Abstract
Despite their paramount importance in molecular ecology and conservation, genetic diversity and structure remain challenging to quantify with traditional genotyping methods. Next-generation sequencing holds great promises, but this has not been properly tested in highly mobile species. In this article, we compared microsatellite and RAD-sequencing (RAD-seq) analyses to investigate population structure in the declining bent-winged bat (Miniopterus schreibersii) across Europe. Both markers retrieved general patterns of weak range-wide differentiation, little sex-biased dispersal, and strong isolation by distance that associated with significant genetic structure between the three Mediterranean Peninsulas, which could have acted as glacial refugia. Microsatellites proved uninformative in individual-based analyses, but the resolution offered by genomic SNPs illuminated on regional substructures within several countries, with colonies sharing migrators of distinct ancestry without admixture. This finding is consistent with a marked philopatry and spatial partitioning between mating and rearing grounds in the species, which was suspected from marked-recaptured data. Our study advocates that genomic data are necessary to properly unveil the genetic footprints left by biogeographic processes and social organization in long-distant flyers, which are otherwise rapidly blurred by their high levels of gene flow.
Similar content being viewed by others
Introduction
A comprehensive picture of the genetic structure of populations is instrumental to any ecological, evolutionary and conservation research1. Patterns of genetic variation, differentiation and admixture inform on the history of populations and thus help understanding the ecological and geographic processes shaping biodiversity in space and time2,3,4. Molecular assessments also offer a swift alternative to traditional monitoring techniques (e.g., mark-recapture), especially when it comes to measuring individual movements, population fragmentation and other key parameters relevant for conservation such as inbreeding and local adaptation5,6,7. As such, genetic surveys have become an essential part of the management of threatened species8,9, by assisting the design of conservation strategies to enhance population connectivity10, restore bottlenecked populations11, and plan translocation/reintroduction efforts in respect to evolutionary units12.
In many instances, however, inferring genetic structure is far from trivial. Patterns of population differentiation may be difficult to detect for mobile organisms, as populations are homogenized by recurrent episodes of gene flow13,14. The issue is also characteristic of genetically impoverished species that experienced range-wide demographic changes, such as refugial bottlenecks and post-glacial expansions, and for which the remaining diversity might be too low to quantify population connectivity4. Moreover, at times in the life cycle, momentary gatherings of individuals from various geographic origins might also blur the appreciations of genetic diversity and between-population differentiation, as seasonal migration confound with dispersal15. In such cases, a high genetic resolution can offer insights into individual ancestries that distinguish seasonal migrators vs dispersers.
In theory, the accuracy of population genetic analyses should depend on the number of loci used in respect to their standing variation. Microsatellites combine the advantages of evolving neutrally16,17 and experiencing high mutation rates that generate myriads of alleles. They accordingly remain popular markers to estimate gene flow and diversity among recently diverged populations ever since the late 1990s18,19. Also highly polymorphic, mitochondrial DNA (mtDNA) is more straightforward and cheaper to analyze, notably across evolutionary distant populations, and has accordingly been preferred by researchers interested in range-wide phylogeographic structure20. However, mtDNA faces inherent limitations with respect to its clonal nature and maternal mode of inheritance, which makes it more sensitive to selection21 and can bias the reconstruction of the demographic history of populations, especially when dispersal is sex-biased22,23,24. In recent years, molecular ecologists benefit from high-throughput sequencing technology to access unprecedented numbers of loci, as prepared by genomic libraries25,26. In particular, restriction-site associated DNA sequencing (RAD-seq27; double-digest ddRAD-seq28) have become a popular method to genotype hundreds to thousands single nucleotide polymorphisms (SNPs) across conspecific or interspecific populations for multiple applications, such as high-resolution phylogeography29 and population genetic studies of weakly structured species4,28. The number of alleles of a single microsatellite locus is much higher than that of SNPs, but microsatellite analyses rarely include more than 10–20 loci due to practical constraints. By generating several hundred times more loci, RAD-seq thus largely compensates for the lower per-locus diversity of SNPs to provide more powerful population genetic inferences, notably in situations when gene exchange is hard to apprehend. However, such improvement has not been satisfactorily quantified, since the few comparative studies have focused on species with relatively strong levels of population differentiation, where structure is already detected efficiently by microsatellites (reviewed by30).
In this study, we assessed how much RAD-seq can clarify subtle patterns of population structure compared to microsatellites in one of Earth’s widespread mammals, the bent-winged bat Miniopterus schreibersii. This species was once considered to extend from Western Europe to Australasia, but taxonomic changes have now restricted its range to the Western Palearctic, where it is mostly found around the Mediterranean Basin (Southern Europe, Asia Minor and North-Africa). Bats in general are models of choice to gauge the resolution of genetic markers. Their high mobility usually combined with female philopatry and sex-biased dispersal31 implies pervasive gene flow and cyto-nuclear discordances32, thus inducing substantial difficulties to retrieve clear signals of population structure from traditionally used loci. In this respect, M. schreibersii is a long-distance traveler33 that conducts fixed seasonal migrations between hibernacula, maternity roosts and mating sites by movements of 40–100 km on average, but up to several hundred kilometers34,35. Nevertheless, young or adult dispersal overall remain rare in this species, so regional rather than long-distance movements have been argued to contribute most of the genetic exchanges36. As a consequence, the composition of nursery and hibernation sites were shown to differ37,38, so individuals of independent origins could temporarily share the same sites.
Despite substantial sampling and genotyping efforts, previous studies reported little genetic structure across the vast range of the species. Mitochondrial analyses found a single mitochondrial clade, with closely related haplotypes shared all over the Mediterranean Basin39, and microsatellite analyses did not detect meaningful geographic patterns of nuclear differentiation40. Both results were interpreted as the consequence of the biogeographic history of the species, namely the massive recolonization of Europe after the last ice age from a single refugium, presumably in Anatolia39,40. While this hypothesis has merit, such strong genetic homogeneity at the continental scale may equally reflect the lack of informativeness of conventional genetic markers to detect shallow levels of population differentiation, for instance, if refugial structure had been compromised by high levels of post-glacial gene flow in this mobile species. Likewise, these may overlook subtle genetic signals caused by the social organization of the species. Precise knowledge of the genetic diversity and genetic structure would also be relevant to guide conservation planning. Despite its wide distribution, M. schreibersii is facing dramatic declines in Europe since the mid-twentieth century, even being extirpated from several countries in recent years, to the point that it is presently considered Vulnerable by the IUCN Red List (VU A2c)41.
