Skip to main content
NCBI home page
Ecology and Evolution logoLink to Ecology and Evolution
. 2020 Jan 15;10(3):1180–1192. doi: 10.1002/ece3.5956

Transcriptomic regulation of seasonal coat color change in hares

Mafalda S Ferreira 1,2, Paulo C Alves 1,2,3, Colin M Callahan 4, Iwona Giska 1, Liliana Farelo 1, Hannes Jenny 5, L Scott Mills 3,6, Klaus Hackländer 7, Jeffrey M Good 3,4, José Melo‐Ferreira 1,2,✉
PMCID: PMC7029059  PMID: 32076506

Abstract

Color molts from summer brown to winter white coats have evolved in several species to maintain camouflage year‐round in environments with seasonal snow. Despite the eco‐evolutionary relevance of this key phenological adaptation, its molecular regulation has only recently begun to be addressed. Here, we analyze skin transcription changes during the autumn molt of the mountain hare (Lepus timidus) and integrate the results with an established model of gene regulation across the spring molt of the closely related snowshoe hare (L. americanus). We quantified differences in gene expression among three stages of molt progression—“brown” (early molt), “intermediate,” and “white” (late molt). We found 632 differentially expressed genes, with a major pulse of expression early in the molt, followed by a milder one in late molt. The functional makeup of differentially expressed genes anchored the sampled molt stages to the developmental timeline of the hair growth cycle, associating anagen to early molt and the transition to catagen to late molt. The progression of color change was characterized by differential expression of genes involved in pigmentation, circadian, and behavioral regulation. We found significant overlap between differentially expressed genes across the seasonal molts of mountain and snowshoe hares, particularly at molt onset, suggesting conservatism of gene regulation across species and seasons. However, some discrepancies suggest seasonal differences in melanocyte differentiation and the integration of nutritional cues. Our established regulatory model of seasonal coat color molt provides an important mechanistic context to study the functional architecture and evolution of this crucial seasonal adaptation.

Keywords: developmental timeline, gene expression, molt cycle, seasonal coat color change, transcriptomics

We analyze skin transcription changes during the autumn color molt of the mountain hare and integrate the results with an established model of gene regulation across the spring molt of the closely related snowshoe hare. Our work provides an important mechanistic context to study the functional architecture and evolution of this crucial seasonal adaptation.

graphic file with name ECE3-10-1180-g004.jpg

1. INTRODUCTION

Seasonal environments impose numerous challenges to organismal survival. To track seasonality, individuals cycle important biological processes, known as phenologies (e.g., reproduction, migration, hibernation, or molt), allowing a better phenotypic match with seasonal selective pressures (Helm et al., 2013; Visser, Caro, Oers, Schaper, & Helm, 2010). Accurate timing of physiological and behavioral changes is important for fitness. Synchronizing phenotypes with the environment requires a precise perception of predictable cues (e.g., variation in day length) that anticipate seasonal change and trigger regulatory cascades and shifts in gene expression, with consequent phenotypic change (Bradshaw & Holzapfel, 2007). Quantifying gene expression changes underlying phenological traits is therefore crucial to understand how organisms synchronize their phenotypes with the surrounding environment (Ferreira et al., 2017; Giska et al., 2019; Jones et al., 2018; Schwartz & Andrews, 2014).

Discrete seasonal molts have evolved as energy‐efficient phenological adaptations to regenerate hair layers and produce coats with insulating properties adequate for each season (Geyfman, Plikus, Treffeisen, Andersen, & Paus, 2015; Hart, 1956; Ling, 1972). In at least 20 vertebrate species, seasonal molts are accompanied by the change from summer brown to winter white coats, allowing camouflage in snow‐covered environments (Mills et al., 2013, 2018). As in other phenological traits, seasonal molting is mostly triggered by annual changes in day length (photoperiod) (Bissonnette & Bailey, 1944; Hoffmann, 1978; Lincoln, Clarke, Hut, & Hazlerigg, 2006; Zimova et al., 2018). Other environmental factors, such as temperature or snow presence, have been suggested to influence the rate and completeness of the molt in different mammals (Zimova et al., 2018 and references therein). However, molt onset seems to have minimal plasticity (Zimova, Mills, Lukacs, & Mitchell, 2014), suggesting that photoperiod is the most important cue entraining annual molt timing. Consequently, since animals coordinate molt with the fixed variation in photoperiod, the reductions of annual snow cover associated with climate change are predicted to lead to increased mismatches between coat and background colors, disrupting camouflage and negatively affecting the fitness of seasonally color changing species (Atmeh, Andruszkiewicz, & Zub, 2018; Zimova, Mills, & Nowak, 2016). Predicting how color changing molts will respond to environmental change requires a better understanding of the internal mechanisms that regulate the change of hair and color (Helm & Stevenson, 2014).

