Biological Conservation 209 (2017) 263–272

Contents lists available at ScienceDirect

Biological Conservation

journal homepage: www.elsevier.com/locate/bioc

Conservation implications of long-distance migration routes: Regional

metapopulation structure, asymmetrical dispersal, and
population declines
Patricia Szczys a,⁎,1, Stephen A. Oswald b,1, Jennifer M. Arnold b,1
a
Department of Biology, Eastern Connecticut State University, 83 Windham Street, Willimantic, CT 06226, USA
b
Division of Science, Pennsylvania State University, Berks Campus, Tulpehocken Road, Reading, PA 19610, USA

a r t i c l e i n f o a b s t r a c t

Article history: Understanding and conserving metapopulations of long-distance migrants is challenging since breeding popula-
Received 10 October 2016 tion structure may not be delimited simply by geography, but also by migration or wintering distributions. We
Received in revised form 18 December 2016 present the first study of the relative importance of breeding, migration and wintering distributions for regional
Accepted 6 February 2017
metapopulation processes in breeding areas. Using a species for which life-history and demography are exten-
Available online xxxx
sively studied (common terns Sterna hirundo), we explored population genetics across eastern North America
Keywords:
to distinguish between these potential drivers of metapopulation structure, understand recent population de-
Dispersal clines in inland areas, and direct appropriate conservation efforts. We analyzed nuclear and mitochondrial
Common tern DNA to determine changes in population structure and dispersal over prehistoric, historical and contemporary
Great Lakes time at regional, sub-regional and local spatial scales. We found evidence for conservation-relevant, hierarchical
Metapopulations metapopulation structure within breeding areas: at (i) local and sub-regional scales resulting from restricted dis-
Population genetics persal distances, and (ii) between inland and coastal regions as a result of non-breeding distributions. Crucially, at
North America regional scales (ii), asymmetrical dispersal rates from inland to coastal colonies have increased ten-fold since the
Sterna hirundo
1960s, contributing to recent, largely-enigmatic inland population declines. Migrating inland common terns pass
over coastal colonies but the reverse is not true. Thus, asymmetrical dispersal from inland to coastal areas pro-
vides the first demonstration that migration routes, not wintering distributions, are drivers of breeding popula-
tion structure in the absence of physical barriers. Our results illustrate the conservation importance of this
determination for long-distance migrants: definitive evidence connecting metapopulation demographics to re-
gional population declines for common terns, previously only speculated at despite N 100 years of banding effort.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction landscape heterogeneity (Olsen et al., 2010; Pavlacky et al., 2012;

Understanding demography of declining populations is of key im-
portance for interpreting and addressing numerical declines (Lande,
1988; Frankham, 2005). Populations rarely exist in isolation, but instead
as a metapopulations where they are in flux between extinction and
colonization and are connected by immigration and emigration
(Hanski, 1991; Harrison, 1991). While classical metapopulations
(“Levins metapopulations”) are relatively rare for vertebrates
(Fronhofer et al., 2012), groups of populations are commonly linked
by dispersal and thus exhibit predictable “metapopulation” dynamics
at both local and regional scales that have important conservation im-
plications (Hanski, 1998; Harrison, 1991). Since metapopulation pro-
cesses act at a variety of spatial scales and often in conjunction with
⁎ Corresponding author.
E-mail addresses: szczysp@easternct.edu (P. Szczys), steve.oswald@psu.edu
(S.A. Oswald), jma25@psu.edu (J.M. Arnold).
1
All authors contributed equally to this work.
Sutherland et al., 2012), this can lead to a hierarchical metapopulations
(population structure and dispersal across multiple scales). Under-
standing such hierarchical metapopulation structure is therefore essen-
tial to inform conservation priorities for declining regional populations.
Empirical demographic studies across metapopulations, such as
multi-site capture-recapture (Lebreton et al., 2003), are logistically
challenging for wide-ranging organisms, especially long-range migrants
with low rates of recruitment and widespread breeding distributions
(Esler, 2000). Terns (Family Sternidae) are a good example of this chal-
lenge and while demographic studies have been achieved for small
metapopulations at small spatial scales (b40 km distant; Breton et al.,
2014; Palestis and Hines, 2015) and occasionally for distances of up to
300 km (Spendelow et al., 1995), even these distances are relatively
small when compared to breeding dispersal documented from banding
recoveries (e.g. up to 950 km for common terns Sterna hirundo in North
America; Nisbet, 2002). Population genetics can be used instead to infer
both population connectivity and diversity (e.g. Bicknell et al., 2012;
Boutilier et al., 2014; Szczys et al., 2012). Moreover, analyses of gene

http://dx.doi.org/10.1016/j.biocon.2017.02.012
0006-3207/© 2017 Elsevier Ltd. All rights reserved.
264 P. Szczys et al. / Biological Conservation 209 (2017) 263–272
flow in current, historical and prehistoric time can provide an important
conservation tool for understanding dispersal between populations and
prioritizing populations for conservation efforts (Dudaniec et al., 2011;
Osborne et al., 2012; Welch et al., 2012).
Unfortunately, typical approaches for regional and landscape-scale
population genetics (Manel and Holderegger, 2013) are not as simple
to apply for long-distance migrants as for resident species (e.g.,
Pavlacky et al., 2012), chiefly because population structure cannot be
defined unequivocally based on geography or life stage (Esler, 2000).
This is especially true for seabirds and waterbirds, since often (but not
always: e.g. Bicknell et al., 2012; Boutilier et al., 2014) barriers to dis-
persal are not static geography (e.g. land barriers, mountain ranges)
but may be caused by ocean regimes (e.g. currents, fronts and mixing),
nonbreeding distributions, foraging distributions, and breeding phenol-
ogy (Friesen, 2015). In the absence of land barriers, nonbreeding distri-
butions have been implicated as the most important factor affecting
population structure and dispersal for migratory seabirds (Friesen,
2015; Friesen et al., 2007) and of importance for shorebirds
(Kraaijeveld, 2008) and landbirds (Kelly and Hutto, 2005). Nonbreeding
distributions include migration routes and wintering areas, but the rel-
ative extent to which each is driving population structure and dispersal
is currently unclear. Few studies differentiate between these life stages
and explicitly investigate, or even speculate, on the potential influence
of each on breeding populations. Breeding populations of Cassin's
auklets (Ptychoramphus aleuticus) and black-browed albatrosses
(Thalassarche melanophris) have been suggested as being structured ac-
cording to their distinct wintering distributions, although the degree to
which migratory route might play a role is unclear (Burg and Croxall,
2001; Wallace et al., 2015). The only previous studies to investigate mi-
gration route are for Eurasian blackcaps (Sylvia atricapilla), where there
is evidence of influences of both wintering distribution and migration
route on population differentiation (Bearhop et al., 2005; Rolshausen
et al., 2009). However, in this case, blackcaps from distinct wintering
grounds also had distinct migration routes and thus it is not possible
to distinguish the individual effects of migration route and wintering
distribution on genetic structure of these breeding populations.
Common terns are long-distance migratory waterbirds that main-
tain breeding colonies in coastal and inland regions across the temper-
ate northern hemisphere and winter in temperate, subtropical and
tropical coasts of the southern hemisphere (Gochfeld et al., 2016).
