Genomic analyses suggest adaptive differentiation of northern European native cattle breeds

Abstract Native domestic breeds represent important cultural heritage and genetic diversity relevant for production traits, environmental adaptation and food security. However, risks associated with low effective population size, such as inbreeding and genetic drift, have elevated concerns over whether unique within‐breed lineages should be kept separate or managed as one population. As a conservation genomic case study of the genetic diversity represented by native breeds, we examined native and commercial cattle (Bos taurus) breeds including the threatened Danish Jutland cattle. We examined population structure and genetic diversity within breeds and lineages genotyped across 770K single nucleotide polymorphism loci to determine (a) the amount and distribution of genetic diversity in native breeds, and (b) the role of genetic drift versus selection. We further investigated the presence of outlier loci to detect (c) signatures of environmental selection in native versus commercial breeds, and (d) native breed adaptation to various landscapes. Moreover, we included older cryopreserved samples to determine (e) whether cryopreservation allows (re)introduction of original genetic diversity. We investigated a final set of 195 individuals and 677K autosomal loci for genetic diversity within and among breeds, examined population structure with principal component analyses and a maximum‐likelihood approach and searched for outlier loci suggesting artificial or natural selection. Our findings demonstrate the potential of genomics for identifying the uniqueness of native domestic breeds, and for maintaining their genetic diversity and long‐term evolutionary potential through conservation plans balancing inbreeding with carefully designed outcrossing. One promising opportunity is the use of cryopreserved samples, which can provide important genetic diversity for populations with few individuals, while helping to preserve their traditional genetic characteristics. Outlier tests for native versus commercial breeds identified genes associated with climate adaptation, immunity and metabolism, and native breeds may carry genetic variation important for animal health and robustness in a changing climate.


| INTRODUC TI ON
Domestication of plants and animal species has permitted significant human population growth and the development of modern human societies (Larson & Burger, 2013;Larson & Fuller, 2014;Marshall, Dobney, Denham, & Capriles, 2014). Domestication is considered to have occurred along three major pathways, termed commensal, prey or directed (reviewed in Larson & Burger, 2013;Larson & Fuller, 2014). They describe the commensal pathway as centred on animal habituation to a human niche, whereas the prey pathway involved animals that humans initially preyed upon and later started to manage. In contrast, Larson and Burger (2013) and Larson and Fuller (2014) note that the more recent directed pathway has been the only deliberate route to domestication, which bypassed the habituation and management phases. There is evidence that sheep (Ovis sp.), goats (Capra sp.) and cattle (Bos sp.) were domesticated 10,500-10,000 years before present via the prey pathway, through various stages of intensive breeding of captive animals and the subsequent development of distinct breeds (Larson & Burger, 2013;Larson & Fuller, 2014). Conservation of native domestic animals and plants is now receiving growing attention. Populations of conservation concern may encompass important cultural heritage and potentially genetic diversity relevant for modern breeding and future food security such as for production traits, adaptation to harsh environments and climate change (Hoffmann, 2013;Iacolina et al., 2016;Kantanen et al., 2015;Kristensen, Hoffmann, Pertoldi, & Stronen, 2015).