How biogeographic history, social organization and recent declines have interacted with philopatry, regional dispersal and long-distance migration to shape the genetic structure of M. schreibersii thus remains an open question, especially given the (lack of) geographic patterns retrieved from previous studies. Genomic analyses would thus be timely, and in turn offer an opportunity to assess the molecular resolution required to detect genetic structure when there is little. To this end, here we analyzed regional sets of population and individual samples across the western and Central European ranges of M. schreibersii with microsatellite loci and SNPs obtained from ddRAD-seq. The outcomes of both approaches were directly compared by quantitative (fixation indices, heterozygosity, pairwise differentiation) and qualitative (clustering assignment) measures. While RAD-seq is generally expected to provide more accurate estimates, it should be particularly more informative than microsatellites to detect phylogeographic and regional structure, even if populations frequently exchange genes and migratory individuals.
Methods
DNA sampling and extraction
A total of 196 adult individuals of M. schreibersii were sampled from 19 localities covering seven European regions during spring and summer 2010, 2011 and 2012 (Table 1). Bats were caught by mist nets or harp traps at the entrance of roosting caves, either upon emergence or by entering the roost during the day. For each individual, blood was sampled by puncturing the uropatagial vein with a 0.5 mm gauge needle (Neolus ®). 10–30 μl volumes were collected with heparinized capillaries (Assistent ®), smeared on filter paper, and dried in an envelope for field storage at room temperature. A homeostatic cotton was applied on the wound until the bleeding has stopped. In addition, a 1 mm Ø biopsy punch was applied on the wing membrane (Stiefel ®) and the tissue was stored in 96% ethanol. Bats were released immediately at the site of capture. All experimental protocols (animal capture and tissue sampling) were approved by relevant committees, namely the Service Vétérinaire du Canton de Vaud for Switzerland (2322), the Ministry of Environment of the Slovak Republic for Slovakia (2598/715/03-5.1 pil), the Departament d’Agricultura, Ramaderia, Pesca, Alimentació i Medi Natural de la Generalitat de Catalunya for Spain (SF/379), the Direction for Nature Protection of the Ministry of Culture for Croatia (URBROJ 532-08-01-04/3-10-02), the Minister dell’Ambiante for Italy (7588), the DREAL of (former) region Aquitaine for France, and the Instituto para a Conservação da Natureza e Biodiversidade for Portugal. All methods were carried out in accordance with relevant guidelines and regulations.
DNA was extracted using the DNeasy Blood & Tissue Kit (Qiagen ®) following the manufacturer’s instructions except for the following particularities. Blood-soaked paper fragments (~ 10 mm in diameter) were cut in small pieces with sterile scissors, and incubated in 360 µL of ATL buffer at 90 °C for 15 min. After cooling, 40 µL of Proteinase K was added to each sample and left for overnight digestion at 56 °C. Elution was performed in 55 µL volumes and by incubating 15 min at 37 °C prior to centrifugation. This step was repeated a second time to maximize DNA yields. For a subset of individuals, additional DNA was extracted from the wing biopsy after rinsing the ethanol-preserved tissues by pure MilliQ water soaking (Millipore ®).
Microsatellites genotyping
Individuals were genotyped at 11 polymorphic microsatellite loci amplified in four multiplex PCRs42,43. Loci, primers, and amplification protocols are detailed in44. Amplicons were quality-checked on an agarose gel, typed on an ABI Prism 3100 genetic analyser (Applied Biosystems ®) and size-scored with GeneMapper (Applied Biosystems). Finally, we checked for large allelic drop out, stuttering and null alleles using Micro-Checker v2.2.345.
ddRAD-seq
For genomic library preparation, extracted samples were first concentrated using a chloropane precipitation. We then prepared a ddRAD library modified from46, as detailed in Supplementary File S1. Three separate libraries were multiplexed and sequenced on two lanes of an Illumina HiSeq2000 (Lausanne Genomic Technologies Facility, Switzerland).
Sequences with low signal-to-noise ratio (< 0.6) were discarded using Illumina’s Chastity filter. In the absence of a reference genome, demultiplexing, stacking and cataloguing of reads was performed with the de-novo pipeline of STACKS 0.99647. Low quality (Phred score below 10 averaged on 15 bp sliding windows) and monomorphic sequences were filtered out, with a minimum coverage of three and stacking up to 3% of divergence between reads.
Custom R scripts were developed for additional filtering and SNP calling. First, loci with more than five SNPs were discarded in an effort to exclude repetitive elements, and a threshold of eight reads per genotype was applied. Second, tri-allelic genotypes, heterozygotes with a rare allele represented by less than 25% of the reads, and singleton were not considered. Third, SNPs were called as genotyped if sequenced in at least 70% of individuals. Finally, to exclude obvious outliers, SNPs with a FIS48 outside the range of − 0.2 and 0.2 were removed with hierfstat49, since extreme heterozygote deficits are not expected in a diploid species with separate sexes.
Population-level analyses
Genetic structure and diversity from the microsatellite and RAD-seq datasets were investigated at the population level using hierfstat49 from standard population genetics parameters50. First, we computed observed heterozygosity (HO), expected heterozygosity (HS, also known as gene diversity), overall heterozygosity over populations (hT), allelic richness (aR), population specific FST (Beta51) and departure from Hardy–Weinberg (FIS). 95% Confidence Intervals (CI) for heterozygosities and F-statistics48 were estimated by bootstrapping 1000 times the variance components. Pearson correlations were used to compare estimates between marker sets, and to assess the influence of missing data on the RAD-seq estimates. To identify any outlier locus, we further investigated the relation between allele frequencies and observed heterozygosity for both marker sets by plotting HS and their 95% confidence intervals given allele frequency and global heterozygote deficiency, computed as HS = 2p(1 − p)(1 − fIT) ± 2[(2p(1 − p)(1 + fIT)/N]0.5.