Banding and geolocator tracking data indicate that common terns
breeding on the Atlantic coast of North America migrate solely offshore
from the Atlantic coast (Austin, 1953; Nisbet et al., 2011). Migration
routes for inland breeders from the Great Lakes to Manitoba are gener-
ally inland to the west of the Florida Panhandle (Austin, 1953; Blokpoel
et al., 1987) but a proportion of these birds also travel up the Atlantic
coast (Austin, 1953). This latter route is confirmed by geolocator track-
ing studies as important for inland breeders, especially during spring
migration (Nisbet et al., in press). Atlantic coast breeding populations
are generally stable (Nisbet et al., in press) (despite declines within
some areas; Palestis and Hines, 2015), however, region-wide declines
across Great Lakes populations have continued since the 1970s, by
18% overall and over 40% in Ontario (Morris et al., 2010, 2012). Further-
more, prevailing views that larger breeding populations in Manitoba
(36–39% of known North American population, Morris et al., 2012)
could compensate for these declines have been overturned by latest
census data that indicate a 57–67% decline in abundance within these
populations in the last 20 years (Wilson et al., 2014). Significant popu-
lation declines over several decades for common terns are not unique
to North America as similar trends were reported in Germany, The
Netherlands, Norway and inland Britain (Becker and Ludwigs, 2004;
Szostek and Becker, 2012). Thus, determining the factors that affect
population structure and dispersal across inland and coastal colonies
is of key importance for directing urgent conservation initiatives for
common terns in inland North America and can inform similar strate-
gies in Europe. Moreover, common terns are one of the most widely
studied waterbird species, with an abundance of life-history and demo-
graphic data. This is especially true in North America, where, in addition
to research (Nisbet, 2002; Nisbet et al., in press), colonies have been
monitored regularly since the 1800s (Nisbet, 2002); regular compre-
hensive censuses of large areas have occurred since the 1960s (e.g.
Kress et al., 1983; Morris et al., 2010); banding has been ongoing since
at least 1909 (Cleaves, 1913) [over 1.5 million common terns have
now been banded and over 32,000 recoveries are available through
the USGS Bird Banding Laboratory (D. Bystrak and I. Nisbet pers.
comm.)], and recent tracking studies have used geolocators to delimit
migration, staging and wintering areas of common terns breeding
across North America (Nisbet et al., 2011, in press; E. Craig, D. J.
Moore, A. Bracey and A. McKellar unpubl. data). However, like many
other diverse avian taxa, they are wide-ranging, long-distance migrants
and thus little is known of their metapopulation demography at region-
al scales because of the logistical challenges described above.
We explore population genetics of breeding common terns across
eastern North America, from Lake St. Joseph, Kenora District, ON in
the west to coastal Newfoundland in the east (a study area
spanning N 2500 km) to assess dispersal and metapopulation structure
that have direct relevance for conservation and management. We inves-
tigate changes in these metapopulation processes at regional
(N1000 km), sub-regional (b400 km), and local (between sites,
~100 km+) spatial scales and across three temporal scales: contempo-
rary (2–3 generations), historical (5 + generations) and prehistoric
(thousands of years). Possible drivers of population structure include
factors acting within the breeding distribution, on migratory routes, or
at wintering grounds. Since the last glaciation, no formidable physical
barriers have existed across our study area: common terns are able to
cross large expanses of open water (and even rest on water for up to
11 h) (Nisbet et al., 2011), the area is permeated by an abundance of
lakes and rivers, and there are no large mountain ranges. Given that,
we predict that any contemporary genetic structure among these
breeding populations will be the product of non-physical barriers, chief-
ly (i) species-specific limits to breeding dispersal distance (generally
b40 km [Breton et al., 2014; Haymes and Blokpoel, 1978; Palestis and
Hines, 2015] but occasionally up to 950 km [Nisbet, 2002]) for local or
sub-regional structure, and (ii) differences in nonbreeding distribution
(either migration routes or wintering distributions) between coastal
and inland breeders for regional structure. If migration routes are the
most important driver of regional population structure, we predict
asymmetrical dispersal from inland regions to the coast (since coastal
breeders migrate only eastward of the Atlantic coast and thus potential
recruits do not encounter inland colonies). Conversely, if mixing in
overlapping wintering areas is most important, we expect symmetrical
dispersal to be evident. Identifying which parts of the annual cycle
shape population structure would be a major step towards understand-
ing the factors behind ongoing population declines.
In summary, we explore the genetic evidence for metapopulation
structure in this region across a range of spatial and temporal scales,
examine the importance of dispersal during breeding, migration and
wintering stages in maintaining this structure, and implicate the poten-
tial factors behind observed temporal changes. So doing, we aim to for-
ward our understanding of inland population declines and formulate
conservation initiatives for this species. More broadly, we provide an
understanding of how non-breeding distributions can impact metapop-
ulation processes and illustrate how our approach can address conser-
vation questions that remain unresolved, even for well-studied species.
2. Materials & methods
2.1. Sampling and genotyping
Blood from 238 non-sibling, common tern chicks was collected fol-
lowing methods of Szczys et al. (2012) at 12 sites throughout the north-
eastern United States and southern Canada between 2008 and 2012, at
 P. Szczys et al. / Biological Conservation 209 (2017) 263–272 265

both inland and coastal colonies (Fig. 1, Table S1.1: Appendix 1 Supple-
mentary data, SD). Although 13 colonies were visited, because of small
sample size (b10 per colony) and close proximity (b 50 km) samples
for two colonies in Lake Michigan (St. Ignace and Hog Shoals, Fig. 1)
were pooled for all analyses. Samples from colonies on two separate
navigational buoys in the St. Lawrence River (within b2 km of one an-
other) were also pooled. Genotypes for all individuals (number per
site [n] = 11–24) were obtained for six microsatellite markers: RBG
13, 27, and 28 (Given et al., 2002), AAT 20 and 27 and AAC 20 (Szczys
et al., 2005), following methods of Szczys et al. (2012). One mitochon-
drial DNA (mtDNA) marker was also assessed: a 464 bp fragment of
the cytochrome-b gene, sequenced following methods of Boutilier et
al. (2014). Although microsatellites measure genetic variation across
only a few generations (see below), mtDNA (such as cytochrome-b)
represents genetic variation accumulating over thousands of years.
2.2. Genetic variation between sites
Allelic richness (a measure of genetic diversity) was calculated for
each site standardizing to a sample size of 10 individuals using FSTAT
2.9.3.2 (Goudet, 2001). We tested all loci to assess violations of assump-
tions of linkage equilibrium and Hardy-Weinberg Equilibrium (HWE)
using the Exact Test implemented in GenePop 4.2 (Raymond and
Rousset, 1995). None of the 120 pairwise tests (12 sites × 10 pairs of
loci) for linkage disequilibrium were significant (P b 0.009). We used
the conservative B-Y correction of critical P-value for all analyses that
included multiple pair-wise comparisons (Narum, 2006). We removed
locus RBG27 from further analysis because FreeNA analysis suggested
extremely high frequency of null alleles in all populations (Chapuis
and Estoup, 2007). The remaining loci were estimated to harbor null al-
leles in a range of frequencies (0.00–0.29) with no general pattern by
locus or site.
Cytochrome-b variation was described by the probability that two
randomly chosen haplotypes in a population are different (haplotype
diversity, h), the mean number of nucleotide differences between two
sequences (nucleotide diversity, π), and the degree of selective neutral-
ity (Tajima's D and Fu's F) using ARLEQUIN 3.5 (Excoffier and Lischer,
2010).
2.3. Population structure between sites
Several analyses were conducted to describe the genetic structure of
populations across the sampled region. For microsatellite data, we used
estimates of FST and Jost's D, Analysis of Molecular Variance (AMOVA),
Isolation by Distance (IBD), and Bayesian clustering analysis. FST and
Jost's D were estimated for all pairs of populations to estimate the de-
gree of population differentiation.
FST values were estimated based on the EM algorithm of Dempster et
al. (1977) to correct for the presence of null alleles (null allele frequency
was estimated using FreeNA; Chapuis and Estoup, 2007). Analysis by
FreeNA indicated no biologically meaningful effect of possible null al-
leles on the global (0.046 uncorrected versus 0.047 corrected) or any
pairwise FST value after correcting for the presence of null alleles
(Table S1.2) so further analysis was conducted on uncorrected genotype
data.

