Genomic Characterization of the Bumblebee Bombus Atratus In

1 Walking the Tightrope: Genomic Characterization of the 2 Bumblebee Bombus atratus in Colombia 3

9 Julian Felipe Peña Polania

15 Advisor: 16 Emilio Realpe, Ph.D. 17

23 Universidad de los Andes 24 Facultad de Ciencias 25 Departamento de Ciencias Biológicas 26

27 28

29 INTRODUCTION

30 METHODS

31 RESULTS

32 DISCUSSION

33 ACKNOWLEDGMENTS

34 LITERATURE CITED

35 SUPLEMMENTARY MATERIALS

50 51 Wallking the Tightrope: Genomic Characterization of the Bumblebee Bombus 52 atratus in Colombia

53 Julian Felipe Peña Polania1, Emilio Realpe1,

54 1. Universidad De Los Andes, Departamento de Ciencias Biológicas. 55

56 Abstract

57 In recent years, habitat loss and climate change have been one of the main causes of the 58 decline of pollinator insects of high importance. However, one of the most affected groups 59 has been the bumblebees of the genus Bombus. In Colombia, the dynamics and 60 processes of dispersion of these species in landscapes affected by humans remain 61 unknown. Since the change in habitat can increase intrinsic threats to populations, 62 including genetic drift and inbreeding, the use of molecular tools in conjunction with 63 population genetics is essential to understand the most basic levels of biological diversity, 64 and identify the genetic status of populations affected by changes in the landscape. We 65 examine the patterns of genetic diversity and how environmental factors can shape it 66 through the use of NextRAD, in a population of bumblebees of the species Bombus 67 atratus, which has a range of distribution that overlaps highly agricultural ecosystems. Our 68 results reveal a low genetic differentiation among sample localities, suggesting a low effect 69 of land use on genetic flow. However, we found a negative correlation between genetic 70 diversity and land use, this indicated that environmental variables at the local level are 71 what are shaping genetic variations. We identified several loci under selection, a subset of 72 which was associated with land use, taken as the percentage of surface surrounding each 73 sampling site. Our findings shed important light on the life history of B. atratus and 74 highlight the role of land use as a phenomenon of change regarding the genetic diversity 75 of these bumblebees. Likewise, our research reveals for the first time patterns of local 76 adaptation in this species in Colombia.

78 Introduction

79 In recent years, the genus Bombus commonly known as bumblebees have shown drastic 80 population declines, more than 70% of the Bombus species worldwide have suffered from 81 processes of population decline or local extinction in the past 10 years (Potts et al., 2010; 82 Martin & Melo, 2010; Arbetman et al., 2017). These bees are one of the most important 83 pollinator groups due to their indispensable ecosystem service in natural and agricultural 84 landscapes (Garibaldi et al., 2013; Fijen et al., 2018). However, these wild bees exhibit a 85 substantial vulnerability as opposed to other groups, due to characteristics related to 86 complex life cycles and different degrees of sociality (Samuelson et al., 2018; Piiroinen & 87 Goulson, 2016; Persson & Smith, 2013; Rasmont & Iserbyt, 2012). For this reason, both 88 Europe and North America have begun to study in depth the causes of the decline of these 89 Apidae (Cameron et al., 2011; Williams & Osborne, 2009; Kerr et al., 2015; Rasmont et al., 90 2015). Although some of the factors related to the decline are still difficult to know (Kent et 91 al., 2018), it seems that different species of bumblebees are mostly affected by changes in 92 the landscape due to agricultural practices and human expansion (Sánchez-Bayo & 93 Wyckhuys, 2019; Jha & Cremen, 2013, Carvalheiro et al., 2013). This factors influences 94 the available resources and nesting places, considerably reshaping important biological 95 aspects of the bumblebees life cycle, promoting the prevalence of parasites and the 96 increase of inbreeding in populations (Goulson, Whitehorn & Fowley, 2012, Woodard et 97 al., 2015).

98 Despite the interest granted to the conservation of these bees in other countries, in 99 Colombia little is known about the population status of different species of bumblebees, a 100 total of nine species are reported with clearly documented records: B. excellens (Smith, 101 1879), B. funebris (Smith, 1854), B. hortulanus (Friese, 1904), B. melaleucus (Handlirschi, 102 1888), B. pullatus (Franklin, 1913), B. robustus (Smith, 1854), B. rubicundus (Smith, 103 1854), B. transversalis (Oliver, 1789), and B. atratus (Franlin, 1913) (Thompson & 104 Oldroyd, 2004; Parra et al., 2016). Moreover, the dynamics and processes of dispersion of 105 these species in landscapes affected by humans remain unknown (Martins et al., 2015). 106 Colombia display one of the greatest potentials in South America to develop the 107 agricultural landscape (De Jaramillo et al., 2017). However, the use of land, must take in to 108 account, how pollinator groups such as bumblebees cannot withstand drastic habitat 109 modifications (Marshall et al., 2018; Reininghaus et al., 2017; Goulson et al., 2010). Since 110 the change in habitat can increase intrinsic threats to populations, including genetic drift 111 and inbreeding (Darvill et al., 2006; 2012), the use of molecular tools in conjunction with 112 population genetics is essential to understand how the bumblebee species founded in 113 these areas are being affected (Zayed, 2009). Consequently, the study of the genetic 114 variation and genetic structure of bumblebees is called for elucidate the most basic levels 115 of biological diversity and identify the genetic status of populations affected by changes in 116 the landscape (Murray et al., 2009; Lozier & Cameron, 2009).

117 A clear example in the field of genomic conservation of bees is the greater availability of 118 genomic resources; Currently, 11 species of bees have a sequenced genome (Elsik et al., 119 2015), within which two species of bumblebees are found, Bombus impatiens and B. 120 terrestris (Sadd et al., 2015), this last two has given the opportunity to investigate in depth 121 the environmental factors that can affect the structure and genetic diversity of this 122 important group of bees. Next-generation sequencing (NGS) is rapidly changing the field 123 of genetic conservation for bees, allowing non-model species to be studied and helping 124 researchers to test new or previous hypotheses about the meaning of genetic variation 125 and the architecture that underlies several ecological and evolutionary processes (Lozier & 126 Zayed, 2017). The field of genomic conservation allows us to evaluate and quantify the 127 genetic diversity of relevant loci and study how species can respond to different 128 environmental threats (Holderegger et al., 2006; Schoville et al., 2012). Likewise, it allows 129 to accurately estimate the levels of genetic diversity and provide novel and relevant 130 information to delimit significant evolutionary units (Funk et al., 2012, Hoffmann et al., 131 2015).

132 Previous studies have investigated intraspecific genetic diversity in wild bees in order to 133 understand how landscape quality, fragmentation, isolation and potential barriers affect the 134 process by which genetic diversity change in a population (Jha, 2015; Darvill et al., 2010; 135 Wilson et al., 2016). These studies have included different estimates of population 136 structure based on; the genetic difference, like Fst (Brown, 1970); distance and resistance 137 isolation (IBD, IBR); tree-based distance and genetic clustering. The incorporation of these 138 estimates has allowed the creation of realistic models, which can identify with greater 139 certainty the factors that influence the structure and genetic variation of bees (Jackson et 140 al., 2018; Kent et al., 2018). Estimates of genetic diversity as effective population size and 141 heterozygosity; give important information about the population's health from genetic and 142 demographic perspectives (Charlesworth, 2009; Boff et al., 2014; Mattila & Seeley, 2007), 143 for this reason these estimates are efficient tools to identify species that may be in danger 144 (Laikere et al., 2010; Cameron et al., 2011; Maebe et al., 2015); and in synergy with NGS, 145 allows to develop optimal investigations, at optimum cost, that contemplate influence of 146 the landscape on genetic diversity understanding how the genetic variation is influenced 147 by environmental patterns (Lozier, 2014; Jaffé et al., 2014; Lozier et al., 2013; Epps & 148 Keyghobadi, 2015).

149 In this study, we use nextRAD to examine the genetic diversity and population structure of 150 the bumblebee Bombus atratus, in a fine geographic scale. This species is widely 151 distributed in South America, presenting one of the largest ranges of distribution compared 152 to other bumblebees (Abrahamovivich et al., 2007). In Colombia, this species has an 153 altitudinal range distribution from 1500 to 3800 meters above sea level, however, they are 154 mostly between 1800 and 2900 m. a. s. l. (Lievano et al., 1991). B. atratus is one of the 155 few species of bumblebees that overlap their range of distribution with areas of extensive 156 agriculture in Colombia (Gonzalez et al., 2004). This makes this bumblebee an appropriate 157 model to understand the processes of genetic variation and isolation due to changes in the 158 landscape. Few studies have examined the processes of genetic variation in high Andean 159 bumblebees at local spatial scales through heterogeneous landscapes (Françoso et al., 160 2019).