In mammals, pathways and genes involved in the perception of the photoperiodic signal, such as the circadian clock and melatonin production and reception, hair growth cycle, and melanogenesis, are expected to be involved in the regulation of seasonal coat color changing molts (Allain & Rougeot, 1980; Balsalobre, 2002; Duncan & Goldman, 1984; Ferreira et al., 2017; Lin et al., 2009; Lincoln et al., 2006; Slominski, Tobin, Shibahara, & Wortsman, 2004). The molecular mechanisms underlying the hair growth cycle (Lin et al., 2009; Schlake, Beibel, Weger, & Boehm, 2004; Slominski et al., 2004) and the basis of static melanin‐based coloration have been thoroughly studied (Allain & Rougeot, 1980; Duncan & Goldman, 1984; Logan & Weatherhead, 1980). However, the specific regulatory mechanisms of coat color changing molts have only been addressed recently. We previously studied the regulatory patterns of seasonal changing molts in snowshoe hares, inferring gene expression changes in the skin during the spring white‐to‐brown molt in the species using RNA‐sequencing (Ferreira et al., 2017). This work established a genic regulatory model for the seasonal molt of snowshoe hares characterized primarily by genes involved in the hair growth cycle, circadian rhythms, and pigmentation. For example, among the differentially expressed genes along the molt, a pulse of expression of the agouti signaling protein (ASIP) gene was detected at the onset of the molt. The ASIP peptide signals the switch between darker and lighter melanin pigment, suggesting a role of the melanin production pathway in the regulation of seasonal color change (Ferreira et al., 2017). Furthermore, cis‐regulatory variation at ASIP has been linked to polymorphism in winter coat color in snowshoe hares (Lepus americanus) (Jones et al., 2018) and mountain hares (L. timidus) (Giska et al., 2019). It remains unclear how generalizable these results are across species.

Here, we study the transcriptional landscape of the autumn molt in the mountain hare (L. timidus) and integrate the results with the previously quantified gene expression landscape of the spring molt of snowshoe hares (Ferreira et al., 2017). Mountain hares are widely distributed across Eurasia, from Western Europe to Eastern Russia, with some isolated populations in the Alps, Ireland, British, and Faroe Islands (Angerbjörn, 2018). Throughout most of their range, mountain hares alternate between summer brown and winter white coats, with some exception, such as in Ireland and the Faroe Islands (Angerbjörn, 2018; Bergengren, 1969; Mills et al., 2018; Schai‐Braun & Hackländer, 2016). Here, we test the generality of the snowshoe hare seasonal molt regulatory model (Ferreira et al., 2017), placing hair color change in the stages of the hair growth cycle, and pin‐pointing seasonal differences in regulatory mechanisms between the two species. While snowshoe and mountain hares are not sister species (Matthee, Vuuren, Bell, & Robinson, 2004; Melo‐Ferreira et al., 2012), it is yet unclear to what degree ancestral variation or ancient introgression may explain sharing of the trait (Jones et al., 2018). We predict that the major expression waves related with pelage molt are conserved in these two species given the conservation of hair growth cycle mechanism in mammals (Lin, Chudova, Hatfield, Smyth, & Andersen, 2004; Schlake et al., 2004; Stenn & Paus, 2001). Possible differences may then be related to species‐specific differences or seasonal differences in hair color and composition, or to the integration of seasonal specific cues onto the regulation of hair growth. By integrating results across both species, we expand the model of genic regulation during both autumn and spring seasonal coat color change in hares and provide a context to understand the functional mechanisms controlling this crucial seasonal adaptation.

2. MATERIALS AND METHODS

2.1. Sampling, laboratory procedures, and sequencing

Sampling was performed following the reasoning of Ferreira et al., (2017). Briefly, the spatial progression of the molt across the body of a single molting individual results in the simultaneous occurrence of different skin patches that are expected to be in different molting stages—proliferative stage of active hair growth “anagen,” the follicle regression stage “catagen” and a final resting stage “telogen” (Geyfman, Gordon, Paus, & Andersen, 2012; Lin et al., 2004; Schlake et al., 2004) (Figure 1). Given an expected lag between expression changes and the visible changes in hair color used as proxy of molting stage (Ferreira et al., 2017), we expect that early molt (“brown”) captures mostly anagen‐related processes and possibly late telogen, and late molt (“white”) captures late anagen and the transition to catagen (Figure 1b). Four mountain hare (L. timidus) specimens were collected from hunting bags during regular permit hunting in Grisons, Switzerland (Table S1). Hares were undergoing pelage molt, from brown to white, which starts on the extremities and ends in the head and dorsum, a pattern that is reversed during spring molt (Hewson, 1958). Sampling different molting stages in the same individual allows controlling for individual variation in gene expression (Ferreira et al., 2017), and only individuals with three identifiable patches of skin corresponding to the stages of molt progression were used in this study: early molt (brown hair), intermediate stage (both brown and white hair), and late molt (white hair) (Figure 1), resulting in a total of 12 tissue samples. Skin patches of 2 cm2 were collected from the animals’ dorsum for each skin type immediately after death, and stored first in RNAlater and then at −80°C until RNA extraction.

