svalbard and western Newfoundland, though not to the same degree (FST = 0.043 and 0.056; p = 0.037 and p < 0.001). In scenario A, the northern
Nunavut sample group had low but significant FST values when compared to Maine
(FST=0.037; p < 0.001) or New Brunswick (FST = 0.026; p = 0.018). In scenario B, the southern Nunavut group had low but significant FST values with Maine (FST=0.031; p =
0.003). North American breeding locations otherwise showed no significant difference in allele frequencies from any wintering population (p = 0.058-0.997). Finally, there was no significant difference between the United Kingdom wintering population and Svalbard (p
= 0.999), or between the United Kingdom and North American breeding populations in either scenario (p = 0.838-0.997). The United Kingdom wintering population was, however, significantly different from Iceland (FST = 0.072; p < 0.001)
80 Jost's D
Overall Jost's values were calculated across breeding populations for both scenarios, and were similar to those found for mtDNA sequences in chapter 2 of this thesis. Jost's D was
0.140 (p < 0.001) for scenario A and 0.137 (p < 0.001) for scenario B, indicating moderate genetic structure. Pairwise Jost's D values (Table 3) were calculated using
GenoDive and compared to pairwise FST. Pairwise Jost's D values between Iceland,
Svalbard and other locations were much higher than the respective FST values (D = 0.113 to 0.299), but otherwise showed a pattern similar to that seen for the FST values.
STRUCTURE clustering
When the STRUCTURE analysis was run using only breeding samples, both with and without a priori population information, the most likely number of clusters was two
(Figure 1). Sixty-four percent of individuals were assigned to a cluster with greater than or equal to 80% probability (76% of non-Svalbard samples). When prior population sources were used, 57% of individuals were assigned to a cluster with greater than or equal to 80% probability (87% of non-Svalbard samples). Individuals from Canada clustered separately from individuals from Iceland, with Svalbard moderately assigned to both.
When all samples were considered without a priori population information, the most likely number of clusters was 4 (Figure 1). However, only 43 individuals (15%) were
81 assigned to a cluster with greater than or equal to 80% probability, 17 of which were birds in Iceland or Svalbard, and 127 (46%) birds were assigned to a cluster with 60% or greater probability. For all samples, Iceland consistently clustered on its own, with
Svalbard grouping weakly in the same cluster (Figure 2). Southern Hudson Bay showed weak assignment to a second cluster, largely due to the two samples from North Twin
Island near the southern tip of Hudson Bay which were assigned to the second cluster with 71% and 91% probability. The remaining three southern Hudson Bay samples were from Long Island farther north along the southeastern coast of Hudson Bay. Northern
Hudson Bay showed relatively strong assignment to a third cluster, and northern Nunavut showed weak assignment to the same. The fourth cluster was not represented in the breeding samples save for a single northern Canadian breeder from Bathurst Island, and may indicate an unsampled breeding population. Individuals in wintering populations showed moderate admixture and assignment to multiple clusters.
Discussion
Levels of genetic diversity in Purple Sandpipers
Avian species in Nearctic and Palearctic regions have lower diversity and less genetic structure on average than birds found elsewhere, likely due to the effects of past ice advances on the species' range (Phillimore and Owens 2006). The last ice advance in the northern latitude, especially in North America, was quite extensive and many species found their ranges reduced to small, isolated pockets of habitable climate called refugia
82 (e.g., Hewitt 2004). When these refugial populations expanded outwards after the ice retreated, new populations would be colonized by a few pioneer individuals and would show correspondingly reduced genetic diversity within populations (Boileau et al. 1992;
Hewitt 1996). Species such as the Ruddy Turnstone (Arenaria interpres; Wenink et al.
1994), Red Knot (Buehler and Baker 2005), Temminck’s Stint (Calidris temminckii,
Rönkä et al. 2012), and Semipalmated Sandpiper (Miller et al. 2013) all show little genetic structure and diversity and are thought to have expanded from a single refugial population after the last ice advance. Species such as Dunlins (Wenink et al. 1996) and
Rock Sandpipers (Pruett and Winker 2005) are thought to have expanded from multiple refugia after the last ice advance, resulting in greater levels of population structure and diversity.
In the present study, there were no reliable signs of a recent bottleneck in any of the
Purple Sandpiper breeding populations, where recent is defined as after 0.25-2.5 times
2Ne generations ago (Cornuet and Luikart 1996). Heterozygosity levels were moderate, and highest in Svalbard. This pattern has been observed in other nuclear markers as well
(Barisas et al. 2015), and may indicate that Icelandic and Canadian breeding populations were established more recently than the Svalbard population.
83 Genetic structure in breeding populations
FST, Jost's D and STRUCTURE all found evidence of population structure among Iceland,
Svalbard and the Canadian breeding populations. STRUCTURE found evidence of some substructure between the southern Hudson Bay breeding population and northern
Canadian populations that was not reflected in pairwise FST values. The two samples from
North Twin Island seemed to drive most of this differentiation, indicating the possibility that the Purple Sandpipers in the south and southeastern part of Hudson Bay are not a panmictic one, but sample sizes are very small and thus this interpretation must be treated cautiously. In all tests, population differentiation among breeding populations appeared to follow a gradient, where Svalbard allelic frequencies were less differentiated from both
Icelandic and Canadian populations than the Icelandic and Canadian populations were from each other.
A recent examination of the proposed subspecies of Purple Sandpipers found that when more breeding populations from the 'intermediate' subspecies C. m. maritima were included in the morphological analysis, differences between subspecies failed to satisfy
Amadon’s rule for the designation of subspecies, which specifies that 75% of individuals from a population be separable from 99% of overlapping populations (Amadon 1949;
Barisas et al. 2015). This same study showed evidence of multilocus genetic differentiation between Iceland, Svalbard and northern Canadian populations despite star-
84 like haplotype networks, though not enough to support the designation of distinct subspecies.
The pattern of pairwise FST values observed between Iceland, Svalbard and Canadian breeding populations is similar to values a recent study found between three Dunlin subspecies, Calidris alpina schinzii, C. a. centralis and the intermediate subspecies C. a. alpina (Marthinsen et al. 2007). However, in the case of Purple Sandpipers, these breeding locations do not correspond with the hypothesized subspecies as proposed by
Engelmoer and Roselaar (1998).
Genetic structure in wintering populations
Pairwise FST between Iceland/Svalbard and most North American wintering locations was large and significant, with the exception of western Newfoundland. Values between western Newfoundland and Iceland/Svalbard were much lower than values between
Iceland/Svalbard and other North American wintering populations, though still significant. Purple Sandpipers from Iceland have been hypothesized to migrate to western
Newfoundland in the winter, and these FST values support a possible connection between this wintering population and Iceland (Hallgrimsson et al. 2012). In addition, low FST values between the United Kingdom and Canada/Svalbard supports evidence of a migration route between Canada and Scotland (Summers et al. 2014) and southern
Svalbard and north-east Scotland (Summers et al. 2010). Migrants from Svalbard are
85 thought to be few in number. The majority of the European breeders that winter in
Scotland come from farther south in Norway and winter mainly in southeastern Scotland
(Hake et al. 1997). Samples from these breeding populations would help determine whether breeding populations in southern Norway are genetically similar to the Svalbard population, and thus how much the low FST values between the United Kingdom and
Svalbard reflects the presence of Svalbard breeders and how much reflects the presence of
Norwegian breeders in general. Conservation efforts should try to maximize genetic diversity in Purple Sandpipers, focusing on breeding populations that are genetically differentiated from each other such as Iceland and Canada. Furthermore, genetic assessment of wintering populations of Purple Sandpipers can shed light on areas that might warrant separate consideration, such as the wintering population of western
Newfoundland.