161 Our objectives in this study were: 1) Evaluate, if habitat loss produced by agricultural 162 growth, is affecting the genetic diversity of B.atratus. We measured land use, 163 heterogeneity and plant richness and dominance to quantify habitat status and determine 164 whether or not these variables are affecting the process that underlies the genetic variation 165 in our bumblebee. We expected to find a negative relation between genetic diversity and 166 habitats with high agricultural use, low heterogeneity and higher plant dominance, as 167 found in other species of bumblebees in danger or decay (Cameron & Sad, 2019), where 168 floral resources are essential for optimal population development; 2) To understand the 169 demographic patterns of this bumblebee species by the estimation of Tajima´s D (Nei &

170 Tajima, 1981) and the estimation of effective population size (Ne) of B. atratus in different 171 temporal periods; we take two points as important in the recent natural history of this 172 genus, and that will help us to understand patterns of decay: the establishment of the 173 species in the eastern mountain range a million years ago (Hines, 2014) and the beginning 174 of the extensive use of land by agricultural activities (a hundred year ago). (Ferreira et al., 175 2015; Vickruck et al., 2017). This is the first study in Colombia that uses NGS to 176 understand the patterns of genetic variation based on environmental variables in a species 177 of bumblebees of high importance, both for its ecosystem service and for its intrinsic 178 diversity. A B

 Sampled Bumblebees

179 180 Figure 1. A. Map of the sampled area, the map was produced using R (Murell, 2018) from the 181 BioClim data base (WorldClim: http://www.worldclim.org) for the raster of altitude, the points show 182 the samples collected. B, C, Predominantly black and yellow respectively forms of B.atratus. 183 184 Materials and Methods

185 Study Species and Sample Collection

186 Bombus atratus (Franklin, 1913), is a species found within the subgenus Fervidobombus 187 and is widely distributed in south America having one of the highest ranges of distribution 188 compared to other bumblebees (Abrahamovich et al. 2007). Current distribution includes 189 Venezuela, Colombia, Ecuador, Perú, Uruguay, Bolivia and Argentina (Abrahamovich et 190 al. 2004). In Colombia they can be found from 1500 to 3800 meters m. a. s. l. (Lievano et 191 al. 1991). They are also one of the few species whose range of distribution overlaps with 192 highly agricultural landscapes (Riaño et al., 2014), which makes it a suitable model to 193 recognizes the processes of genetic variation and isolation due to habitat loss.

194 We sampled 90 female workers of Bombus atratus from ten localities across two of the 195 Colombia departments with highest use of land use, Cundinamarca and Boyacá (DANE, 196 2017; Figure 1, supplementary Table 1). Bumblebee collection were made actively 197 searching in different types of flowers, always seeking the independence of the samples, 198 to avoid catching individual from the same nest, each sample was collected with a 199 minimum distance of 3 km due to the displacement patterns of the workers of this species 200 of bumblebees (Krauss et al., 2009; Pardo & Jimenez, 2006). Once collected, the samples 201 were stored in 97% alcohol and placed at -20 ° C until DNA extraction.

202 DNA Extration

203 DNA was extracted from the third pair of bumblebees legs with punctured metacoxa, using 204 the DNeasy Blood & Tissue Kit from Qiagen (Hilden, Germany). All the extractions were 205 checked by 2.5% agarose gels and then quantified with Qbit 2.0 Fluorometer (Invitrogen). 206 The selected samples had a concentration of ~10ng/ul. After, all the samples were sent to 207 SNPsaurus (SNPsaurus, LLC) for nextRAD sequencing. Voucher specimens were housed 208 at the Museo de Historia Natural ANDES, Universidad de los Andes, Bogotá, Colombia.

209 NextRAD Sequencing

210 Genomic DNA was converted into Nextera-tagmented reductively-amplified DNA 211 sequencing (NextRAD), and then genotyping-by-sequencing libraries as described by 212 Russello, Waterhouse, Etter, and Johnson (2015). Briefly, genomic DNA was made into 213 whole genome genotyping libraries with Nextera DNA Flex reagent (Illumina, San Diego, 214 CA, USA), this one ligates short adapter sequences to the ends of the fragments. The 215 Nextera reaction was scaled for fragmenting 10 nanograms of genomic DNA. Fragmented 216 DNA was then amplified, with one of the primers matching the adapter and extending nine 217 nucleotides into the genomic DNA with the selective sequence. Therefore, only fragments 218 starting with a sequence that can be hybridized by the selective sequence of the primer 219 were efficiently amplified by PCR. The WGS libraries were sequenced on a HiSeq X with 220 one lane of paired-end 150 bp reads. Custom scripts (SNPsaurus, LLC) were used to 221 assembly a de novo low quality reference genome using the reads from four samples, of 222 different localities, with abyss-pe and a k of 86. The resulting contigs were length filtered to 223 a minimum of 250 bp, then cleaned of contaminating species by blastn to the NCBI nt 224 database and removing blast hits to bacteria, plants and fungi. The gene annotation was 225 created with augustus using species=Bombus terrestris (Stanke & Morgenstern, 2005).

226 The genotyping analysis used custom scripts (SNPsaurus, LLC) that trimmed the reads 227 using bbduk (BBMap tools, http://sourceforge.net/projects/bbmap/): All reads were 228 mapped to the reference genome, with an alignment identity threshold of .95 using bbmap 229 (BBMap tools). Genotype calling was done using callvariants (BBMap tools). 230 SNP Filtering

231 We used VCFtools (0.1.6; Danecek et al. 2011) for filtering the variants and ensure that 232 VCF data set contained high quality, single-copy loci. We exclude genotypes with 233 genotype quality scores < 30. Sites with > 5% missing data were removed. Next, we 234 calculate average sequencing depth per site (~66), however a small fraction of loci had 235 much higher coverage; these could indicate paralogous or multi-copy regions (Koch et al., 236 2014), therefore sites with mean sequencing depth > 132 were removed from the data. 237 Then we filter the loci for strong deviations from Hardy-Weinberg equilibrium (hwe, 0.001). 238 The remaining sites, those with minor allele frequency of 0.025% across all individuals 239 were removed. After, due to the reproduction patterns of the bumblebees were sex 240 determination typically leads to males arising from unfertilized eggs (haploid) and females 241 from fertilized eggs (diploid), we exclude organisms from the same colony in our data 242 through a kinship analysis performed using the function of Plink1.9 rel-cutoff (Purcell, 243 2007), this excludes one member of each pair of samples with observed genomic 244 relatedness greater than the given cutoff value, this function is based on the algorithm 245 presented by Manichaikul and collaborators (2010), were the inference criteria goes from 0 246 to 0.5 (Monozygotic twin: 0.5; Parent–offspring: 0.25; Full sib: 0.25; 2nd Degree: 0.125; 247 3rd Degree:0.0625 Unrelated: 0).

248 Population Structure and Genetic Diversity 249 After filtering, to assess population structure two different approaches were used: the first 250 one, Admixture was used in Plink 1.9 (Purcell, 2007) and the second one the SNMF- 251 function of the LEA (v2.0) package (Frichot & François, 2015; Frichot, Mathieu, Trouillon, 252 Bouchard, & François, 2014). Linkage disequilibrium (LD) in bumblebees breaks down 253 very quickly (Maebe et al., 2016), so, we randomly sampled one variant from each RADtag 254 to ensure independence of loci prior to running SNMF. The number ancestral populations 255 K was allowed to vary between 1 and 10, keeping in mind the locations sampled, with 10 256 replicate runs for each K‐value, and the best K was chosen based on cross‐entropy and 257 cross‐validation errors (Frichot et al., 2014). Full R scripts of SNMF can be found in the 258 LEA website (http://members-timic.imag.fr/Olivier.francois/LEA/index.htm). Then, to 259 visualize genomic similarity among individuals, data was prepared for R using vcfR (Knaus 260 & Grünwald 2017) and the adegenet package (Jombart, 2008) was used to tested for 261 isolation by distance (IBD) using a Mantel test between a matrix of genetic distances 262 (Euclidian distance) and a matrix of geographic distances. Because correlation between 263 genetic and geographical distances can occur under various biological scenarios, such as 264 gradual change due to environmental conditions; or differentiation by comparison of distant 265 populations; a scatterplot was performed showing the consistency between previously 266 created matrices. After, we use VCFtools to calculated genetic diversity metrics for our 267 sample data. These, included nucleotide diversity per RADtag (π), observed

268 heterozygosity (HO) and expected heterozygosity (HE) per individual; we calculated the 269 inbreeding coefficient based on the observed versus expected number of homozygous 270 genotypes (F). Finally, we calculated Tajima’s D per loci (Ferretti et al., 2016), to represent 271 de differences between the mean number of pairwise variation and the number of 272 segregating sites. 273

274 Traditionally, Effective population size Ne has been estimated from demographic or 275 pedigree data (Murray et al., 2009, Pollack, 1983), nevertheless the use of high throughput 276 genotyping techniques, has allowed methods based on linkage disequilibrium (LD) 277 measures (Waples et al., 2016). As, LD between a pair of markers is determined by the

278 product of Ne, the frequency of recombination and the number of generations since the

279 mutation, the genetic distances between loci can be used to infer Ne at any point in time 280 from a single samples collected in the same generation (Dellicour et al., 2017). First, LD 281 was calculated using vcftools to determine the speed of genetic segregation, then the 282 proportion of recombination 4.4 cM/Mb was chosen as this one, is the same of the 283 organism with which the genome annotation was performed, B. terrestris. The natural 284 history of Bombus atratus is recent compared to old-world bumblebees (Hines, 2008), the 285 exponential growth of the eastern mountain range in Colombia in the last 4 million years 286 (Gregory-Wodzicki), the last maximum glaciation in the Pleistocene (Martini et al., 2017; 287 Vuilleumier, 1971), and the beginning of extensive agriculture in the mountain range 288 (León-Sicard et al., 2017), could shape the current distribution and genetic diversity of this 289 bumblebee, for this reason the effective population size was calculated at 3 points: 1 290 million years ago, 100 years ago and recently.