Figure 1.

Figure 1

Open in a new tab

Mountain hare autumn molt and sampling scheme. (a) Three patches of skin were sampled from individuals undergoing autumn molt, classified according to the color of guard hairs: “brown,” “intermediate” (both brown and white hairs), and “white.” (b) Parallel between the molt stages of our sampling scheme and the stages of the hair growth cycle, as established for the spring molt in Ferreira et al., (2017), and hypothesized for the autumn molt (this work)

From each skin sample, total RNA was extracted using the RNeasy® Mini Kit (QIAGEN) and strand‐specific cDNA libraries were synthetized using the SureSelect Strand‐Specific RNA Library Prep for Illumina Multiplexed Sequencing (Agilent Technologies). Library sizes were inferred using Bioanalyzer 2100 and pooled in equal molar concentrations using the KAPA Library quantification kit (KAPA BIOSYSTEMS). Pooled libraries were sequenced in two separate Illumina HiSeq runs: ~3% of two lanes of 100 bp paired‐end reads of an HiSeq 2000, at University of California Berkeley, and then three lanes of 101 bp paired‐end reads in a HiSeq 2500 at Utah Microarray Core.

2.2. Quality control and transcriptome assembly

Reads failing quality control, flagged by the Illumina Casava‐1.8 Fastq Filter (Gordon, 2011), were discarded and adapters removed using Cutadapt (1.3). Trimmomatic (0.32) was used to retain a minimum phred quality score of 15 across 4 bp sliding windows, to trim the first 13 base pairs from every read, which showed a nucleotide composition bias (Hansen, Brenner, & Dudoit, 2010), and finally to remove trimmed reads smaller than 23 bp.

A de novo skin transcriptome assembly for L. timidus was performed to allow an independent unbiased comparison with the results of Ferreira et al., (2017). The transcriptome assembly was performed with Trinity (2.6.6) (Grabherr et al., 2011), using filtered and properly paired reads and specifying “RF” for strand‐specific assembly. Transrate (1.0.2) (Smith‐Unna, Boursnell, Patro, Hibberd, & Kelly, 2016) was used to assess the quality of the transcriptome, filtering erroneously assembled contigs. We then annotated this filtered transcriptome using the rabbit (OryCun2.0) and mouse (GRCm38.p6) ENSEMBL 92 peptide references applying the reciprocal blast implemented in Transrate. The mouse annotation was only used when no rabbit annotation was produced for all analyses, except for the Reactome pathway enrichment analysis (see below).

2.3. Differential expression analysis

Reads were mapped to the de novo assembled transcriptome using bowtie2 (2.3.4.1) with the following flags: ‐q, ‐‐phred33, ‐D 20, ‐R 3, ‐N 1, ‐L 20, ‐i S,1,0.50, ‐‐dpad 0, ‐‐gbar 9,999,999, ‐‐mp 1,1, ‐‐np 1, ‐‐score‐min L,0,‐0.1, ‐I 1, ‐X 1,000, ‐‐no‐mixed, ‐‐no‐discordant, ‐‐nofw, ‐p 1, ‐k 200, ‐‐fr, ‐x. To generate bowtie2 indices, we used the extract‐transcript‐to‐gene‐map‐from‐trinity script and then rsem‐prepare‐reference script from RSEM (1.3.0). Finally, the outputs from bowtie2 were used as inputs for rsem‐calculate‐expression to generate relative abundances for all genes across the transcriptome. The transcriptome annotation was used to filter the RSEM output, removing Trinity “genes” (putative genes) with annotation to multiple genes (i.e., assembled putative isoforms annotated to different ENSEMBL genes), or with no annotation.