Summary
The gradual change in allele frequencies seen among populations from Canada, Svalbard and Iceland may indicate gene flow between Svalbard and the other two breeding populations. While Canada and Svalbard are not geographically adjacent to each other,
Purple Sandpipers are known to winter in populations that contain birds from multiple breeding locations (Corse and Summers 1999). While the majority of Svalbard breeders are thought to migrate to the Norwegian coast and western Sweden (Hake et al. 1997), small numbers have been observed wintering in north-east Scotland (Summers et al.
86 2010), near a population of relatively long-billed Purple Sandpipers now known to be from Canada (Summers et al. 2014). Gene flow between these two groups of wintering birds may account for the lower FST value between Svalbard and Canadian populations. A similar pattern of gene flow is found in birds that choose mates before migrating to their breeding grounds (e.g., Lesser Snow Geese; Quinn 1992). Natal dispersal is extremely difficult to detect through banding, however, a shared wintering ground might facilitate the migration of naive birds to a different breeding population. Wintering grounds of
Icelandic breeding Purple Sandpipers are unknown, though they are not thought to winter in Britain (Summers et al. 1988). Evidence against wintering birds migrating to a different breeding location are the differences in arrival time observed between morphologically- defined groups in Scotland. 'Long-billed' birds from Canada arrive much later
(October/November vs July) and may depart later than the 'short-billed' Norwegian population (Nicoll et al. 1988; Corse and Summers 1999). Alternatively, the pattern of differentiation could be the result of incomplete lineage sorting due to recent expansion from a single refugium after the last glaciation. The high diversity values seen in nuclear markers here and in other studies may indicate a dispersal pattern of individuals radiating outward from Svalbard to Iceland and Canada, resulting in the observed pattern of differentiation. Further sampling of additional breeding populations would shed more light on the genetic structure present in the rest of the range. The pattern of differentiation seen between breeding populations of Purple Sandpipers in this study, however, does not support the recognition of distinct subspecies within the Purple Sandpiper species.
87 Identification of a genetic marker that has undergone selection due to local adaptation, such as often seen in large SNP panels, may allow for greater discrimination between breeding populations than neutral markers.
88 Tables
Table 5. Average success and error rates of Purple Sandpiper microsatellite amplification reactions, according to the tissue from which the DNA was isolated across all loci, and also according to loci across all tissue types. Success rate was calculated using the initial PCR performed for each sample loci. Percent reamplification ratio was calculated using total successful PCR genotypes. ______No. PCR % Error PCRs success Reamplified Rate
Tissue type Museum Tissue 200 75% (150/200) 96% (191/200) 5% (9/191) Liver 250 89% (223/250) 50% (124/250) 4% (5/124) Feathers 620 54% (332/620) 23% (113/498) 13% (15/113) Blood 1749 91% (1586/1749) 9% (145/1701) 6% (37/573)
Loci 26681 282 81% (228/282) 24% (65/267) 3% (2/65) 29881 282 84% (238/282) 26% (72/272) 8% (6/72) 7051 282 70% (197/282) 15% (36/237) 6% (2/36) 9971 282 77% (217/282) 36% (94/269) 2% (2/94) 2961 282 83% (233/282) 21% (57/269) 7% (4/57) 30071 282 87% (245/282) 23% (63/273) 6% (4/63) 16691 282 57% (161/282) 13% (32/254) 9% (0/32) 21981 281 89% (249/281) 21% (56/267) 20% (11/56) 14222 282 92% (259/282) 17% (46/276) 7% (3/52) 35472 282 94% (264/282) 19% (52/273) 6% (3/52)
1 samples primarily amplified in duplexes, 2 samples amplified in singleplexes
89 Table 6. Diversity estimates for 10 microsatellite loci of breeding and wintering populations of Purple Sandpipers (Calidris maritima), as well as for all samples pooled together. Breeding populations are Iceland, Norway, northern Nunavut, and southern Nunavut. Five samples from northern Hudson bay are grouped with northern Nunavut in scenario A, and with southern Nunavut in scenario B. Standard deviation is indicated in brackets. ______
N AT AR HO HE
Iceland 18 4.6 3.38 0.60 (0.18) 0.62 (0.19)
Norway 20 5.6 3.92 0.66 (0.18) 0.68 (0.15)
Scenario A N. Nunavut 15 4.6 3.35 0.57 (0.17) 0.61 (0.12) S. Nunavut 5 3.3 3.30 0.54 (0.23) 0.60 (0.16)
Scenario B N. Nunavut 10 4.3 3.43 0.63 (0.15) 0.63 (0.11) S. Nunavut 10 4.0 3.25 0.49 (0.20) 0.60 (0.12)
Breeding Total 58 4.5 3.49 0.59 0.63 Maine 115 6.8 3.34 0.58 (0.15) 0.60 (0.12) New Brunswick 21 5.0 3.35 0.58 (0.10) 0.61 (0.11) Nova Scotia 25 5.1 3.39 0.56 (0.14) 0.60 (0.13) Newfoundland 16 4.6 3.47 0.60 (0.18) 0.63 (0.13) United Kingdom 42 5.8 3.68 0.58 (0.24) 0.64 (0.14)
Wintering Total 221 5.5 3.45 0.58 0.62
Grand total 279 5.0 3.46 0.59 0.62
N = sample size, AT = mean number of alleles, AT = mean number of alleles standardized to smallest population size (n = 5), HO = observed heterozygosity, HE = expected heterozygosity. Nucleotide diversity (π), and haplotype diversity (ĥ) are shown with standard deviations in parentheses. Breeding totals and grand totals were calculated using data for scenario A results.
90
Table 7. Pairwise estimates of FST for breeding and wintering populations of Purple Sandpipers (Calidris maritima). Values of FST are above the
diagonal and p values are below. Significant values (P < 0.05) are indicated in bold.
______
Scenario A
Scenario B
Maine NB NS Nfl West Nfl UK Iceland Svalbard N. Nun. S. Nun. N. Nun. S.
Nun.