Second, pairwise FST between populations48 were estimated and compared between marker sets by Mantel tests52, then used to test for isolation-by-distance (IBD53). Chord distances54 were computed (genet.dist function of hierfstat), translated in a neighbour-joining tree (bionj function of ape55). Furthermore, hierarchical analyses of genetic variance were carried out to test for the effects of countries and populations. To this aim, FST was partitioned between populations, within country (FSC), and between countries (FCT).
Third, we tested for sex-biased dispersal using the mAIc index implemented in hierfstat (sexbias.test function56), which was compared between marker sets by a Pearson correlation.
Individual-level analyses
Genetic structure among individual genotypes was first examined by Principal Component Analyses (PCA) on each marker set, using the indpca function of hierfstat50.
Second, we applied the snmf clustering algorithm implemented in LEA57 to estimate in how many clusters the dataset can be summarized, and compute individual ancestries of each individual among these clusters. For each marker set, chains were run for K = 1–7 clusters, with 20 repetitions for each K, and blinding 10% of the data for the cross-entropy estimation (function snmf). Most likely Ks are reflected by low cross-entropy values. The alpha parameter was kept to the default 100 (as recommended by the authors), especially as it did not affect the results in preliminary runs. To appreciate the influence of missing data on the RAD-seq dataset, runs were also performed on a reduced dataset based only on individuals featuring above 70% of SNP matrix completeness (n = 172). Finally, we further explored how the number of loci influenced the retrieved patterns of genetic structure by repeating the admixture analysis on randomly selected subsets of 3,000, 2,000, 1,000, 500, 200 and 100 SNPs.
Results
In this study, we obtained and compare genotypes from 11 microsatellite loci and 4,994 SNPs obtained by RAD-seq for the same 196 individuals sampled over 19 populations of M. schreibersii across seven European countries (Table 1).
Population level analyses
Population-based statistics (summarized in Table 1) showed mostly consistent patterns of diversity and structure. Microsatellite loci were more polymorphic than SNPs obtained by RAD-seq. Microsatellites bear between 3 and 18 alleles, a mean observed heterozygosity HO of 0.537 (0.383–0.671), a mean expected heterozygosity HS of 0.557 (0.395–0.694) and an overall heterozygosity HT of 0.588 (0.418–0.734). These values are much larger than the diversity measured at bi-allelic SNPs, with a mean HO of 0.122 (0.119–0.126), mean HS of 0.129 (0.125–0.133) and mean HT of 0.139 (0.134–0.143). Departure from Hardy Weinberg equilibrium overlapped between marker sets (SNPs FIS = 0.053 [0.045–0.057]; microsatellite FIS = 0.035 [0.011–0.058]). Most SNPs and microsatellite alleles show good concordance between observed and expected heterozygosities (Fig. 1). For the RAD dataset, HO and FIS were sensitive to the completeness of the SNP genotype matrix, as reflected by lower heterozygosity and higher FIS in populations with higher proportions of missing data (Supplementary File S2). Nevertheless, the effect was mainly influenced by a single population that featured nearly 50% of missing data (loc. 9, France), and was not significant for HS (Supplementary File S2).
Microsatellite and RAD-seq estimates were significantly correlated for HS (r = 0.62; P = 0.004) and population specific FST (r = 0.63; P = 0.004) (Fig. 2). The correlation was also large for ar (r = 0.68; P = 0.001), as expected since ar co-varies with hs. In contrast, the relationships were not significant for fis and ho (Fig. 2).
Pairwise population differentiation (fst) was stronger when estimated from SNPs (mean FST = 0.068 [0.065–0.071]) than from microsatellites (mean FST = 0.053 [0.041–0.066]). There was a significant correlation between microsatellite- and SNPs-based FST estimates (Mantel test, Fig. 3), but a significant pattern of isolation-by-distance was recovered only from the SNP dataset (Mantel tests, Fig. 3). Furthermore, both sets of fst produced similar trees based on Chord distances, where populations are mostly grouped by countries (Fig. 4).
Accordingly, analyses of genetic variance revealed stronger differentiation between countries (fct) than between sub-populations within countries (fsc), especially when estimated from SNPs: for microsatellites, fsc was 0.023 (0.013–0.034) and fct was 0.035 (0.026–0.043), whereas for SNPs, fsc was 0.019 (0.017–0.020) and fct was 0.058 (0.056–0.061).
Both datasets showed no evidence for sex-biased dispersal (microsatellites: t = − 0.91, P = 0.37; SNPs: t = − 0.62, P = 0.53). The corrected mAIc for microsatellites and SNPs are strongly correlated (r = 0.49, P < 0.001).
Individual-level analyses
In contrast to the mostly congruent population-level analyses, the SNPs dataset provided a much higher resolution than microsatellites in individual-based analyses.
The microsatellite PCA showed no clear patterns of structure, as individuals of various origins overlap on the first axes. In contrast, axes 1 and 2 of the PCA built from SNP genotypes sorted most individuals by their country of origin (Fig. 4).
Accordingly, clustering analyses by the snmf algorithm showed radically different patterns of genetic structure (Fig. 5). With microsatellites, cross-entropy hardly highlighted a particular number of clusters. With SNPs, it clearly shows that at least four clusters are needed to explain the structure present in the dataset. Like the PCAs, the microsatellite ancestry coefficients did not convey a meaningful geographic pattern while individuals were assigned according to their populations of origin and country based on SNPs (Fig. 5).
Admixture analyses conducted on subsets of SNPs recovered the “true” signal of genetic structure (as obtained from the full dataset), depending on the number of retained markers and the complexity of the clustering scheme (number of K groups) (Fig. 6). With large datasets (≥ 500–1000 SNPs), broadly similar results are obtained, even for estimating ancestries among many groups. The picture becomes blurry with only 200 SNPs (especially for higher values of K), and cross-entropy suggests that two groups mostly exist in the dataset. Finally, the analysis with only 100 SNPs was essentially uninformative.