291 We followed the procedure of Saura and collaborators (2015) to estimate Ne with closely 292 separated SNPs providing information about the distant past (10,000 to 1,000,000 years

293 ago). Then, Ne of the other two points was estimated employing NeEstimator 2.0.1 (Do et 294 al., 2014). We used the LDNe method with the monogamous mating model, which is 295 approximately correct for bumblebees (Duchateau et al., 1994) to estimate effective 296 population size Ne in the past century, and finally ran the heterozygote excess and

297 molecular coancestry methods within NeEstimator to estimate very recent Ne. 298 299 Landscape Genetic Analyses 300 In order to assess the influence of landscape on genetic variation, we obtained the 301 following high resolution rasters: 1) A continuous cultivated and managed Vegetation 302 cover map; all the data layers from the cover maps contain unsigned 8-bit values and the 303 valid values range from 0-100, representing the consensus prevalence in percentage; 2) A 304 categorical first order range global land cover map; 3) A categorical high resolution second 305 order (Simpson and Shannon) homogeneity map; this two based on the textural features 306 of Enhanced Vegetation Index (EVI), texture measures are statistics describing the 307 frequency distribution of EVI values and measuring compositional variability within an 308 area. The second-order texture measures are statistics of the occurrence probabilities of 309 different EVI combinations among pixel pairs within an area and thus also reflect spatial 310 arrangement and dependency of the EVI values (JetzLab: http://www.earthenv.org, 311 Tuanmu & Jetz, 2014). 4) A continuous high-resolution digital elevation map (DEM) for the 312 whole study area (WorldClim: http://www.worldclim.org/), where every pixel contained an 313 elevation value expressed in meters. This last ones, since genetic differentiation has been 314 found to be influenced by elevation in bumblebees (Lozier et al. 2011). With the purpose to 315 accomplish our goal of comparing the genetic diversity with our sample data, we extracted 316 the values of the rasters at the locations were each bumblebee was captured. Taking into 317 account the distance of foraging and dispersion of B. atratus, boundary regions (called a 318 buffer) were created representing the spatial extent of each sample. Then, we extracted all 319 values pixels that fall within the buffer for each individual and used to estimate the 320 environmental information.

321 Because there could be a spatial correlation between the variables (HE, F), two mantel 322 tests were carried out to corroborate correlation effects, the VEGAN package was used in 323 R creating matrices of Euclidean distance of the variables with a 999 permutation value, to 324 compare them (Supplementary Material 1). Finally, we computed percentage of landscape

325 variation, as previously described, and genetic diversity (HO, HE, F) for each sample, we

326 used linear models (LM) with the package lmer, to account for genetic variation (HE, F) as 327 response of the habitat variables predictors described before. After, Akaike information 328 criterion (AICc) was used to compare models with different correlation structures, fitted 329 with restricted maximum likelihood. All models were validated checking for residual 330 autocorrelation and plotting residual versus fitted values. 331 332 Results 333 Genotyping and Filtering 334 Our variants were filtered to produce a total of 30505 SNPs from 74483 for B. atratus. The 335 relatedness analysis resulted in the removal of 20 samples (Table 1). The final set of data 336 comprised 70 bumblebees from 10 different localities, however, not all locations 337 maintained the same numbers of bumblebees.

Table 1 Filter summaries for the data with vcf and plink commands; “minQ” includes only sites with quality value above the threshold; “min-max alleles” include only sites with a number of alleles less, greater or equal to the value; “max-missing” exclude sites on the basis of the proportion of missing data; “max-meanDp” includes only sites with mean depth values; “hwe” assesses sites for Hardy-Weinberg Equilibrium using an exact test, sites with a p-value below the threshold defined by this option are taken to be out of HWE, and therefore excluded; “rel-cutoff” excludes one member of each pair of samples with observed genomic relatedness greater than the given cutoff value max- max- rel- Total SNPs / Total Min-max hwe minQ (10) missing meanDp cutoff Indv Alleles (2) (0.001) (0.95) (132) (0.2) 70320 / 45827 / 30505 / 30505 / 74483 / 90 74483 / 90 73410 / 90 90 90 90 70 338

339 Population Structure and Genetic Diversity 340 Patterns of genetic differentiation varied among samples and locations, and are likely 341 associated with the distribution and niche across a landscape gradient. The heterozygosity 342 observed and expected, did not have a significant difference, however the variation 343 observed is greater than expected, the inbreeding factor F presented a negative value 344 consistent with the heterozygosity observed, however there are samples with high values 345 of inbreeding and low levels of heterozygosity (Table 2, supplementary Table 2). Average 346 π per SNP in B atratus population ranged from 0.095 - 0.153. The index of Tajima’s D 347 presented a deviation from the neutral expectation, with a negative value, this may indicate 348 that our bumblebee population may be passing for a bottleneck process or an event of 349 variability removal.

Table 2. Genetic diversity estimates for Bombus atratus population. Effective population size (Ne); Observed and expected heterozygocity respectively (Ho; He); Inbreedinf factor (F); Nucleotide diversity measured as the degree of differences per site between the DNA sequences (π). Tajimas´D measured as the number of pairwise differences within the number of segregating sites. Total Ne Ho He F π Tajima’s D

6.4 0.11873 0.1118 (-)0.06095 0.011259 (-)0.6187 70 (SD±0.013) (SD±0.013) (SD±0.0002) (SD±0.1224) (SD±0.0036) (SD±0.020) 350

351 We detected significant genetic differentiation relative to geographic distance (IBD) across 352 our populations of B. atratus (r2=0.196, p-value << 0.01). Nevertheless, we did not detect 353 genetic clusters using the two different clustering approaches. Significant genetic 354 structuring across the locations of B. atratus were not found, though cross entropy (ΔK) 355 statistic was greatest at K = 2, with less explanatory power gained by additional K 356 (Supplementary Figure 1.A). However, even at K = 2, our bumblebee species lack 357 population structuring, which is corroborated by the scatterplot correlation across 358 geographic distance and genetic distances (Supplementary Figure 1.A). The distribution of 359 genetic variation among genotyped individuals in relation to the neutral loci filtered can be 360 visualized in the figure 2. In this analysis using 30505 informative SNPs, the ten locations 361 just formed one distinct cluster, still seems that there would be a gradient between north 362 and south where there are some SNPs fixed at highlands and others in the lowlands. The 363 fairly weak clustering observed from structure suggested a potential ongoing gene flow 364 among geographic regions (Figure 2.B and C). A B

365 366 Figure 2. A. Genetic assignment of localities (sampling sites) to K=2 populations for the bumblebee 367 B. atratus in the altiplano of the departments of Boyacá and Cundinamarca. The pie slice of each 368 circle represent the average genetic assignment of all individuals in each location (10) to one of K 369 populations. B. Admixture proportions of each individual from K = 2 and K=3, obtained in Structure 370 for our data set, localities are assigned from 1 to 10 (1. Arcabuco 2. Choconta 3. Duitama 4. 371 Facatativa 5. Guasca 6. Morro 7. Pacho 8. Samaca 9. Tenjo 10. Villa Pinzon). C. Principal 372 component analysis showing maximum explained variances (30%) for B. atratus. 373

374 The estimation of Ne based on LD between SNPs showed a large decline in effective 375 population size that likely occurred within the past 100 years, consistent with ecological 376 evidence of decline for bumblebee species. The values reported are for MAF = 0.05; Each

377 estimates of Ne at a different time in the past, giving the three time points are show in 378 Figure 3A. The average r2 between adjacent SNP was 0.09 (±0.17) and the mean distance 379 between adjacent SNP was 1026.25 bp. Figure 3B represents the decline in r2 between 380 SNP pairs with increasing distance, lower levels of LD and faster rates of LD decay are 381 typically found in large, outbred populations, whereas in small bottlenecked populations 382 background levels of LD are higher and LD decays much more slowly (Knigth et al., 2008). 383

A B

Log10(Ne)

384 385 386 Figure 3. A. Estimated Ne-100 (last 100 gen Ne, green), recent Ne (blue), and ancestral Ne (red) 387 show orders of magnitude declines from ancient to recent to previous generation timescales. The 388 error bars lie with the points show 95% confidence intervals. B. Average linkage disequilibrium 389 plotted against distance between SNPs. Average linkage disequilibrium (solid line) measured as r2. 390 391 Landscape genetics 392 393 Habitat use and altitude was found to be associated with heterozygosity an inbreeding 394 (Figure 4; 5). Predictors such as: heterogeneity; diversity and abundance of plants 395 categorized by EVI turned out not to have a significant relationship with respect to genetic 396 diversity and inbreeding. In spite of this, the interactions between the variables provided a 397 better explanation for He and F (Table 3, 4), where the diversity trends were consistent 398 across the full range of environmental variables. However, 50% of the variance was 399 explained just by habitat use and altitude.