Maine * 0.002 -0.002 0.014 0.013 -0.038 0.142 0.076 0.037 -0.004 0.013 0.031
New Brunswick 0.279 * -0.005 0.001 0.001 -0.052 0.132 0.064 0.026 -0.024 0.004 0.014
Nova Scotia 0.660 0.825 * -0.002 0.017 -0.051 0.156 0.068 0.016 -0.012 -0.003 0.011
Newfoundland 0.044 0.490 0.611 * 0.015 -0.070 0.146 0.057 -0.003 0.002 -0.014 0.000
West Newfound. 0.264 0.493 0.232 0.326 * -0.074 0.061 0.043 0.055 -0.023 0.026 0.036
United Kingdom 0.999 0.999 0.999 0.998 0.914 * 0.072 -0.046 -0.047 -0.041 -0.070 -0.041
9 Iceland 0.000 0.000 0.000 0.000 0.022 0.000 * 0.062 0.156 0.126 0.142 0.148
1
Svalbard 0.000 0.000 0.000 0.000 0.037 0.999 0.000 * 0.046 0.049 0.034 0.049
A
N. Nunavut 0.000 0.018 0.058 0.586 0.540 0.997 0.000 0.002 * 0.013 -0.011
S. Nunavut 0.579 0.925 0.812 0.507 0.789 0.838 0.000 0.020 0.327 *
N. Nunavut 0.077 0.353 0.610 0.824 0.168 0.997 0.000 0.010 B
S. Nunavut 0.003 0.185 0.209 0.577 0.219 0.972 0.000 0.002 0.790
Table 8. Pairwise estimates of Jost's D for breeding and wintering populations of Purple Sandpipers (Calidris maritima).
______
______
Scenario A
Maine NB NS Nfl West Nfl UK Iceland Svalbard N. Nun. S. Nun.
Maine *
New Brunswick 0.003 *
Nova Scotia -0.003 -0.010 *
Newfoundland 0.014 -0.007 -0.004 *
West Newfound. 0.010 -0.010 0.018 0.016 *
United Kingdom 0.052 0.041 0.039 0.015 0.004 *
Iceland 0.275 0.253 0.297 0.230 0.104 0.201 *
9
Svalbard 0.143 0.123 0.128 0.102 0.087 0.048 0.133 *
2
A N. Nunavut 0.059 0.042 0.023 0.012 0.051 0.041 0.299 0.089 *
S. Nunavut -0.015 -0.045 -0.026 -0.021 0.044 0.025 0.209 0.090 0.012 *
N. Nunavut 0.024 0.009 -0.005 -0.011 0.095 0.023 0.282 0.075 B
S. Nunavut 0.042 0.015 0.011 0.002 -0.056 0.034 0.260 0.086 -0.024
Table 9. P-values from Wilcoxon and mode-shift tests for recent bottlenecks within breeding and wintering populations of Purple Sandpipers (Calidris
maritima), under the following mutational models: Infinite alleles (IAM), stepwise (SMM), and two-phase model with several parameters (indicated in the
format V-P, where V is model variance and P is percent of mutations that follow the stepwise model).
______
IAM SMM 4-60 4-80 9-60 9-80 16-60 16-80 25-60 25-80 36-60 36-80 Mode
Shift
Nova Scotia 0.1934 0.1934 0.9219 0.4922 1.0000 0.6250 0.8457 0.7695 0.8457 0.9219 0.8457 0.9219 N
Newfoundland 0.1602 0.5567 0.8457 1.0000 0.5566 1.0000 0.5566 1.0000 0.5566 0.9219 0.5566 0.7695 N
New Brunswick 0.0244 0.0840 0.7695 0.2754 0.9219 0.4922 0.8457 0.7695 0.8457 0.8457 0.8457 0.9219 N
Maine 0.2754 0.0049 0.1934 0.0322 0.3223 0.1309 0.6250 0.1602 0.6953 0.2324 0.7695 0.2754 N
United Kingdom 0.2324 0.1602 0.8457 0.4922 1.0000 0.6953 1.0000 0.8457 1.0000 0.9219 0.9219 1.0000 N 9
3 Iceland 0.2324 0.9219 0.7695 0.8457 0.6953 0.8457 0.6250 0.7695 0.6250 0.7695 0.6250 0.7695 N Norway 0.0137 0.3750 0.5566 1.0000 0.4316 0.8457 0.3223 0.6953 0.3223 0.6953 0.2754 0.6250 N Scenario A S. Nunavut 0.6250 0.6250 1.0000 0.8457 0.92188 0.9219 0.9219 0.9219 0.9219 0.9219 0.9219 0.9219 Y* N. Nunavut 0.2754 0.0186 0.3223 0.1602 0.55664 0.2324 0.6250 0.2324 0.6250 0.3223 0.6250 0.3750 N Scenario B S. Nunavut 0.6250 0.6250 1.0000 0.8457 0.92188 0.9219 0.9219 0.9219 0.9219 0.9219 0.9219 0.9219 N N. Nunavut 0.2754 0.0186 0.3223 0.1602 0.55664 0.2324 0.6250 0.2324 0.6250 0.3223 0.6250 0.3750 N
* Population is under the minimum number of samples required for the mode-shift test. Figures
Figure 9. Mean Ln Likelihood values for number of genetic clusters inferred from A) 279 Purple Sandpipers taken from wintering and breeding locations, and B) 58 Purple Sandpipers taken from breeding locations only. Values were calculated using 10 microsatellite loci and the admixture model in STRUCTURE.
94
9 5
Figure 10. Below: Individual assignment of samples from breeding and wintering populations of Purple Sandpipers (Calidris maritima), to four genetic clusters calculated in STRUCTURE. Above: Overall assignment of each location to the four clusters. 96 CHAPTER 4
General Discussion and Conclusions
I examined mtDNA and nuDNA from samples collected in four breeding and five wintering locations to assess the level of genetic divergence in breeding and wintering populations of Purple Sandpipers, and to determine whether observed divergence correlated with putative subspecies designations in this species. I found evidence of recent divergence in both mtDNA and nuDNA markers between breeding populations of Purple
Sandpipers, however divergence was not sufficient to support the recognition of Icelandic and southern Hudson Bay Purple Sandpipers as distinct subspecies. I summarize my interpretations of the data collected below.
F-statistics performed on mitochondrial markers and nuclear microsatellite markers were largely in agreement. In particular, most values between Icelandic breeders and North
American wintering populations were similar between the two genomes. Differentiation between Svalbard breeders and North American wintering populations was greater in mtDNA, but also displayed similar patterns. The exception to this was wintering birds taken from western Newfoundland, which showed much higher FST values with Iceland in mtDNA than in nuclear DNA. These four samples had two mitochondrial haplotypes between them: one was the haplotype common to all sampled populations, and the other was a haplotype shared by other North American wintering birds but no European
97 breeders, suggesting that these birds may not be of Icelandic origin. In addition, F- statistics for both mtDNA and nuDNA did not show a significant difference between western Newfoundland and other North American wintering populations. Nevertheless, microsatellite clustering for this population was noticeably different than other North
American wintering populations, and further sampling would determine whether this difference is indeed biologically significant.
F-statistics showed no differentiation between the northern Canadian group and southern
Canadian group for both scenarios. Differentiation between these groups and wintering populations, however, changed for Maine and Western Newfoundland. In Scenario A, where northern Hudson Bay samples were grouped with northern Canadian breeders, the northern Canadian group showed higher, significant FST values with both Maine and western Newfoundland. In Scenario B, where northern Hudson Bay samples were grouped with southern Hudson Bay, it was the southern Canadian group that showed higher FST values. The difference between scenarios was greater in mtDNA than in nuclear microsatellites. This combined with differences in cluster assignment of northern
Hudson bay samples in microsatellite STRUCTURE analysis, may indicate northern
Hudson Bay breeding Purple Sandpipers are a separate population, and do not winter in
Maine or western Newfoundland.
