A N I M A L M O V E M E N T O N S H O R T A N D L O N G

T I M E S C A L E S A N D T H E E F F E C T O N G E N E T I C

D I V E R S I T Y I N C O L D - A D A P T E D S P E C I E S

Vendela Kempe Lagerholm

Animal movement on short and long time

scales and the effect on genetic diversity in

cold-adapted species

Vendela Kempe Lagerholm

©Vendela Kempe Lagerholm, Stockholm University 2016

ISBN 978-91-7649-421-9

Cover by Kerstin Kempe

Printed in Sweden by Holmbergs, Malmö 2016

Distributor: Department of Zoology, Stockholm University

The thesis is based on the following papers, which are referred to

in the text by their Roman numerals:

I Lagerholm VK, Norén K, Ehrich D, Ims RA, Killengreen ST,

Abramson NI, Angerbjörn A, Henttonen H, & Dalén L. (2016). Run to

the hills: gene flow among mountain areas leads to low genetic

differentiation in the Norwegian lemming. – Manuscript

II Lagerholm VK, Sandoval-Castellanos E, Ehrich D, Abramson NI,

Nadachowski A, Kalthoff DC, Germonpré M, Angerbjörn A, Stewart

JR, & Dalén L. (2014). On the origin of the Norwegian lemming.

Molecular Ecology 23(8):2060-2071

III Lagerholm VK, Sandoval-Castellanos E, Vaniscotte A, Potapova O,

Tomek T, Bochenski ZM, Shepherd P, Barton N, Van Dyck M-C,

Miller R, Höglund J, Yoccoz NG, Dalén L, & Stewart JR. (2016).

Range shifts or extinction? Ancient DNA and distribution modelling

reveal past and future responses to climate warming in cold-adapted

birds. – Manuscript

IV Smith S, Sandoval-Castellanos E, Lagerholm VK, Napierala H, Sablin

M, Nyström J, Fladerer FA, Germonpré M, Wojtal P, Stewart JR, &

Dalén L. (2014). Non-receding hare lines: genetic continuity since the

Late Pleistocene in European mountain hares (Lepus timidus). –

Manuscript

Candidate contributions to thesis articles*

I II III IV

Conceived the

study

Minor Minor

Designed the

study

Substantial Substantial Substantial

Collected the

data

Substantial Substantial Substantial Significant

Analysed the

data

Substantial Substantial Substantial Substantial

Manuscript

preparation

Substantial Substantial Substantial Substantial

* Contribution Explanation

Minor: contributed in some way, but contribution was limited.

Significant: provided a significant contribution to the work.

Substantial: took the lead role and performed the majority of the work.

I am also a co-author of the following article, which is not included in this

thesis but was published during my doctoral studies at Stockholm

University:

Elmhagen B, Destouni G, AngerbjörnA, Borgström S, Boyd E, Cousins

SAO, Dalén L, Ehrlén J, Ermold M, Hambäck PA, Hedlund J, Hylander K,

Jaramillo F, Lagerholm VK, Lyon SW, Moor H, Nykvist B, Pasanen-

Mortensen M, Plue J, Prieto C, Van der Velde Y, & Lindborg R. (2015).

Interacting effects of change in climate, human population, land use, and

water use on biodiversity and ecosystem services. Ecology and Society

20(1): 23

Contents

Introduction ................................................................................................ 10

The past holds the key to the future ................................................................... 11

Objectives ................................................................................................................. 14

Study species ........................................................................................................... 15

Methods ....................................................................................................... 19

Samples and DNA extraction................................................................................. 19

DNA amplification and sequencing ....................................................................... 21

Ancient DNA precautions ....................................................................................... 22

Radiocarbon dating ................................................................................................. 23

Data analyses ........................................................................................................... 23

Results ......................................................................................................... 25

Movement on a short time scale .......................................................................... 25

Movement over longer time scales ...................................................................... 26

Discussion ................................................................................................... 33

Sammanfattning ........................................................................................ 38

Acknowledgments ..................................................................................... 40

References .................................................................................................. 41

10

Introduction

The genetic diversity within species and the way it is geographically

structured is strongly affected by population size and movement across the

landscape. In small and isolated populations, neutral variation is lost over

time through the random process of genetic drift (Nei et al. 1975) and this

stochasticity may also cause differentiation between different parts of the

range. Mutations and, most importantly, gene flow are counteracting forces

that bring new variation to a population and can disrupt local genetic

structures over large areas.

Many northern species today exhibit cyclic population dynamics, with

substantial fluctuations in numbers between years. This could have complex

genetic outcomes, where on one hand the pronounced drift during the low

phase leads to comparatively low diversity (Elton 1924; Nei et al. 1975),

while an increased dispersal during peak years potentially could act as pulses

of gene flow that maintains a high genetic variation (e.g. Ehrich & Stenseth

2001; Norén & Angerbjörn 2014).

Current patterns of variation are not only shaped by contemporary gene

flow, but have also been influenced by movement on longer time scales.

This is perhaps most obvious in the arctic and alpine regions that were

covered by continental glaciers during the height of the last Ice Age, ca 20

thousand years ago. Consequently, modern populations should all be

descendants to immigrants from the surrounding areas that colonised these

regions following the ice retreat. The mode of these post-glacial

colonisations have had a great impact on modern species diversities, and to

get a more comprehensive understanding of taxa it can thus be important to

look deeper into their histories (Hewitt 2000).

11

The past holds the key to the future

The last 2.6 million years of the earth’s history, called the Quaternary, has

been characterised by large-scale climatic fluctuations. Long ice ages have

been repeatedly interrupted by shorter interglacials, in transitions that during

the more recent part of the Quaternary (Fig. 1) came to occur in 100

thousand year cycles (Imbrie et al. 1992). These followed a saw-toothed

pattern, where full glacial conditions were reached after long periods of

gradual cooling and build-up of ice volume, while the transitions to

interglacial conditions have been characterised by rapid warming (Broecker

& van Donk 1970). The last switch from the Pleistocene glacial period to the

current interglacial period, the Holocene, occurred 11,700 years before the

present (yr BP), and ice core data have shown that half of the total 15 ºC

temperature increase occurred within less than 15 years (Taylor et al. 1997).

Continued climate warming, this time induced by humans since the start of

the industrial revolution, is now being suggested to have pushed the Earth

into the Anthropocene epoch (Waters et al. 2014).

Figure 1. Past climate changes. (A): Temperature fluctuations over the last four glacial

cycles, obtained from the Vostok ice core in Antarctica (modified from Petit et al. 1999). The

top shaded area indicates the Pleistocene period in blue, with the Late Pleistocene in darker

blue, as well as the Holocene period in red. (B): Northern hemisphere temperature

fluctuations over the last 140 thousand years, obtained from several Greenland ice cores

(modified from Johnsen et al. 2001).

A

B

12

The Quaternary climate fluctuations caused substantial rearrangements of

habitat and species distributions. These range dynamics have generally been

regarded as expansions and contractions, where taxa advanced and retreated

in a north/south direction over time (Hewitt 1999; Hewitt 2000). However,

climate changes often result in a combination of the two contrasting

processes, and what determines if the total distribution of a taxa will

increase or decrease is simply whether the expanding range is more

successful than the contracting one (Jackson & Sax 2009). For instance,

cold-adapted species are expected to have had their greatest distributions

during periods of climate cooling when they expanded their ranges to more

southern and lowland areas, but this would also have been coupled with

retreats from some regions at the northern or alpine limits of their

distribution as these became covered by the growing ice sheets (Hewitt

2001; Stewart et al. 2010). These increases and declines in species

distributions also led to changes in connectivity, and it has been proposed

that the recurrent population fragmentation and isolation in refugia led to

accelerated evolutionary rates during the Quaternary (Hewitt 1996; Rand

1948). However, this has been difficult to evaluate since modern genetic data

and fossil morphology often lack sufficient resolution to specify the exact

locations and timings of past speciation events (Ho et al. 2011; Hofreiter &

Barnes 2010; Lister 2004).