Using the filtered RSEM output, differential expression was inferred between molt stages (“brown,” “intermediate,” and “white”) using edgeR (3.20.9) (Robinson, McCarthy, & Smyth, 2010). Only Trinity genes expressed at a minimum of one count per million (CPM) mapped reads in at least 4 samples were retained. The trimmed mean of M‐values was used to normalize data across libraries, and a Cox‐Reid profile‐adjusted likelihood was used to calculate common, trended, and tagwise dispersions, as described in Robinson et al. (2010). The biological coefficient of variation (BCV) was estimated as the square root of the common dispersion (McCarthy, Chen, & Smyth, 2012). To inspect the relationship between expression counts in each sample, we plotted a multidimensional scaling plot (MDS plot) in R using the 500 genes with highest dispersion between each sample pair. Given an aberrant expression pattern estimated for one individual, which did not result in the individualization of molting stages (Figure S1) expected from our sampling strategy (Ferreira et al., 2017), we evaluated the BCV and estimated the expected power to detect gene expression changes in the 4 versus 3‐individual dataset using RNASeqPower (Hart, Therneau, Zhang, Poland, & Kocher, 2013). For comparison, we also estimated the expected power for the snowshoe hare dataset from Ferreira et al., (2017).

The inspection of the MDS plot showed that replicates tend to cluster primarily by molting stage and secondarily by individual (Figure S2), suggesting that individual is a significant contributor of variation to gene expression differences. To incorporate this factor in the analysis, we used “individual” as a blocking factor in the statistical test for differential expression, to control for individual variation in gene expression. Differential expression between molting stages was inferred using a likelihood ratio test for 3 pairwise comparisons. After a Benjamin‐Hochberg multiple test correction, all genes with a false discovery rate (FDR) smaller than 0.05 were considered differentially expressed. We tested for significant overlap between mountain hare and snowshoe hare set of upregulated genes in each molt stage, by collapsing the upregulated genes for the same stage from each pairwise comparison, and using a Fisher exact test in R, considering as background all common unique ENSEMBL gene annotations between the transcriptomes of the two species.

A clustering analysis was then performed to identify major expression patterns across genes. We converted CPM levels for each gene to fragments per kilobase per million reads mapped (FPKM) and further log2 transformed and mean centered these values. The log2 mean‐centered FPKM value for each differentially expressed gene was used as input for two clustering methods. First, a complete linkage hierarchical clustering was performed using the Pearson's correlation coefficient between samples and Euclidian distances between genes with the R package gplots (3.0). Second, a partition clustering analysis was done using a gap statistic to calculate the optimal number of clusters (k) using clustgap( ) from the R package cluster (2.0.7). The optimal k of 9 was used in a partitioning around medoids (pam) clustering analysis using pam( ) from the R package cluster (Figure S3). The k clusters were then summarized in 3 major general patterns of expression, and the pam clustering was repeated with k = 3. Finally, the expression levels and mean expression level per cluster were plotted using a modified version of the plot_expression_patterns.pl script from the Trinity pipeline.

2.4. Functional enrichment analysis

The functions of genes within each cluster of distinct expression signature were investigated using a Gene Ontology enrichment analysis using Ontologizer (2.1) (Bauer, Grossmann, Vingron, & Robinson, 2008). A custom gene association file (GAF) was built using custom scripts make_input_4_gaf.py and do_Gaf_file_new_version.py (available at https://github.com/evochange/coat_color_change_transcriptomics) and using the GO term ID for each ENSEMBL ID present in the transcriptome assembly. Enrichment was tested (FDR < 0.05) using a parent‐child‐union test with the entire skin transcriptome as background and the Benjamin–Hochberg correction for multiple tests. Furthermore, a Reactome pathway enrichment analysis was done using the Reactome database (version 65; https://reactome.org/PathwayBrowser/#TOOL=AT) and the mouse annotation for each gene (the rabbit is not supported in the database). Enrichment was tested using the binomial test.

Finally, we investigated the expression profiles of genes annotated with Gene Ontology terms potentially related with seasonal coat color change: “molting cycle” (GO:0042303), “circadian rhythm” (GO:0007623), “pigmentation” (GO:0043473), or one of the child terms. The child terms were identified using the GOOSE tool at the Gene Ontology Consortium website (http://amigo.geneontology.org/goose).