98 MtDNA nucleotide diversity was higher in Canadian populations, likely due to the presence of haplotypes from Clade B. This is in contrast to patterns from the nuclear diversity measures presented here and in another recent study (Barisas et al. 2015) that consistently found higher diversity in Svalbard. Clade B could represent the presence of
Purple Sandpipers from two refugial populations in Canada. If this is the case, the genetic distance between the two mtDNA clusters is unusually low (compared to Dunlins,
Buehler and Baker 2005; Temmink's Stints, Sonsthagen et al. 2011; and Common Eiders,
Rönkä et al. 2012), but not unheard of (e.g., Rock Sandpipers; Pruett and Winker 2005).
Both North American and United Kingdom wintering populations were represented in
Clade B haplotypes, indicating it likely does not correspond to different migratory routes.
As intra-specific genetic analyses become more widespread, there is growing awareness that avian subspecies, which are defined largely on morphology, behavioural, or ecological traits, often do not correspond to genetically distinctive groups. Zink (2004) argued that conservation efforts should be focused on historical, genetically significant groups, measured through reciprocal monophyly in the mtDNA genome. That is, all individuals in a subspecies or conservation unit should share an ancestor that is common to them but should not include any individuals recognized as being from another subspecies or conservation unit (Zink 2004). None of the breeding populations of Purple
Sandpipers satisfied this criterion, in either mtDNA or nuDNA studied, and there was
99 thus not sufficient evidence to support recognition of the putative subspecies proposed by
Engelmoer and Roselaar (1998).
The analyses presented here do not support a scenario in which Purple Sandpipers would be considered a panmictic population. Differentiation observed between breeding populations may be the result of isolation since the last deglaciation. Wintering samples in
North America indicate the presence of a genetic population not sampled in this study, possibly from Greenland or an unsampled region of northern Canada. Sampling from additional breeding populations may provide a clearer picture of patterns of genetic structure within Purple Sandpipers. An analysis using a sufficient number of SNP-based genetic markers, which may be cost-effective in the near future, may also be able to differentiate among breeding populations by including loci that reflect adaptive selection
(Freamo et al. 2011).
100 Literature Cited
Amadon, D. 1949. The seventy-five per cent rule for subspecies. Condor 51:250–258. Andres, B. A., P. A. Smith, R. I. G. Morrison, C. L. Gratto-Trevor, S. C. Brown, and C. A. Friis. 2012. Population estimates of North American shorebirds, 2012. Wader Study Gr. Bull. 119:178–194. Arandjelovic, M., K. Guschanski, G. Schubert, T. R. Harris, O. Thalmann, H. Siedel, and L. Vigilant. 2009. Two-step multiplex polymerase chain reaction improves the speed and accuracy of genotyping using DNA from noninvasive and museum samples. Mol. Ecol. Resour. 9:28–36. Arbabi, T., J. Gonzalez, H.-H. Witt, R. Klein, and M. Wink. 2014. Mitochondrial phylogeography of the Eurasian Reed Warbler Acrocephalus scirpaceus and the first genetic record of A. s. fuscus in Central Europe. Ibis 156:799–811. Atkinson, N. K., R. W. Summers, M. Nicoll, and J. J. D. Greenwood. 1981. Population, movements and biometrics of the Purple Sandpiper Calidris maritima in eastern Scotland. Ornis Scand. 12:18–27. Austin, G. E., M. P. Collier, and M. M. Rehfisch. 2008. 2007 Non-estuarine Coastal Waterbird Survey: Population estimates and broad comparisons with previous surveys. BTO Research Report No. 501. Avise, J. C. 2000. Phylogeography: The history and formation of species. Harvard Press, Cambridge, MA. Avise, J. C. 2009. Phylogeography: retrospect and prospect. J. Biogeogr. 36:3–15. Avise, J. C., J. Arnold, R. M. Ball, E. Bermingham, T. Lamb, J. E. Neigel, C. a Reeb, and N. C. Saunders. 1987. Intraspecific phylogeography: the mitochondrial DNA bridge between population genetics and systematics. Annu. Rev. Ecol. Syst. 18:489–522. Avise, J. C., D. Walker, and G. C. Johns. 1998. Speciation durations and Pleistocene effects on vertebrate phylogeography. Proc. Biol. Sci. 265:1707–1712. Ball, R. M., and J. C. Avise. 1992. Mitochondrial DNA phylogeographic differentiation among avian populations and the evolutionary significance of subspecies. Auk 109:626–636. Ballard, J. W. O., and M. C. Whitlock. 2004. The incomplete natural history of mitochondria. Mol. Ecol. 13:729–744. Barisas, D. A. G., J. Amouret, G. T. Hallgrímsson, R. W. Summers, and S. Pálsson. 2015. A review of the subspecies status of the Icelandic Purple Sandpiper Calidris maritima littoralis. Zool. J. Linn. Soc. 1–11.
101 Bart, J., S. Brown, B. Harrington, and R. I. Guy Morrison. 2007. Survey trends of North American shorebirds: Population declines or shifting distributions? J. Avian Biol. 38:73–82. Boileau, M. G., P. D. N. Heber, and S. S. Schwartz. 1992. Non-equilibrium gene frequency divergence: persistent founder effects in natural populations. J. Evol. Biol. 5:25–39. Buehler, D. M., and A. J. Baker. 2005. Population divergence times and historical demography in Red Knots and Dunlins. Condor 107:497–513. Buehler, D. M., A. J. Baker, and T. Piersma. 2006. Reconstructing palaeoflyways of the late Pleistocene and early Holocene red knot (Calidris canutus). Ardea 94:485–498. Burton, N. H. K., J. Blew, K. Colhoun, J. Cortes, B. Deceuninck, K. Devos, F. Hortas, L. Nilsson, D. Radović, M. M. Rehfisch, M. Van Roomen, C. Soldatini, O. Thorup, and D. A. Stroud. 2008. Population status of waders wintering on Europe’s non-estuarine coasts. Eur. Non-Estuarine Coast. Waterbird Surv. Int. Wader Stud. 18 95–101. Burton, N. H. K., and P. R. Evans. 2001. Aggressive behaviours and correlates of dominance in Purple Sandpipers Calidris maritima at a communal winter roost. Ibis 143:248–254. Bush, K. L., M. D. Vinsky, C. L. Aldridge, and C. A. Paszkowski. 2005. A comparison of sample types varying in invasiveness for use in DNA sex determination in an endangered population of greater Sage-Grouse (Centrocercus uropihasianus). Conserv. Genet. 6:867–870. Chapuis, M. P., and A. Estoup. 2007. Microsatellite null alleles and estimation of population differentiation. Mol. Biol. Evol. 24:621–631. Clark, N. A., C. D. T. Minton, J. W. Fox, K. Gosbell, R. B. Lanctot, R. R. Porter, and S. Yezerinac. 2010. The use of light-level geolocators to study wader movements. Wader Study Gr. Bull. 117:173–178. Clement, M., D. Posada, and K. A. Crandall. 2000. TCS : a computer program to estimate gene genealogies. Mol. Ecol. 9:1657–1659. Corl, A., and H. Ellegren. 2013. Sampling strategies for species trees: the effects on phylogenetic inference of the number of genes, number of individuals, and whether loci are mitochondrial, sex-linked, or autosomal. Mol. Phylogenet. Evol. 67:358–66. Cornuet, J. M., and G. Luikart. 1996. Description and power analysis of two tests for detecting recent population bottlenecks from allele frequency data. Genetics 144:2001– 2014. Corse, C. J., and R. W. Summers. 2009. Numbers of coastal birds on the Orkney Islands during winter 2006/07. Scottish Birds 29:22–32.