An underlying assumption to these expansion/contraction dynamics,

originally made by Darwin (1859) and later revived by Eldredge (1989), is

that taxa are able to track their habitat as it is shifting in space due to

climatic changes, thus avoiding the need for adaptation. However, this

simplified view has recently been challenged, and it is uncertain to what

degree taxa are really able to keep pace when the rate of habitat

rearrangements is too high (e.g. Kerr et al. 2015). One important pre-

requisite for successful shifts to new areas could be the geographic

connectivity between suitable habitat patches (Hodgson et al. 2012),

especially for species with low dispersal capacity. Further, inter- and

intraspecific interactions might also affect the ability to track shifts in

habitat, and to do so with small mismatches in relation to climate. The

13

potential rate of dispersal for taxa is consequently not always equal to their

actual rate of range shift. For instance, the expansion of a species into a new

area might be slowed down by the preceding response of its prey or the

persistence of declining resident taxa (Jackson & Sax 2009), thereby forcing

it to linger in areas of unfavourable environmental conditions and increasing

the risk of population extinctions. Similarly, the early colonisers could

obstruct the subsequent arrival of conspecific individuals from more distant

parts of the range through competition (Hewitt 1996; Hewitt 1999). A

climate-induced expansion of a dominant species can conversely also drive

the contraction of a competitor’s range (Hersteinsson & MacDonald 1992).

The concept of habitat tracking is also problematic from a genetic

perspective, especially when it comes to declines in available habitat. More

specifically, we cannot assume that a range contraction equals to a

population contraction, in the sense that the genetic variation at the trailing

range margin survives as the species’ range decreases (Hewitt 1996). For

cold-adapted species, this means that populations that have been at the

leading edge during a glacial expansion are not necessarily involved in the

shifts back to northern areas during the transition from glacial to interglacial

conditions. Climate warming might then rather cause extinctions of these

southern populations, and instead it is only the individuals at the species’

northern margin that are involved in the shift back to refugial areas (Bennett

et al. 1991). The trailing edge is in this case vanishing, more than

contracting, and assuming that the genetic variation within the given species

is geographically structured, such a pattern of local extinctions and

colonisation by few founders would deplete the gene pool (e.g. Hewitt

1996).

A parameter that has greatly influenced the genetic diversity in cold-adapted

species is consequently how they coped with the rapid climate warming at

the last glacial/interglacial transition. This a key question to resolve, not only

to improve the knowledge of our current biota but also to ensure reliable

predictions of future changes, since the effects of on-going climate warming

are expected to be most severe in northern biomes (Pithan & Mauritsen

2014; Urban 2015). It has historically been difficult to study the genetic

effects of past range contractions, since modern phylogeographic analyses

14

only provides information about the history of surviving lineages. However,

recent years’ development in the field of ancient DNA has started to open

the black box, and a growing number of studies have retrieved genetic

information from the fossil remains of cold-adapted animals that had

expanded populations at the end of the last glaciation. Apart from the famous

megafauna extinctions that occurred during this time period (e.g. Barnosky

et al. 2004), it is becoming clear that also the surviving arctic species were

severely affected by the post-glacial habitat transformations (Hofreiter &

Barnes 2010). In fact, many studies have revealed unique lineages in the

non-refugial areas that did not endure until the present, and large losses of

genetic diversity following the glacial/interglacial transition, suggesting that

local extinctions of southern populations were the dominant process for

cold-adapted species during past climate warming (Campos et al. 2010;

Dalén et al. 2007; Palkopoulou et al. 2016).

Objectives

To ensure the future viability of cold-adapted species in an era of climate

warming, it is crucial to understand not only their present status and habitat

requirements, but also how they respond to environmental alterations. The

last transition from glacial to interglacial conditions here provides a unique

case study, enabling researchers to analyse genetic changes over time and

how these are related to the palaeoclimate record. However, a majority of

these studies have focused on charismatic large mammals, and the inferences

might then be complicated by the coinciding expansion of humans across

Eurasia. Small herbivores, which likely were much less affected by

prehistoric hunting, therefore constitute an appealing complement that

allows analyses of the pure climate effect on past population changes. In this

thesis, I have focused on three types of cold-adapted herbivores; true

lemmings (Lemmus), ptarmigan (Lagopus), and hares (Lepus), whose

modern distributions stretch from the exposed tundra to the subarctic

moorlands and taiga. With their complementing ecologies, they together

make a good model system to investigate the effects of climate-induced

range shifts in the northern biome. Although their remains are abundant in

15

the late glacial fossil record of Europe, they have so far received little

interest from palaeogeneticists. In fact, birds in general have often been

overlooked as a potential source of genetic information due to that their

hollow bones are expected to contain poorly preserved DNA. Their flight

capability, however, adds an important dimension to the study of habitat

tracking, since it could enable a less restricted dispersal across fragmented

landscapes.

Paper I of this thesis investigates the importance of lemming dispersal on a

short time scale in shaping modern patterns of genetic diversity, while

climate-induced dispersal and isolation on longer time scales have been the

topic for Paper II (lemmings), III (ptarmigan) and IV (hares).

Study species

Lemmings constitute the basic resource in many northern ecosystems, and

together with co-occurring prey species like ptarmigan and mountain hares

they are vital for the function of the Arctic ecosystems. During the last

glacial period, these taxa had widespread populations across the vast tundra-

steppes of midlatitude Europe, and the rich collections of predator-

assembled fossil remains suggest that they played a key role in the food

chain also during the last Ice Age (e.g. Averianov 2001; Nadachowski 1982;

Tyrberg 1998). While morphological species identification is possible for the

ptarmigan (Lagopus) and hare (Lepus) bones, the identity of the remains

from true lemmings (Lemmus) is often unclear.

The Norwegian lemming (L. lemmus) is one of two Lemmus species that

lives in Europe today. It inhabits the Fennoscandian mountain tundra

(Hansson 1999) where it is the only endemic mammal. This restricted range

(Fig. 2) has for a long time caused speculations about the species’ origin,

since the region is thought to have been completely glaciated during the Last

Glacial Maximum, LGM (Ekman 1922). Although this would imply that the

Norwegian lemming originated during post-glacial times, the species is

morphologically and genetically too distinct to have originated from its

closest living relative, the Siberian lemming L. sibiricus, at the end of the

16

Pleistocene (e.g. Fedorov & Stenseth 2001). Apart from the mysterious

history, the Norwegian lemming is also famous for its dramatic population

cycles. The very rapid increases during peak years are facilitated by the

species’ ability to breed all year round which is likely an adaptation to the

long winters of the north (Ims et al. 2011). These peaks can occur in

synchrony across much of its range, with individuals sometimes moving

over 100 km down into forest and farmland regions (Hansson 1999;

Henttonen & Kaikusalo 1993). Although dispersing lemmings may

overwinter in such lowland valleys, there are no records of uphill return

movements to their core tundra habitat, something which would be necessary

to enable gene flow among mountain tundra areas separated by forest

(Østbye et al. 1993). Another special feature of the Norwegian lemming is

the local migrations that usually take place between their summer and winter

habitats (Henttonen & Kaikusalo 1993). This has been explained by the

strong seasonality in its mountainous habitat and does not occur in the other

Lemmus species, which live in the relatively flat and moist arctic tundra of

Russia and America that allows suitable environmental conditions during

both summer and winter. In fact, Ekman proposed in 1920 that this

behaviour evolved during the last glacial period when the species was

hypothesised to have survived in ice-free coastal areas of Norway (Ekman

1920). Since that time, the Norwegian lemmings’ high mobility has been

intensively investigated, but no studies have analysed its impact on the

genetic structure and variation of the species.