Geographic patterns
Despite the overall low genetic differentiation (see above), the SNPs dataset recovered clear patterns of genetic structure across the sampled range of M. schreibersii (Figs. 4, 5, 6, 7). Specifically, the four main clusters identified by the preferred snmf analysis roughly distinguished bats from Portugal (loc. 1–4; blue in Fig. 7); northwestern Spain (loc. 5–8; green in Fig. 7); Italy (loc. 11–13; orange in Fig. 7); and Croatia (loc. 14–17; red in Fig. 7). Individuals from the other populations showed ancestries from several clusters (Fig. 7), and were accordingly placed at intermediate positions on the PCA’s first two axes (Fig. 4): those from France (loc. 9) and Switzerland (loc. 10) are mostly “Spanish”, with substantial contributions from all other clusters; individuals from one Italian population (loc. 12) showed equal contributions from the “Italian” and the “Croatian” clusters; individuals from the two Slovakian populations are mostly “Croatian”, with substantial contributions from the “Italian” and “Spanish” cluster. Separate analyses limited to 172 individuals with less than 30% of missing data retrieved nearly identical results (Supplementary File S3).
Detailed inspections of the admixture runs with up to seven groups revealed remarkable substructure among the main Portuguese, Italian and Croatian clusters that allow tracking individual movements at the regional level. In Portugal, bats caught at loc. 3 and 4 differ genetically (blue and yellow clusters in Fig. 7), and pure individuals of both sub-clusters were found in syntopy at loc. 1 and 2. In Italy, loc. 12 individuals were identified as a different sub-cluster (pink in Fig. 7), except for two that seemingly originated from the main Italian group found in loc. 11 and 13 (orange in Fig. 7). In Croatia, loc. 17 represents a distinct population (purple in Fig. 7), but loc. 14–16 also host some of its individuals. These patterns are confirmed by inspection of additional PCA axes (Supplementary File S4).
Genetic diversity (HS, which was not influenced by missing data), varied greatly among populations, being generally higher in Portugal, northwestern Spain and Croatia (Fig. 7).
Discussion
Flying vertebrates are expected to feature shallow genetic differentiation between populations due to their high movement capabilities58,59, but at the same time, complex patterns of regional substructure may arise typically because of philopatric and migratory behaviors31, combined with the biogeographic history of populations. With respect to these aspects, here we aimed to estimate genetic structure in the mobile, migratory and putatively philopatric bat M. schreibersii across Europe, comparing approaches based on few highly polymorphic loci (microsatellites) vs many bi-allelic loci (SNPs obtained by RAD-seq).
Little yet genuine phylogeographic structure in M. schreibersii
Both sets of markers recovered low fST estimates across the European range of M. schreibersii, namely 0.05 (microsatellites) or 0.07 (SNPs) on average. This is in accordance with previous microsatellite genotyping that found little genetic differentiation across the range of the species40, as well as the fact that a single mitochondrial lineage inhabits Europe39. As these studies concluded, the lack of phylogenetic divergence between populations separated by thousands of kilometers, and the marked isolation-by-distance, suggest high connectivity.
Unlike previous studies, however, we did recover genetic structure across Europe with genome-wide SNPs. We also found similar levels of diversity between populations, especially between the eastern- and westernmost ones (Portugal and Croatia). Both results are somewhat surprising if one assumes that the species post-glacially expanded across Europe from a single Anatolian refugium—the currently accepted biogeographic scenario, based on higher mitochondrial and microsatellite diversity in Anatolia39,40. Under this scenario, a single genetic cluster and a westward longitudinal gradient of genetic variation would have been expected across the whole continent (i.e., lower diversity in Iberia compared to the Balkans). Whether the glacial ranges of M. schreibersii also encompassed Mediterranean Europe, in addition to Anatolia, thus remains an open question. In fact, the structure we identified in Europe would even be consistent with distinct historical refugia across the Iberian, Apennine and Balkan Peninsula, perhaps subdivided into micro-refugia (e.g., Iberia). In this respect, the puzzling ancestry of Slovakian bats could reflect recent admixture between the Croatian group and a gene pool that we have not sampled i.e., from another separate refugium, for instance, Anatolia. The last lines on the biogeography of M. schreibersii are thus yet to be written. A genomic survey encompassing both European and extra-European populations to thoroughly infer range-wide patterns of genetic diversity, structure, and admixture, perhaps combined with environmental niche reconstruction to track habitats suitable for the species during the last glacial maximum, will hold key to disentangle among the one refugium vs several refugia hypotheses.
Suitable glacial habitats were modelled in southern Europe for other species of bats with roughly similar distributions, e.g., Nyctalus leisleri60 or Plecotus austriacus61. Like M. schreibersii, European bats in general show little phylogeographic structure60,62,63, or sometimes recent geographic isolation resulting from the last glacial stage, as in Rhinolophus ferrumequinum64 and Rhinolophus hipposideros65. Yet, the lack of clear genetic structure between candidate refugial regions does not necessarily imply a single refugium scenario, as here previously assumed for M. schreibersii39,40, as well as in other species, e.g., Myotis bechsteinii66. In mobile organisms, genetic signals for multiple refugia (e.g., several glacial lineages) can quickly disappear in the event of gene flow and lineage fusion67, and become undetectable with few bi-parentally inherited microsatellite loci. Long-lasting signals may persist in the mitochondrion, but only if females are more philopatric than males. In a similar fashion, multiple refugia exist in the European Barn Owl (Tyto alba)—a fast-dispersing bird with a similar circum-Mediterranean distribution—but it was likewise only retrieved from genomic data, namely whole-genome sequencing59, and not microsatellites68. Classic biogeographic paradigms characteristic of terrestrial vertebrates, such as the long-term persistence in separate Mediterranean refugia69,70 and their regional differentiation in refugia-within-refugia71 might thus apply to bats and birds even if microsatellite studies have so far suggested otherwise. That said, some widespread Holarctic bat species may truly be genetically homogenous. For instance, the north-American Myotis lucifugus lacks population differentiation, as measured from both conventional (microsatellites + mtDNA72) and genomic markers (RAD-seq73; low coverage genome74).
Regional genetic structure by social organization
Despite the relatively weak range-wide differentiation, the high-resolution offered by our RAD-seq dataset recovered regional signals of population sub-structuring, which can be interpreted in the light of the social habits of M. schreibersii. In bats, mating and breeding colonies are often spatially separated37,75, and their average genetic diversity and relatedness vary accordingly66. In addition, females are usually the philopatric sex, and males the dispersing sex, which should affect mitochondrial vs nuclear patterns of genetic structure31.