Fig. 1. Location of sampling sites (ABBREVIATION [designated Sub-region for analysis, see text]; n = number of individuals; latitude, longitude): Lake St. Joseph, Kenora District, Ontario
(ONT [BOR]; n = 11; 51°05′ N, 90°38′ W); St. Ignace Pier, St. Ignace, Michigan (MI [UGL]; n = 10; 45°52′ N, 84°43′ W) and Hog Island Reef Shoals, Lake Michigan (MI [UGL]; n = 8; 45°49′
N, 85°18′ W); Henry Island, North Channel of Lake Huron, Ontario (LH [UGL]; n = 23; 45°55′ N, 82°47′ W); Young Squaw Island, Northeastern Manitoulin, Ontario (YS [UGL]; n = 24;
45°51′ N, 81°27′ W); Gull Island, Presqu'ile Provincial Park, Ontario (PR [LGL]; n = 24; 43°59′ N, 77°45′ W); Tommy Thompson Park, Toronto, Ontario (TT [LGL]; n = 23; 43°38′ N,
79°20′ W); Little Island, Oneida Lake, New York (OL [LGL]; n = 23; 43°14′ N, 76°00′ W); Upper St. Lawrence River, New York (SLR [SLR]; n = 20; 44°46′ N, 75°23′ W); Bird Island,
Marion, Massachusetts (BI [USCoast]; n = 24; 41°40′ N, 70°43′ W); Pettit Island, Barnegat Bay, New Jersey (PI [USCoast]; n = 24; 39°40′ N, 74°11′ W); Country Island, Tor Bay, Nova
Scotia (CI [CACoast]; n = 11; 45°09′ N, 61°27′ W); and O'Donnells Lagoon, O'Donnells, Newfoundland (OD [CACoast]; n = 13; 47°04′ N, 53°34′ W).
266 P. Szczys et al. / Biological Conservation 209 (2017) 263–272

AMOVA was evaluated by in GenAIEx 6.5.01 (Peakall and Smouse,
2012, 2006) to explore the partitioning of genetic variation within and
among sites across the sampled region.
IBD was evaluated by the Isolde option (Rousset, 2000) in GenePop
4.2 (Raymond and Rousset, 1995) and regression analysis in GenAnIEx
6.5.01.
Lastly, we used Bayesian clustering analysis implemented by Struc-
ture 2.3.4 (Pritchard et al., 2000) to estimate the number of genetic clus-
ters (K) represented in the data. We followed the methods of
Puechmaille (2016) to ensure accurate estimate of K considering un-
even sampling among sites (n = 11–24) by calculating MedianMeanK,
MaximumMeanK, MedianMedianK, and MaximumMedianK along with
the posterior probability for each K and delta K as reported by STRUC-
TURE HARVESTER (Earl, 2012) and Clumpak (Kopelman et al., 2015)
(see Appendix 2 for details). We use the maximum value of the four pa-
rameters above as estimates of the true number of clusters in the data.
For mtDNA, genetic structure was explored by examining popula-
tion differentiation based on the cytochrome-b gene characterized by
φST pairwise values in ARLEQUIN 3.5 (Excoffier and Lischer, 2010). Ad-
ditionally, the minimum spanning network was generated in PopArt
v1.7 (http://popart.otago.ac.nz; Bandelt et al., 1999). This network is a
visual representation of the mutational relationship between different
haplotype sequences and represents the frequency of each haplotype
at each site.
longitudinal partitioning of populations into categorized sub-regions
(Fig. 3b).
Pairwise φST values based on cytochrome-b variation resulted in no
significant differences between sub-regions (Table 1) or between inland
and coastal regions (φST = 0.004, P = 0.23) indicating weak population
structure in prehistoric time. The star-like shape of the minimum
spanning network of 14 cytochrome-b haplotypes (Fig. S1.1) suggests
a recently-expanding population.
3.2. Contemporary and historical dispersal among regions and sub-regions
Analysis of m (proportion of a population consisting of individuals
from another population per generation) between inland and coastal
regions indicated high rates of dispersal from inland colonies to coastal
colonies, but no detectable dispersal (95% CI overlaps zero) in the oppo-
site direction (Fig. 3, Table 2). Historical dispersal rates were ten-fold
lower but the signature of inland to coastal directional dispersal was
still apparent (Fig. 3, Table 2).
Sub-region Bayesian analysis was consistent across all iterations
(Fig. 3, Table S1.4) and indicated significant and asymmetrical dispersal
rates from Lower Great Lakes sub-region [LGL] to (i) Boreal [ONT] and
(ii) St. Lawrence River [SLR] but not to the coastal regions. Significant bi-
directional dispersal was noted only among Lower Great Lakes [LGL]

2.4. Contemporary and historical dispersal among regions and sub-regions

Table 1
Measures of population differentiation between six sub-regions during (a) contemporary
We used BAYESASS 3.3 (Wilson and Rannala, 2003) to infer m (the time: Pairwise uncorrected FST values (below diagonal) and D values (above diagonal);
proportion of a population consisting of individuals from another popu- and (b) during prehistoric time: Pairwise φST values. All FST values are significantly differ-
lation, per generation) as a measure of contemporary dispersal (gene ent between sites with B-Y correction for multiple tests (P b 0.02); all others: significant
differences between sites (with B-Y correction for multiple tests) are indicated in bold at
flow within the past 2–3 generations) (i) between inland (all inland
the P b 0.01 level and by * at the P = 0.025 level. Shading indicates sub-regions in the
sites) and coastal (all coastal sites) regions and (ii) among six sub-re- coastal region.
gions (Boreal Ontario [BOR], Upper Great Lakes [UGL], Lower Great
Lakes [LGL], Saint Lawrence River [SLR], US Atlantic Coast [USCoast], a) BOR UGL LGL SLR USC CAC
Canadian Atlantic Coast [CACoast]; Fig. 1). These six sub-regions were
chosen based on proximity and shared geographic locale. BAYESASS BOR - 0.09 0.06* 0.10 0.13 0.12
uses a Bayesian approach and MCMC to infer m over the last few (gen-
erally 2–3) generations (see Appendix 2 for analysis parameters). UGL 0.028 - 0.03 0.12 0.07 0.07
Estimation of historical dispersal (gene flow occurring N 5 genera-
tions previously) (i) among regions and (ii) between inland and coastal LGL 0.021 0.011 - 0.08 0.10 0.09
regions was achieved using Migrate-n (Beerli, 2006) (see Appendix 2).
SLR 0.040 0.040 0.029 - 0.10 0.08
3. Results
USC 0.049 0.024 0.036 0.040 - 0.04
3.1. Regional and sub-regional scale population structure
CAC 0.037 0.022 0.029 0.031 0.013 -
Between inland and coastal regions, weak but significant population
structure was found at the regional scale (FST = 0.013 and D = 0.06;
P = 0.001). At the sub-regional scale, differentiation was stronger but
still low-to-moderate (FST = 0.011–0.049 and D = 0.06–0.13; all
P b 0.03), with generally the largest differentiation occurring between b) BOR UGL LGL SLR USC CAC
inland and coastal sub-regions (Table 1). At local scales (between indi-
vidual sites), differentiation was more variable (FST = 0.016–0.088 and BOR
D = 0.01–0.41; Table S1.2). Analysis of Molecular Variance confirmed
hierarchical population structure that varied from weak-to-moderate, UGL 0.052
being weaker at larger spatial scales: 84% of genetic variation was repre-
sented within individuals, 11% within sites, 4% among sites, and only 1% LGL 0.044 0.004
among regions.
For IBD analysis, linear geographic distance did not predict popula-
SLR 0.009 -0.024 -0.014
tion differentiation either when FST or D were considered (both R2 =
0.02; P = 0.12). Bayesian clustering analysis supported either three
USC 0.015 0.012 0.030 -0.040
clusters (groups of individuals between which gene flow is obstructed;
K = 3, Fig. S1.4, using the method of Evanno et al., 2005) or a maximum
CAC 0.101 -0.014 0.032 -0.020 0.023
of four clusters (Table S1.3, using the newer method of Puechmaille,
2016). The four clusters partitioned the UGL and LGL sub-regions
(Fig. 2) and both three and four cluster diagrams did indicate a weak
 P. Szczys et al. / Biological Conservation 209 (2017) 263–272
Fig. 2. Three and four genetic clusters (K = 3 and K = 4) inferred through Bayesian clustering analysis in Structure. Each cluster is represented by a different color. Each column represents
an individual's proportional membership in each cluster. Individuals are ordered by sampling site and bars delineate sub-regions (Figs. 1 and 3).(For interpretation of the references to
color in this figure legend, the reader is referred to the web version of this article.)
and Upper Great Lakes [UGL] sub-regions and was higher for contempo-
rary time periods.