Land use (km)2

400 401 Figure 4. Individual average heterocigocity and inbreeding for B. atratus with respect to agricultural 402 use km2 and altitude. Solid lines are trend lines for each variable. (Table 4; 5)

Table 3

Summary statistics for the Linear regression models for Heterozygosity(LRM)

Explained Predictor and Formula (Model) AIC ΔAIC LogLik P Variance Heterozygosity ~ Land Use*** + << -441.46 0 0.506 225.04 Altitude ** 0.001 Heterozygosity ~ Heterogeneity + -436.93 4.53 0.015 233.2 0.6 Altitude Heterozygosity~(CF*EVIsim*EVIsha) -393.1 48.37 0.14 200.86 0.03 + Altitude

Heterozygosity ~ All predictors -387.5 53.96 0.609 205.61

403 Table 4

Summary statistics for the Linear regression models for Inbreeding(LRM)

Explained Predictor and Formula (Model) AIC ΔAIC LogLik P Variance Inbreeding ~ Land Use*** + << -134.8 0 0.506 71.71 Altitude ** 0.001 Inbreeding ~ Heterogeneity + -130.25 4.54 0.015 79.86 0.6 Altitude Inbreeding~(CF*EVIsim*EVIsha) -86.42 48.37 0.14 47.52 0.03 + Altitude

Inbreeding ~ All predictors -161 53.96 0.609 52.27

404 405 Discussion 406 We studied the process of differentiation of a bumblebee population at a fine geographic 407 scale of 300 km, due to factors that affect its habitat such as an extensive agricultural use 408 and landscape heterogeneity. Our study revealed a minimal gradual change in genetic 409 structure across the sampled distribution of B. atratus, with no identifiable clusters. This 410 low levels of genetic structure that were found for our bumblebee species and in 411 congruence with the analysis of IBD most likely reflect the high dispersal abilities, in 412 accordance with results from other continental bumblebee populations in Europe and 413 North America (Spevak, Jepson & Williams, 2015) this is not surprising as reproductive 414 bumblebees caste usually can move for a range of 3 km and a maximum of 15 km when 415 there is a perfect condition for dispersion (Lepais et al., 2010).The lack of evidence for 416 strongly isolated populations certainly raises questions regarding the factors that could be 417 playing an essential role in the genetic variation of these bumblebees, it may thus be the 418 case that substantial habitat heterogeneity is required to produce detectable genetic 419 structure in bumble bees collected at local geographic scales. 420 421 Heterozygosity and inbreeding was influenced by land use of habitat and altitude (Figure. 422 4), but not for habitat heterogeneity; this agree with previous studies that found that 423 heterogeneous landscapes in generalist and large distributed bumblebees, as B. atratus, 424 may not have a broad effect on genetic diversity due to the behavior patterns of these 425 bees (Laiolo et al., 2018; Rodriguez & Kouki, 2017; Kraus et al., 2009; Osborne et al., 426 2008). However, the land use is affecting the processes underlying genetic variation, we 427 observed that in places where there is a greater use of the landscape the inbreeding factor 428 per individual is greater, as well, the heterozygosity decreases in relation to the mean land 429 used (Figure 4). As described by Cameron and Sadd (2019) and corroborated in this 430 investigation, the extensive land use can cause irreparable damage to the resources 431 necessary for the development and function of bumblebees (Colgan et al., 2019), making 432 parasitic diseases more prevalent, decreasing the nesting sites and making floral resource 433 deficient in quality. A clear example, is reflected in our data when we compared genetic 434 diversity with respect to plant dominance (Table 3, 4), when there is a high dominance of a 435 type of plant due to extensive agricultural use, genetic diversity tends to decline, despite 436 that B. atratus is a generalist bumblebee, it has been shown that there are inflection points 437 regarding the availability of resources that can shape the genetic stability of a population 438 (Harmon‐Threatt et al., 2017). 439 440 The other important factor with significance in the genetic diversity was the altitude, 441 although our samples did not vary dramatically, there is a clear pattern which may be 442 related to physiological responses to environmental pressures in the highlands as low 443 temperature, low oxygen and low air density (Dillon & Lozier, 2019); Cold temperatures at 444 high altitudes may shut down bumblebees for large parts of the day and night, also they 445 rely in large part on aerobic respiration to supply energy demands (Dillion et al., 2006), this 446 makes daily forages and flight patterns much shorter compared to lowland individuals 447 (Oyen et al., 2016). Because of these challenges, we could be watching reductions in 448 abundance and diversity of this bumblebees, and the patters of SNPs fixation founded in 449 this study (Figure 2). The movement of mountain bumblebees in search of more temperate 450 places (Sirois-Delisle, 2018), has been mostly reflected in bumblebees that do not have a 451 wide altitudinal range, in contrast with our bumblebee species (Rasmont et al., 2015). 452 Although the imposition of climate change on bumblebee populations with wide ranges 453 cannot be ruled out (Kerr et al., 2015), a detailed study is necessary to determine if climate 454 change in the last decade may be affecting this group of bees. 455 456 Following the results from Packer and Owen (2001), our mean F value suggest that up to 457 10% of the male population may be diploid. Complementary sex determination increases 458 extinction risk of bees due to the diploid male vortex, inbred bee populations produce 459 higher numbers of infertile diploid males instead of workers and gynes (Zayed and Packer, 460 2005; Duchateau et al., 1994), and these males are not fertile or produce non-viable 461 offspring, reducing the effective population size. As an important parameter in 462 conservation genetics, effective population size, is a powerful tool that allows to infer 463 demographic patterns that explain the processes through which a population is going

464 through (Soule, 1987). As social insects, bumblebees Ne will be very low in relation to the 465 observed number of specimens, as most bumblebee species are monoandrous and their 466 colonies consist mostly out of only one founder queen. 467

468 We were able to estimate the ancestral long-term harmonic mean Ne in the period 469 corresponding to one million years ago and ten thousand years (Figure 3), which presents

470 a range of 100.000 and 200.000. Although other studies calculate ancestral Ne values 471 between 300,000 to 600,000 (Maebe et al., 2016; Kent et al., 2011); our unusual low

472 ancestral Ne with respect to 1 million years ago it is probably related to a composite of a 473 bottleneck due to the recent growth of the eastern cordillera (4M years ago) and the post- 474 glacial expansions in the Pleistocene (Mo-llanes et al., 2019; Montemayor et al., 2018) 475 These phenomes may have caused a selective sweep processes on past populations of B.

476 atratus reflecting an ancestral low Ne compared to other species (Maebe et al., 2016; Ellis 477 et al., 2006), and in turn forging the current genetic variations that we can see today in this 478 population. Our methods which estimate recent Ne from recent times gave results in

479 orders of magnitude: around Ne = 1840 for the LDNe method (last 100 generations) and 480 Neb = 6.2 for the molecular ancestry method (last generation). These dramatically smaller 481 estimates suggest a recent major reduction in effective population size of B. atratus, and 482 are consistent with ecological evidence of decline for bumblebee species worldwide. 483 Although extended bottleneck processes usually show values of Tajima’s D normally 484 positives (Tajima, 1989), we observe mean D = -0.06, this may be indicating the beginning 485 of a process of loss of genetic variability in a short period of time, corresponding to 486 changes in the landscape due to agricultural growth (Shmack et al., 2019; Zayed, 2004) 487 488 In general, habitat loss, is found to be associated with changes in foraging processes 489 (greater effort) and the prevalence of parasites, which in turn leads to the disruption of 490 metabolic functions related to immunity genes (Barribeau & Schmid‐Hempelp, 2017; 491 Barribea et al., 2011). Although the strength of responses varies in ways that may reflect 492 interesting evolutionary dynamics, and in our case is showing the loss genetic diversity, 493 the potential for adaptation to habitat loss, or local adaptation to existing abiotic variation, 494 requires further study for B. atratus. However, it is clear that these patterns are important 495 to understand that local adaptation processes are unique (Lowe, & Hoffmann, 2011), and 496 that the genetic diversity of these bumblebees should be preserved to maintain 497 evolutionary potential of this species (Hoffmann & Sgrò, 2011). 498 499 Together, our results suggest that the processes that are affecting genetic variation may 500 be related to local factors and possible physiological restrictions, as the use of land and 501 altitude respectively. Our findings shed important light on the life history of B. atratus and 502 highlight the role of land use as a phenomenon of change regarding the genetic diversity 503 of these bumblebees. Likewise, our research reveals for the first time patterns of local 504 inbreeding and loss of genetic diversity in this species in Colombia. This knowledge could 505 help guide future conservation actions such as avoiding future schemes of bumblebees 506 importation, and better emphasize resources in conserving and restoring mountain 507 ecosystems that may be supporting a genetic diversity still unknown. Considering the high 508 biological and economic importance of this native pollinator, future work should test 509 whether and how loci variants are of functional relevance for adaptation to life in a 510 constantly changing ecosystem due to anthropogenic growth. 511 512 Acknowledgments 513 We thank to the Rufford Foundation by the supporting grant (26114-1), to the to the 514 research fund of the science faculty of Universidad de los Andes (Convocatoria 2018-1 515 Financiación de proyectos de Investigación y presentación de resultados en eventos 516 académicos). To the Ph.D Jeffrey Lozier for the advice throughout the investigation, and 517 to all the families that allowed us to sample in their lands. 518 519 References 520 Abrahamovich, A., Díaz, N., & Lucia, M. (2007). Identificación de las “abejas sociales” del 521 género Bombus (Hymenoptera, Apidae) presentes en la Argentina: clave pictórica, 522 diagnosis, distribución geográfica y asociaciones florales. Revista de la Facultad de 523 Agronomía, La Plata, 106(2), 165-176.