The Lagopus genus is represented by two sister species in Europe, the

willow (L. lagopus) and rock (L. muta) ptarmigan, which are thought to have

diverged approximately 1 million years ago (Drovetski 2003). Their modern

ranges (Fig. 2) mainly comprise high latitude regions, where rock ptarmigan

occupies the more exposed rocky terrains with sparse vegetation, and willow

ptarmigan the lower altitudes where birch forest and willow shrubs dominate

the landscape (Svensson 1999). In addition, rock ptarmigan also have

isolated mountain populations in the high-alpine regions of Southern Europe.

Both species are sedentary ground-dwelling birds, and except for willow

ptarmigan in the British Isles (L. lagopus scotica) they all moult to a white

winter plumage (Freeland et al. 2007). This loss in the British Isles is

17

probably a recent Holocene adaptation to the milder winters in this region

where a snow camouflage is often disadvantageous. Interestingly, modern

ptarmigan of both species are generally lighter built than their Ice Age

counterparts, possibly suggesting differences in diet or degree of

sedentariness (Bochenski 1985; Stewart 1999). So far, however, it has not

been resolved whether this is due to adaptive changes to post-glacial

environmental conditions, or if the fossil remains in fact represent

specialised glacial lineages that are now extinct.

The mountain hare (Lepus timidus) has a disjunct modern European

distribution alike that of the rock ptarmigan, although the hare has a broader

ecological niche that also includes the taiga zone (Fig. 2). Further, all hares

except the Irish (L. timidus hibernicus) also develop a white winter coat

during the autumn. Along its southern range margin, the cold-adapted

mountain hare overlaps in distribution with the temperate European brown

hare (L. europaeus), which often results in competitive exclusion of L.

timidus by its dominant cousin (Thulin 2003). Having had a continuous

glacial distribution across the Mammoth steppe biome (Kahlke 1999), the

species has thus experienced large range shifts and contractions during post-

glacial times (Thulin 2003). However, even though having a fragmented

distribution today and despite having low natal dispersal rates and high site

fidelity (Dahl & Willebrand 2005), the mountain hare has a large genetic

diversity with surprisingly little structure between the isolated populations

(Melo-Ferreira et al. 2007). This has sparked a debate regarding the

phylogeographic history of the species (Hamill et al. 2006).

18

Figure 2. Modern European ranges of Norwegian lemmings (A), mountain hares (B), willow

ptarmigan (C) and rock ptarmigan (D).

A

C D

B

19

Methods

Samples and DNA extraction

For Paper I, modern Norwegian lemming samples were collected from

throughout the species’ complete Fennoscandian distribution (Fig. 3A). The

material mainly consisted of tissue samples stored in ethanol that were taken

either from trapped individuals or lemming carcasses between the years

1997 and 2012. In addition, the data set was complemented with museum

samples from areas were no fresh material was available, and consisted of

bone or tissue material collected between 1960 and 2007. Samples were also

gathered from the Norwegian lemmings’ sister species, the Siberian

lemming (Lemmus sibiricus), to serve as a reference to the level of intra-

specific differentiation within Fennoscandia.

For Papers II, III, and IV, we collected bone remains of lemmings,

ptarmigan and hares from a range of palaeontological sites across mid-

latitude Europe, situated in the former tundra-steppe region that stretched

between the Scandinavian and the southern alpine ice sheets during the Late

Pleistocene glacial period (Fig. 3B). These had previously been identified to

species or genus level, and spanned an age of between 48 thousand (k) years

before the present (yr BP) and up until the final glacial/interglacial

transition, ca 10 kyr BP. These remains were found in areas outside, or in

exceptional cases at the edge, of the study species’ current distributions and

might thus either represent extinct lineages, or direct ancestors of modern

arctic and alpine populations that succeeded in tracking their habitats as

these shifted at the end of the Pleistocene. In addition to the midlatitude Late

Pleistocene samples, we also included early to mid-Holocene (3-8 kyr BP)

lemming bones from a northern, previously glaciated, cave site located

within the modern range of the Norwegian lemming.

20

Figure 3. (A) Geographic locations of all modern Norwegian lemming sample sites that

produced nuclear DNA for the data analyses of Paper I. (B) Palaeontological sample sites

from which successful ancient mitochondrial DNA could be retrieved (Paper II-IV). Lemming

sites are shown in yellow, willow and rock ptarmigan in red and blue and mountain hare in

green. The shaded areas indicate the maximum extent of the major Scandinavian and alpine

ice sheets during the Last Glacial Maximum (Mangerud et al. 2011; Svendsen et al. 2004).

A subset of the Norwegian and Siberian lemming samples included in Paper

I was used as modern reference material for the analyses in Paper II. For

Papers III and IV, published sequence data from earlier phylogeographic

studies provided a comprehensive representation of modern ptarmigan and

hare populations across the Holarctic. In addition, I further complemented

the ptarmigan dataset with new tissue samples from areas where no

published sequences were available.

DNA from the modern tissue material was extracted using either a

GeneMole MG10-000 robot with the MoleStrips DNA Tissue kit (Mole

Genetics), or manually using Qiagen’s QIAamp DNA mini kit or DNeasy

tissue kit. The bone samples were first powdered with a pestle and mortar or

sampled with a multi-tool drill. DNA was then extracted from the bone

powder following the protocols described in Palkopoulou et al. (2013) or

Fernandez et al. (2006). Due to post-mortem degradation, the DNA that is

preserved in fossil remains (called ancient DNA, aDNA) is fragmented and

occurs in few copy numbers. These extractions therefore followed protocols

A B

21

specifically developed to purify low quality samples and minimise the

presence of inhibitors in later amplification steps.

DNA amplification and sequencing

Microsatellites were chosen as the most suitable marker for Paper I to

investigate the genetic structure in modern Norwegian lemming populations.

These are short non-coding repeats that occur across the nuclear genome,

and their high mutation rates have made them popular to study neutral levels

of intraspecific variation. Since there were only a limited number of markers

available from related species, we developed twelve new microsatellite

primer pairs at Ecogenics (Zurich-Schlieren, Switzerland) from a number of

representative Norwegian lemming samples according to the method

described by Abdelkrim et al. (2009). The extracted DNA was then

amplified with the polymerase chain reaction (PCR) in three multiplexes,

each containing four fluorescently labelled primer pairs, using Qiagen’s

Type-it Microsatellite PCR Kit. Genotyping of the products was performed

on an ABI 3130xl with the GeneScan 500 LIZ size standard, and the alleles

were scored using GeneMapper v.4.0 (Applied Biosystems). After

evaluations of the obtained genotypes, only nine of the twelve developed

loci were used for the final data analyses.

The genetic analyses in Papers II, III and IV are based on mitochondrial

(mt) DNA, since this occurs in higher copy numbers than nuclear DNA and

thus is very suitable for ancient material (Hofreiter et al. 2001). It is

maternally inherited as a single locus, and since variation only arises through

new mutations it gives a poor resolution to detect individual-level

differences. However, it has been the marker of choice for investigating

deeper phylogenetic relationships within species, including those described

in this thesis. To enable comparisons with these previously published

modern sequences, we targeted the same mtDNA regions for our aDNA

datasets. For the ptarmigan study (Paper III) this meant parts of the

mitochondrial control region (CR) while for the lemming and hare studies

(Papers II and IV) parts of both the CR and the cytochrome b gene (CytB)

were analysed. However, due to the fragmented nature of aDNA it was not

22

possible to use the methods developed for modern samples, and instead the

chosen regions had to be amplified in shorter overlapping fragments using

specifically designed primers and PCR protocols developed for material with

low quantity DNA. The modern samples that had been collected to

complement the published sequence datasets were also amplified using these

new primers. All PCR products were purified and sequenced for both the

forward and reverse strands using an ABI 3130xl (Applied Biosystems) or a

MegaBace 1000 (GE Healthcare).