3.3. Genetic diversity
Genetic variability and gene diversity were high in all sites: allelic
richness ranged from 4.8 to 6.4 per locus (mean 5.2) and HE ranged
from 0.66 to 0.76 (mean = 0.71; Table 1). Fourteen cytochrome-b
haplotypes were characterized (GenBank Accession numbers forth-
coming). Nine of 14 haplotypes were unique to one site each and
the most frequent haplotype was common to all regions (Fig. S1.1).
Haplotype and nucleotide diversity (π) were generally higher within
the Great Lakes sub-regions (UGL, LGL) than in Boreal Ontario or
Coastal sub-regions, suggesting that the Great Lakes sites are the
oldest populations.
4. Discussion
We found metapopulation connectivity for common terns at all
scales explored: local (between sites, ~ 100 km +), sub–regional
(b400 km) and regional (N1000 km, inland versus coastal). The
existence of such hierarchical metapopulation structure is a situation
rarely reported for birds, but described for fish in freshwater habitats
as a consequence of dispersal at two different scales (Olsen et al.,
2010; Roberts et al., 2013). In the same way, common terns in eastern
North America exhibit local-scale breeding dispersal, rarely exceeding
40 km from natal site (Breton et al., 2014; Haymes and Blokpoel,
1978; Palestis and Hines, 2015), which we confirm but also find strong
evidence for regional dispersal as a consequence of non-breeding
distributions. Dispersal among freshwater fish populations is influenced
by physical habitat: locally by linear habitats (rivers) and long-distance
through lakes (Olsen et al., 2010; Roberts et al., 2013). However,
for long-distance migrants such as common terns (for which no
discernable physical barriers to dispersal exist) our results indicate
underlying constraints to be more behavioral, acting in non-breeding
areas to drive regional-scale structure and in breeding areas to drive
structure at sub-regional and local scales. Further, only by assessing
genetic differences and variation at multiple scales are we able to
267
determine the impacts of hierarchical metapopulation structure on
spatial demography and population trends.
4.1. Drivers and conservation relevance of regional scale structure
Use of discrete areas outside of the breeding season can be a mech-
anism that promotes genetic structure of breeding populations (Esler,
2000; Friesen, 2015; Friesen et al., 2007). The relative importance of mi-
gration routes or wintering grounds for genetic structure has yet, how-
ever, to be determined. Our results show that asymmetrical dispersal
through existing migration routes, rather than mixing in wintering
areas, drives contemporary, regional-scale differentiation between
breeding populations of common terns in eastern North America.
Weak population structure between inland and coastal sites in Eastern
North America (FST = 0.013), can be explained by increased asymmet-
rical dispersal from inland sites to the coastal regions between historical
(N 5 generations, prior to ~1960) and contemporary time (2–3 genera-
tions, since ~ 1990 based on a generation time of 10 years; Sæther et
al., 2013; Fig. 3). Mixing at wintering grounds, which would produce
symmetrical dispersal, is therefore not a driver of the current population
structure for North American common terns.
Our results clearly demonstrate a recent, ten-fold increase in dis-
persal rates from our Great Lakes and other inland colonies to colonies
along the Atlantic coast between historical and contemporary time pe-
riods. During the same period, there was a relatively constant and
much lower rate of dispersal in the opposite direction (Table 2, Fig. 3),
which matches all analyses of banding returns between these areas:
more terns banded in the Great Lakes have been recorded on the coast
than vice versa (Austin, 1953; Haymes and Blokpoel, 1978; Blokpoel
and Courtney, 1982; Cuthbert et al., 2003; Nisbet et al., in press).
This asymmetrical movement from inland to coastal areas parallels
continued declines in nesting numbers and breeding sites in inland
North America, particularly the Great Lakes where numbers of breeding
common terns have declined by an average of 18% since the 1970s (as
high as N40% in some areas; Morris et al., 2010, 2012). Across the
Great Lakes, both breeding abundance and distribution have contracted
since the 1960s, with the most rapid reductions occurring prior to 2000
(Morris et al., 2010, 2012). This is in contrast to the generally stable or
 268
P. Szczys et al. / Biological Conservation 209 (2017) 263–272
Fig. 3. Bayesian estimates of contemporary (2–3 generations, dotted line) and historical (N5 generations, solid line) dispersal between (a) inland and coastal regions (dotted ellipses), and (b) among six subregions (solid ellipses): Boreal [BOR], Upper
Great Lakes [UGL], Lower Great Lakes [LGL], St. Lawrence River [SLR], US Coast [USCoast], and CA Coast [CACoast]. Individual sites are indicated and correspond to those labeled in Fig. 1. Mean (contemporary) or median (historical) proportion of focus
population that are migrants from another population (L95% CI - U95% CI) are only presented when m equals or exceeds either 0.05 (contemporary) or 0.04 (historical). See Table S1.2 for values for all comparisons. Thickness of arrow indicates relative
magnitude of dispersal.
 P. Szczys et al. / Biological Conservation 209 (2017) 263–272
Table 2
Estimates of (a) contemporary and (b) historical migration between inland regions (eight
sites) versus coastal regions (four sites) from Bayesian analysis of microsatellites. Lower
and upper 95% confidence intervals are shown, respectively. Data indicate significant
and asymmetrical migration (m) during both historical and contemporary time periods.
Historical migration rate (M) was converted to m for ease of comparison. Asymmetrical
gene flow is indicated by bold values in region contributing migrants to another region.
Migration from
Inland region Coastal region
To m 95% CI m 95% CI
a from prehistoric panmixia (Tables 1, S1.2), indicating greater barriers
Inland 0.99 0.98 1.01 0.01 −0.01
Coast 0.16 0.07 0.25 0.84 0.75
b rary views that common tern breeding colonies persist as local
Inland – – – 0.010 0.003
Coast 0.017 0.015 0.024 – –
increasing population trajectories of the majority (but not all; Burger
and Gochfeld, 2016; Palestis and Hines, 2015) of the Atlantic coast
(Nisbet et al., in press). Thus, movement away from Great Lakes
breeding areas is further exasperating inland population declines in
this and other inland regions (Morris et al., 2012, 2010; Wilson et al.,
2014).
Our mtDNA analyses suggest population expansion and near-
panmixia for common terns in prehistoric time. This matches that
of similar species and has been interpreted as expansion from previ-
ous glacial refugia during the last ice age (Oomen et al., 2011).
Furthermore, haplotype diversity and distribution indicates the
Lower Great Lakes to be the oldest population since this recoloniza-
tion (Fig. S1.1). This result supports the proposal by Arnold et al.
(2016) that freshwater breeding grounds represent ideal habitat
for this species.
The large increase in asymmetrical dispersal rates and migration to
coastal areas between historical and contemporary time periods sug-
gests a degradation in the suitability of this freshwater habitat concom-
itant with anthropogenic land-use changes and increases in human
population densities along many shorelines, especially in the Lower
Great Lakes. Predominant factors reducing the suitability of inland
breeding sites for common terns are high levels of predation, interspe-
cific competition for nest sites, and high water levels leading to flooding
of nests or colony sites and forcing larger, denser breeding colonies that
are more attractive to predators and can suffer negative density depen-
dence (Cuthbert et al., 2003; Nisbet et al., in press). Predation, or the
presence of potential predators, is probably the most common cause
of large-scale reproductive failure and colony abandonment at inland
locations (Cuthbert et al., 2003) and colony productivity is generally
much lower (below estimated replacement: N 1 chick fledged per pair;
DiCostanzo, 1980) across inland colonies than in coastal colonies
(Nisbet et al., in press). Additionally, major population expansion by
ring-billed gulls (Larus delawarensis) that occurred throughout the
Great Lakes between 1976 and 1990 (Morris et al., 2011) precludes
common terns from favored nesting areas (Courtney and Blokpoel,
1983; Cuthbert et al., 2003; Morris and Hunter, 1976). Flooding impacts
can then be exasperated as terns are pushed to nest in suboptimal hab-
itats (Courtney and Blokpoel, 1983; Cuthbert et al., 2003). Substantial
changes in forage fish abundance and the presence of toxic pollutants
in fish and bird tissues during the period of rapid population declines
(Fox, 2001; Gilbertson et al., 1976) may also have impacted populations
of common terns throughout the Great Lakes. All these factors contrib-
ute to the degradation of habitat suitability for terns (hereafter “habitat
degradation”) and are affected by human activities, either directly (e.g.
regulated water levels in Lakes Ontario and Superior) or indirectly (e.g.
artificially elevated populations of mammalian predators).