| Genetic diversity within and among native breeds
Native domestic breeds are known to share several conservation concerns with populations of wild species at risk, such as low effective population size (N E ), which in turn reduces the effectiveness of selection and increases the impacts of genetic drift and inbreeding (Kantanen et al., 2000;Leroy et al., 2013;Pertoldi et al., 2014;Taberlet et al., 2008). Numerous livestock breeds have gone extinct or are threatened (FAO, 2007(FAO, , 2015. Genetic variation within and between breeds is rapidly lost and, for many breeds, we have little information about levels of genetic variation, N E , and adaptation to past and present local environmental conditions. Past selection for conditions such as the ability to survive on food with limited nutritional content could be important for adaptation to climate change, and for the use of native breeds in habitat management. To maintain sustainable populations in the short term, an N E of at least 50 is sometimes recommended in conservation genetics to keep inbreeding rates at acceptable levels, and an N E >500 is needed to allow maintenance of evolutionary potential over time (i.e., across hundreds of generations) (Frankham, Bradshaw, & Brook, 2014;Hoffmann, Sgrò, & Kristensen, 2017). Because domestic animal breeds typically have N E <100 (Leroy et al., 2013), current breeding practices raise long-term concerns for the evolutionary potential of many of these populations.
Genomic methods and reproductive techniques are rapidly developing and can help provide answers to many important questions in conservation genetics. These include levels of inbreeding and genetic drift, genetic uniqueness, identification of genomic regions under selection, and creation of genetic rescue programmes or breeding schemes to maintain adaptive genetic variation in domestic animals more efficiently than pedigree-based breeding (Kantanen et al., 2015;Kukučková et al., 2017;Porto-Neto et al., 2014;Williams et al., 2015). Improved communication between the fields of research and management concerning experimental results on genetic rescue and other conservation actions is therefore important for an efficient management of populations with small and declining N E (Hoffmann, Merilä, & Kristensen, 2016;Kristensen et al., 2015). Vesterbølle (1968)(1969) Kortegaard (1989)(1990) Oregaard (1986) indicated an overall N E for the Jutland cattle breed of around 40 (http://www.fao.org/dad-is/browse-by-country-and-species/en/), suggesting strong drift and high rates of inbreeding. The low N E for the Jutland cattle breed was supported by Pertoldi et al. (2014) who analysed genome-wide profiles with 50K single nucleotide polymorphism (SNP) markers in the Jutland Kortegaard-lineage. This native livestock breed thus offers an informative case study of how genome-wide profiles can inform conservation genetic management of small populations at risk.

| Genetic drift and selection in native breeds
Breeders and managers working with small populations at risk are often concerned with the genetic uniqueness of these populations (Ginja, Gama, & Penedo, 2010;Kantanen et al., 2000;Withen, Brüniche-Olsen, Pedersen, European Cattle Genetic Diversity Consortium, & Gravlund, 2011). This issue is relevant for wild and domestic species, including native livestock breeds where survival has been influenced by local environmental conditions. Genomic profiles from such breeds can advance evolutionary research and conservation by improving our understanding of (i) the role of genetic drift versus selection in contemporary within-breed lineages, (ii) signatures of selection in native versus commercial breeds and (iii) adaptation to different landscapes in native breeds.
For the Jutland cattle, a long-standing discussion among farmers and managers in Denmark has been whether lineages should be kept separate or managed as one population. Analyses of genomic profiles from all four contemporary Jutland cattle lineages will thus allow us to determine the amount and distribution of diversity within the breed and among individual lineages. This issue has relevance across livestock breeds where managers acknowledge that rapid conservation actions may be needed to preserve small and declining populations, yet they have concerns that admixture of within-breed lineages may risk further loss of unique genetic variation, with potential negative implications for locally adapted traits (Hoffmann, 2013;Taberlet et al., 2008). Vital management considerations include the costs and benefits of maintaining separate lineages with few remaining individuals, and preservation of characteristics potentially limiting productivity (e.g., milk, wool, meat) but augmenting survival in harsh environments such as areas with extreme temperatures or precipitation levels, or in habitats with poor-quality food sources (Hoffmann, 2013;Pariset, Joost, Marsan, & Valentini, 2009).
Preservation of semi-natural and cultural landscapes has been recognized as a key priority in Europe (Halada, Evans, Romao, & Petersen, 2011;Timmermann, Damgaard, Strand, & Svenning, 2014), and native breeds appear well-suited to extensive agriculture and biodiversity maintenance in the form of grazing, which can simultaneously offer suitable conditions for in situ natural selection by means of exposure to changing environmental conditions (Hoffmann, 2013). Northern European countries bordering the Atlantic Ocean have cool and humid climates, which may exert selective pressures on domestic species (Pariset et al., 2009). Yet within this region, there are substantial differences in landscape form and terrain ruggedness. In this study, we compare the Jutland cattle from Denmark, dominated by a relatively flat terrain, to native breeds from rugged landscapes in Norway and the Faroe Islands, enabling an investigation of adaption to local conditions (Bailey et al., 2015;Raqiz, Tareen, & Verdier, 2011).