Ancient DNA precautions

Due to the degraded nature of aDNA there is an increased risk of

contamination as well as sequence errors. To ensure reliable results it is

therefore necessary to take certain precautions. All pre-PCR work on the

glacial remains was carried out in the ancient DNA laboratory at the

Swedish Museum of Natural History. I contributed to designing and building

this laboratory as an initial part of my thesis work, and consequently I did

not have to deal with the potential risk of contamination that may result from

previous studies of related taxa. Further, the laboratory is physically isolated

from the modern DNA and post-PCR facilities, and all lab equipment and

working surfaces were frequently sterilised with UV light, bleach or

hydrochloric acid. Blanks were extensively used in both extraction and PCR

reactions in order to monitor possible contamination, and results from a

minimum of two independent amplifications were required to enable the

identification of nucleotide misincorporations that otherwise lead to

erroneous DNA sequences. The pre-PCR work on the Holocene lemming

bones from Norway was performed in the ancient DNA laboratory at

Laboratoire d’Ecologie Alpine in Grenoble, France. No rodent samples had

been analysed there prior to this study, and equivalent sterilisation

procedures and use of negative controls were carried out as in the Swedish

laboratory. The sequences obtained from these samples only displayed

variation in nucleotide positions that had also been observed to be variable in

extant lemming populations. Since such a pattern is unlikely to be caused by

23

random PCR misincorporation, these samples were therefore not considered

necessary to replicate through multiple amplifications.

Radiocarbon dating

The bone remains analysed in this thesis all came from excavated sites

whose deposits had been previously studied by palaeontologists and

archaeologists. Consequently, the stratigraphic layers were often well

described and age determined, both by direct radiocarbon dating of selected

remains as well as inferred from the general faunal and palynological

composition of the deposits. Nevertheless, we chose to send eleven lemming

bones that had produced good quality DNA (included in Paper II) for dating

at the Oxford Radiocarbon Accelerator Unit. Due to their small size though,

only five samples contained enough collagen to give reliable age estimates.

Data analyses

The modern lemming dataset analysed in Paper I was divided into different

geographic sub-regions, for which a number of standard population genetic

statistics were calculated. To investigate how well these pre-defined areas

represented the true genetic structure of the species we performed different

clustering analyses, both using only the “blind” genetic data as well as

including geographic coordinates for all the samples. These coordinates were

further used to analyse patterns of isolation-by-distance across the

Norwegian lemming´s range, and the so called “genetic patch size” which

describes the geographic extent of detectable genetic structure.

The mtDNA analyses performed in Papers II, III and IV aimed to clarify

the evolutionary relationships between the modern populations and the old

bone remains. I performed phylogenetic tree reconstructions and skyline

plots of temporal demographic changes using the software BEAST

(Drummond et al. 2005) and/or built haplotype networks with the R-script

TempNet (Prost & Anderson 2011). The approximate ages of the old

samples were used as priors in the BEAST analyses, which allowed for

24

internal calibrations of the molecular clock (Drummond et al. 2002). This is

an additional advantage of using ancient DNA data, since it produces more

accurate divergence time estimates than traditional inferences that rely on

fossil-calibrated mutation rates (Ho et al. 2007).

In order to evaluate the most likely evolutionary histories of all the study

species, we used simulations coupled with Approximate Bayesian

Computation (ABC) to evaluate different population scenarios. Just as the

BEAST analyses, these simulations are based on coalescent theory, which

traces individual sequences backwards in time to the point when they

coalesce (Hudson 1990). In brief, we simulated datasets consisting of a large

number of genealogies under different evolutionary models. Summary

statistics from these datasets were then compared to our observed data, and

using a rejection procedure and the proportions of accepted simulations in

each model, we could select and further analyse the most probable historic

scenario (Beaumont et al. 2002).

The genetic analyses in Paper III were further complemented with species

distribution modelling (SDM), in order to investigate in more detail the

connections between past habitat alterations and ptarmigan population

responses. Climate data from throughout the species’ current ranges, as well

as from palaeontological sites with Lagopus spp. remains, was used to

calibrate the models and predict temporal changes in available habitat over

the last 20 thousand years. Further, we also used SDM to look forward in

time and make projections of ptarmigan distributions in a world of

continuing climate warming.

25

Results

Movement on a short time scale

The modern microsatellite study (Paper I) revealed that Norwegian

lemmings have a high genetic variation, and a surprisingly weak geographic

structure among the analysed sub-regions. The differentiation was mainly

seen as a south-west to north-east gradient across Fennoscandia, and the

clustering programs identified two major genetic clusters located in

southernmost Norway and Sweden, on one hand, and in the Kola Peninsula,

north eastern Norway and Finland on the other. The central parts of the

lemmings’ distribution comprised a broad transition zone consisting of

individuals with gradually admixed assignment proportions (Fig. 4). The

existence of a genetic cline was further supported by a pattern of isolation-by

distance observed across the species’ range. Spatial autocorrelation analyses

suggested a genetic patch size of approximately 100 km, meaning that

lemmings generally are more similar to each other within this range than to a

randomly sampled individual from the complete distribution. When

comparing the scale of genetic differentiation between the Fennoscandian

sub-regions, I found that for a given geographic distance the southern

populations were more similar to each other than the populations were in the

north-eastern part of the range. In summary, these results are consistent with

that lemming dispersal during peak years acts as pulses of gene flow that

homogenise the gene pool over vast geographic areas. Further, we propose

that the higher differentiation found between northern subpopulations can be

explained by the longer periods of interrupted cyclicity reported from this

region (Angerbjörn et al. 2001), which have increased the isolation of

lemmings in different mountain tundra areas.

26

Figure 4. Results from the genetic cluster analyses. The upper panel shows the barplot

obtained from the software Structure for the selected value of K=2, with individuals ordered

along a south-west to north-east gradient. The lower panel geographically display the

admixture proportions in the three genetic clusters inferred from the software TESS, with the

X and Y axis indicating longitude and latitude.

Movement over longer time scales

Paper II investigated the evolutionary history of the Norwegian lemming,

by comparing mitochondrial genetic variation in modern Fennoscandian (L.

lemmus) and Siberian (L. sibiricus) individuals with that retrieved from 23

European lemming bone remains of up to 48,000 yr BP in age. The results

showed that the glacial populations had a large mtDNA diversity that

appeared to be geographically structured, but that they were genetically

divergent from the two modern species. Further, both the phylogenetic

reconstructions and the coalescent simulations supported a pre-LGM

divergence time between the midlatitude populations and modern Norwegian

lemmings, therefore making a postglacial colonisation of Fennoscandia from

these surrounding areas highly unlikely (Fig. 5). Whereas the star-like L.

lemmus haplogroup appears to have been absent in the areas south of the

Scandinavian ice sheet up until the final de-glaciation, it is found in a cave in

Norway less than 3.5 thousand years later, often in the form of the modal

haplotype. Bone remains of Lemmus sp. dating back to a previous

interstadial ca 36 kyr BP have previously been reported from Norway

27

(Larsen et al. 1987), and the aDNA results from this study strongly suggests

that the endemic Norwegian lemming descends from a small population that

locally survived the LGM in one or more restricted ice free areas. One could

speculate that such a prolonged refugial isolation under extreme climate

conditions might have caused rapid evolutionary changes (Mayr 1963; Orr &

Smith 1998), and might potentially even represent the speciation event of L.

lemmus.