269
Furthermore, breeding declines for common terns are greatest within
the most heavily-developed watersheds, such as the shores of Lakes
Erie and Ontario (Morris et al., 2012, 2010).
4.2. Drivers and conservation relevance of sub-regional and local scale
structure
Significant differentiation (albeit low-to-moderate; FST: 0.01 to 0.05)
also exists between sub-regions. This represents a dramatic change
0.02 to dispersal across inland North America as a result of broader distribu-
0.93
tions and/or habitat fragmentation. This is consistent with contempo-
0.012 metapopulations across several proximate colonies, with little breeding
–
dispersal beyond distances exceeding 40 km (Breton et al., 2014;
Haymes and Blokpoel, 1978; Palestis and Hines, 2015). For our sub-re-
gional analysis, we pooled these structured local metapopulations into
larger sub-regions (150–400 km) likely diluting the signal of this differ-
entiation, explaining why it was only moderate at most. Indeed, pair-
wise FST values between sites (Table S1.2a) generally support stronger
differentiation except for adjacent sites, however because pair-wise
site comparisons include distant sites as well as local ones values in
Table S1.2 are more variable. Since structure at both sub-regional and
local scales results from the same mechanism (restricted breeding
dispersal distances), we limit further discussion of these scales to sub-
regional results only because these show more consistent trends. Our
finding of metapopulations at this scale is consistent with that common-
ly reported or assumed for both landbirds and seabirds (Esler, 2000;
Munilla et al., 2016; Oro, 2003).
At the sub-regional level, we find that terns are also dispersing away
from the areas of high habitat degradation. Lower Great Lakes colonies
appear to be supplying immigrants to the rapidly expanding St. Law-
rence River colonies (mainly artificial sites, managed against flooding
and competition, L. Harper pers. comm.) and the more pristine, boreal
sites (Fig. 3). Such large-scale emigration helps to explain why Lower
Great Lakes common terns have exhibited disproportionally high rates
of decline (Morris et al., 2010). We also report a history of bidirectional
movement between the Upper and Lower Great Lakes colonies (which
was previously not detected because of limited banding returns;
Haymes and Blokpoel, 1978) that has intensified with time (Fig. 3,
Table S1.4), suggesting reduced philopatry as a result of habitat
degradation.
4.3. Genetic diversity and spatial demography
All sampled populations, both in inland and coastal regions, exhibit-
ed high genetic variability (HE = 0.66 to 0.76) and gene diversity similar
to that found for black terns Chlidonias niger in Europe (HE = 0.55 to
0.74; Szczys et al., 2016) but greater than found previously for common
terns in Europe (HE = 0.35 to 0.038; (Sruoga et al., 2006) or for a range
of tern species across North America (HE = 0.28 to 0.76; Boutilier et al.,
2014; Draheim et al., 2012; Miller et al., 2013). Although at first glance,
Great Lakes populations are not sufficiently differentiated from coastal
populations to warrant protective legislation from genetic distinctness
alone, for example as a race or subspecies (c.f. Szczys et al., 2012), the
asymmetrical movement we document represents an erosion of inland
populations in North America and has important conservation implica-
tions (see below). Furthermore, causes of extinction are generally deter-
ministic (e.g. habitat loss, exploitation) or stochastic (e.g. demographic,
environmental) (Duncan and Blackburn, 2007; Simberloff, 1986) rather
than genetic. Thus, while the levels of gene flow are currently sufficient
to avoid harmful genetic effects (e.g. inbreeding depression), they are
unlikely to reverse current rates of decline in the Great Lakes
(−1.04%/year basin-wide), especially in Ontario where this rate is al-
most double (Morris et al., 2010).
270 P. Szczys et al. / Biological Conservation 209 (2017) 263–272
5. Conservation implications
We show that for long-distance migrants, seasonal migration routes
can be responsible for regional-scale population differentiation and lead
to hierarchical metapopulation structure. Thus, an understanding of mi-
gration patterns can be used to predict population trends and direct crit-
ical conservation decisions. Our study highlights the importance of
discriminating between the influence of barriers to dispersal within
breeding areas, migration routes and wintering distributions to formu-
late regional-scale conservation efforts. In our study, dispersal of birds
from two breeding regions (inland and coastal) that differed in their mi-
gration routes was asymmetrical and therefore contributed to inland
population declines. Summaries of recent tracking data using attached
geolocators (Nisbet et al., in press), has expanded our understanding
of why banding returns from Great Lakes breeders are more common
from Atlantic coast areas than vice-versa (Austin, 1953; Haymes and
Blokpoel, 1978; Blokpoel and Courtney, 1982; Cuthbert et al., 2003).
These seasonal migration tracks of almost 50 breeders from five inland
colonies indicate that the spring migration routes from Great Lakes'
breeders often passed close to Atlantic coast breeding colonies (Nisbet
et al., in press; E. Craig, D. J. Moore, A. Bracey and A. McKellar unpubl.
data). However, Atlantic coast breeders were never close to the Great
Lakes' colonies on migration. We propose that asymmetrical dispersal
to coastal sites in our study was caused by large numbers of first-time
breeders (e.g. Bicknell et al., 2014), or even established breeders, not
returning to inland areas of eastern North America. Instead, a propor-
tion of these birds recruited to coastal Atlantic colonies near to their mi-
gration route rather than returning to degraded inland colonies.
Since major migration routes appear inflexible for common terns,
being utilized since before the end of the last ice-age (Austin, 1953), dis-
persal along migration routes has three important implications for con-
servation efforts. Firstly, despite management efforts that successfully
restore colonies at sub-regional scales (e.g. St. Lawrence River colonies,
L. Harper pers. comm.), future dispersal from coastal-to-inland areas is
unlikely to be sufficient to reverse inland population declines because
birds that breed along the Atlantic coast do not pass near inland colonies
during their seasonal migrations (Austin, 1953; Nisbet et al., 2011; Fig.
3). Secondly, since other inland populations that also use the same in-
land migration routes (Nisbet, 2002; Nisbet et al., in press; Wilson et
al., 2014) are declining they are unlikely to provide demographic rescue
for birds breeding in the Great Lakes. Thirdly, over time declining inland
populations will become increasingly differentiated and isolated, which
has potential negative implications for genetic diversity and persistence.
Taken together with other evidence: (i) banding studies that suggest
longer-range dispersal events for common terns breeding in the Great
Lakes than in coastal areas Austin, 1953, Haymes and Blokpoel, 1978,
Blokpoel and Courtney, 1982, Cuthbert et al., 2003, Nisbet et al., in
press), (ii) high gene flow between colonies in Minnesota and Wiscon-
sin (Burson, 1990) and high turnover of breeding colonies in the Great
Lakes (Morris et al., 2010); and (iii) differences in demography and
physiology at inland colonies (Arnold et al., 2016; Arnold and Oswald,
2013); our results indicate strong demographic differences between in-
land common terns and their coastal counterparts, that exhibit stable,
local metapopulations with little sub-regional movement (Breton et
al., 2014; Palestis and Hines, 2015). As other authors have contended
(Burson, 1990; Courtney and Blokpoel, 1983; Cuthbert et al., 2003),
such sub-regional demographics and lower site tenacity (Haymes and
Blokpoel, 1978) may result from the use of temporally-unstable breed-
ing habitats in inland areas (Burson, 1990; Shugart and Scharf, 1983;
Szczys et al., 2016). Thus, even though not genetically distinct, common
tern populations in inland North America are ecologically and demo-
graphically dissimilar from stable coastal populations and thus their de-
cline represents an important erosion of diversity within this species.