| The potential role of cryopreservation in preserving native breeds
The carrying capacity of a population can be increased artificially (i.e., without expanding the in situ population) by supplying genetic material from less related earlier generations of animals. This strategy can be applied to native breeds of cattle and other species by cryopreservation of spermatozoa and oocytes in a gene bank (Curry, 2000;Su et al., 2012). An increasing number of species will, in the near future, be managed with the help of cryopreservation techniques (Charlton et al., 2018), including initiatives such as The Frozen Ark Project (https://www.frozenark.org/). In Denmark, cryopreservation is now used for native livestock breeds including the Jutland cattle, and these collections can provide genetic material for ongoing conservation efforts. Specifically, Hertz et al. (2016) simulated a supplementation in the Jutland Kortegaard-lineage where one male was added to the population every 5 years, representing cryopreserved semen from previous generations. The simulation suggested that such supplementation can postpone the time to extinction and reduce inbreeding levels (Hertz et al., 2016).
Emerging genomic methods provide exciting prospects for identification of genetic variation and selection in native breeds, yet additional efforts are needed for native breed conservation management to benefit fully from these new approaches (Bruford et al., 2015). The objective of our study was to use within-breed lineages of the Danish Jutland cattle, native breeds from neighbouring countries that may have been subject to different selective pressures (artificial and natural selection), and commercial dairy breeds selected for high productivity, to investigate (a) the amount and dis-

| Genetic diversity within and among native breeds
We compared genomes from the Danish Jutland cattle to native breeds from other northern European countries and to commercial breeds originating from northern Europe ( Table 1). The Jutland cattle is an indigenous breed that descends from original black and grey pied cattle from the 16th-18th century, and although the Jutland cattle was once widespread in Denmark, it declined following competition and near-replacement with larger more productive breeds (Brüniche-Olsen et al., 2012;Kantanen et al., 2000;Sørensen & Nielsen, 2015). In 1949, the breeding association for the Jutland cattle decided to accept Dutch and German black pied cattle (Friesian) into the studbook, and the resulting highly successful crosses formed the Danish black pied cattle that within a decade had replaced almost all original Jutland cattle (Brüniche-Olsen et al., 2012;Sørensen & Nielsen, 2015). Subsequently, genetic material from Holstein cattle in North America, where this European breed had been imported and further developed, was introduced into the global population of black pied cattle, and this breed constitutes the modern Holstein (also named Holstein-Friesian) breed (Sørensen & Nielsen, 2015).
This process also occurred in Denmark, however; a Friesian population with limited contribution of Holstein genes has been preserved and this lineage is named SDM-1965(DAD-IS, 2017b Figure S1).
We pruned the data for loci in linkage disequilibrium (LD), and because the Jutland cattle lines have small populations and low genetic diversity where LD is expected to be high, we filtered the data to remove highly linked SNPs. For filtering, we used the PLINK formula (-indep 50 5 2), where 50 is the size of the sliding window (i.e., 50 bp at a time are examined for linked loci), 5 is the number of SNPs shifted in each step and 2 is the variance inflation factor (see further details at http://zzz.bwh.harvard.edu/plink/summary.shtml).
We then examined genetic structure in the data with PCA in the adegenet package (Jombart, 2008;Jombart & Ahmed, 2011) in R 2.14.2 (R Development CoreTeam, 2012) and with ADMIXTURE (Alexander, Novembre, & Lange, 2009). In ADMIXTURE, the number of population clusters (K) is determined with a cross-validation procedure, where the optimal K-value has the lowest cross-validation error relative to alternate K-values (Alexander, Shringarpure, Novembre, & Lange, 2015). We examined a range of K-values from 1 to 15 and used 20 cross-validations for each K-value and 1,000 bootstrap replicates. Subsequently, to provide another measure of differentiation among breeds and lineages, we calculated pairwise F ST in Genepop (4.6) (Raymond & Rousset, 1995;Rousset, 2008). For each pairwise comparison, we evaluated the statistical significance of population differentiation by permutations in GenoDive v.2.0b23 (Meirmans & van Tienderen, 2004) with 50,000 randomly selected SNP loci. We applied the Bonferroni correction for multiple comparisons (Rice, 1989 (Do et al., 2014) with the LD method (Waples & Do, 2008) and the data set pruned for LD. The same data set was used to calculate the number of private alleles (P A ) per population in R with a script (https://johnbhorne.wordpress. com/2017/07/12/identifying-private-snps-in-r/) for the HierfStat package (Goudet, 2005).