Figure 5. Temporal haplotype network (upper left) and phylogenetic tree (lower right) of

Holocene and Late Pleistocene European Lemmus spp. Black dots in the network represent

missing haplotypes, while empty circles correspond to a haplotype that is absent from that

temporal layer but present in the other. The dashed circles and connecting lines between the

two layers indicate the absence of the Fennoscandian haplogroup in the Late Pleistocene data

set. In the phylogeny, the positions of the Holocene Fennoscandian samples are highlighted

with yellow (modern Norwegian lemmings, L. lemmus) or brown (3 to 8 kyr BP old Lemmus

sp. from Norway) branches. A high mutation rate of 30 % Myr-1 has been used in the analysis,

and internal nodes with posterior probabilities above 0.8 are marked with asterix. The time

scale is presented in years before the present, with the ages of all samples being proportional

to their branch lengths.

28

In contrast to the lemming paper, I found that the modern European

haplogroups of both ptarmigan species (Paper III) were present also in the

Late Pleistocene European population (represented by 42 samples), with the

modal haplotypes being identical between the time periods (Fig. 6). This

pattern is consistent with the type of genetic founder effects expected during

postglacial colonisation waves (Hewitt 1999). Moreover, temporal genetic

continuity in Europe was also the most highly supported scenario in the

analyses of molecular variance (AMOVA). In addition, the habitat tracking

scenario also received the highest support in the rock ptarmigan ABC

analyses, but unfortunately the willow ptarmigan dataset was not informative

enough to give consistent support to any of the simulated scenarios. The

species distribution modelling revealed that this long-term survival of both

species might have been facilitated by a continuous availability of ptarmigan

habitats over the last 20 millennia, even though these habitats have shifted

substantially in space since the height of the last glaciation (Fig. 7). The

range shifts might also have been coupled with adaptive changes, as shown

by the differences in morphology (e.g. Stewart 1999) but not in mtDNA

between the Late Pleistocene and today. However, the predictions for the

future suggests that at least one third of willow ptarmigans’ current habitat,

and nearly half of that available for rock ptarmigan, will disappear by 2070-

2100, and the remaining areas will also become increasingly fragmented

(Figs. 6 and 7).

29

Figure 6. Temporal haplotype networks of modern and Late Pleistocene willow ptarmigan

(upper) and rock ptarmigan (lower). Barplots to the right show estimated changes in the

proportion of available habitat and the degree of habitat fragmentation in Europe, illustrated

as number of patches and median patch area, for the respective species. The future

projections, averaged for 2070 to 2100, are shown for both IPCC climate scenario B1 (white

bar) and A2 (black bar)

30

Figure 7. Back-casted European distributions of willow and rock ptarmigan for the last 20

thousand years, with occurrence probabilities ranging from 0 (grey) to 1 (red). Also shown is

the Lagopus spp. fossil sites used in the model calibrations, and the approximate extent of

glaciated land during the different time periods. Panels to the right show future projections of

ptarmigan distributions for the years 2070-2100, based on the two different climate scenarios.

The palaeogenetic analyses of 19 European mountain hares (Paper IV)

revealed that the complicated phylogeographic structure and high diversity

found in the modern species also existed during the Late Pleistocene period.

In the less variable cytochrome b gene (CytB), the modal haplotypes were

identical in the modern and glacial datasets, whereas a majority of the

individuals in both time periods carried unique variants of the control region

(CR) sequence. However, the ancient haplotypes were nested within the

diversity of the contemporary ones, and the reconstructed phylogeny shows

that the glacial hares are intermingled with modern individuals throughout

31

the evolutionary tree (Figs. 8 and 9). We therefore propose that the ancient

midlatitude bone remains belonged to representatives of the same diverse

hare population that is currently found in various locations across northern

and alpine Europe. Further, the demographic analyses indicated that the

mountain hare’s population size has fluctuated over time in concordance

with the glacial cycles, with expansions and declines occurring during past

periods of climate cooling and warming, respectively. In that sense, the

mountain hare consequently appears to be a textbook example of a cold-

adapted species, even though it stands out from the majority of previously

studied Arctic taxa in that it successfully survived the Pleistocene-Holocene

transition without any major declines in diversity.

Figure 8. Temporal haplotype networks of modern and Late Pleistocene mountain hares, with

CytB variation to the left and CR variation to the right.

32

Figure 9. Mountain hare phylogenetic tree based on concatenated sequences from both the

CytB and CR, and a mutation rate of 7.7 % Myr-1. The European samples are colour coded

according to geographic sub-region, and the position of the ancient samples in the phylogeny

are indicated with black ovals. The scale and posterior probability levels are identical to those

described in Fig. 5.

33

Discussion

Animals often spend a substantial part of their life on the move, whether it is

in search of new resources and potential mates or to avoid competition and

unfavourable environmental conditions. These individual movements can

occur on different geographic scales depending on landscape barriers and

species’ physical abilities as well as behavioural constraints. A well-known

type of movement in northern regions is the recurring migration between

summer and winter habitats, which allows animals to follow seasonal

changes in climate and resource availability. On a short time scale, these

periodic shifts can cover vast geographic areas, but since individuals

generally return to the same breeding grounds that they left, there are no

apparent changes in the longer perspective. However, although migrations

do not involve gene flow per se, these periodic movements can sometimes

develop into dispersal events. This is seen during peak years of Norwegian

lemmings (Lemmus lemmus), when their local shifts between wintering and

summer habitats continues as mass movements across the mountain tundra

and further down into forested valleys (Hansson 1999; Henttonen &

Kaikusalo 1993). It should be noted though, that dispersal is in turn just a

pre-requisite for gene flow and to have a permanent impact on coming

generations, the dispersing individuals also need to locate a new area that

fulfil their environmental requirements, as well as to successfully set up a

territory and reproduce. For Norwegian lemmings, there are no observations

of uphill movements to indicate that the dispersing individuals ever return to

their core tundra habitat (Østbye et al. 1993). However, the genetic analyses

in Paper I revealed that this small rodent displays a pattern of

geographically widespread genetic clusters similar to that described for

Scandinavian brown bears and moose (Manel et al. 2004; Wennerström et

al. 2016) which strongly suggests that mass movements during peak years

indeed result in long-distance gene flow between mountain areas separated

by forest.

34

While there might be substantial amounts of movement and gene flow

between different sites across a species’ range, it normally occurs within

stable geographic boundaries. However, on a longer time scale,

environmental changes can lead to geographic shifts of habitats that require

abilities for directional dispersal over many generations. This type of

successional movement has been the focus of Papers II-IV in this thesis,

where I have investigated post-glacial recolonization dynamics in three types

of important prey species in the northern biome. Previous ancient DNA

studies on cold-adapted mammal species have generally shown an

unexpected lack of habitat tracking during past climate warming, sometimes

despite very high dispersal abilities in modern populations (e.g. Dalén et al.

2007). The second paper in this thesis (Paper II) conforms to these

discouraging findings, and show that the midlatitude glacial lemming

(Lemmus sp.) populations were unable spread northwards at a sufficiently

high rate to keep pace with the rapid post-glacial alterations. Together with

the previously reported extinctions of Dicrostonyx lemmings (Brace et al.

2012), this lack of habitat tracking in the past presumably would have had

large bottom-up effects on the tundra ecosystem, and one could further

speculate on the importance of their disappearance in the failure of European

arctic fox populations to survive until the present (Dalén et al. 2007).

The intriguing support for a long-term northern glacial survival of lemmings

in restricted ice-free refugia highlights the alternative to range shifts, namely

staying in place and enduring the new climate. The persistence of such a

remnant population would most likely have included severe genetic

bottlenecks, and given that genetic diversity is considered vital for long-term

survival it could be questioned whether this putative persistence during the

whole LGM period is at all possible. However, new genomic data from

Channel Island foxes (Urocyon littoralis) have revealed that survival of very

small and isolated populations is indeed possible for several thousands of

years, despite a near absence of genetic variation and accumulation of

deleterious alleles (Robinson et al. 2016). Similar to our suggestions (see

Paper II), this remarkable persistence was explained by the habitats’

absence of predators and competitors, and the authors propose that such

35

decreased environmental stress can allow populations to tolerate higher

levels of genetic load.