Our results represent the first definitive evidence of how metapopu-
lation demographics connect to negative population trends for common
terns in the Great Lakes, previously only speculated at based on limited
regional banding recoveries (Courtney and Blokpoel, 1983; Haymes and
Blokpoel, 1978), despite being vital to inform appropriate conservation
strategies (Cuthbert et al., 2003; Morris et al., 2012). Our results confirm
the fears of Morris et al. (2012) that, barring a reverse in the population
trends that have persisted over the last 40 years, the long-term viability
of Great Lakes common terns is in jeopardy, even without considering
projected changes in average water levels and ice cover under forecast
climate scenarios (Gronewold et al., 2013; Trumpickas et al., 2009)
and any possible impacts on habitat and species-interactions. Freshwa-
ter colonies are the major stronghold for common tern populations
globally (Wetlands International, 2006), the ancestral population center
in North America (Fig. S1.1), and may be this species' historically ideal
habitat (Arnold et al., 2016; Fig. S1.1) further underscoring the need to
protect these colonies. As yet, no legislation exists in Ontario, which
now supports the majority of the Great Lakes breeding population and
which has undergone the largest declines of this region (Morris et al.,
2012) and we urge that provincial protection be considered for com-
mon terns in this province (Morris et al., 2012, 2010; Wilson et al.,
2014). We also support stronger efforts throughout inland areas to im-
prove habitat suitability and bolster the widespread poor reproductive
output of colonies within inland regions (Nisbet et al., in press) through
successful methods (e.g. Blokpoel et al., 1997; Morris et al., 1992; Quinn
et al., 1996) and to increase colonies to sizes that do not require exten-
sive management (e.g. against predators; Hernández-Matías et al.,
2003), before population sizes are so low that recovery is not possible.
Given our detection of hierarchical metapopulation structure, the
noted tendency of inland breeding common terns to use temporally-un-
stable breeding habitats (Burson, 1990; Shugart and Scharf, 1983), and
the few natural sites that have a long-term history of occupancy
(Morris et al., 2010), local scale management must focus on multiple
appropriate sites within ~ 40 km of one another, rather than single
colonies, to build back up the regional population.
Role of the funding source
CSU-AAUP Faculty Research Grants to PS and various internal grants
from Penn State University to JMA funded all molecular analysis and
most field sample collection.
Acknowledgements
We thank colleagues who contributed samples to this study includ-
ing Francie Cuthbert, Lee Harper, Dave Moore, Brian Palestis, Cynthia
Pekarik, Erin Roche, Charles Smith and Sabina Wilhelm. We also thank
the following for their assistance in the field: Jessica Amenta, Jane
Graves, Lee Harper, Dave Moore, Carolyn Mostello, Milo Richmond,
Erin Roche, Charles Smith and Chip Weseloh; and Peter Castagna, Erin
Conn, and Mackenzie Robert for assistance in laboratory work. We are
grateful to two anonymous reviewers whose comments improved the
initial manuscript.
Appendix A. Supplementary data
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.biocon.2017.02.012.
References
Arnold, J.M., Oswald, S.A., 2013. First confirmed record of a Common Tern Sterna hirundo
breeding at one year of age. Bird Study 60:275–279. http://dx.doi.org/10.1080/
00063657.2013.768590.
Arnold, J.M., Nisbet, I.C.T., Oswald, S.A., 2016. Energetic constraint of non-monotonic mass
change during offspring growth: a general hypothesis and application of a new tool.
J. Anim. Ecol. 85:476–486. http://dx.doi.org/10.1111/1365-2656.12467.
Austin, O.L., 1953. The migration of the Common Tern (Sterna hirundo) in the western
hemisphere. Bird-Banding 24, 39–55.
Bandelt, H.J., Forster, P., Röhl, A., 1999. Median-joining networks for inferring intraspecific
phylogenies. Mol. Biol. Evol. 16, 37–48.
 P. Szczys et al. / Biological Conservation 209 (2017) 263–272
Bearhop, S., Fiedler, W., Furness, R.W., Votier, S.C., Waldron, S., Newton, J., Bowen, G.J.,
Berthold, P., Farnsworth, K., 2005. Assortative mating as a mechanism for rapid evo-
lution of a migratory divide. Science 310, 502–504.
Becker, P.H., Ludwigs, J.-D., 2004. Common Tern (Sterna hirundo). BWP Updat. 6, 91–137.
Beerli, P., 2006. Comparison of Bayesian and maximum-likelihood of population genetic
parameters. Bioinformatics 22, 341–345.
Bicknell, A.W.J., Knight, M.E., Bilton, D., Reid, J.B., Burke, T., Votier, S.C., 2012. Population
genetic structure and long-distance dispersal among seabird populations: Implica-
tions for colony persistence. Mol. Ecol. 21:2863–2876. http://dx.doi.org/10.1111/j.
1365-294X.2012.05558.x.
Bicknell, A.W.J., Knight, M.E., Bilton, D.T., Campbell, M., Reid, J.B., Newton, J., Votier, S.C.,
2014. Intercolony movement of pre-breeding seabirds over oceanic scales: implica-
tions of cryptic age-classes for conservation and metapopulation dynamics. Divers.
Distrib. 20:160–168. http://dx.doi.org/10.1111/ddi.12137.
Blokpoel, H., Courtney, P.A., 1982. Immigration and recruitment of ring-billed gulls
and common terns on the lower great lakes. Can. Wildl. Serv. Prog. Notes 133,
1–12.
Blokpoel, H., Tessier, G.D., Harfenist, A., 1987. Distribution during post-breeding dispersal,
migration, and overwintering of common terns color-marked on the lower Great
Lakes. J. Ornithol. 58, 207–217.
Blokpoel, H., Tessier, G.D., Andress, R.A., 1997. Successful restoration of the Ice Island com-
mon tern colony requires on-going control of ring-billed gulls. Colon. Waterbirds 20,
98–101.
Boutilier, S.T., Taylor, S.A., Morris-Pocock, J.A., Lavoie, R.A., Friesen, V.L., 2014. Evidence for
genetic differentiation among Caspian tern (Hydroprogne caspia) populations in
North America. Conserv. Genet. 15:275–281. http://dx.doi.org/10.1007/s10592-013-
0536-1.
Breton, A.R., Nisbet, I.C.T., Mostello, C.S., Hatch, J.J., 2014. Age-dependent breeding dispers-
al and adult survival within a metapopulation of Common Terns Sterna hirundo. Ibis
(Lond. 1859) 156:534–547. http://dx.doi.org/10.1111/ibi.12161.
Burg, T.M., Croxall, J.P., 2001. Global relationships amongst black-browed and grey-head-
ed albatrosses: analysis of population structure using mitochondrial DNA and
microsatellites. Mol. Ecol. 10, 2647–2660.
Burger, J., Gochfeld, M., 2016. Habitat, Population Dynamics, and Metal Levels in Colonial
Waterbirds: A Food Chain Approach. CRC Press.
Burson, S.L., 1990. Population genetics and gene flow of the Common Tern. Condor 92,
182–192.
Chapuis, M.-P., Estoup, A., 2007. Microsatellite null alleles and estimation of population
differentiation. Mol. Biol. Evol. 24, 621–631.
Cleaves, H., 1913. What the American Bird Banding Association has accomplished during
1912. Auk 30, 248–261.
Courtney, P.A., Blokpoel, H., 1983. Distribution and numbers of common terns on the
lower Great Lakes during 1900–1980: a review. Colon. Waterbirds 6:107–120.
http://dx.doi.org/10.2307/1520977.
Cuthbert, F.J., Wires, L.R., Timmerman, K., 2003. Status assessment and conservation. Rec-
ommendations for the Common Tern (Sterna hirundo) in the Great Lakes Region. Ft.
Snelling, MN.
Dempster, A., Laird, N., Rubin, D., 1977. Maximum likelihood from incomplete data via the
EM algorithm. J. R. Stat. Soc. Ser. B 39, 1–38.