| Genetic drift and selection in native breeds
To evaluate whether genomic profiles of different breeds showed signs of selection, we performed tests of outlier loci in the pcadapt package (Luu, Bazin, & Blum, 2017) Table S2 for a list of all genes) and chose genes for indepth investigation (henceforth focal genes) using the bibliography from NCBI Gene and additional references obtained from literature searches.
For focal genes, we examined possible enrichment in gProfiler (Reimand et al., 2016) with the Gene Ontology (GO) Biological Processes and Human Phenotype Ontology databases (https://biit. cs.ut.ee/gprofiler/index.cgi). We followed the approach of Caniglia et al. (2018) and limited the size of functional categories to maximum 500 terms, to focus our analyses on more specific genome regions.
We chose the gProfiler g:SCS significance threshold for multiple testing as recommended by the program authors (Reimand et al., 2016).
We examined and compared runs of homozygosity (ROH) in PLINK for representative groups including two native breeds from different environments and one commercial breed. These were

| Genetic diversity within and among native breeds
Quality screening of the data produced 710,471 SNPs. After equalizing sample sizes (Table 1) Table S3 and Supporting Information Figure   S3). The results for N E exhibited relatively broad variation with the smallest value found for the Westergaard-lineage and the highest estimate observed in the Holstein cattle (Table 4). The values for P A ranged from none in the old bulls post-1980 to 221 being found in the Western Norwegian Red-polled cattle (Table 4) and Westergaard (n = 9).
Comparisons of the outlier results show that the majority of genes identified were associated with production traits (growth/ meat) and reproduction. The distribution of results was uneven, ranging from T2 where we identified one flanking gene (IGF1) linked to production traits, milk and reproduction to T7 where we detected numerous genes associated with production traits (n = 27), reproduction (n = 11) and other traits ( Examination of ROH found that none were shared among all individuals within breeds. We plotted ROH shared among at least six individuals per breed, and focal genes found within outlier flanking regions, across all autosomal chromosomes (Figure 6a,b).

| The potential role of cryopreservation in preserving native breeds
Cryopreserved samples exhibited high genetic diversity, with seven P A found in the pre-1980 sample of SDM-1965 cattle (

| Genetic diversity within and among native breeds
Our results suggest the Danish Jutland cattle offer an informative case study on the threats facing many native breeds around the world (FAO, 2015), and the resources these breeds represent for long-term preservation of genetic variation in domestic animals (Hoffmann, 2013). Our findings on genetic structure and diversity agree with earlier results (Brüniche-Olsen et al., 2012) showing Jutland cattle as a distinct breed (see also Supporting Information Appendix S1).
Although earlier microsatellite analyses produced mixed results for Notes. N E and P A were calculated with LD-pruned data. Further IBD details are provided in Supporting Information Table S3 and Supporting Information Figure S3. a Including only cryopreserved semen samples of n = 14 Danish Jutland cattle recognized in the studbook.
TA B L E 5 Focal genes found in the 3,000-bp flanking regions of outlier single nucleotide polymorphism loci in native and commercial cattle Notes. Details on tests T1-T7 and populations are given in Table 2; gene information and references are provided in Supporting Information Appendix S2 and Supporting Information Table S4. data from a lower-density chip with 50K SNPs reported lower genetic diversity in the Kortegaard-lineage than for a range of commercial breeds (Pertoldi et al., 2014). Moreover, Pertoldi et al. (2014) reported  (69.9% and 0.327, respectively). However, in a broader perspective, the polymorphism and IBD values for the investigated native breeds suggest these comprise important diversity relative to their considerably smaller census population sizes than those of commercial breeds (Table 1). Altogether 1,019 native cattle breeds have been reported globally, with 369 in Europe and the Caucasus (FAO, 2015).
Understanding and preserving the genetic diversity of native breeds, especially those with small population sizes, is thus a priority, including assessments of epigenetic processes linked to environmental factors (e.g., diet) (FAO, 2015). Adaptation to changing environmental conditions will become increasingly important (Hoffmann, 2013), and the existing diversity within the many native cattle breeds could help facilitate more rapid adaptation.
Diverse patterns of variation may emerge from the use of different markers (Brandström & Ellegren, 2008)