In M. schreibersii, ringing data have suggested that both males and females were philopatric36, and we also recovered no evidence for sex-biased dispersal. Male-biased dispersal was however detected by a regional population genetic study focusing on the Portuguese colonies38. The discrepancy probably lies in the larger geographic scale of our study compared to the latter: sex-biased dispersal can only be retrieved if the dataset contains a fair amount of dispersers56, which are obviously in lower proportions at the continental compared to the regional scale, especially if dispersal remains rare overall.
Furthermore, we detected syntopic individuals with unrelated ancestries in several nursing sites of Portugal (loc. 1–2), Italy (loc. 12), and Croatia (loc. 15–16), yet with little signs of local admixture—which would have been flagged by intermediate ancestry coefficients. These sites thus act as “regional hubs” where individuals of various origins temporarily meet yet remain genetically differentiated. Such observations are consistent with the idea that M. schreibersii is a keen seasonal migrator (moving a lot between colonies), but a weak disperser (exchanging little genes between them)36. Hence, the social organization of the species momentarily shapes the genetic diversity of its populations, which may differ between nursing, mating and hibernating sites. The factors triggering genome-wide divergence between co-occurring nursing individuals remain elusive, and may include isolation-by-distance, assortative mating, or a meta-population structure—which has been documented in bats76,77. More insights will come from assessing fine-scale geographic patterns of genetic diversity from the mating sites i.e., the “true” origins of individuals, and test whether these are genetically isolated.
Inferring genetic structure when there is little
In recent years, several studies have compared traditional genetics vs genomics methods to infer population diversity and structure, especially in the context of conservation78,79,80,81. In their synthesis, Sunde et al.30 concluded that RAD-seq performed as well or better than microsatellites, as similar trends of clustering and heterozygosity were always recovered from both types of markers (see their Table 1). However, most previous work was biased towards aquatic organisms (especially fish), with relatively pronounced genetic structures reflected by higher pairwise fST e.g., > 0.1 in30,81 and 0.4 on average in78.
Unlike previous comparative studies, here the generally low genetic differentiation among European populations of M. schreibersii offers a framework to evaluate the performance of genotyping methods to recover shallow patterns of genetic structure. In this respect, microsatellites were out-performed by SNPs obtained from RAD-seq in most analyses. While the difference in population-based F-statistics was not so obvious (Figs. 1, 2, 3, 4), individual-based analyses (PCA, snmf clustering) were much clearer when inferred from SNPs (Figs. 4, 5). This result is somewhat expected given the weak divergence among our populations, which may have maintained high effective sizes in large (although potentially composite) refugia (see above). In such cases, drift does not counteract the effect of mutation at fast-evolving microsatellites, which results in high yet shared diversity (homoplasy) between regions, hence uninformative of population structure. By capturing less diversity at multiple, putatively more conserved loci, SNPs clearly provided a more reliable picture of the entire genome: not only could they recover the subtle patterns of range-wide phylogeographic differentiation, but they also informed on local population structure. Moreover, because estimates of allele frequencies based on a few individuals will have higher variance for multi-allelic compared to bi-allelic markers, genomic SNPs may be particularly more informative than microsatellites when samples sizes are modest. Microsatellite analyses may require up to 25–30 individuals per population for accurate population genetic inferences82, which is often unrealistic due to biological and field constraints (e.g., 5–15 samples per population here). Therefore, our study extends the conclusions of Sunde et al.30 that SNPs provide a significant upgrade to microsatellite-based population genetics, even more so in mobile organisms with shallow genetic structure, as well as for tracking individual movements.
Going beyond the molecular approach itself, can we estimate what numbers of genomic loci are needed to reliably infer population differentiation in weakly structured organisms? In a study on red mangroves, Hodel et al.79 reported gradual improvements with different filtering stringencies that increased their datasets from ~ 200 to ~ 25′000 SNPs (but also increased missing data from 11 to 78%). Our sub-setting analyses largely agrees with their conclusions. While our 100–200 SNPs datasets were mostly uninformative, 500 SNPs (10 times less than our initial dataset) still recovered most of the shallow structure of M. schreibersii, and at least 1,000 SNPs were required to investigate the regional substructure (Fig. 6). Hence, depending on the biological question and organisms, meaningful patterns might be recovered from lesser sequencing efforts (i.e., retrieving fewer SNPs) or genetically impoverished species (i.e., containing fewer SNPs to begin with). Special precautions should nevertheless be taken to acknowledge missing data, especially when inferring population genetic estimates based on heterozygosity. For a given coverage threshold, heterozygous sites are typically discarded more often than homozygous sites by bioinformatics filters, because fewer reads per allele are sequenced. In turn, heterozygous SNPs are more difficult to retain than homozygote SNPs in low-quality samples, which contributes to artificially deplete HO and inflate FIS. While this issue can be circumvented by exploring the filtering parameters, an alternative is to consider genetic diversity based on allelic richness and HE, which both have the advantage to rely on the number of alleles per population rather than the number of heterozygotes.
Conservation implications
Our study provides applied lessons for the conservation of M. schreibersii, which has sustained massive die-offs in Spain, France and Portugal in 2002, associated to the emergence of a new filovirus, the Lloviu virus (LLOV)83,84. As previously recommended36, the genetic differentiation documented at the regional scale calls for protecting all the sites used by the species, both to ensure regional connectivity between the hubs visited by seasonal migrants and to preserve the genetic uniqueness of the different demes, as it might convey components of local adaptation. Moreover, some of the most isolated sites in Portugal (loc. 4) and France (loc. 9) show the lowest genetic variation, which may reflect genetic bottlenecks ensuing the population collapses from the 2000s83. These populations should thus remain under constant care, especially if their eroded genetic diversity negatively affect fitness85. Besides, recovery programs have proven very efficient in M. schreibersii86. Finally, the genetic groups and individual movements identified inform on the risk assessment of future disease outbreaks, in the context of the possible re-emergence of the Lloviu virus, which has been re-detected since 2016 in an Hungarian colony87 and was shown potential to infect human cells 88. The spreading may be fast within regions—where bat populations are either panmictic, or in close contact during nursing—but could be slower at the continental scale given the structuration between European countries.