DiCostanzo, 1980. Population dynamics of a common tern colony. J. Ornithol. 51,
229–243.
Draheim, H.M., Baird, P., Haig, S.M., 2012. Temporal analysis of mtDNA variation reveals
decreased genetic diversity in Least Terns. Condor 114:145–154. http://dx.doi.org/
10.1525/cond.2012.110007.
Dudaniec, R.Y., Schlotfeldt, B.E., Bertozzi, T., Donnellan, S.C., Kleindorfer, S., 2011. Genetic
and morphological divergence in island and mainland birds: informing conservation
priorities. Biol. Conserv. 144:2902–2912. http://dx.doi.org/10.1016/j.biocon.2011.08.
007.
Duncan, R.P., Blackburn, T.M., 2007. Causes of extinction in island birds. Anim. Conserv.
10:149–150. http://dx.doi.org/10.1111/j.1469-1795.2007.00110.x.
Earl, D.A., 2012. STRUCTURE HARVESTER: a website and program for visualizing
STRUCTURE output and implementing the Evanno method. Conserv. Genet. Resour.
4, 359–361.
Esler, D., 2000. Applying metapopulation theory to conservation of migratory birds.
Conserv. Biol. 14:366–372. http://dx.doi.org/10.1046/j.1523-1739.2000.98147.x.
Evanno, G., Regnaut, S., Goudet, J., 2005. Detecting the number of clusters of individuals
using the software STRUCTURE: a simulation study. Mol. Ecol. 14, 2611–2620.
Excoffier, L., Lischer, H., 2010. Arlequin suite ver 3.5: a new series of programs to perform
population genetics analyses under Linux and Windows. Mol. Ecol. Notes 10,
564–567.
Fox, G.A., 2001. Wildlife as sentinels of human health effects in the Great Lakes—St. Law-
rence basin. Environ. Health Perspect. 109, 853–861.
Frankham, R., 2005. Genetics and extinction. Biol. Conserv. 126:131–140. http://dx.doi.
org/10.1016/j.biocon.2005.05.002.
Friesen, V.L., 2015. Speciation in seabirds: why are there so many species … and why
aren't there more? J. Ornithol. 156:27–39. http://dx.doi.org/10.1007/s10336-015-
1235-0.
Friesen, V.L., Burg, T.M., McCoy, K.D., 2007. Mechanisms of population differentiation in
seabirds. Mol. Ecol. 16, 1765–1785.
Fronhofer, E.A., Kubisch, A., Hilker, F.M., Hovestadt, T., Poethke, H.J., 2012. Why are meta-
populations so rare? Ecology 93, 1967–1978.
Gilbertson, M., Morris, R.D., Hunter, R.A., 1976. Abnormal chicks and PCB residue
levels in eggs of colonial birds on the Lower Great Lakes (1971–1973). Auk 93,
434–442.
Given, A., Mills, J., Baker, A., 2002. Isolation of polymorphic microsatellite loci from the
red-billed gull (Larus novaehollandiae scopulinus). Mol. Ecol. Notes 2, 416–418.
271
Gochfeld, M., Burger, J., Christie, D.A., 2016. Common Tern (Sterna hirundo). In: del Hoyo,
J., Elliott, A., Sargatal, J., Christie, D.A., de Juana, E. (Eds.), Handbook of the Birds of the
World Alive. Lynx Edicions, Barcelona.
Goudet, F., 2001. FSTAT, a Program to Estimate and Test Gene Diversities and Fixation In-
dices (Version 2.9.3).
Gronewold, A., Fortin, V., Lofgren, B., Clites, A., Stow, C., 2013. Coasts, water levels, and cli-
mate change: a Great Lakes perspective. Clim. Chang. 120, 697–711.
Hanski, I., 1991. Single-species metapopulation dynamics: concepts, models and observa-
tions. Biol. J. Linn. Soc. 42, 17–38.
Hanski, I., 1998. Metapopulation dynamics. Nature 396, 41–49.
Harrison, S., 1991. Local extinction in a metapopulation context: an empirical evaluation.
Biol. J. Linn. Soc. 42, 73–88.
Haymes, G.T., Blokpoel, H., 1978. Seasonal distribution and site tenacity of the Great Lakes
common tern. Bird-Banding 49, 142–151.
Hernández-Matías, A., Jover, L., Ruiz, X., 2003. Predation on common tern eggs in relation
to sub-colony size, nest aggregation and breeding synchrony. Waterbirds 26,
280–289.
Kelly, J.F., Hutto, R.L., 2005. An east-west comparison of migration in North American
wood warblers. Condor 107, 197–211.
Kopelman, N., Mayzel, J., Jakobsson, M., Rosenberg, N., Mayrose, I., 2015. Clumpak: a pro-
gram for identifying clustering modes and packaging population structure inferences
across K. Ecol. Res. 15, 1179–1191.
Kraaijeveld, K., 2008. Non-breeding habitat preference affects ecological speciation in mi-
gratory waders. Naturwissenschaften 95, 347–354.
Kress, S., Weinstein, E., Nisbet, I., Shugart, G., Scharf, W.C., Blokpoel, H., Smith, G.A.,
Karwowski, K., Maxwell, G.R.I., Chapdelaine, G., Montevecchi, W.A., Lock, A.R.,
Smith, C.F., Miller, E., Spendelow, J.A., Gochfeld, M., Burger, J., Erwin, R.M., 1983. The
status of tern populations in northeastern United States and adjacent Canada.
Colon. Waterbirds 6, 84–106.
Lande, R., 1988. Genetics and demography in biological conservation. Science 241,
1455–1460.
Lebreton, J.D., Hines, J.E., Pradel, R., Nichols, J.D., Spendelow, J.A., 2003. Estimation by
capture-recapture of recruitment and dispersal over several sites. Oikos 101,
253–264.
Manel, S., Holderegger, R., 2013. Ten years of landscape genetics. Trends Ecol. Evol. 28:
614–621. http://dx.doi.org/10.1016/j.tree.2013.05.012.
Miller, M.P., Mullins, T.D., Haig, S.M., 2013. Genetic structure, diversity and
subspecies status of gull-billed terns (Gelochelidon nilotica) from the United
States terns (Gelochelidon nilotica) from the United States. Waterbirds 36,
310–318.
Morris, R.D., Hunter, R.A., 1976. Factors influencing desertion of colony sites by common
terns (Sterna hirundo). Can. Field. Nat. 90, 137–143.
Morris, R.D., Blokpoel, H., Tessier, G.D., 1992. Management efforts for the conservation of
common tern Sterna hirundo colonies in the Great Lakes: two case histories. Biol.
Conserv. 60, 7–14.
Morris, R.D., Weseloh, D.V., Cuthbert, F.J., Pekarik, C., Wires, L.R., Harper, L., 2010. Distribu-
tion and abundance of nesting common and Caspian terns on the North American
Great Lakes, 1976 to 1999. J. Great Lakes Res. 36, 44–56.
Morris, R.D., Weseloh, D.V., Wires, L.R., Pekarik, C., Cuthbert, F.J., Moore, D.J., 2011. Popu-
lation trends of ring-billed gulls breeding on the North American Great Lakes, 1976 to
2009. Waterbirds 34, 202–212.
Morris, R.D., Pekarik, C., Moore, D.J., 2012. Current status and abundance trends of Com-
mon Terns breeding at known coastal and inland nesting regions in Canada. Water-
birds 35, 194–207.
Munilla, I., Genovart, M., Paiva, V.H., Velando, A., 2016. Colony foundation in an oceanic
seabird. PLoS One 11, 1–24.
Narum, S.R., 2006. Beyond Bonferroni: less conservative analyses for conservation genet-
ics. Conserv. Genet. 7, 783–787.
Nisbet, I.C.T., 2002. In: Poole, A., Gill, F. (Eds.), Common tern (Sterna hirundo). The Birds of
North America, Inc., Philadelphia, PA.
Nisbet, I.C.T., Mostello, C.S., Veit, R.R., Fox, James, W., 2011. Migrations and winter quarters
of five common terns tracked using geolocators. Waterbirds 34, 791–803.