102 Corse, C. J., and R. W. Summers. 1999. The seasonal pattern of numbers, population structure and migration of Purple Sandpipers Calidris maritima in Orkney. Ringing Migr. 19:275–282. Dąbrowski, M. J., M. Pilot, M. Kruczyk, M. Żmihorski, H. M. Umer, and J. Gliwicz. 2014. Reliability assessment of null allele detection: inconsistencies between and within different methods. Mol. Ecol. Resour. 14:361–73. Darriba, D., G. L. Taboada, R. Doallo, and D. Posada. 2012. jModelTest 2: more models, new heuristics and parallel computing. Nat. Methods 9:772. De Barba, M., and L. P. Waits. 2010. Multiplex pre-amplification for noninvasive genetic sampling: is the extra effort worth it? Mol. Ecol. Resour. 10:659–65. Deane, P. E., K. D. McCoy, R. J. Robertson, T. P. Birt, and V. L. Friesen. 2013. Minimal genetic structure in the cerulean warbler despite evidence for ecological differentiation among populations. Condor 115:178–185. Defaveri, J., H. Viitaniemi, E. Leder, and J. Merilä. 2013. Characterizing genic and nongenic molecular markers: Comparison of microsatellites and SNPs. Mol. Ecol. Resour. 13:377–392. Dharmarajan, G., W. S. Beatty, and O. E. Rhodes. 2013. Heterozygote deficiencies caused by a Wahlund effect: Dispelling unfounded expectations. J. Wildl. Manage. 77:226–234. Di Rienzo, A., A. C. Peterson, J. C. Garza, A. M. Valdes, M. Slatkin, and N. B. Freimer. 1994. Mutational processes of simple-sequence repeat loci in human populations. Proc. Natl. Acad. Sci. U. S. A. 91:3166–3170. Dierschke, V. 1998. Site fidelity and survival of Purple Sandpipers Calidris maritima at Helgoland (SE North Sea). Ringing Migr. 19:41–47. Dowling, D. K., U. Friberg, and J. Lindell. 2008. Evolutionary implications of non- neutral mitochondrial genetic variation. Trends Ecol. Evol. 23:546–54. Edwards, S., and S. Bensch. 2009. Looking forwards or looking backwards in avian phylogeography? A comment on Zink and Barrowclough 2008. Mol. Ecol. 18:2930– 2933. Engelmoer, M., and K. Roselaar. 1998. Geographic variation in waders. Kluwer Academic Publ., London. Excoffier, L., and G. Heckel. 2006. Computer programs for population genetics data analysis: a survival guide. Nat. Rev. Genet. 7:745–58. Excoffier, L., and H. E. L. Lischer. 2010. Arlequin suite ver 3.5: a new series of programs to perform population genetics analyses under Linux and Windows. Mol. Ecol. Resour. 10:564–7.
103 Excoffier, L., P. E. Smouse, and J. M. Quattro. 1992. Analysis of Molecular Variance Inferred From Metric Distances Among DNA Haplotypes: Application Application to Human Mitochondrial DNA Restriction Data. Genetics 491:479–491. Freamo, H., P. O’Reilly, P. R. Berg, S. Lien, and E. G. Boulding. 2011. Outlier SNPs show more genetic structure between two Bay of Fundy metapopulations of Atlantic salmon than do neutral SNPs. Mol. Ecol. Resour. 11:254–267. Fu, Y. 1997. Statistical tests of neutrality of mutations against population growth, hitchhiking and background selection. Genetics 147:915–925. Galtier, N., F. Depaulis, and N. H. Barton. 2000. Detecting bottlenecks and selective sweeps from DNA sequence polymorphism. Genetics 155:981–987. Galtier, N., B. Nabholz, S. Glémin, and G. D. D. Hurst. 2009. Mitochondrial DNA as a marker of molecular diversity: a reappraisal. Mol. Ecol. 18:4541–4550. Gill, R. E., P. S. Tomkovich and B. J. Mccaffery. 2002. Rock Sandpiper (Calidris ptilocnemis), The Birds of North America Online (A. Poole, Ed.). Ithaca: Cornell Lab of Ornithology; Retrieved from the Birds of North America Online: http://bna.birds.cornell.edu/bna/species/686 Goudet, J. 1995. FSTAT (version 1.2): a computer program to calculate F-statistics. J. Hered. 86:458–486. Guindon, S., and O. Gascuel. 2003. A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Syst. Biol. 52:696–704. Haig, S. M., W. M. Bronaugh, R. S. Crowhurst, J. D’Elia, C. A. Eagles-Smith, C. W. Epps, B. Knaus, M. P. Miller, M. L. Moses, S. Oyler-McCance, W. D. Robinson, and B. Sidlauskas. 2011. Genetic applications in avian conservation. Auk 128:205–229. Hake, M., D. Blomqvist, E. P. Pierce, T. Jaras, and O. C. Johansson. 1997. Population size, migration routes and breeding origin of Purple Sandpipers Calidris maritima wintering in Sweden. Ornis Svecica 7:121–132. Hallgrimsson, G. T., R. W. Summers, B. Etheridge, and B. R. L. Swann. 2012. The winter range of Nearctic Purple Sandpipers Calidris maritima on the East Atlantic flyway. Ardea 100:13–18. Harpending, H. C. 1994. Signature of ancient population growth in a low-resolution mitochondrial DNA mismatch distribution. Hum. Biol. 66:591–600. Hedmark, E., and H. Ellegren. 2005. A test of the multiplex pre-amplification approach in microsatellite genotyping of wolverine faecal DNA. Conserv. Genet. 7:289–293. Hewitt, G. M. 1996. Some genetic consequences of ice ages, and their role in divergence and speciation. Biol. J. Linn. Soc. 58:247–276.