Weak genetic differentiation at the global scale despite rare dispersal and marked structure at the local scale may appear counter-intuitive, yet such pattern is not unheard of among geographically-scattered species. An analogous situation was documented in the common hippopotamus, which expanded across Africa in a single wave (homogenizing its rich genetic diversity), before contemporary habitat fragmentation created barriers to dispersal and drove genetic bottlenecks89. In a similar fashion, the ongoing fragmentation of M. schreibersii’s range could have contributed to promote the genetic differentiation between the Portuguese, Spanish, French, Swiss and Croatian populations. We argue that the lack of resolution offered by traditional population genetics potentially hamper the identification of subtle patterns of genetic structure in phylogeographically shallow taxa, and that the situation of the declining M. schreibersii might be more common that previously assumed.
Data availability
Microsatellite and SNP matrices are available at the following link: https://doi.org/10.5061/dryad.4xgxd25d2. R scripts are available upon demand to Jérôme Goudet (jerome.goudet@unil.ch) and Christophe Dufresnes (christophe.dufresnes@hotmail.fr).
References
Charlesworth, B. & Charlesworth, D. Population genetics from 1966 to 2016. Heredity 118, 2–9 (2017).
Orsini, L., Vanoverbeke, J., Swillen, I., Mergeay, J. & Meester, L. Drivers of population genetic differentiation in the wild: Isolation by dispersal limitation, isolation by adaptation and isolation by colonization. Mol. Ecol. 22, 5983–5999 (2013).
Vendrami, D. L. J. et al. RAD sequencing resolves fine-scale population structure in a benthic invertebrate: Implications for understanding phenotypic plasticity. R. Soc. Open Sci. 4, 160548 (2017).
Dufresnes, C., Rodrigues, N. & Savary, R. Slow and steady wins the race: Contrasted phylogeographic signatures in two Alpine amphibians. Integr. Zool. 17, 181–190 (2021).
Frankham, R. Genetics and extinction. Biol. Conserv. 126, 131–140 (2005).
Schwartz, M. K., Luikart, G. & Waples, R. S. Genetic monitoring as a promising tool for conservation and management. Trends in Ecol. Evol. 22, 25–33 (2007).
Ottewell, K. M., Bickerton, D. C., Byrne, M. & Lowe, A. J. Bridging the gap: A genetic assessment framework for population-level threatened plant conservation prioritization and decision-making. Divers. Distrib. 22, 174–188 (2016).
Frankham, R., Bradshaw, C. J. A. & Brook, B. W. Genetics in conservation management: Revised recommendations for the 50/500 rules, red list criteria and population viability analyses. Biol. Conserv. 170, 56–63 (2014).
Hohenlohe, P. A., Funk, C. W. & Rajora, O. P. Population genomics for wildlife conservation and management. Mol. Ecol. 30, 62–82 (2020).
Angelone, S. & Holderegger, R. Population genetics suggests effectiveness of habitat connectivity measures for the European tree frog in Switzerland. J. Appl. Ecol. 46, 879–887 (2009).
Griffiths, S. M., Taylor-Cox, E. D., Behringer, D. C., Butler, M. J. IV. & Preziosi, R. F. Using genetics to inform restoration and predict resilience in declining populations of a keystone marine sponge. Biodivers. Conserv. 29, 1383–1410 (2020).
Moritz, C. Conservation units and translocations: Strategies for conserving evolutionary processes. Hereditas 130, 217–228 (1999).
Bohonak, A. J. Dispersal, gene flow, and population structure. Q. Rev. Biol. 74, 21–45 (1999).
Arguedas, N. & Parker, P. G. Seasonal migration and genetic population structure in house wrens. Condor 102, 517–528 (2000).
Quillfeldt, P. et al. Does genetic structure reflect differences in non-breeding movements? A case study in small, highly mobile seabirds. BMC Evol. Biol. 17, 160 (2017).
Charlesworth, B., Sniegowski, P. & Stephan, W. The evolutionary dynamics of repetitive DNA in eukaryotes. Nature 371, 215–220 (1994).
Schlötterer, C. Evolutionary dynamics of microsatellite DNA. Chromosoma 109, 365–371 (2000).
Ellegren, H. Microsatellites: Simple sequences with complex evolution. Nat. Rev. Genet. 5, 435–445 (2004).
Hodel, R. G. J. et al. The report of my death was an exaggeration: A review for researchers using microsatellites in the 21st century. Appl. Plant Sci. 4, 1600025 (2016).
Dufresnes, C. & Litvinchuk, S. N. Diversity, distribution and molecular species delimitation in frogs and toads from the Eastern Palearctic. Zool. J. Linn. Soc. 195, 695–760 (2022).
Galtier, N., Nabholz, B., Glémin, S. & Hurst, G. D. D. Mitochondrial DNA as a marker of molecular diversity: A reappraisal. Mol. Evol. 18, 4541–4550 (2009).
Zink, R. M. & Barrowclough, G. Mitochondrial DNA under siege in avian phylogeography. Mol. Ecol. 17, 2107–2121 (2008).
Toews, D. P. L. & Brelsford, A. The biogeography of mitochondrial and nuclear discordance in animals. Mol. Ecol. 21, 3907–3930 (2012).
Bonnet, T., Leblois, R., Rousset, F. & Crochet, P.-A. A reassessment of explanations for discordant introgressions of mitochondrial and nuclear genomes. Evolution 71, 2140–2218 (2017).
Davey, J. W. & Blaxter, M. L. RADSeq: Next-generation population genetics. Brief Funct. Genomics 9, 416–423 (2010).
Lexer, C. et al. ‘Next generation’ biogeography: Towards understanding the drivers of species diversification and persistence. J. Biogeogr. 40, 1013–1022 (2013).
Baird, N. A. et al. Rapid SNP discovery and genetic mapping using sequenced RAD markers. PLoS ONE 3, e3376 (2008).
Peterson, B. K., Weber, J. N., Kay, E. H., Fisher, H. S. & Hoekstra, H. E. Double Digest RADseq: An inexpensive method for de novo SNP discovery and genotyping in model and non-model species. PLoS ONE 7, e37135 (2012).
Dufresnes, C. et al. Phylogeography of a cryptic speciation continuum in Eurasian spadefoot toads (Pelobates). Mol. Ecol. 28, 3257–3270 (2019).
Sunde, J., Yıldırım, Y., Tibblin, P. & Forsman, A. Comparing the performance of microsatellites and RADseq in population genetic studies: Analysis of data for pike (Esox Lucius) and a synthesis of previous studies. Front. Genet. 11, 218 (2020).