524 Arbetman, M. P., Gleiser, G., Morales, C. L., Williams, P., & Aizen, M. A. (2017). Global 525 decline of bumblebees is phylogenetically structured and inversely related to species 526 range size and pathogen incidence. Proceedings of the Royal Society B: Biological 527 Sciences, 284(1859), 20170204. 528 Barribeau, S. M., & Schmid‐Hempel, P. (2017). Sexual healing: mating induces a protective 529 immune response in bumblebees. Journal of evolutionary biology, 30(1), 202-209.

530 Barribeau, S. M., Schmid-Hempel, P., & Sadd, B. M. (2016). Royal decree: gene expression 531 in trans-generationally immune primed bumblebee workers mimics a primary immune 532 response. PLoS One, 11(7), e0159635.

533 Boff, S., Soro, A., Paxton, R. J., & Alves-dos-Santos, I. (2014). Island isolation reduces 534 genetic diversity and connectivity but does not significantly elevate diploid male 535 production in a neotropical orchid bee. Conservation genetics, 15(5), 1123-1135.

536 Brown, A. H. D. (1970). The estimation of Wright's fixation index from genotypic 537 frequencies. Genetica, 41(1), 399-406.

538 Bushnell, B., Rood, J., & Singer, E. (2017). BBMerge–accurate paired shotgun read merging 539 via overlap. PLoS One, 12(10), e0185056. 540 Cameron, S. A., & Sadd, B. M. (2019). Global Trends in Bumble Bee Health. Annual review 541 of entomology, 65.

542 Cameron, S. A., Lozier, J. D., Strange, J. P., Koch, J. B., Cordes, N., Solter, L. F., & 543 Griswold, T. L. (2011). Patterns of widespread decline in North American bumble 544 bees. Proceedings of the National Academy of Sciences, 108(2), 662-667. Williams, P. 545 H., & Osborne, J. L. (2009). Bumblebee vulnerability and conservation world- 546 wide. Apidologie, 40(3), 367-387.

547 Carvalheiro, L. G., Kunin, W. E., Keil, P., Aguirre‐Gutiérrez, J., Ellis, W. N., Fox, R., & Van 548 de Meutter, F. (2013). Species richness declines and biotic homogenization have 549 slowed down for NW‐European pollinators and plants. Ecology letters, 16(7), 870-878.

550 Charlesworth, B. (2009). Effective population size and patterns of molecular evolution and 551 variation. Nature Reviews Genetics, 10(3), 195.

552 Colgan, T. J., Fletcher, I. K., Arce, A. N., Gill, R. J., Ramos Rodrigues, A., Stolle, E., & Wurm, 553 Y. (2019). Caste‐and pesticide‐specific effects of neonicotinoid pesticide exposure on 554 gene expression in bumblebees. Molecular ecology, 28(8), 1964-1974.

555 Danecek, P., Auton, A., Abecasis, G., Albers, C. A., Banks, E., DePristo, M. A., & McVean, 556 G. (2011). The variant call format and VCFtools. Bioinformatics, 27(15), 2156-2158.

557 Darvill, B., Ellis, J. S., Lye, G. C., & Goulson, D. (2006). Population structure and inbreeding 558 in a rare and declining bumblebee, Bombus muscorum (Hymenoptera: 559 Apidae). Molecular Ecology, 15(3), 601-611.

560 Darvill, B., Lepais, O., Woodall, L. C., & Goulson, D. (2012). Triploid bumblebees indicate a 561 direct cost of inbreeding in fragmented populations. Molecular Ecology, 21(16), 3988- 562 3995.

563 Darvill, B., O’Connor, S., Lye, G. C., Waters, J., Lepais, O., & Goulson, D. (2010). Cryptic 564 differences in dispersal lead to differential sensitivity to habitat fragmentation in two 565 bumblebee species. Molecular Ecology, 19(1), 53-63. 566 Dellicour, S., Kastally, C., Varela, S., Michez, D., Rasmont, P., Mardulyn, P., & Lecocq, T. 567 (2017). Ecological niche modelling and coalescent simulations to explore the recent 568 geographical range history of five widespread bumblebee species in Europe. Journal of 569 biogeography, 44(1), 39-50.

570 Dillon, M. E., Frazier, M. R., & Dudley, R. (2006). Into thin air: physiology and evolution of 571 alpine insects. Integrative and Comparative Biology, 46(1), 49-61.

572 Dillon, M. E., & Lozier, J. D. (2019). Adaptation to the abiotic environment in insects: the 573 influence of variability on ecophysiology and evolutionary genomics. Current opinion in 574 insect science. 575 Do, C., Waples, R. S., Peel, D., Macbeth, G. M., Tillett, B. J., & Ovenden, J. R. (2014). 576 NeEstimator v2: re‐implementation of software for the estimation of contemporary 577 effective population size (Ne) from genetic data. Molecular ecology resources, 14(1), 578 209-214.

579 Duchateau, M. J., Hoshiba, H., & Velthuis, H. H. W. (1994). Diploid males in the bumble bee 580 Bombus terrestris: sex determination, sex alleles and viability. Entomologia 581 experimentalis et applicata, 71(3), 263-269.

582 Ellis, J. S., Knight, M. E., Darvill, B., & Goulson, D. (2006). Extremely low effective population 583 sizes, genetic structuring and reduced genetic diversity in a threatened bumblebee 584 species, Bombus sylvarum (Hymenoptera: Apidae). Molecular Ecology, 15(14), 4375- 585 4386.

586 Elsik, C. G., Tayal, A., Diesh, C. M., Unni, D. R., Emery, M. L., Nguyen, H. N., & Hagen, D. 587 E. (2015). Hymenoptera Genome Database: integrating genome annotations in 588 HymenopteraMine. Nucleic acids research, 44(D1), D793-D800.

589 Epps, C. W., & Keyghobadi, N. (2015). Landscape genetics in a changing world: 590 disentangling historical and contemporary influences and inferring change. Molecular 591 ecology, 24(24), 6021-6040.

592 Ferreira, P. A., Boscolo, D., Carvalheiro, L. G., Biesmeijer, J. C., Rocha, P. L., & Viana, B. 593 F. (2015). Responses of bees to habitat loss in fragmented landscapes of Brazilian 594 Atlantic Rainforest. Landscape ecology, 30(10), 2067-2078.

595 Ferretti, L., Schmiegelt, B., Weinreich, D., Yamauchi, A., Kobayashi, Y., Tajima, F., & Achaz, 596 G. (2016). Measuring epistasis in fitness landscapes: the correlation of fitness effects 597 of mutations. Journal of theoretical biology, 396, 132-143.

598 Fijen, T. P., Scheper, J. A., Boom, T. M., Janssen, N., Raemakers, I., & Kleijn, D. (2018). 599 Insect pollination is at least as important for marketable crop yield as plant quality in a 600 seed crop. Ecology letters, 21(11), 1704-1713.

601 Françoso, E., Zuntini, A. R., & Arias, M. C. (2019). Combining phylogeography and future 602 climate change for conservation of Bombus morio and B. pauloensis (Hymenoptera: 603 Apidae). Journal of Insect Conservation, 23(1), 63-73. 604 Frichot, E., & François, O. (2015). LEA: an R package for landscape and ecological 605 association studies. Methods in Ecology and Evolution, 6(8), 925-929.

606 Garibaldi, L. A., Steffan-Dewenter, I., Winfree, R., Aizen, M. A., Bommarco, R., 607 Cunningham, S. A., & Bartomeus, I. (2013). Wild pollinators enhance fruit set of crops 608 regardless of honey bee abundance. Science, 339(6127), 1608-1611.

609 Gonzalez, V. H., Mejia, A., & Rasmussen, C. (2004). Ecology and nesting behavior of 610 Bombus atratus Franklin in Andean highlands (Hymenoptera: Apidae). Journal of 611 Hymenoptera Research, 13(2), 234-242.

612 Goulson, D., Lepais, O., O’Connor, S., Osborne, J. L., Sanderson, R. A., Cussans, J., & 613 Darvill, B. (2010). Effects of land use at a landscape scale on bumblebee nest density 614 and survival. Journal of Applied Ecology, 47(6), 1207-1215.

615 Goulson, D., Whitehorn, P., & Fowley, M. (2012). Influence of urbanization on the 616 prevalence of protozoan parasites of bumblebees. Ecological Entomology, 37(1), 83- 617 89.

618 Gregory-Wodzicki, K. M. (2000). Uplift history of the Central and Northern Andes: a review. 619 Geological Society of America Bulletin, 112(7), 1091-1105.

620 Harmon‐Threatt, A. N., de Valpine, P., & Kremen, C. (2017). Estimating resource 621 preferences of a native bumblebee: the effects of availability and use–availability 622 models on preference estimates. Oikos, 126(5), 633-641.