Nisbet, I.C.T., Arnold, J.M., Oswald, S.A., 2017. Common tern (Sterna hirundo) [revision]. In:
Rodewald, P. (Ed.), The Birds of North America. Cornell Lab of Ornithology, Ithaca, NY
(in press).
Olsen, J.B., Beacham, T.D., Wetklo, M., Seeb, L.W., Smith, C.T., Flannery, B.G., Wenburg, J.K.,
2010. The influence of hydrology and waterway distance on population structure of
Chinook salmon Oncorhynchus tshawytscha in a large river. J. Fish Biol. 76:
1128–1148. http://dx.doi.org/10.1111/j.1095-8649.2010.02554.x.
Oomen, R.A., Reudink, M.W., Nocera, J.J., Somers, C.M., Green, M.C., Kyle, C.J., 2011. Mito-
chondrial evidence for panmixia despite perceived barriers to gene flow in a widely
distributed waterbird. J. Hered. 102:584–592. http://dx.doi.org/10.1093/jhered/
esr055.
Oro, D., 2003. Managing seabird metapopulations in the Mediterranean: constraints and
challenges. Sci. Mar. 67, 13–22.
Osborne, M., Sharp, A., Monzingo, J., Propst, D.L., Turner, T.F., 2012. Genetic analysis sug-
gests high conservation value of peripheral populations of Chihuahau chub (Gila
nigrescens). Conserv. Genet. 13, 1317–1328.
Palestis, B.G., Hines, J.E., 2015. Adult survival and breeding dispersal of Common Terns
(Sterna hirundo) in a declining population. Waterbirds 38, 221–228.
Pavlacky, D.C.J., Possingham, H.P., Lowe, A.J., Prentis, P.J., Green, D.J., Goldizen, A.W., 2012.
Anthropogenic landscape change promotes asymmetric dispersal and limits regional
patch occupancy in a spatially structured bird population. J. Anim. Ecol. 81:940–952.
http://dx.doi.org/10.1111/j.1365-2656.2012.01975.x.
Peakall, R., Smouse, P.E., 2006. GenAlEx 6: genetic analysis in Excel. Population genetic
software for teaching and research. Mol. Ecol. Notes 6, 288–295.
272 P. Szczys et al. / Biological Conservation 209 (2017) 263–272
Peakall, R., Smouse, P., 2012. GenAlEx 6.5: genetic analysis in Excel. Population
genetic software for teaching and research – an update. Bioinformatics 28,
2537–2539.
Pritchard, J.K., Stephens, M., Donnelly, P.J., 2000. Inference of population structure using
multilocus genotype data. Genetics 155, 945–949.
Puechmaille, S.J., 2016. The program STRUCTURE does not reliably recover the correct
population structure when sampling is uneven: subsampling and new estimators al-
leviate the problem. Mol. Ecol. Resour. 16, 608–627.
Quinn, J., Morris, R., Blokpoel, H., 1996. Design and management of bird nesting habitat:
tactics for conserving colonial waterbird biodiversity on artificial islands in Hamilton
Harbour, Ontario. Can. J. Fish. Aquat. Sci. 53, 45–57.
Raymond, M., Rousset, F., 1995. GENEPOP (version 1.2): population genetics software for
exact tests and ecumenicism. Heredity (Edinb). 86, 248–249.
Roberts, J.H., Angermeier, P.L., Hallerman, E.M., 2013. Distance, dams and drift: what
structures populations of an endangered, benthic stream fish? Freshw. Biol. 58:
2050–2064. http://dx.doi.org/10.1111/fwb.12190.
Rolshausen, G., Segelbacher, G., Hobson, K.A., Schaefer, H.M., 2009. Contemporary evolu-
tion of reproductive isolation and phenotypic divergence in sympatry along a migra-
tory divide. Curr. Biol. 19, 2097–2101.
Rousset, F., 2000. Genetic differentiation between individuals. J. Evol. Biol. 13, 58–62.
Sæther, B.E., Coulson, T., Grøtan, V., Engen, S., Altwegg, R., Armitage, K.B., Barbraud, C.,
Becker, P.H., Blumstein, P.T., Dobson, F.S., Festa-Bianchet, M., Gaillard, J.M., Jenkins,
A., Jones, C., Nicoll, C., Norris, K., Oli, M.K., Ozgul, A., Weimerskirch, H., 2013. How
life history influences population dynamics in fluctuating environments. Am. Nat.
182, 743–759.
Shugart, G.W., Scharf, C., 1983. Common terns in the northern Great Lakes: current status
and population trends. J. Ornithol. 54, 160–169.
Simberloff, D., 1986. The proximate causes of extinction. In: Raup, D.M., Jablonski, D.
(Eds.), Patterns and Processes in the History of Life. Springer, Berlin/Heidelberg,
pp. 259–276.
Spendelow, J.A., Nichols, J.D., Nisbet, I.C.T., Hays, H., Cormons, G.D., 1995. Estimating annu-
al survival and movement rates of adults within a metapopulation of Roseate Terns.
Ecology 76, 2415–2428.
Sruoga, A., Butkauskas, D., Prakas, P., Paulauskas, A., 2006. Evaluation of the genetic struc-
ture of the breeding Common Tern (Sterna hirundo) population by means of micro-
satellite markers. Biologija 2006, 47–52.
Sutherland, C., Elston, D.A., Lambin, X., 2012. Multi-scale processes in metapopulations:
contributions of stage structure, rescue effect, and correlated extinctions. Ecology
93, 2465–2473.
Szczys, P., Hughes, C.R., Kesseli, R.V., 2005. Novel microsatellite markers used to deter-
mine the population genetic structure of the endangered Roseate Tern, Sterna
dougallii, in Northwest Atlantic and Western Australia. Conserv. Genet. 6:461–466.
http://dx.doi.org/10.1007/s10592-004-4975-6.
Szczys, P., Nisbet, I.C.T., Wingate, D.B., 2012. Conservation genetics of the Common Tern
(Sterna hirundo) in the North Atlantic region; implications for the critically endan-
gered population at Bermuda. Conserv. Genet. 13:1039–1043. http://dx.doi.org/10.
1007/s10592-012-0351-0.
Szczys, P., Lamothe, K.A., Druzyaka, A., Poot, M.J.M., Siokhin, V., van der Winden, J., 2016.
Range-wide patterns of population differentiation of Eurasian black terns (Chlidonias
niger niger) related to use of discrete post-nuptial staging sites. J. Ornithol. http://dx.
doi.org/10.1007/s10336-016-1408-5.
Szostek, Becker, P.H., 2012. Terns in trouble: demographic consequences of low bredding
success and recruitment on a common tern population in the German Wadden Sea.
J. Ornithol. 153, 313–326.
Trumpickas, J., Shuter, B., Minns, C., 2009. Forecasting impacts of climate change on Great
Lakes surface water temperatures. J. Great Lakes Res. 35, 454–463.
Wallace, S.J., Wolf, S.G., Bradley, R.W., Harvey, L.L., Friesen, V.L., 2015. The influence of bio-
geographical barriers on the population genetic structure and gene flow in a coastal
Pacific seabird. J. Biogeogr. 42, 390–400.
Welch, A.J., Fleischer, R.C., James, H.F., Wiley, A.E., Ostrom, P.H., Adams, J., Duvall, F.,
Holmes, N., Hu, D., Penniman, J., Swindle, K.A., 2012. Population divergence and
gene flow in an endangered and highly mobile seabird. Heredity (Edinb). 109, 19–28.
Wetlands International, 2006. Waterbird Population Estimates. fourth ed. (Wageningen,
Netherlands).
Wilson, G.A., Rannala, B., 2003. Bayesian inference of recent migration rates using
multilocus genotypes. Genetics 163, 1177–1191.
Wilson, S., Bazin, R., Calvert, W., Doyle, T.J.T.J., Earsom, S.D., Oswald, S.A., Arnold, J.M.,
2014. Abundance and trends of colonial waterbirds on the large lakes of southern
Manitoba. Waterbirds 37:233–244. http://dx.doi.org/10.1675/063.037.0302.