| Genetic drift and selection in native breeds
Our analyses of outliers indicated that the breeds have evolved in consequence of both artificial and natural selection. The relatively high number of genes found associated with production traits and reproduction is likely influenced by the importance of these features for commercial breeds and thus for development of the SNP panel used in our study. The enrichment results also seem to suggest strong research focus on genomic regions important for livestock production (meat and milk) and potentially also for other areas within biology and evolution (hearing, behaviour and cognition). The overall distribution of results among outlier tests appears to support our expectations of broad genome-wide differences among the investigated populations and the evolutionary uniqueness of native breeds.
Danish native cattle exhibited differentiation from commercial breeds, which has also been reported from other regions (Iso-Touru et al., 2016;Kawahara-Miki et al., 2011;Lim et al., 2016). Pairwise tests between native breeds also showed signs of divergent selection, which is in accordance with findings from other parts of the world (Iso-Touru et al., 2016;Stucki et al., 2017). Certain characteristics, such as olfaction (McRae et al., 2013), hormonal cues concerning reproduction (Jiang et al., 2006), and cognitive features including visual map development (Xu et al., 2011) and spatial memory (Bailey et al., 2015;Qiu et al., 2010) (see also Supporting Information Tables   S4 and S5) could be vital for conserving native livestock and their role in extensive agriculture, where native breeds may be subject to in situ natural selection while contributing to grazing and biodiversity maintenance priorities (Bailey et al., 2015;Halada et al., 2011;Timmermann et al., 2014).
Our results suggest native breeds may carry genetic variants important for adaptation to a rapidly changing environment due to climate change and other anthropogenic activities, and such variants could be critical for their preservation and contribute to increased environmental tolerance in commercial breeds via modern genomic techniques (O'Neill, Swain, & Kadarmideen, 2010;Rauw & Gomez-Raya, 2015). Such advances could affect livestock survival and development related to important environmental stressors including infectious disease and parasite resistance (Kadarmideen, Ali, Thomson, Müller, & Zinsstag, 2011;O'Neill et al., 2010;Porto-Neto et al., 2014). Our findings may also indicate trade-offs between production gains and animal health (O'Neill et al., 2010;Rauw & Gomez-Raya, 2015;Takasuga et al., 2015). For example, genes related to metabolism and ketosis suggest potential conflicts between humaninduced artificial selection towards high milk production and the optimal energy balance of individual animals (Mulligan & Doherty, 2008 There are limiting factors that should be considered for interpretation of our results, including potential ascertainment bias in the bovine array developed with a focus on commercial breeds (Supporting Information Appendix S1). Moreover, we cannot exclude the possibility that small sample sizes may have affected our results, including the analyses of outlier loci. Reports from earlier simulations with pcadapt indicated benefits from increasing sample sizes from 20 to 60 individuals (Luu et al., 2017). However, as certain Jutland cattle lineages now have few remaining individuals (e.g., we obtained only 16 from Westergaard and 20 from Vesterbølle), we focused on tests with equalized sample sizes and a high number of LD-pruned loci.
Accordingly, for these small populations we are confident of having included most of the genomic variation (included on the SNP array) that is still in existence.

| The potential role of cryopreservation in preserving native breeds
A possibility emerging from our findings, consistent with recommendations from earlier studies (Charlton et al., 2018), is the potential to infuse variability into genetically depauperate lineages by means of original genetic diversity from older cryopreserved samples. These older semen samples could represent some of the existing variability prior to dramatic declines in census size and may be considered for inclusion in conservation breeding plans.
For the Jutland cattle, this would appear especially relevant for the Westergaard and Vesterbølle-lineages, where it is difficult to see management of genetically closed populations as a long-term viable option.

| CON CLUS ION
Our analyses demonstrate the potential use of genomics for investigating the genetic structure and unique variation of native domestic breeds, and how these may differ from commercial breeds in traits relevant for production and environmental tolerance. Our results, using northern European extensive native cattle as a case study, also underline how these differences may have implications for humaninduced artificial and natural selection of both native and commercial breeds. We show the potential for addressing basic evolutionary inquiries and applied conservation questions that can help managers create conservation plans to preserve genetic uniqueness while maintaining in situ selection in native domestic breeds, so these can continue to be shaped by their local environmental conditions even if the N E is low (and thus, selection has limited ability to act on the frequency of SNPs under potential selection). Furthermore, the implementation of an appropriate breeding plan will help increase the N E , which will in turn reduce the fluctuations in allelic frequencies across generations. At the same time, assortative mating strategies may be chosen for individuals that carry genes of interest, to augment the relevant allelic frequencies and reduce the risk of their loss from the population.

ACK N OWLED G EM ENTS
We

CO N FLI C T O F I NTE R E S T
None declared.