623 Hines, H. M. (2008). Historical biogeography, divergence times, and diversification patterns 624 of bumble bees (Hymenoptera: Apidae: Bombus). Systematic biology, 57(1), 58-75.

625 Hodson de Jaramillo, E., Castaño, J., Poveda, G., Roldán, G., & Chavarriaga, P. (2017). 626 Seguridad alimentaria y nutricional en Colombia. La Red Interamericana de Academias 627 de Ciencias (IANAS).

628 Hoffmann, A. A., & Sgro, C. M. (2011). Climate change and evolutionary 629 adaptation. Nature, 470(7335), 479.

630 Hoffmann, A., Griffin, P., Dillon, S., Catullo, R., Rane, R., Byrne, M., & Lockhart, P. (2015). 631 A framework for incorporating evolutionary genomics into biodiversity conservation and 632 management. Climate Change Responses, 2(1), 1.

633 Holderegger, R., Kamm, U., & Gugerli, F. (2006). Adaptive vs. neutral genetic diversity: 634 implications for landscape genetics. Landscape Ecology, 21(6), 797-807.

635 Jackson, J. M., Pimsler, M. L., Oyen, K. J., Koch‐Uhuad, J. B., Herndon, J. D., Strange, J. 636 P., & Lozier, J. D. (2018). Distance, elevation and environment as drivers of diversity 637 and divergence in bumble bees across latitude and altitude. Molecular ecology, 27(14), 638 2926-2942. 639 Jaffé, R., Castilla, A., Pope, N., Imperatriz-Fonseca, V. L., Metzger, J. P., Arias, M. C., & 640 Jha, S. (2016). Landscape genetics of a tropical rescue pollinator. Conservation 641 Genetics, 17(2), 267-278.

642 Jha, S. (2015). Contemporary human‐altered landscapes and oceanic barriers reduce 643 bumble bee gene flow. Molecular Ecology, 24(5), 993-1006.

644 Jha, S., & Kremen, C. (2013). Urban land use limits regional bumble bee gene 645 flow. Molecular Ecology, 22(9), 2483-2495.

646 Jombart, T. (2008). adegenet: a R package for the multivariate analysis of genetic 647 markers. Bioinformatics, 24(11), 1403-1405.

648 Jouganous, J., Long, W., Ragsdale, A. P., & Gravel, S. (2017). Inferring the joint 649 demographic history of multiple populations: beyond the diffusion 650 approximation. Genetics, 206(3), 1549-1567.

651 Kass, J. M., Vilela, B., Aiello‐Lammens, M. E., Muscarella, R., Merow, C., & Anderson, R. 652 P. (2018). Wallace: A flexible platform for reproducible modeling of species niches and 653 distributions built for community expansion. Methods in Ecology and Evolution, 9(4), 654 1151-1156.

655 Kent, C. F., Dey, A., Patel, H., Tsvetkov, N., Tiwari, T., MacPhail, V. J., & Colla, S. R. 656 (2018). Conservation genomics of the declining North American bumblebee Bombus 657 terricola reveals inbreeding and selection on immune genes. Frontiers in genetics, 9, 658 316.

659 Kerr, J. T., Pindar, A., Galpern, P., Packer, L., Potts, S. G., Roberts, S. M., & Wagner, D. 660 L. (2015). Climate change impacts on bumblebees converge across 661 continents. Science, 349(6244), 177-180.

662 Kerr, J. T., Pindar, A., Galpern, P., Packer, L., Potts, S. G., Roberts, S. M., ... & Wagner, D. 663 L. (2015). Climate change impacts on bumblebees converge across 664 continents. Science, 349(6244), 177-180.

665 Knaus, B. J., & Grünwald, N. J. (2017). vcfr: a package to manipulate and visualize variant 666 call format data in R. Molecular Ecology Resources, 17(1), 44-53.

667 Koch, V., Nissen, I., Schmitt, B. D., & Beye, M. (2014). Independent evolutionary origin of 668 fem paralogous genes and complementary sex determination in hymenopteran 669 insects. PLoS One, 9(4), e91883.

670 Kraus, F. B., Wolf, S., & Moritz, R. F. A. (2009). Male flight distance and population 671 substructure in the bumblebee Bombus terrestris. Journal of Animal Ecology, 78(1), 672 247-252.

673 Laikre, L., Allendorf, F. W., Aroner, L. C., Baker, C. S., Gregovich, D. P., Hansen, M. M., & 674 Olivieri, I. (2010). Neglect of genetic diversity in implementation of the convention on 675 biological diversity. Conservation Biology, 24(1), 86-88. 676 Laiolo, P., Pato, J., & Obeso, J. R. (2018). Ecological and evolutionary drivers of the 677 elevational gradient of diversity. Ecology letters, 21(7), 1022-1032.

678 León-Sicard, T., De Prager, M. S., & Acevedo Osorio, Á. (2017). Toward a history of 679 agroecology in Colombia. Agroecology and Sustainable Food Systems, 41(3-4), 296- 680 310.

681 Lepais, O., Darvill, B. E. N., O’connor S., Osborne, J. L., Sanderson, R. A., Cussans, J., & 682 Goulson, D. (2010). Estimation of bumblebee queen dispersal distances using sibship 683 reconstruction method. Molecular Ecology, 19(4), 819-831.

684 Li, A., & Barber, R. F. (2019). Multiple testing with the structure‐adaptive Benjamini– 685 Hochberg algorithm. Journal of the Royal Statistical Society: Series B (Statistical 686 Methodology), 81(1), 45-74.

687 Lievano, A., Ospina, R., & Nates, G. (1991). Distribución altitudinal del género bombus en 688 Colombia (Hymenoptera: Apidae). Trianea, 4(1), 541-550.

689 Lozier, J. D., & Cameron, S. A. (2009). Comparative genetic analyses of historical and 690 contemporary collections highlight contrasting demographic histories for the bumble 691 bees Bombus pensylvanicus and B. impatiens in Illinois. Molecular Ecology, 18(9), 692 1875-1886.

693 Lozier, J. D., & Zayed, A. (2017). Bee conservation in the age of genomics. Conservation 694 Genetics, 18(3), 713-729.

695 Lozier, J. D., Strange, J. P., & Koch, J. B. (2013). Landscape heterogeneity predicts gene 696 flow in a widespread polymorphic bumble bee, Bombus bifarius (Hymenoptera: 697 Apidae). Conservation Genetics, 14(5), 1099-1110.

698 Knight, M. E., Martin, A. P., Bishop, S., Osborne, J. L., Hale, R. J., Sanderson, R. A., & 699 Goulson, D. (2005). An interspecific comparison of foraging range and nest density of 700 four bumblebee (Bombus) species. Molecular Ecology, 14(6), 1811-1820.

701 Maebe, K., Meeus, I., Ganne, M., De Meulemeester, T., Biesmeijer, K., & Smagghe, G. 702 (2015). Microsatellite analysis of museum specimens reveals historical differences in 703 genetic diversity between declining and more stable Bombus species. PloS one, 10(6), 704 e0127870.

705 Maebe, K., Meeus, I., Vray, S., Claeys, T., Dekoninck, W., Boevé, J. L., ... & Smagghe, G. 706 (2016). A century of temporal stability of genetic diversity in wild bumblebees. Scientific 707 reports, 6, 38289.

708 Marshall, L., Biesmeijer, J. C., Rasmont, P., Vereecken, N. J., Dvorak, L., Fitzpatrick, U., & 709 Pawlikowski, T. (2018). The interplay of climate and land use change affects the 710 distribution of EU bumblebees. Global change biology, 24(1), 101-116.

711 Martini, M. A., Kaplan, M. R., Strelin, J. A., Astini, R. A., Schaefer, J. M., Caffee, M. W., & 712 Schwartz, R. (2017). Late Pleistocene glacial fluctuations in Cordillera Oriental, 713 subtropical Andes. Quaternary Science Reviews, 171, 245-259. 714 Martins, A. C., & Melo, G. A. (2010). Has the bumblebee Bombus bellicosus gone extinct in 715 the northern portion of its distribution range in Brazil?. Journal of Insect 716 Conservation, 14(2), 207-210.

717 Martins, A. C., Silva, D. P., De Marco, P., & Melo, G. A. (2015). Species conservation under 718 future climate change: the case of Bombus bellicosus, a potentially threatened South 719 American bumblebee species. Journal of insect conservation, 19(1), 33-43.

720 Mattila, H. R., & Seeley, T. D. (2007). Genetic diversity in honey bee colonies enhances 721 productivity and fitness. Science, 317(5836), 362-364.

722 Montemayor, S. I., Melo, M. C., Scattolini, M. C., Pocco, M. E., del Río, M. G., Dellapé, G., 723 & Dellapé, P. M. (2017). The fate of endemic insects of the Andean region under the 724 effect of global warming. PloS one, 12(10), e0186655.

725 Moo-Llanes, D., Pechamay, A., Ibarra C., Rebollar, E., & Ramsey, J. (2019). Inferring 726 distributional shifts of epidemiologically important North and Central American sandflies 727 from Pleistocene to future scenarios. Medical and Veterinary Entomology, 33, 31-43.