2.5. Quantitative PCR for Agouti gene expression

Given its importance as a marker of anagen and for coat color regulation (Ferreira et al., 2017; Giska et al., 2019; Jones et al., 2018), we further investigated ASIP’s expression pattern. We quantified the expression of ASIP hair cycle isoform and reference housekeeping genes SDHA and ACTB for the three individuals and nine skin samples included in the differential gene expression analyses. Total RNA was extracted from 30 mg of each skin sample with the RNeasy® Mini Kit (QIAGEN). Tissue was homogenized in a rotor‐stator homogenizer (Mixer Mill MM400, Retsch) at 30 Hz for 10 min. The RNA integrity was checked with the Agilent Technologies TapeStation for a minimum RIN > 7. First‐strand cDNA synthesis was performed using oligo (dT) primers and the GRS cDNA synthesis kit (GRISP) using 400 ng of RNA from each skin sample. Relative expression was quantified based on quantitative PCR (qPCR) of the gene of interest normalized across individuals to the reference genes (primers are indicated in Table S2). Their amplification efficiencies were calculated based on the slope of a regression line fitted to Ct values of five samples from 2‐fold serial dilutions (E = 10(−1/slope) − 1), using two replicates per sample. Then, to quantify ASIP expression across the molt, three replicates of qPCR reactions per skin sample were performed using 1X iTaqTM universal SYBR® green supermix (Bio‐Rad), 0.4 µmol/L each primer and 1 µl cDNA (2‐fold dilution of stock cDNA), total volume 10 µl, and the thermal conditions shown in Table S3. The raw qPCR results, presented as the threshold cycle (Ct), were used to calculate expression of the ASIP hair cycle isoform, normalized by the expression of the ACTB or SDHA gene according to the formula 2 -ΔCt, where ΔCt=CtAgouti-Ctrefgene. We tested for differences in expression between molt stages using the linear regression (pcr_lm()) test from R package pcr (Ahmed & Kim, 2018), using “individual” and “molt stage” as factors, and ACTB as the reference gene.

3. RESULTS

3.1. Assembly, annotation, and differential expression

A total of 139,005,571 paired‐end reads were obtained, averaging 11,583,798 read pairs per individual (range = 10,974,557–12,255,152). After adapter and quality‐based trimming, the final dataset comprised 124,393,164 read pairs (averaging = 10,366,097 read pairs per individual, range = 9,916,123–10,946,710), varying in length from 23 to 87 base pairs (Table S4). These reads were then used to produce a strand‐specific assembly using Trinity, and the resulting raw transcriptome consisted of 173,848 Trinity “genes” and 233,182 Trinity “transcripts” (Table S5). After quality control using Transrate, the transcriptome contained 159,769 “genes” and 202,135 “transcripts” (Tables S5 and S6). Finally, the annotation‐based filter resulted in 36,101 Trinity “genes” (Table S6).

The inspection of the MDS plot revealed that the skin samples of one individual consistently deviated from the remaining samples and did not capture the individualization of the sampled molt stages aimed with our sampling strategy (Figure S1) (Ferreira et al., 2017). This suggests an artificial bias during sampling that increased the variance in the dataset. The 3‐individuals’ dataset provided higher power (0.82 vs. 0.72 to detect a 1.5‐fold change for a probability of false positives of 0.05) and lower BCV (0.16 vs. 0.21) when compared to the 4‐individuals’ dataset. This estimated power is comparable to that of the snowshoe hare dataset from Ferreira et al., (2017) (0.85 to detect 1.5‐fold changes in expression), and adequate to detect differential expression for lowly expressed genes (Todd, Black, & Gemmell, 2016). Therefore, hereafter we report the results from the analyses of the 3‐individuals’ dataset. After filtering for multiple annotated and lowly expressed genes, our transcriptome consisted of 17,270 genes, corresponding to 12,070 annotated ENSEMBL genes (11,223 rabbit genes and 847 mouse genes) (Tables S6 and S7). Of the annotated genes, 75% (9,091) were also present among the 10,345 genes annotated in the snowshoe hare transcriptome (Ferreira et al., 2017).

Differential expression was tested in three pairwise comparisons between the “brown,” “intermediate,” and “white” molt stages, and 632 genes, corresponding to 514 unique ENSEMBL annotations, were found differentially expressed in at least one pairwise test (Figure 2; Table S8). Of these, 131 (25%) were also found differentially expressed in the snowshoe hare dataset (total of 754 unique ENSEMBL annotations) in any stage, representing a significant overlap (Fisher exact test, odds ratio = 4.37, p = 2.2E‐16; Table S9). A significant overlap was also found among genes upregulated early in molt in both mountain and snowshoe hares (102 unique ENSEMBL annotations of 379 and 547 upregulated early in mountain and snowshoe hare, respectively; Fisher exact test, odds ratio = 6.83, p = 2.20E‐16), but no overlap or significant overlap among genes upregulated in intermediate or late in molt, respectively (p > .05; Table S9). As in Ferreira et al., (2017), most upregulated genes were found in the early molt (“brown” in the autumn molt), whereas “intermediate” was the stage with the lowest number of upregulated genes (Figures 2, 3a, and S4).

Figure 2.

Figure 2

Open in a new tab

Smear plots showing the number of differentially expressed genes (FDR < 0.05) in three pairwise comparisons between molt stages. Average log counts per million for each gene (dots) are plotted against log fold change. For each comparison, the number of upregulated genes in each stage is shown

Figure 3.