104 Hewitt, G. M. 2004. The structure of biodiversity – insights from molecular phylogeography. Front. Zool. 1:4. Höglund, J., T. Johansson, A. Beintema, and H. Schekkerman. 2009. Phylogeography of the Black-tailed Godwit Limosa limosa: Substructuring revealed by mtDNA control region sequences. J. Ornithol. 150:45–53. Holsinger, K., and B. Weir. 2009. Genetics in geographically structured populations: defining, estimating and interpreting FST. Nat. Rev. Genet. 10:639–650. Jeratthitikul, E., T. Hara, M. Yago, T. Itoh, M. Wang, S. Usami, and T. Hikida. 2013. Phylogeography of Fischer’s Blue, Tongeia fischeri, in Japan: Evidence for introgressive hybridization. Mol. Phylogenet. Evol. 66:316–26. Jost, L. 2008. GST and its relatives do not measure differentiation. Mol. Ecol. 17:4015– 4026. Kalinowski, S. T., and M. L. Taper. 2006. Maximum likelihood estimation of the frequency of null alleles at microsatellite loci. Conserv. Genet. 7:991–995. Latch, E. K., G. Dharmarajan, J. C. Glaubitz, and O. E. Rhodes. 2006. Relative performance of Bayesian clustering software for inferring population substructure and individual assignment at low levels of population differentiation. Conserv. Genet. 7:295–302. Leblanc, N. M. 2013. Phylogeographic analysis of genetic structure present in Purple Sandpipers. Unpublished Honours thesis, Acadia University, Wolfville, NS. Leng, L., and D. E.-X. Zhang. 2011. Measuring population differentiation using GST or D? A simulation study with microsatellite DNA markers under a finite island model and nonequilibrium conditions. Mol. Ecol. 20:2494–509. Librado, P., and J. Rozas. 2009. DnaSP v5: a software for comprehensive analysis of DNA polymorphism data. Bioinformatics 25:1451–2. Lischer, H. E. L., and L. Excoffier. 2012. PGDSpider: an automated data conversion tool for connecting population genetics and genomics programs. Bioinformatics 28:298–9. Lounsberry, Z. T., J. B. Almeida, T. Grace, R. B. Lanctot, J. Liebezeit, B. K. Sandercock, K. M. Strum, S. Zack, and S. M. Wisely. 2013. Range-wide conservation genetics of Buff-breasted Sandpipers (Tryngites subruficollis). Auk 130:429–439. Marthinsen, G., L. Wennerberg, and J. T. Lifjeld. 2007. Phylogeography and subspecies taxonomy of Dunlins (Calidris alpina) in western Palearctic analysed by DNA microsatellites and amplified fragment length polymorphism markers. Biol. J. Linn. Soc. 92:713–726. Martin, D. P., P. Lemey, and D. Posada. 2011. Analysing recombination in nucleotide sequences. Mol. Ecol. Resour. 11:943–55.
105 Matschiner, M., and W. Salzburger. 2009. TANDEM: integrating automated allele binning into genetics and genomics workflows. Bioinformatics 25:1982–3. McGuire, J. A., C. W. Linkem, M. S. Koo, D. W. Hutchison, a. K. Lappin, D. I. Orange, J. Lemos-Espinal, B. R. Riddle, and J. R. Jaeger. 2007. Mitochondrial introgression and incomplete lineage sorting through space and time: Phylogenetics of crotaphytid lizards. Evolution 61:2879–2897. McKay, B. D., and R. M. Zink. 2010. The causes of mitochondrial DNA gene tree paraphyly in birds. Mol. Phylogenet. Evol. 54:647–650. Meirmans, P. G., and P. W. Hedrick. 2011. Assessing population structure: F(ST) and related measures. Mol. Ecol. Resour. 11:5–18. Meirmans, P. G., and P. H. Van Tienderen. 2004. GENOTYPE and GENODIVE: Two programs for the analysis of genetic diversity of asexual organisms. Mol. Ecol. Notes 4:792–794. Miller, M. P., C. Gratto-trevor, S. M. Haig, S. David, M. M. Mitchell, and T. D. Mullins. 2013. Population genetics and evaluation of genetic evidence for subspecies in the Semipalmated sandpiper (Calidris pusilla). Waterbirds 36:166–178. Miller, M. P., S. M. Haig, T. D. Mullins, K. J. Popper, and M. Green. 2012. Evidence for Population bottlenecks and subtle genetic structure in the Yellow Rail. Condor 114:100–112. Mittelhauser, G. H., L. Tudor, and B. Connery. 2013. Abundance and distribution of Purple Sandpipers (Calidris maritima) wintering in Maine. Northeast. Nat. 20:219– 228. Mittelhauser, G. H., L. Tudor, and B. Connery. 2006. Distribution and ecology of Purple Sandpipers wintering in the Acadia National Park Region, Maine 2001-2004. Mittelhauser, G. H., L. Tudor, and B. Connery. 2012. Within-year movements and site fidelity of Purple Sandpipers during the nonbreeding season. J. F. Ornithol. 83:32–40. Technical Report NPS/NER/NRTR-2006/048, National Park Service, Boston, Massachusetts. Morrison, R. I. G., Y. Aubry, R. W. Butler, G. W. Beyersbergen, G. M. Donaldson, C. L. Gratto-Trevor, P. W. Hicklin, V. H. Johnston, and R. K. Ross. 2001. Declines in North American shorebird populations. Bull. Study Gr. 94:34–38. Morrison, R. I. G., B. J. Mccaffery, R. E. Gill, S. K. Skagen, S. L. Jones, G. W. Page, C. L. Gratto-trevor, and B. A. Andres. 2006. Population estimates of North American shorebirds, 2006. Wader Study Gr. Bull. 111. Narum, S. R., and J. E. Hess. 2011. Comparison of FST outlier tests for SNP loci under selection. Mol. Ecol. Resour. 11:184–194.
106 Nei, M. 1987. Molecular Evolutionary Genetics. Columbia University Press, New York, NY. Nicoll, M., R. Summers, L. Underhill, K. Brockie, and R. Rae. 1988. Regional, seasonal and annual variations in the structure of Purple Sandpiper Calidris maritima populations in Britain. Ibis 130:221–233. Oddou-Muratorio, S., G. G. Vendramin, J. Buiteveld, and B. Fady. 2009. Population estimators or progeny tests: What is the best method to assess null allele frequencies at SSR loci? Conserv. Genet. 10:1343–1347. Oring, L. W., and D. B. Lank. 1984. Breeding area fidelity, natal philopatry, and the social systems of sandpipers. Pp. 125–147 in Shorebirds: Breeding, Behavior and Populations, edited by J. Burger and B. L. Olla. Springer US. Ottvall, R., J. Höglund, S. Bensch, and K. Larsson. 2005. Population differentiation in the redshank (Tringa totanus) as revealed by mitochondrial DNA and amplified fragment length polymorphism markers. Conserv. Genet. 6:321–331. Payne, L. X. and Elin P. Pierce. 2002. Purple Sandpiper (Calidris maritima), The Birds of North America Online (A. Poole, Ed.). Ithaca: Cornell Lab of Ornithology; Retrieved from the Birds of North America Online: http://bna.birds.cornell.edu/bna/species/706 Pearce, J. M., S. L. Talbot, B. J. Pierson, M. R. Petersen, K. T. Scribner, D. L. Dickson, and A. Mosbech. 2004. Lack of spatial genetic structure among nesting and wintering King Eiders. Condor 106:229–240. Phillimore, A. B., and I. P. F. Owens. 2006. Are subspecies useful in evolutionary and conservation biology? Proc. Biol. Sci. 273:1049–1053. Pielou, E. C. 1991. After the ice age: The return of life to glaciated North America. University of Chicago Press, 2008. Piersma, T. 2003. “Coastal” versus “inland” shorebird species: interlinked fundamental dichotomies between their life- and demographic histories? Wader Study Gr. Bull. 100:5–9. Piry, S., G. Luikart, and J. M. Cornuet. 1999. BOTTLENECK: A computer program for detecting recent reductions in the effective population size using allele frequency data. J. Hered. 90:502–503. Pritchard, J. K., M. Stephens, and P. Donnelly. 2000. Inference of population structure using multilocus genotype data. Genetics 155:945–959. Provan, J., and K. D. Bennett. 2008. Phylogeographic insights into cryptic glacial refugia. Trends Ecol. Evol. 23:564–71.