728 Murray, T. E., Kuhlmann, M., & Potts, S. G. (2009). Conservation ecology of bees: 729 populations, species and communities. Apidologie, 40(3), 211-236. 730 Murrell, P. (2018). R graphics. CRC Press.

731 Nei, M., & Tajima, F. (1981). DNA polymorphism detectable by restriction 732 endonucleases. Genetics, 97(1), 145-163.

733 Osborne, J. L., Martin, A. P., Carreck, N. L., Swain, J. L., Knight, M. E., Goulson, D., & 734 Sanderson, R. A. (2008). Bumblebee flight distances in relation to the forage 735 landscape. Journal of Animal Ecology, 77(2), 406-415.

736 Oyen, K. J., Giri, S., & Dillon, M. E. (2016). Altitudinal variation in bumble bee (Bombus) 737 critical thermal limits. Journal of Thermal Biology, 59, 52-57.

738 Packer, L., & Owen, R. (2001). Population genetic aspects of pollinator 739 decline. Conservation Ecology, 5(1). 740 Pardo, L., & Jimenez, L. (2006). Observación de rangos de vuelo de Bombus atratus 741 (Hymenoptera: Apidae) en ambientes urbanos. Acta Biológica Colombiana, 11(2), 131- 742 136.

743 Parra, G., et al. (2016). Iniciativa Colombiana de Polinizadores Capítulo Abejas -ICPA.

744 Pérez‐Angel, L. C., & Molnar, P. (2017). Sea surface temperatures in the Eastern Equatorial 745 Pacific and surface temperatures in the Eastern Cordillera of Colombia during El Niño: 746 Implications for pliocene conditions. Paleoceanography, 32(11), 1309-1314.

747 Persson, A. S., & Smith, H. G. (2013). Seasonal persistence of bumblebee populations is 748 affected by landscape context. Agriculture, ecosystems & environment, 165, 201-209. 749 Piiroinen, S., & Goulson, D. (2016). Chronic neonicotinoid pesticide exposure and parasite 750 stress differentially affects learning in honeybees and bumblebees. Proceedings of the 751 Royal Society B: Biological Sciences, 283(1828), 20160246.

752 Pollak, E. (1983). A new method for estimating the effective population size from allele 753 frequency changes. Genetics, 104(3), 531-548.

754 Potts, S. G., Biesmeijer, J. C., Kremen, C., Neumann, P., Schweiger, O., & Kunin, W. E. 755 (2010). Global pollinator declines: trends, impacts and drivers. Trends in ecology & 756 evolution, 25(6), 345-353.

757 Purcell, S., Neale, B., Todd-Brown, K., Thomas, L., Ferreira, M. A., Bender, D., ... & Sham, 758 P. C. (2007). PLINK: a tool set for whole-genome association and population-based 759 linkage analyses. The American journal of human genetics, 81(3), 559-575.

760 Rasmont, P., & Iserbyt, S. (2012, January). The Bumblebees Scarcity Syndrome: Are heat 761 waves leading to local extinctions of bumblebees (Hymenoptera: Apidae: Bombus)?. 762 In Annales de la Société entomologique de France (Vol. 48, No. 3-4, pp. 275-280). 763 Taylor & Francis Group.

764 Rasmont, P., Franzén, M., Lecocq, T., Harpke, A., Roberts, S. P., Biesmeijer, J. C., & 765 Gonseth, Y. (2015). Climatic risk and distribution atlas of European bumblebees (Vol. 766 10, pp. 1-236). Pensoft Publishers.

767 Reininghaus, H., Udy, K. L., Treanore, E., Tscharntke, T., & Scherber, C. (2017). Reversed 768 Importance of Local vs. Landscape Flower Resources for Bumblebee Foraging and 769 Colony Performance along Farmland-Urban Gradients. Scaling of Animal Communities: 770 From Local and Landscape to Global Processes, 38.

771 Riaño, D., Veloza, M., Cure, J. R., & Almanza, M. T. (2014). Desarrollo de dos colonias de 772 Bombus atratus (Hymenoptera: Apidae) mantenidas bajo dos modos de 773 alimentación. Revista Facultad de Ciencias Básicas, 10(2), 132-141.

774 Rodríguez, A., & Kouki, J. (2017). Disturbance‐mediated heterogeneity drives pollinator 775 diversity in boreal managed forest ecosystems. Ecological applications, 27(2), 589-602.

776 Russello, M. A., Waterhouse, M. D., Etter, P. D., & Johnson, E. A. (2015). From promise to 777 practice: pairing non-invasive sampling with genomics in conservation. PeerJ, 3, e1106. 778 Sadd, B. M., Barribeau, S. M., Bloch, G., De Graaf, D. C., Dearden, P., Elsik, C. G., & 779 Robertson, H. M. (2015). The genomes of two key bumblebee species with primitive 780 eusocial organization. Genome biology, 16(1), 76.

781 Samuelson, A. E., Gill, R. J., Brown, M. J., & Leadbeater, E. (2018). Lower bumblebee 782 colony reproductive success in agricultural compared with urban 783 environments. Proceedings of the Royal Society B: Biological Sciences, 285(1881), 784 20180807. 785 Sánchez-Bayo, F., & Wyckhuys, K. A. (2019). Worldwide decline of the entomofauna: A 786 review of its drivers. Biological conservation, 232, 8-27.

787 Schmack, J. M., Brenton-Rule, E. C., Veldtman, R., Wenseleers, T., Beggs, J. R., Lester, P. 788 J., & Bulgarella, M. (2019). Lack of genetic structuring, low effective population sizes 789 and major bottlenecks characterise common and German wasps in New 790 Zealand. Biological Invasions, 21(10), 3185-3201.

791 Schoville, S. D., Bonin, A., François, O., Lobreaux, S., Melodelima, C., & Manel, S. (2012). 792 Adaptive genetic variation on the landscape: methods and cases. Annual Review of 793 Ecology, Evolution, and Systematics, 43, 23-43.

794 Sgro, C. M., Lowe, A. J., & Hoffmann, A. A. (2011). Building evolutionary resilience for 795 conserving biodiversity under climate change. Evolutionary Applications, 4(2), 326-337.

796 Sirois-Delisle, C., & Kerr, J. T. (2018). Climate change-driven range losses among 797 bumblebee species are poised to accelerate. Scientific reports, 8(1), 14464. 798 Soulé, M. E. (Ed.). (1987). Viable populations for conservation. Cambridge university press.

799 Spevak, E., Jepson, S., & Williams, P. (2015). Bumblebee conservation worldwide within 800 the IUCN framework. Pollination Services to Agriculture: Sustaining and Enhancing a 801 Key Ecosystem Service, 180-199.

802 Stanke, M., & Morgenstern, B. (2005). AUGUSTUS: a web server for gene prediction in 803 eukaryotes that allows user-defined constraints. Nucleic acids research, 33(suppl_2), 804 W465-W467.

805 Tajima, F. (1989). The effect of change in population size on DNA 806 polymorphism. Genetics, 123(3), 597-601.

807 Theodorou, P., Radzevičiūtė, R., Kahnt, B., Soro, A., Grosse, I., & Paxton, R. J. (2018). 808 Genome-wide single nucleotide polymorphism scan suggests adaptation to 809 urbanization in an important pollinator, the red-tailed bumblebee (Bombus lapidarius 810 L.). Proceedings of the Royal Society B: Biological Sciences, 285(1877), 20172806.

811 Thompson, G. J., & Oldroyd, B. P. (2004). Evaluating alternative hypotheses for the origin 812 of eusociality in corbiculate bees. Molecular Phylogenetics and Evolution, 33(2), 452– 813 456.

814 Vickruck, J. L., & Richards, M. H. (2017). Nesting habits influence population genetic 815 structure of a bee living in anthropogenic disturbance. Molecular ecology, 26(10), 2674- 816 2686.

817 Vray, S., Rollin, O., Rasmont, P., Dufrêne, M., Michez, D., & Dendoncker, N. (2019). A 818 century of local changes in bumblebee communities and landscape composition in 819 Belgium. Journal of Insect Conservation, 23(3), 489-501.

820 Vuilleumier, B. S. (1971). Pleistocene changes in the fauna and flora of South 821 America. Science, 173(3999), 771-780. 822 Waples, R. K., Larson, W. A., & Waples, R. S. (2016). Estimating contemporary effective 823 population size in non-model species using linkage disequilibrium across thousands of 824 loci. Heredity, 117(4), 233.

825 Waples, R. S. (2016). Making sense of genetic estimates of effective population 826 size. Molecular ecology, 25(19), 4689-4691.

827 Wilson, M. C., Chen, X. Y., Corlett, R. T., Didham, R. K., Ding, P., Holt, R. D., & Laurance, 828 W. F. (2016). Habitat fragmentation and biodiversity conservation: key findings and 829 future challenges.

830 Woodard, S. H., Lozier, J. D., Goulson, D., Williams, P. H., Strange, J. P., & Jha, S. (2015). 831 Molecular tools and bumble bees: revealing hidden details of ecology and evolution in 832 a model system. Molecular Ecology, 24(12), 2916-2936.