Moussy, C. et al. Migration and dispersal patterns of bats and their influence on genetic structure. Mammal Rev. 43, 183–195 (2013).
Berthier, P., Excoffier, L. & Ruedi, M. Recurrent replacement of mtDNA and cryptic hybridization between two sibling bat species Myotis myotis and Myotis blythii. Proc. R. Soc. B: Biol. Sci. 273, 3101–3109 (2007).
Wright, P. G. R. et al. Hydrogen isotopes reveal evidence of migration of Miniopterus schreibersii in Europe. BMC Ecol. 20, 52 (2020).
Schnetter, W. Beringungsergebnisse an der Langflügelfledermaus (Miniopterus schreibersi Kühl) im Kaiserstuhl. Bonn. Zool. Beitr. 11, 150–165 (1960).
Rodrigues, L. Miniopterus schreibersii. In The Atlas of European Mammals (eds Mitchell-Jones, A. J. et al.) 154–155 (Academic Press, 1999).
Rodrigues, L., Ramos Pereira, M. J., Rainho, A. & Palmeirim, J. M. Behavioral determinants of gene flow in the bat Miniopterus schreibersii. Behav. Ecol. Sociobiol. 64, 835–843 (2010).
Rodrigues, L. & Palmeirim, J. M. Migratory behaviour of Miniopterus schreibersii (Chiroptera): When, where, and why do cave bats migrate in a Mediterranean region?. J. Zool. 274, 116–125 (2008).
Ramos Pereira, M. J., Salgueiro, P., Rodrigues, L., Coelho, M. M. & Palmeirim, J. M. Population structure of a cave-dwelling bat, Miniopterus schreibersii: Does it reflect history and social organization?. J. Hered. 100, 533–544 (2009).
Bilgin, R. et al. Circum-Mediterranean phylogeography of a bat coupled with past environmental niche modeling: A new paradigm for the recolonization of Europe?. Mol. Phylogenet. Evol. 99, 323–336 (2016).
Gürün, K. et al. A continent-scale study of the social structure and phylogeography of the bent-wing bat, Miniopterus schreibersii (Mammalia: Chiroptera), using new microsatellite data. J. Mammal. 100, 1865–1878 (2019).
Gazaryan, S., Bücs, S., Çoraman, E. Miniopterus schreibersii (errata version published in 2021). The IUCN Red List of Threatened Species 2020: e.T81633057A195856522 (2020).
Miller-Butterworth, C. M., Jacobs, D. S. & Harley, E. H. Isolation and characterization of highly polymorphic microsatellite loci in Schreibers’ long-fingered bat, Miniopterus schreibersii (Chiroptera: Vespertilionidae). Mol. Ecol. Notes 2, 139–141 (2002).
Wood, R., Weyeneth, N. & Appleton, B. Development and characterisation of 20 microsatellite loci isolated from the large bent-wing bat, Miniopterus schreibersii (Chiroptera: Miniopteridae) and their cross-taxa utility in the family Miniopteridae. Mol. Ecol. Resour. 11, 675–685 (2011).
Witsenburg, F. et al. How a haemosporidian parasite of bats gets around: The genetic structure of a parasite, vector and host compared. Mol. Ecol. 24, 926–940 (2015).
Van Oosterhout, C., Hutchinson, W. F., Wills, D. P. M. & Shipley, P. micro-checker: Software for identifying and correcting genotyping errors in microsatellite data. Mol. Ecol. Notes 4, 535–538 (2004).
Parchman, T. L. et al. Genome wide association mapping of an adaptive trait in lodgepole pine. Mol. Ecol. 21, 2991–3005 (2012).
Catchen, J. M., Amores, A., Hohenlohe, P., Cresko, W. & Postlethwait, J. H. Stacks: Building and genotyping loci de novo from short-read sequences. G3 1, 171–182 (2011).
Weir, B. S. & Cockerham, C. C. Estimating F-statistics for the analyses of population structure. Evolution 38, 1358–1370 (1984).
Goudet, J. hierfstat, a package for r to compute and test hierarchical F-statistics. Mol. Ecol. Notes 5, 184–186 (2005).
Frankham, R., Ballou, J. D. & Briscoe, D. A. A Primer of Conservation Genetics (Cambridge University Press, 2004).
Weir, B. S. & Goudet, J. A unified characterization of population structure and relatedness. Genetics 206, 2085–2103 (2017).
Mantel, N. A. The detection of disease clustering and a generalized regression approach. Cancer Res. 27, 209–220 (1967).
Wright, S. Isolation by distance. Genetics 28, 114–138 (1943).
Cavalli-Sforza, L. L. & Edwards, A. W. F. Phylogenetic analysis: Model and estimation procedures. Am. J. Hum. Genet. 19, 233–257 (1967).
Paradis, E. & Schliep, K. ape 5.0: An environment for modern phylogenetics and evolutionary analyses in R. Bioinformatics 35, 526–528 (2019).
Goudet, J., Perrin, N. & Waser, P. Tests for sex-biased dispersal using bi-parentally inherited genetic markers. Mol. Ecol. 11, 1103–1114 (2002).
Frichot, E. & François, O. lea: An r package for landscape and ecological association studies. Methods Ecol. Evol. 6, 925–929 (2015).
Yannic, G. et al. High connectivity in a long-lived High-Arctic seabird, the ivory gull Pagophila eburnea. Polar Biol. 39, 221–236 (2016).
Cumer, T. et al. Landscape and climatic variations of the Quaternary shaped multiple secondary contacts among barn owls (Tyto alba) of the Western Palearctic. Mol. Biol. Evol. 39, msab343 (2022).
Boston, E. S. M., Montgomery, W. I., Hynes, R. & Prodöhl, P. A. New insights on postglacial colonization in western Europe: The phylogeography of the Leisler’s bat (Nyctalus leisleri). Proc. R. Soc. B: Biol. Sci. 282, 20142605 (2015).
Razgour, O. et al. The shaping of genetic variation in edge-of-range populations under past and future climate change. Ecol. Lett. 16, 1258–1266 (2013).
Petit, E., Balloux, F. & Goudet, J. Sex-biased dispersal in a migratory bat: A characterization using sex-specific demographic parameters. Evolution 55, 635–640 (2001).