The results from the two Lemmus studies (Paper I and II) illustrate that

there are different factors controlling dispersal between existing populations,

and the type of directional movement that involves the successional

expansion of a range into new territories. In fact, the Lemmus populations

that inhabited midlatitude Europe during the last glaciation might not have

exhibited the strong cyclicity and subsequent mass movements that we see in

modern Norwegian lemmings (L. lemmus). Further, a pre-requisite for the

development of lemming peaks is successful winter breeding, which is in

turn is dependent on a long period of continuous snow cover (Ims et al.

2011). The climate deteriorations at the last glacial/interglacial transition

could thus have led to irregular or even absent outbreaks and consequently

less frequent long-distance dispersal events across areas of unsuitable

habitat.

Whereas post-glacial warming appears to have caused local extinctions of

lemming populations outside of the northern refugium, the aDNA analyses

of ptarmigan (Paper III) and mountain hares (Paper IV) have both revealed

a temporal genetic continuity in Europe that supports the hypothesis of

habitat tracking. The two latter genera also distinguished themselves from

the glacial lemmings by exhibiting no clear phylogeographic structure in the

ancient datasets, suggesting that they were not experiencing any major

barriers to gene flow in the Late Pleistocene landscape. It could therefore be

speculated that the comparatively restricted dispersal of lemmings, as

indicated by the observed genetic structure among the Pleistocene

midlatitude populations, coupled with a rapid post-glacial fragmentation of

the midlatitude tundra habitat contributed to the failure of Lemmus

populations to shift their ranges. Further, competition with a resident

northern population could potentially also have obstructed the southern

lemmings to establish in Fennoscandia. The contrasting long-term survival

of ptarmigan in Europe could most likely be attributable to their flight

capability, which would have made them less dependent on corridors

between suitable habitat patches and enabled a more efficient movement

across heterogeneous landscapes. While mountain hares are known to have

36

low natal dispersal rates and high site fidelity (Dahl & Willebrand 2005), it

has the broadest ecological niche of the taxa investigated here. This ability to

survive in a variety of different habitats was presumably very important

during the environmental deteriorations at the Pleistocene/Holocene

transition, and could be the key to why mountain hares have successfully

shifted their distribution without any losses of unique mtDNA lineages.

As a complement to the genetic analyses, we also included back-casted

species distribution modelling (Paper III) to investigate how climate-

induced changes in suitable habitat affected the ability of ptarmigan

(Lagopus spp.) to shift their ranges. Although these models were based on

occurrence data for the two bird species, it might be worth noting that rock

ptarmigan (L. muta) share roughly the same modern habitat preferences as

the Norwegian lemming (Lemmus lemmus), despite the much more restricted

distribution of the latter species (Fig. 2). Consequently, the range changes

predicted for L. muta could give a hint on how areas suitable for lemmings

have shifted over the last 20 millennia. A similar comparison could perhaps

also be made between the willow ptarmigan (Lagopus lagopus) and the

mountain hare (Lepus timidus), where the predicted occurrence of willow

ptarmigan can give an approximation of the minimum amount of areas

suitable for hares. For L. timidus, however, the inference is complicated by

the close interaction with the European brown hare (L. europaeus), which is

proposed to have led to recent competitive exclusion of mountain hares from

many areas along its southern range margin. The back-casted range

projections showed that the type of tundra habitat favoured by rock

ptarmigan and lemmings only comprised ca 25 % of the comparable areas

suitable for willow ptarmigan and mountain hares, and the former habitat

type also seemed to have been constantly more fragmented (Figs. 6 and 7).

Such a patchy Lemmus environment is in line with the previously discussed

genetic differentiation between the glacial midlatitude European sub-regions,

and this fragmentation might also have obstructed a subsequent post-glacial

range shift. Whereas neither the rock ptarmigan, or the more common

willow ptarmigan and mountain hare, appear to have been restricted in their

movement by this past habitat structure, they could face problems in the

future. In many parts of their current European distributions, they now reside

37

in isolated Arctic regions or southern alpine “sky islands” (Hampe & Jump

2011) that are already at the geographic limits of possible areas to colonise.

As a result of anthropogenic climate warming, our forecasts of the coming

100 years show that these regions will only have remnant, if any, areas left

that fulfil the environmental requirements of ptarmigan. These resident birds,

and presumably also the corresponding mountain hare populations, will

consequently need to adapt in order to survive. Interestingly, both the willow

ptarmigan and mountain hare have populations in the British Isles that do not

develop the regular white winter colour, which could indicate such an

ongoing selection in both species to the milder winters of this region.

Further, the previously reported morphological difference between modern

and Late Pleistocene ptarmigan (Bochenski 1985; Stewart 1999) suggests

that the birds might also have adapted to climate warming in the past,

presumably through changes in their diet or the degree of sedentariness

(Stewart 1999).

The Norwegian lemming inhabits a region that seems to be comparatively

less affected by future climate changes (Fig. 7). However, the increased

habitat fragmentation, predicted for both L. muta/L. lemmus and L.

lagopus/L. timidus (Figs. 6 and 7), in combination with a dampened

frequency and amplitude of lemming peaks due to milder winter conditions

(Ims et al. 2011), could still be expected to increase the genetic

differentiation between Norwegian lemming populations in the future.

38

Sammanfattning

Den genetiska variationen hos nutida arter är starkt kopplad till det genflöde

som sker mellan olika populationer, vilket i sin tur påverkas av barriärer i

landskapet samt individernas spridningsförmåga och beteendemönster.

Dagens diversitet har dock även formats av spridningsprocesser över längre

tidshorisonter, såsom de successiva utbredningsförändringar som skett under

historiens gång till följd av klimatförändringar. Denna avhandling fokuserar

på köldanpassade arter och en aspekt som har haft stor betydelse för deras

genetiska variation är hur väl de hanterade den kraftiga uppvärmning och

habitatförflyttning som skedde vid den senaste istidens slut. Jag har

undersökt tre grupper av små växtätare: äkta lämlar (Lemmus), ripor

(Lagopus) och harar (Lepus), vilkas moderna utbredning sträcker sig från

kalfjället till lågalpina hedar och taigaområden. I den första studien

analyserade jag den genetiska strukturen hos nutida fjällämlar (Lemmus

lemmus) vilka har en starkt cyklisk populationsdynamik. Resultaten tyder på

att de lämmeltåg som sker under toppår fungerar som en drivkraft för

genflöde mellan olika bergsmassiv, vilket har blandat upp genpoolen över

förvånansvärt stora geografiska områden. När jag analyserade gammalt

DNA från istida lämlar för att undersöka deras förmåga till långsiktig, riktad

spridning fann jag dock att de populationer som då levde i Mellaneuropa

tycks ha varit inkapabla att förändra sin utbredning när den post-glaciala

uppvärmningen försköt deras tundrahabitat till nordliga trakter. Våra resultat

tyder istället på att den endemiska fjällämmeln härstammar från en isolerad

nordlig population som överlevde den senaste perioden av maximal

nedisning i ett begränsat isfritt område i Skandinavien. Till skillnad från

istidens lämlar, liksom en majoritet av tidigare studerade arktiska däggdjur,

avslöjade de genetiska analyserna av både ripor (L. lagopus och L. muta) och

skogsharar (L. timidus) en långsiktig kontinuitet i Europa. Deras

mellaneuropeiska populationer verkar således ha klarat av att följa de snabba

klimatförändringarna som skedde vid övergången till den nuvarande

mellanistiden och förflytta sina utbredningsområden till nordliga och

39

högalpina trakter. Dessa olika utfall skulle kunna härledas till ripornas

flygförmåga, vilken möjliggör en friare spridning över fragmenterade

områden, samt till att skogshararnas generalistiska natur minskar deras

sårbarhet för habitatförändringar. Utbrednings-modelleringar för de

kommande 100 åren visade dock att en fortsatt global uppvärmning kan leda

till att klimatet i vissa isolerade utbredningsområden blir olämpligt i

framtiden, vilket kräver att populationerna där anpassar sig till de nya

förutsättningarna för att undgå lokala utdöenden.