833 Zayed, A. (2004). Effective population size in Hymenoptera with complementary sex 834 determination. Heredity, 93(6), 627.

835 Zayed, A. (2009). Bee genetics and conservation. Apidologie, 40(3), 237-262. 836 837 Supplementary Tables

838

Sup. Table 1

Name of the samples with location and coordinates

Sample_ID Location Lat Long Providence

V1__sorted VillaPinzon 5.2015545 -73.53361 Cun

V2__sorted VillaPinzon 5.2463423 -73.51799 Cun

V3__sorted VillaPinzon 5.2420756 -73.54682 Cun

V4__sorted VillaPinzon 5.2205323 -73.56484 Cun

V5__sorted VillaPinzon 5.2624143 -73.48536 Cun

V6__sorted VillaPinzon 5.199523 -73.48587 Cun

V7__sorted VillaPinzon 5.2297623 -73.50047 Cun

V8__sorted VillaPinzon 5.2319342 -73.47255 Cun

G8A2__sorted Guasca 4.8851234 -73.86743 Cun

G2__sorted Guasca 4.8319142 -73.91326 Cun

G3__sorted Guasca 4.8690243 -73.90228 Cun G4__sorted Guasca 4.8437143 -73.88992 Cun

G5__sorted Guasca 4.8435443 -73.86674 Cun

G6__sorted Guasca 4.8089243 -73.92064 Cun

G7__sorted Guasca 4.8148234 -73.89343 Cun

G8H2__sorted Guasca 4.81522 -73.8625 Cun

F1__sorted Facatativa 4.8794323 -74.42136 Cun

F2__sorted Facatativa 4.8498365 -74.41742 Cun

F3__sorted Facatativa 4.8325565 -74.41693 Cun

F4__sorted Facatativa 4.8496646 -74.39202 Cun

F5__sorted Facatativa 4.8212623 -74.39459 Cun

T1__sorted Tenjo 4.8203465 -74.12419 Cun

T2__sorted Tenjo 4.8203443 -74.1242 Cun

T3__sorted Tenjo 4.8394239 -74.1114 Cun

U1__sorted Choconta 5.158352 -73.78243 Cun

U2__sorted Choconta 5.1624732 -73.75596 Cun

U3__sorted Choconta 5.1844645 -73.76869 Cun

U4__sorted Choconta 5.1967823 -73.73877 Cun

U5__sorted Choconta 5.180145 -73.73659 Cun

U6__sorted Choconta 5.1604623 -73.72371 Cun

U7__sorted Choconta 5.1717452 -73.69587 Cun

U8__sorted Choconta 5.1907323 -73.69194 Cun

B1__sorted Morro 5.6674935 -73.10643 Boy

B2__sorted Morro 5.6520353 -73.13149 Boy

B3__sorted Morro 5.639735 -73.11313 Boy

B4__sorted Morro 5.6347753 -73.1539 Boy

T4__sorted Tenjo 4.7932734 -74.14437 Cun

T5__sorted Tenjo 4.7999467 -74.16135 Cun

D1__sorted Duitama 5.9191164 -72.97925 Boy

D2__sorted Duitama 5.8938466 -72.95865 Boy D3__sorted Duitama 5.8692535 -72.93668 Boy

D4__sorted Duitama 5.883634 -73.01293 Boy

D5__sorted Duitama 5.9423334 -72.87282 Boy

D6__sorted Duitama 5.9771635 -72.96483 Boy

D7__sorted Duitama 5.9464335 -72.92432 Boy

D8__sorted Duitama 5.8337375 -72.97445 Boy

P1__sorted Pacho 5.1500136 -74.11499 Cun

P2__sorted Pacho 5.1876235 -74.11533 Cun

P3__sorted Pacho 5.1807934 -74.08066 Cun

P4__sorted Pacho 5.2081435 -74.13078 Cun

P5__sorted Pacho 5.2153234 -74.09928 Cun

P6__sorted Pacho 5.2495134 -74.09714 Cun

P7__sorted Pacho 5.2546434 -74.04222 Cun

P8__sorted Pacho 5.2163435 -74.06212 Cun

M1__sorted Arcabuco 5.7484775 -73.51881 Boy

M2__sorted Arcabuco 5.7758335 -73.42935 Boy

M3__sorted Arcabuco 5.7239135 -73.31742 Boy

M4__sorted Arcabuco 5.6848835 -73.36137 Boy

M5__sorted Arcabuco 5.707513 -73.39227 Boy

M6__sorted Arcabuco 5.6685753 -73.41561 Boy

M7__sorted Arcabuco 5.7478235 -73.47124 Boy

M8__sorted Arcabuco 5.7143534 -73.48084 Boy

S1__sorted Sotaquira 5.5225853 -73.41875 Boy

S2__sorted Sotaquira 5.4702935 -73.39025 Boy

S3__sorted Sotaquira 5.4822653 -73.43866 Boy

S4__sorted Sotaquira 5.511353 -73.48088 Boy

S5__sorted Sotaquira 5.5147653 -73.51556 Boy

S6__sorted Sotaquira 5.436853 -73.45961 Boy

S7__sorted Sotaquira 5.4702935 -73.49702 Boy S8__sorted Sotaquira 5.4593635 -73.55264 Boy 839

Sup. Table 2

Samples with values of inbreeding and observed heterosygocity

Sample_ID F Ho

V1__sorted -0.13021 0.12631182

V2__sorted -0.05293 0.11765074

V3__sorted -0.08227 0.12096623

V4__sorted -0.13071 0.12634358

V5__sorted 0.18086 0.09152921

V6__sorted 0.08051 0.10275161

V7__sorted 0.05579 0.1054889

V8__sorted 0.0127 0.11028084

G8A2__sorted 0.00367 0.11132497

G2__sorted -0.08017 0.12075992

G3__sorted -0.17993 0.13182332

G4__sorted 0.11921 0.09841762

G5__sorted 0.15696 0.09419931

G6__sorted 0.12137 0.09817663

G7__sorted 0.08946 0.1017498

G8H2__sorted -0.0172 0.11367309

F1__sorted -0.05936 0.11840952

F2__sorted -0.01129 0.11300879

F3__sorted -0.07743 0.12042444

F4__sorted -0.02139 0.1141328 F5__sorted -0.12735 0.12592872

T1__sorted -0.07428 0.12004966

T2__sorted -0.09409 0.12229093

T3__sorted -0.15067 0.12856519

U1__sorted -0.12901 0.12617173

U2__sorted -0.22839 0.13734755

U3__sorted -0.17049 0.13082291

U4__sorted -0.36477 0.15260807

U5__sorted -0.09278 0.12210974

U6__sorted -0.03872 0.11609253

U7__sorted 0.06197 0.10484773

U8__sorted 0.19372 0.09009121

B1__sorted 0.00658 0.11106272

B2__sorted 0.0712 0.103797

B3__sorted -0.13722 0.12703868

B4__sorted -0.15619 0.12917847

T4__sorted -0.18753 0.13274392

T5__sorted 0.00563 0.11111111

D1__sorted -0.13773 0.12713461

D2__sorted -0.32959 0.14865126

D3__sorted -0.02175 0.11416122

D4__sorted 0.11437 0.09895833

D5__sorted -0.08957 0.12175436

D6__sorted -0.10256 0.12323484

D7__sorted -0.20186 0.13440213

D8__sorted 0.04275 0.10696317

P1__sorted -0.12951 0.12627922 P2__sorted -0.29694 0.14496036

P3__sorted -0.26232 0.14110199

P4__sorted -0.31735 0.14717819

P5__sorted -0.23993 0.13864587

P6__sorted -0.22833 0.13724008

P7__sorted 0.05245 0.10587139

P8__sorted 0.11191 0.09923096

M1__sorted -0.09588 0.12248227

M2__sorted 0.06562 0.10441094

M3__sorted -0.0362 0.11585939

M4__sorted -0.01343 0.1132476

M5__sorted -0.1065 0.12367314

M6__sorted -0.15311 0.12885136

M7__sorted -0.02044 0.11403618

M8__sorted 0.02669 0.10873218

S1__sorted -0.08246 0.12098038

S2__sorted -0.04968 0.11737795

S3__sorted -0.17691 0.13154253

S4__sorted -0.0678 0.11932563

S5__sorted -0.00685 0.11251091

S6__sorted 0.13054 0.09716096

S7__sorted -0.00153 0.1119403

S8__sorted -0.008 0.11262392

840

Sup Table 3. Mantel test values of

correlation

Significance Predictor and Formula (Model) R2 P value Heterozygosity ~ Geographic 0.878 -0.045 distance

Inbreeding ~ Geographic distance 0.862 -0.047

Land Use~ Geographic distance 0.381 0.010

841 842 Supplementary Figures A B

843 844 Supplementary Figure 1. A. Cross-entropy plot for the number of cluster K = 1-10. The 845 retained value of K is K = 2. B. Scatterplot clearly showing one single consistent 846 cloud of points between genetic and geographic distances 847

848

849

850

nbreeding

I I 851

852

853 Mean land Use Geographic distance 854

855

856

857

Heterozygosity Heterozygosity 858 859 Mean land Use Geographic distance 860 Supplementary Figure 2. Added variable plots showing the correlation between the 861 variables (He and F) and geographic distance.

862

863