Moussy, C. et al. Population genetic structure of serotine bats (Eptesicus serotinus) across Europe and implications for the potential spread of bat rabies (European bat lyssavirus EBLV-1). Heredity 115, 83–92 (2015).
Rossiter, S. J., Benda, P., Dietz, C., Zhang, S. & Jones, G. Rangewide phylogeography in the greater horseshoe bat inferred from microsatellites: Implications for population history, taxonomy and conservation. Mol. Ecol. 16, 4699–4714 (2007).
Dool, S. E. et al. Phylogeography and postglacial recolonization of Europe by Rhinolophus hipposideros: Evidence from multiple genetic markers. Mol. Ecol. 22, 4055–4070 (2013).
Kerth, G. et al. Communally breeding Bechstein’s bats have a stable social system that is independent from the postglacial history and location of the populations. Mol. Ecol. 17, 2368–2381 (2008).
Garrick, R. C., Banusiewicz, J. D., Burgess, S., Hyseni, C. & Symula, R. E. Extending phylogeography to account for lineage fusion. J. Biogeogr. 46, 268–278 (2019).
Burri, R. et al. The genetic basis of color-related local adaptation in a ring-like colonization around the Mediterranean. Evolution 70, 140–153 (2016).
Taberlet, P., Fumagalli, L., Wust-Saucy, A.-G. & Cosson, J.-F. Comparative phylogeography and postglacial colonization routes in Europe. Mol. Ecol. 7, 453–464 (1998).
Hewitt, G. M. Post-glacial re-colonization of European biota. Biol. J. Linn. Soc. 68, 87–112 (1999).
Gómez, A. & Lunt, D. H. Refugia within refugia: Patterns of phylogeographic concordance in the Iberian Peninsula. In Phylogeography of Southern European Refugia (eds Weiss, S. & Ferrand, N.) 155–188 (Springer, 2007).
Vonhof, M. J., Russell, A. L. & Miller-Butterworth, M. Range-wide genetic analysis of little brown bat (Myotis lucifugus) populations: Estimating the risk of spread of white-nose syndrome. PLoS ONE 10, e0128713 (2015).
Auteri, G. G. & Knowles, L. L. Decimated little brown bats show potential for adaptive change. Sci. Rep. 10, 3023 (2020).
Gignoux-Wolfsohn, S. A. et al. Genomic signatures of selection in bats surviving white-nose syndrome. Mol. Ecol. 30, 5643–5657 (2021).
Rivers, N. M., Butlin, R. K. & Altringham, J. D. Autumn swarming behaviour of Natterer’s bats in the UK: Population size, catchment area and dispersal. Biol. Conserv. 127, 215–226 (2006).
Reis, N. R., Fregonezi, M. N., Peracchi, A. L. & Rossaneis, B. K. Metapopulation in bats of Southern Brazil. Braz. J. Biol. 72, 605–609 (2012).
Humphrey, S. R. & Oli, M. K. Population dynamics and site fidelity of the cave bat, Myotis velifer, Oklahoma. J. Mammal. 96, 946–956 (2015).
Jeffries, D. L. et al. Comparing RADseq and microsatellites to infer complex phylogeographic patterns, an empirical perspective in the Crucian carp, Carassius carassius. L. Mol. Ecol. 25, 2997–3018 (2016).
Hodel, R. G. J. et al. Adding loci improves phylogeographic resolution in red mangroves despite increased missing data: Comparing microsatellites and RAD-Seq and investigating loci filtering. Sci. Rep. 7, 17598 (2017).
Lemopoulos, A. et al. Comparing RADseq and microsatellites for estimating genetic diversity and relatedness—Implications for brown trout conservation. Ecol. Evol. 9, 2106–2120 (2019).
Zimmerman, S. J., Aldridge, C. L. & Oyler-McCance, S. J. An empirical comparison of population genetic analyses using microsatellite and SNP data for a species of conservation concern. BMC Genom. 21, 382 (2020).
Hale, M. L., Burg, T. M. & Steeves, T. E. Sampling for microsatellite-based population genetic studies: 25 to 30 individuals per population is enough to accurately estimate allele frequencies. PLoS ONE 7, e45170 (2012).
Quetglas, J., Gonzalez, F. & Paz, O. Estudian la extraña mortandad de miles de murcielago de cuevas. Quercus 203, 50 (2003).
Negredo, A. et al. Discovery of an ebolavirus-like filovirus in Europe. PLoS Pathog. 7, e1002304 (2011).
Reed, D. H. & Frankham, R. Correlation between fitness and genetic diversity. Conserv. Biol. 17, 230–237 (2003).
Alcalde, J. T., Artácoz, A. & Meijide, F. Recuperación de la colonia de Miniopterus schreibersii de la cueva de Cueva de Ágreda (Soria). Barbastella 5, 32–35 (2012).
Kemenesi, G. et al. Re-emergence of Lloviu virus in Miniopterus schreibersii bats, Hungary, 2016. Emerg. Microbes Infect. 7, 66 (2018).
Kemenesi, et al. Isolation of infectious Lloviu virus from Schreiber’s bats in Hungary. Nat. Commun. 13, 1706 (2022).
Stoffel, C. et al. Genetic consequences of population expansions and contractions in the common hippopotamus (Hippopotamus amphibius) since the late Pleistocene. Mol. Ecol. 24, 2507–2520 (2015).
Acknowledgements
This study was supported by the Fondation Herbette, as well as Grants 31003A_179358 and 31003A_138187 from the Swiss National Science Foundation.
Author information
Authors and Affiliations
Contributions
P.C. and J.G. designed research. L.D., F.G.-W., L.C., A.L.-B., J.P., I.P., D.S., M.Š. and P.C. conducted fieldwork. L.D., A.B. and L.C. conducted labwork; C.D., L.D. and J.G. analyzed the data; C.D., L.D., P.C., J.G. drafted the manuscript, which was critically improved by all coauthors.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Dufresnes, C., Dutoit, L., Brelsford, A. et al. Inferring genetic structure when there is little: population genetics versus genomics of the threatened bat Miniopterus schreibersii across Europe. Sci Rep 13, 1523 (2023). https://doi.org/10.1038/s41598-023-27988-4
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-023-27988-4
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.