Figure 3

Open in a new tab

A model of the transcriptional landscape of seasonal coat color change. Left panels show the results of this work on the autumn molt of the mountain hare and right panels show the corresponding results on the spring molt of the snowshoe hare, adapted from Ferreira et. al (2017). (a) Patterns of expression of differentially expressed genes in the mountain hare and snowshoe hare datasets are shown as heatmaps of log2 transformed, mean‐centered FPKM levels of expression per differentially expressed gene (row) and sample (column; a–c and a–e). Dendrograms represent the complete linkage hierarchical clustering of Pearson's correlation coefficient between samples. Clusters result from PAM clustering of individual gene expression levels for k = 3 (B—“brown”; I—“intermediate”; W—“white”). The gray line connects the mean expression value of genes’ expression in “brown” (B), “intermediate” (I), and “white” (W). (b) Functional enrichment and candidate gene expression at early and late molt. For clusters in (a), a Gene Ontology and Reactome enrichment analysis was performed, and the presence of candidate genes, known to have functions related with molting cycle, circadian rhythms, and pigmentation regulation was inspected. Terms and genes specific to mountain hare (this work) and snowshoe hare (Ferreira et al., 2017) are highlighted in brown and green, respectively. Bold black indicates common terms to both species

To inspect the pattern of expression of these genes across the molt, we plotted mean‐centered log2 FPKM expression levels in a heatmap (left panel of Figure 3a), showing that most genes tend to be upregulated early in the molt and that levels of expression tend to decrease toward the end of molt. However, a smaller proportion of genes showed higher levels of expression at the end of molt. The hierarchical clustering analyses showed that sampled “brown” and “white” skin tissues clustered according to molt stage, while “intermediate” samples were split between the two clusters. Furthermore, the partitioning clustering analysis resulted in three clusters of genes with similar expression patterns: 1) 239 genes that decrease their expression gradually toward the end of molt, from “brown” to “white”; 2) 239 genes decrease their expression from “brown” to “intermediate” and retain a constant level of expression toward “white”; and 3) 154 genes increase their expression gradually from “brown” to “white” (left panel of Figure 3a and S4). Both clusters 1 and 2 group genes with higher expression early in the molt, which was also found in snowshoe hares (right panel of Figure 3a).

3.2. Gene Ontology term and Reactome pathway analyses

To understand if genes with different expression patterns could be distinguished by their functions, we tested the enrichment of Gene Ontology (GO) terms and Reactome pathways in each gene cluster obtained from partitioning clustering analysis (Tables S10–S15). We found 65, 4, and 2 GO biological process terms enriched in cluster 1, 2, and 3, respectively (Figure 3b, Tables S10–S12). Furthermore, 10 Reactome pathways were found enriched in cluster 1, 18 in cluster 2 and 9 in cluster 3 (Figure 3b, Tables S13–S15). In general, the enrichment analysis revealed functions and pathways that are expected in the molting skin (Ferreira et al., 2017). Both clusters with higher levels of expression in “brown” (1 and 2; Figure 3b) showed a similar enrichment of GO terms that were related with muscle development, cytoskeleton organization, transmembrane transport, circulatory system development, cell–cell communication and adhesion and regulation of gene expression and protein synthesis (Figure 3b, Tables S10 and S11). Pathway enrichment in both clusters was consistent with the GO enrichment; muscle‐related pathways enriched in cluster 1, and metabolism of proteins and RNA, and signaling transduction were enriched in cluster 2 (Figure 3b, Tables S13 and S14). In cluster 3 (genes increasing their expression toward the end of molt), only two GO biological process terms were found enriched: “lipid localization” and “regulation of behavior” (Figure 3b, Table S12). Pathways found enriched in cluster 3 are involved in keratinization, lipid metabolism, the retinoid cycle, and transport of small molecules (Figure 3b, Table S15). Next, we compared patterns of functional enrichment detected in mountain hares with patterns found during the snowshoe hare molt (Ferreira et al., 2017). Overlapping GO terms and Reactome pathways were only detected for gene clusters showing higher expression early in the molt cycle including muscle tissue development, ion transport, cytoskeleton organization, and glycogen breakdown (Figure 3b and S5).

In addition, we inspected the expression of genes with functions expected to be involved in seasonal coat color change. We found nine candidate genes that were differentially expressed across the molt with the GO terms “pigmentation,” “molting cycle,” and “circadian rhythm” (12 Trinity genes; Figure 3b, Table S16). Six of these genes (EN1, SPARC, MEF2C, DKK4, KRT71, EDAR) showed higher expression at early molt (cluster 2) and were found to be upregulated in “brown,” while three (Foxq1, KRT16, MC3R) increased their expression toward the end of molt (cluster 3) and were up regulated either in “white” or “intermediate” (Figure 3b, Table S16). Genes Foxq1, DKK4, KRT71, KRT16, EDAR are putatively involved in molting cycle‐related functions. Genes EN1, SPARC, EDAR, and MEF2C are annotated to the Gene Ontology term “pigmentation” and the child term “melanosome differentiation,” being all upregulated in “brown.” Finally, one gene, MC3R, annotated to child terms of “circadian rhythm” (“circadian regulation of gene expression” and “locomotor rhythm”), showed higher expression in “white” and “intermediate.” KTR71 was the only candidate gene found differentially expressed both here and in Ferreira et al., (2017) (an overlap not expected by chance; odds ratio = 66.3, p = .02; Table S9), though its late molt upregulation contrasts with the higher expression early in the molt in the snowshoe hare (Figure 3b).