40

Acknowledgments

I am grateful to my supervisor Love Dalén for giving the opportunity to

work with these exiting projects and for being such a great support during all

these years. You have always taken the time to sit down and discuss things,

and I really appreciate your positive attitude and encouragement during

times of doubt. I also want to thank my co-supervisor Anders Angerbjörn

and the members of my evaluation committee, Per Ericsson and Bodil

Elmhagen, for valuable advice during my PhD studies.

To members of the ancient DNA research group who have come and left

during my time as a PhD student: Edson, Yvonne, Jean-Luc, Elle, Erik,

Patricia, Davíd, George and Johanna. Thank you for interesting scientific

discussions and entertaining fika breaks. Elle and Patricia also deserve extra

thanks for taking care of various plants during my parental leaves. These

probably never did so well as in my absence. The last stage of this thesis

work was supported by a regular provision of excellent cakes, thank you

Patricia! I am also grateful to the staff of the molecular lab at the museum

for technical assistance, as well as to Pia and Anna-Lena for help with

administrative issues.

I want to thank Jacob Höglund for taking me on as a master student, despite

my lack of previous molecular experience, and introducing me to the life of

science.

A big thank you goes to my family for taking care of the kids so that I could

finalise this thesis. In particular dad, who have kindly volunteered as a nanny

and in return gotten repeatedly infected with influenza. Kerstin, thank you

for drawing such an awesome cover for this thesis! Finally, I am immensely

grateful to Richard. Without your endless support and patience, this work

would not have been possible. Ruben and Miranda: You have not been

helpful at all, but I love you anyway.

41

References

Abdelkrim J, Robertson B, Stanton JA, Gemmell N (2009) Fast, cost-effective development of species-specific microsatellite markers by genomic sequencing. Biotechniques, 46, 185-192.

Angerbjörn A, Tannerfeldt M, Lundberg H (2001) Geographical and temporal patterns of lemming population dynamics in Fennoscandia. Ecography, 24, 298-308.

Averianov AO (2001) Lagomorphs (Mammalia) from the Pleistocene of Eurasia. Paleontological Journal, 35, 191-199.

Barnosky AD, Koch PL, Feranec RS, Wing SL, Shabel AB (2004) Assessing the Causes of Late Pleistocene Extinctions on the Continents. Science, 306, 70-75.

Beaumont MA, Zhang WY, Balding DJ (2002) Approximate Bayesian computation in population genetics. Genetics, 162, 2025 - 2035.

Bennett KD, Tzedakis PC, Willis KJ (1991) Quaternary Refugia of North European Trees. Journal of Biogeography, 18, 103-115.

Bochenski Z (1985) Osteological Differentiation in Willow Grouse. Fortschritte der Zoologie, 69-72

Brace S, Palkopoulou E, Dalén L, et al. (2012) Serial population extinctions in a small mammal indicate Late Pleistocene ecosystem instability. Proceedings of the National Academy of Sciences, 109, 20532-20536.

Broecker WS, van Donk J (1970) Insolation changes, ice volumes, and the O18 record in deep-sea cores. Rev. Geophys., 8, 169-198.

Campos PF, Willerslev E, Sher A, et al. (2010) Ancient DNA analyses exclude humans as the driving force behind late Pleistocene musk ox (Ovibos moschatus) population dynamics. Proceedings of the National Academy of Sciences of the United States of America, 107, 5675-5680.

Dahl F, Willebrand T (2005) Natal dispersal, adult home ranges and site fidelity of mountain hares Lepus timidus in the boreal forest of Sweden. Wildlife Biology, 11, 309-317.

Dalén L, Nyström V, Valdiosera C, et al. (2007) Ancient DNA reveals lack of postglacial habitat tracking in the arctic fox. Proceedings of the National Academy of Sciences of the United States of America, 104, 6726-6729.

Darwin C (1859) On the Origin of Species by Means of Natural Selection, 5 edn. J. Murray, London.

Drovetski SV (2003) Plio-Pleistocene climatic oscilations, Holarctic biogeography and speciation in an avian subfamily. Journal of Biogeography, 30, 1173-1181.

42

Drummond AJ, Nicholls GK, Rodrigo AG, Solomon W (2002) Estimating mutation parameters, population history and genealogy simultaneously from temporally spaced sequence data. Genetics, 161, 1307-1320.

Drummond AJ, Rambaut A, Shapiro B, Pybus OG (2005) Bayesian Coalescent Inference of Past Population Dynamics from Molecular Sequences. Mol Biol Evol, 22, 1185-1192.

Ehrich D, Stenseth NC (2001) Genetic structure of Siberian lemmings (Lemmus sibiricus) in a continuous habitat: large patches rather than isolation by distance. Heredity, 86, 716-730.

Ekman S (1920) Der skandinavische Lemming (Lemmus lemmus) als uberrest einer interglazialen skandinavischen Fauna. Festschrift fur Zschokke (Basel), 2, 1-12.

Ekman S (1922) Djurvärldens utbredningshistoria på den Skandinaviska halvön Albert Bonniers Förlag, Stockholm.

Eldredge N (1989) Macroevolutionary dynamics: species, niches, and adaptive peaks McGraw-Hill.

Elton CS (1924) Periodic Fluctuations in the Numbers of Animals: Their Causes and Effects. Journal of Experimental Biology, 2, 119-163.

Fedorov VB, Stenseth NC (2001) Glacial survival of the Norwegian lemming (Lemmus lemmus) in Scandinavia: inference from mitochondrial DNA variation. Proceedings of the Royal Society B: Biological Sciences, 268, 809-814.

Fernández H, Hughes S, Vigne J-D, et al. (2006) Divergent mtDNA lineages of goats in an Early Neolithic site, far from the initial domestication areas. Proceedings of the National Academy of Sciences, 103, 15375-15379.

Freeland J, Anderson S, Allen D, Looney D (2007) Museum samples provide novel insights into the taxonomy and genetic diversity of Irish red grouse. Conservation Genetics, 8, 695-703.

Hamill RM, Doyle D, Duke EJ (2006) Spatial patterns of genetic diversity across European subspecies of the mountain hare, Lepus timidus L. Heredity, 97, 355-365.

Hampe A, Jump AS (2011) Climate Relicts: Past, Present, Future. Annual Review of Ecology, Evolution, and Systematics, 42, 313-333.

Hansson L (1999) Lemmus lemmus. In: The atlas of European mammals (eds. Mitchell-Jones AJ, Amori G, Bogdanowicz W, et al.). Academic Press, London, UK.

Henttonen H, Kaikusalo A (1993) Lemming movements. In: The Biology of lemmings (eds. Stenseth NC, Ims RA). Published for the Linnean Society of London by Academic Press, London; San Diego.

Hersteinsson P, MacDonald DW (1992) Interspecific Competition and the Geographical Distribution of Red and Arctic Foxes Vulpes Vulpes and Alopex lagopus. Oikos, 64, 505-515.

Hewitt GM (1996) Some genetic consequences of ice ages, and their role in divergence and speciation. Biological Journal of the Linnean Society 58, 247-276.