107 Pruett, C. L., and K. Winker. 2005. Biological impacts of climatic change on a beringian endemic: cryptic refugia in the establishment and differentiation of the rock sandpiper (Calidris ptilocnemis). Clim. Change 68:219–240. Putman, A. I., and I. Carbone. 2014. Challenges in analysis and interpretation of microsatellite data for population genetic studies. Ecol. Evol. 4:4399–4428. Quinn, T. W. 1992. The genetic legacy of Mother Goose--phylogeographic patterns of lesser snow goose Chen caerulescens caerulescens maternal lineages. Mol. Ecol. 1:105–117. Rae, R., M. Nicoll, and R. W. Summers. 1986. The distribution of Hardangervidda Purple Sandpipers outwith the breeding season. Scottish Birds 14:68–73. Raeymaekers, J. A. M., L. Lens, F. Van Den Broeck, S. Van Dongen, and F. A. M. Volckaert. 2012. Quantifying population structure on short timescales. Mol. Ecol. 21:3458–3473. Ramírez-Soriano, A., S. E. Ramos-Onsins, J. Rozas, F. Calafell, and A. Navarro. 2008. Statistical power analysis of neutrality tests under demographic expansions, contractions and bottlenecks with recombination. Genetics 179:555–567. Ramos-Onsins, S. E., and J. Rozas. 2002. Statistical properties of new neutrality tests against population growth. Mol. Biol. Evol. 19:2092–2100. Rehfisch, M. M., S. J. Holloway, and G. E. Austin. 2003. Population estimates of waders on the non-estuarine coasts of the UK and the Isle of Man during the winter of 1997– 98: An updated estimate is presented and major declines in the internationally important populations of several species identified. Bird Study 50:22–32. Remsen, J. V. Jr. 2010. Subspecies as a meaningful taxonomic rank in avian classification. Ornithol. Monogr. 67:62–78. Renaut, S., A. W. Nolte, S. M. Rogers, N. Derome, and L. Bernatchez. 2011. SNP signatures of selection on standing genetic variation and their association with adaptive phenotypes along gradients of ecological speciation in lake whitefish species pairs (Coregonus spp.). Mol. Ecol. 20:545–559. Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43:223–225. Rodríguez, F., J. Albornoz, and A. Domínguez. 2007. Cytochrome b pseudogene originated from a highly divergent mitochondrial lineage in genus Rupicapra. J. Hered. 98:243–9. Rogers, A. R., and H. Harpending. 1992. Population growth makes waves in the distribution of pairwise genetic differences. Mol. Biol. Evol. 9:552–569. Rokas, A., E. Ladoukakis, and E. Zouros. 2003. Animal mitochondrial DNA recombination revisited. Trends Ecol. Evol. 18:411–417.
108 Rönkä, A., L. Kvist, J. Karvonen, K. Koivula, V.-M. Pakanen, D. Schamel, and D. M. Tracy. 2007. Population genetic structure in the Temminck’s stint Calidris temminckii, with an emphasis on Fennoscandian populations. Conserv. Genet. 9:29–37. Rönkä, N., L. Kvist, V.-M. Pakanen, A. Rönkä, V. Degtyaryev, P. Tomkovich, D. Tracy, and K. Koivula. 2012. Phylogeography of the Temminck’s Stint (Calidris temminckii): historical vicariance but little present genetic structure in a regionally endangered Palearctic wader. Divers. Distrib. 18:704–716. Ronquist, F., and J. P. Huelsenbeck. 2003. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19:1572–1574. Ryman, N., and O. Leimar. 2009. GST is still a useful measure of genetic differentiation — a comment on Jost’s D. Mol. Ecol. 1:2084–2087. Seeley, R. H., and W. H. Schlesinger. 2012. Sustainable seaweed cutting? The rockweed (Ascophyllum nodosum) industry of Maine and the Maritime Provinces. Ann. N. Y. Acad. Sci. 1249:84–103. Seutin, G., B. N. White, and P. T. Boag. 1991. Preservation of avian blood and tissue samples for DNA analyses. Can. J. Zool. 69:82–90. NRC Research Press. Sievers, F., A. Wilm, D. Dineen, T. J. Gibson, K. Karplus, W. Li, R. Lopez, H. McWilliam, M. Remmert, J. Söding, J. D. Thompson, and D. G. Higgins. 2011. Fast, scalable generation of high-quality protein multiple sequence alignments using Clustal Omega. Mol. Syst. Biol. 7:539. Slatkin, M., and R. R. Hudson. 1991. Pairwise comparisons of mitochondrial DNA sequences in stable and exponentially growing populations. Genetics 129:555–562. Smith, R. D., and R. W. Summers. 2005. Population size, breeding biology and origins of Scottish Purple Sandpipers. Br. Birds 98:579–588. Sonsthagen, S. A., S. L. Talbot, K. T. Scribner, and K. G. Mccracken. 2011. Multilocus phylogeography and population structure of common eiders breeding in North America and Scandinavia. J. Biogeogr. 38:1368–1380. Strann, K., R. Summers, and R. Rae. 2006. Population structure and origins of Purple Sandpipers Calidris maritima in north Norway during winter. Ringing Migr. 23:95– 100. Strum, K. M., M. J. Hooper, K. a. Johnson, R. B. Lanctot, M. E. Zaccagnini, and B. K. Sandercock. 2010. Exposure of nonbreeding migratory shorebirds to cholinesterase- inhibiting contaminants in the Western Hemisphere. Condor 112:15–28. Summers, R. W., C. J. Corse, M. Nicoll, R. Smith, and D. P. Whitfield. 1988. The biometrics and wintering area of Icelandic Purple Sandpipers. Ringing Migr. 9:133– 138.