Using qPCR, we note a tendency for higher expression of ASIP hair cycle isoform early in the molt of individual A, a marginal difference between early and the other stages for individual B, and no difference for individual C, though none of these patterns was significant (Figure S6, Tables S17 and S18). No significant differences in ASIP’s expression along the molt were detected based on RNA‐sequencing data (Figure S6).

4. DISCUSSION

Understanding how changes at different levels of molecular organization (e.g., gene, protein, or pathway) affect phenotype variation is critical to deciphering the evolution of adaptive phenotypes (e.g., Dalziel, Rogers, & Schulte, 2009). When possible, mechanistic studies of phenotypes should be based on natural populations so that molecular variation can be tied to the natural conditions that individuals experience (Alvarez, Schrey, & Richards, 2014) and, ultimately, to fitness (Barrett & Hoekstra, 2011). However, field‐based quantitative studies are challenging given the multiplicity of uncontrolled factors, including individual variation, that may hinder biologically meaningful patterns (Todd et al., 2016). We leveraged the ability to sample different stages of the developmental timeline of seasonal coat color change in single molting hares to control for individual variation and then combine transcriptional data from mountain and snowshoe hares to build a working model of the genic regulation of this crucial seasonal adaptation.

4.1. The transcriptional landscape of the molt in the context of the hair growth cycle

In keeping with the results in snowshoe hares (Ferreira et al., 2017), we found that most genes show a pattern of expression that is correlated with early (“brown”) or late (“white”) molt (Figure 3a), underlining the difficulty in discretizing the “intermediate” stage (Figures 2, 3 and S4). As in the snowshoe hare spring molt, most of the upregulated genes were detected early in the molt. This pattern of induction is expected during the proliferative stage of the hair growth cycle, anagen (Lin et al., 2004; Schlake et al., 2004), when melanogenesis occurs (Slominski & Paus, 1993). Genes upregulated in brown/early molt (cluster 1 and 2 of Figure 3a) showed enrichment of functional categories that are characteristic of anagen (Figure 3b and S10, S11, S13, and S14). For example, genes in cluster 1 and 2 were enriched in functions related with tissue organization and morphogenesis (e.g., “cytoskeleton organization,” “supramolecular fiber organization,” “tissue development”; Lin et al., 2004), gene expression and protein synthesis (e.g., “positive regulation of gene expression,” “metabolism of RNA”; Geyfman et al., 2012) cell–cell communication and adhesion (e.g., “regulation of cell communication by electrical coupling”; pathway “integrin cell surface interactions”; Botchkarev & Kishimoto, 2003; Mayer et al., 1993), and metabolism (e.g., “glycogen metabolism”; Geyfman et al., 2015; Shipman, Chase, & Montagna, 1955). An inspection of candidate genes for molt, circadian rhythm and pigment regulation, revealed that genes with known anagen‐related expression were also highly expressed early in the molt, such as hair keratins (KRT71; Stark, Breitkreutz, Limat, Bowden, & Fusenig, 1987; Porter et al., 2001), hair proliferation and shape‐determining genes (EDAR and DKK4; Millar, 2002; Fliniaux, Mikkola, Lefebvre, & Thesleff, 2008; Adhikari et al., 2016), genes involved in the incorporation of collagen to maintain skin integrity (SPARC, Table S16; collagen genes IIA1, IIIA1 and VIA2 annotated to “integrin cell surface interactions,” Table S14; Parakkal, 1969; Mayer et al., 1993; Strandjord, Madtes, Weiss, & Sage, 1999), and pigmentation genes (EN1, SPARC, EDAR and MEF2C; Table S16; Crocker & Cattanach, 1979; Slominski & Paus, 1993; Loomis, Kimmel, Tong, Michaud, & Joyner, 1998; Agarwal et al., 2011; Ditommaso et al., 2014).