43

Hewitt GM (1999) Post-glacial re-colonization of European biota. Biological Journal of the Linnean Society, 68, 87-112.

Hewitt GM (2000) The genetic legacy of the Quaternary ice ages. Nature, 405, 907-913.

Hewitt GM (2001) Speciation, hybrid zones and phylogeography — or seeing genes in space and time. Molecular Ecology, 10, 537-549.

Ho SYW, Kolokotronis S-O, Allaby RG (2007) Elevated substitution rates estimated from ancient DNA sequences. Biology Letters, 3, 702-705.

Ho SYW, Lanfear R, Bromham L, et al. (2011) Time-dependent rates of molecular evolution. Molecular Ecology, 20, 3087-3101.

Hodgson JA, Thomas CD, Dytham C, Travis JMJ, Cornell SJ (2012) The Speed of Range Shifts in Fragmented Landscapes. PLoS ONE, 7.

Hofreiter M, Barnes I (2010) Diversity lost: are all Holarctic large mammal species just relict populations? BMC Biology, 8, 46.

Hofreiter M, Serre D, Poinar HN, Kuch M, Pääbo S (2001) Ancient DNA. Nat Rev Genet, 2, 353-359.

Hudson RR (1990) Gene genealogies and the coalescent process. Oxford surveys in evolutionary biology, 7, 44.

Imbrie J, Boyle EA, Clemens SC, et al. (1992) On the Structure and Origin of Major Glaciation Cycles 1. Linear Responses to Milankovitch Forcing. Paleoceanography, 7, 701-738.

Ims RA, Yoccoz NG, Killengreen ST (2011) Determinants of lemming outbreaks. Proceedings of the National Academy of Sciences, 108, 1970–1974.

Jackson ST, Sax DF (2009) Balancing biodiversity in a changing environment: extinction debt, immigration credit and species turnover. Trends in Ecology & Evolution, 25, 153-160.

Johnsen SJ, Dahl-Jensen D, Gundestrup N, et al. (2001) Oxygen isotope and palaeotemperature records from six Greenland ice-core stations: Camp Century, Dye-3, GRIP, GISP2, Renland and NorthGRIP. Journal of Quaternary Science, 16, 299-307.

Kahlke R-D (1999) The History of the Origin, Evolution and Dispersal of the Late Pleistocene Mammuthus–Coelodonta Faunal Complex in Eurasia (Large Mammals) Mammoth Site of Hot Springs, SD.

Kerr JT, Pindar A, Galpern P, et al. (2015) Climate change impacts on bumblebees converge across continents. Science, 349, 177-180.

Larsen E, Gulliksen S, Lauritzen SE, et al. (1987) Cave stratigraphy in western Norway; multiple Weichselian glaciations and interstadial vertebrate fauna. Boreas, 16, 267-292.

Lister AM (2004) The impact of Quaternary Ice Ages on mammalian evolution. Philosophical Transactions of the Royal Society B: Biological Sciences, 359, 221-241.

Manel S, Bellemain E, Swenson JE, François O (2004) Assumed and inferred spatial structure of populations: the Scandinavian brown bears revisited. Molecular Ecology, 13, 1327-1331.

44

Mangerud J, Gyllencreutz R, Lohne Ø, Svendsen JI (2011) Glacial history of Norway. In: Quaternary glaciations - Extent and chronology. A closer look (eds. Ehlers J, Gibbard PL, Hughes PD), pp. 279-298. Elsevier, Amsterdam.

Mayr E (1963) Animal species and evolution Harvard University Press, Cambridge.

Melo-Ferreira J, Boursot P, Randi E, et al. (2007) The rise and fall of the mountain hare (Lepus timidus) during Pleistocene glaciations: expansion and retreat with hybridization in the Iberian Peninsula. Molecular Ecology, 16, 605-618.

Nadachowski A (1982) Late Quaternary rodents of Poland with special reference to morphotype dentition analysis of voles Panstwowe Wydawnictwo Naukowe, Warszawa.

Nei M, Maruyama T, Chakraborty R (1975) The Bottleneck Effect and Genetic Variability in Populations. Evolution, 29, 1-10.

Norén K, Angerbjörn A (2014) Genetic perspectives on northern population cycles: bridging the gap between theory and empirical studies. Biological Reviews, 89, 493-510.

Orr MR, Smith TB (1998) Ecology and speciation. Trends in Ecology & Evolution, 13, 502-506.

Palkopoulou E, Baca M, Abramson NI, et al. (2016) Synchronous genetic turnovers across Western Eurasia in Late Pleistocene collared lemmings. Global Change Biology, 22, 1710-1721.

Palkopoulou E, Dalén L, Lister AM, et al. (2013) Holarctic genetic structure and range dynamics in the woolly mammoth. Proc Biol Sci, 280, 7.

Petit JR, Jouzel J, Raynaud D, et al. (1999) Climate and atmospheric history of the past 420,000 years from the Vostok ice core, Antarctica. Nature, 399, 429-436.

Pithan F, Mauritsen T (2014) Arctic amplification dominated by temperature feedbacks in contemporary climate models. Nature Geosci, 7, 181-184.

Prost S, Anderson CNK (2011) TempNet: a method to display statistical parsimony networks for heterochronous DNA sequence data. Methods in Ecology and Evolution, 2, 663-667.

Rand AL (1948) Glaciation, an isolating factor in speciation. Evolution, 2, 314-321.

Robinson Jacqueline A, Ortega-Del Vecchyo D, Fan Z, et al. (2016) Genomic Flatlining in the Endangered Island Fox. Current Biology.

Stewart JR (1999) Intraspecific Variation in Modern and Quaternary European Lagopus. In: Avian Paleontology at the Close of the 20th Century. Proc. of the 4th International Meeting of the Society of Avian Paleontology and Evolution (ed. Olson SL). Smithsonian Contributions to Paleobiology 89: 159-168, Washington, D.C.

Stewart JR, Lister AM, Barnes I, Dalén L (2010) Refugia revisited: individualistic responses of species in space and time. Proceedings of the Royal Society B: Biological Sciences, 277, 661-671.

45

Svendsen JI, Alexanderson H, Astakhov VI, et al. (2004) Late Quaternary ice sheet history of northern Eurasia. Quaternary Science Reviews, 23, 1229-1271.

Svensson S (1999) Svensk Fågelatlas Ekologiska institutionen, Lunds universitet, ArtDatabanken, Sveriges lantbruksuniversitet.

Taylor KC, Mayewski PA, Alley RB, et al. (1997) The Holocene-Younger Dryas Transition Recorded at Summit, Greenland. Science, 278, 825-827.

Thulin C-G (2003) The distribution of mountain hares Lepus timidus in Europe: a challenge from brown hares L. europaeus? Mammal Review, 33, 29-42.

Tyrberg T (1998) Pleistocene birds of the Palearctic: a catalogue Nuttall Ornithological Club Cambridge.

Urban MC (2015) Accelerating extinction risk from climate change. Science, 348, 571-573.

Waters CN, Zalasiewicz JA, Williams M, Ellis MA, Snelling AM (2014) A stratigraphical basis for the Anthropocene? Geological Society, London, Special Publications, 395, 1-21.

Wennerström L, Ryman N, Tison J-L, et al. (2016) Genetic landscape with sharp discontinuities shaped by complex demographic history in moose (Alces alces). Journal of Mammalogy, 97, 1-13.

Østbye E, Engh C-E, Lien L, et al. (1993) Regional distribution of lemmings (Lemmus lemmus) during cyclic highs in the Hallingdalen Valley, southern Norway, 1966-1985 In: The Biology of Lemmings (eds. Stenseth NC, Ims RA). Published for the Linnean Society of London by Academic Press, London; San Diego.