109 Summers, R. W., S. Smith, M. Nicoll, and N. K. Atkinson. 1990. Tidal and sexual differences in the diet of Purple Sandpipers Calidris maritima in Scotland. Bird Study 37:187–194. Summers, R. W. 1994. The migration patterns of the Purple Sandpiper Calidris maritima. Ostrich 65:167–173. Summers, R., T. Piersma, K. Strann, and P. Wiersma. 1998. How do Purple Sandpipers Calidris maritima survive the winter north of the Arctic Circle? Ardea 86:51–58. Summers, R. W., M. Nicoll, and W. Peach. 2001. Numbers, migration phenology and survival of Purple Sandpipers Calidris maritima at Gourdon, eastern Scotland. Bird Study 48:139–146. Summers, R. W., G. T. Hallgrimsson, D. Aiton, B. Etheridge, and B. R. L. Swann. 2009. Population structure, biometrics and moult of migrant Purple Sandpipers Calidris maritima in southwest Iceland in spring. Bird Study 56:357–368. Summers, R. W., C. J. Corse, B. Etheridge, J. Heaton, R. Rae, and R. L. Swann. 2010. Movements to Iceland and Svalbard by Purple Sandpipers wintering in Scotland. Scottish Birds 30:194–200. Summers, R. W., S. Foster, B. Swann, and B. Etheridge. 2012a. Local and global influences on population declines of coastal waders: Purple Sandpiper Calidris maritima numbers in the Moray Firth, Scotland. Estuar. Coast. Shelf Sci. 102- 103:126–132. Summers, R. W., K. Strann, R. Rae, J. Heggås, K. Strann, and J. Heggas. 2012b. Wintering Purple Sandpipers Calidris maritima in Troms County, Northern Norway. Ornis Scand. 21:248–254. Summers, R. W., K. Strann, R. Rae, J. Heggås, K. Strann, and J. Heggas. 2012c. Wintering Purple Sandpipers Calidris maritima in Troms County, Northern Norway. Ornis Scand. 21:248–254. Summers, R. 2014. The population dynamics of wintering Purple Sandpipers in Britain. in Ecology and conservation of birds in upland and alpine habitats, BOU Proceedings, BOU Annual Meeting, University of Leicester. Summers, R. W., H. Boland, K. Colhoun, N. Elkins, B. Etheridge, S. Foster, J. W. Fox, K. Mackie, L. R. Quinn, and R. L. Swann. 2014. Contrasting trans-atlantic migratory routes of Nearctic Purple Sandpipers Calidris maritima associated with low pressure systems in spring and winter. Ardea 102:139–152. Taberlet, P., and J. Bouvet. 1991. A single plucked feather as a source of DNA for bird genetic studies. Auk 108:959–960. Tajima, F. 1989. Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics 123:585–95.
110 Tamura, K., D. Peterson, N. Peterson, G. Stecher, M. Nei, and S. Kumar. 2011. MEGA5: molecular evolutionary genetics analysis using maximum likelihood, evolutionary distance, and maximum parsimony methods. Mol. Biol. Evol. 28:2731–9. Tamura, K., G. Stecher, D. Peterson, A. Filipski, and S. Kumar. 2013. MEGA6: Molecular Evolutionary Genetics Analysis version 6.0. Mol. Biol. Evol. 30:2725–9. Toews, D. P. L., and A. Brelsford. 2012. The biogeography of mitochondrial and nuclear discordance in animals. Mol. Ecol. 21:3907–3930. Toews, D. P. L., M. Mandic, J. G. Richards, and D. E. Irwin. 2014. Migration, mitochondria, and the yellow-rumped warbler. Evolution 68:241–255. Trimbos, K. B., C. Doorenweerd, K. Kraaijeveld, C. J. M. Musters, N. M. Groen, P. Knijff, T. Piersma, and G. R. De Snoo. 2014. Patterns in nuclear and mitochondrial DNA reveal historical and recent isolation in the black-tailed godwit (Limosa limosa). PLoS One 9: e83949. doi:10.1371/journal.pone.0083949. U.S. Shorebird Conservation Plan. 2004. High Priority Shorebirds ─ 2004. Unpublished Report, U. S. Fish and Wildlife Service, 4401 N. Fairfax Dr., MBSP 4107, Arlington, VA, 22203 U.S.A. 5 pp. Van Oosterhout, C., W. F. Hutchinson, D. P. M. Wills, and P. Shipley. 2004. Micro- Checker: software for identifying and correcting genotyping errors in microsatellite data. Mol. Ecol. Notes 4:535–538. Verity, R., and R. A. Nichols. 2014. What is genetic differentiation, and how should we measure it - GST, D, neither or both? Mol. Ecol. 23:4216–4225. Villesen, P. 2007. FaBox: an online toolbox for fasta sequences. Mol. Ecol. Notes 7:965– 968. Waterhouse, A. M., J. B. Procter, D. M. a Martin, M. Clamp, and G. J. Barton. 2009. Jalview Version 2--a multiple sequence alignment editor and analysis workbench. Bioinformatics 25:1189–91. Webster, M. S., P. P. Marra, S. M. Haig, S. Bensch, and R. T. Holmes. 2002. Links between worlds: Unraveling migratory connectivity. Trends Ecol. Evol. 17:76–83. Wenink, P. W., A. J. Baker, and M. G. J. Tilanus. 1993. Hypervariable-control-region sequences reveal global population structuring in a long-distance migrant shorebird, the Dunlin (Calidris alpina). Proc. Natl. Acad. Sci. U. S. A. 90:94–98. Wenink, P. W., A. J. Baker, and M. G. J. Tilanus. 1994. Mitochondrial control-region sequences in two shorebird species, the Turnstone and the Dunlin, and their utility in population genetic studies. Mol. Biol. Evol. 11:22–31.
111 Wenink, P. W., A. J. Baker, H. Rosner, and M. G. J. Tilanus. 1996. Global mitochondrial DNA phylogeography of holarctic breeding Dunlins (Calidris alpina). Evolution 50:318–330. Wennerberg, L. 2001. Breeding origin and migration pattern of Dunlin (Calidris alpina) revealed by mitochondrial DNA analysis. Mol. Ecol. 10:1111–1120. Wennerberg, L., G. Marthinsen, and J. T. Lifjeld. 2008. Conservation genetics and phylogeography of southern Dunlins Calidris alpina schinzii. J. Avian Biol. 39:423– 437. Wilson, A. C., R. L. Cann, S. M. Carr, M. George, U. B. Gyllensten, K. M. Helm- Bychowski, R. G. Higuchi, S. R. Palumbi, E. M. Prager, R. D. Sage, and M. Stoneking. 1985. Mitochondrial DNA and two perspectives on evolutionary genetics. Biol. J. Linn. Soc. 26:375–400. Zink, R. M., G. F. Barrowclough, J. L. Atwood, and R. C. Blackwell-Rago. 2000. Genetics, threatened taxonomy, California and conservation of the Gnatcatcher. Conservation 14:1394–1405. Zink, R. M. 2004. The role of subspecies in obscuring avian biological diversity and misleading conservation policy. Proc. Biol. Sci. 271:561–564. Zink, R. M., and G. F. Barrowclough. 2008. Mitochondrial DNA under siege in avian phylogeography. Mol. Ecol. 17:2107–2121. Zink, R. M., A. Pavlova, S. Drovetski, and S. Rohwer. 2008. Mitochondrial phylogeographies of five widespread Eurasian bird species. J. Ornithol. 149:399–413.
112