Thus, the brown early molt stage largely recovers anagenic processes. However, we also detected some clear exceptions, such as the upregulation of pathways related to cell death usually characteristic of catagen (Lin et al., 2004; Stenn & Paus, 2001) and the upregulation of muscle‐related genes at the beginning of molt (Tables S10, S11, S13, and S14). Notably, muscle‐related genes were also differentially upregulated early in the spring molt of the snowshoe hare (Ferreira et al., 2017; Figure 3b), which could be related with the remodeling of tissue types that surround the hair follicle during hair growth cycle (Schmidt & Horsley, 2012).

Our white/late molt skin samples showed expression signatures consistent with hair follicles transitioning into late anagen and possibly catagen (Figures 2 and 3, Tables S12 and S15), mirroring patterns of the snowshoe hare molt (Figure 3; Ferreira et al., 2017). For instance, genes important for hair shaft differentiation and development (Table S16 ; Lin et al., 2009), keratinization (Pinkus, Iwasaki, & Mishima, 1981), and lipid‐related functions (Tables S12 and S15) (Lin et al., 2009) were all enriched for late upregulation. Finally, catagen induction is signaled by the enrichment of retinoid related pathways (Table S15) and late molt upregulation of STRA6 (annotated to “regulation of behavior”; Table S12), responsible for the uptake of vitamin A into the cell which seems to be important for catagen initiation (Everts, 2012; Foitzik, Spexard, Nakamura, Halsner, & Paus, 2005).

Despite the natural noise of expression studies performed in wild specimens, we were able to recover strong signatures that associate visible markers of molt progression (i.e., change in hair color) with the underlying hair cycle stages of anagen and catagen. Among this and the snowshoe hare study (Ferreira et al., 2017), we found significant overlaps among genes expressed in the molting skin (75%), among genes with stronger expression early in the molt (25% overlap), and among candidate genes for regulation of molt functions. These results suggest conservatism of the pathways and genes involved in skin processes, molt phenology, hair cycle activation (early molt) and regulation (candidate genes) in mountain and snowshoe hares, despite differences in the sampled species and molting season. The conservation of regulatory processes underlying pelage molts in these species is expected given the general conservatism of the hair growth cycle mechanism in mammals (Stenn & Paus, 2001) and the recent divergence between mountain and snowshoe hares (4 million years (Matthee et al., 2004); ~1% uncorrected pairwise sequence divergence, (Jones et al., 2018)) (Uebbing et al., 2016; Voolstra, Tautz, Farbrother, Eichinger, & Harr, 2007). The developmental framework used here should be transferable to other seasonally molting species, including other species that have evolved seasonal camouflage, thus contributing to the general understanding of the regulatory underpinnings of this adaptive phenotype across taxa (Manceau, Domingues, Linnen, Rosenblum, & Hoekstra, 2010; Rosenblum, Parent, & Brandt, 2014).

4.2. Candidate genes associated with circadian rhythm and pigmentation

To better understand the mechanistic regulation of brown to white autumn color change in mountain hares, we also inspected the expression of candidate genes associated with circadian rhythm regulation and pigmentation. While circadian gene expression is usually associated with late telogen and early anagen (Geyfman et al., 2015; Lin et al., 2009), in our work only one gene with “circadian rhythm” GO term annotation, MC3R, was found differentially expressed and upregulated toward the end of molt (Figure 3b; Table S16). MC3R is not a central clock gene and rather acts as a reinforcer of rhythm behavior (reviewed in Girardet, Begriche, Ptitsyn, Koza, and Butler (2014)) by regulating the expression of clock genes in response to nutrient availability and light cues (Sutton et al., 2008). Likewise, genes related to rhythmic gene expression in response to metabolic cues (Helfer et al., 2012; Oh et al., 2016; Slominski et al., 2003; Xu et al., 2009) were differentially expressed and upregulated late in molt (e.g., APOE, STRA6, and HTR2B in Table S12). Collectively, these genes may integrate the response to external or internal cues, such as food intake or nutritional condition, and cyclic phenotypic changes in the body, such as the molt cycle. In snowshoe hares, most of the differentially expressed circadian rhythm genes were upregulated early in the molt, and only one is known to be involved in the regulation of the circadian clock in response to nutrient intake (GFPT1 in Figure 3b; Ferreira et al., 2017).

Given that the autumn molt of mountain hares is accompanied by growing hair changing from brown to white, we looked for expression changes of known pigmentation genes. We first focus on the pigmentation gene ASIP (Lu et al., 1994; Manceau, Domingues, Mallarino, & Hoekstra, 2011), which tracks the onset of melanogenesis during the spring molt of the snowshoe hare (Ferreira et al., 2017) and determines the expression of alternative winter coat colors in snowshoe and mountain hares (Giska et al., 2019; Jones et al., 2018). ASIP’s expression trended higher early in the molt (Figure S6), consistent with activation of melanogenesis (Slominski & Paus, 1993). However, there was no significant difference in ASIP’s expression among stages, both using RNA‐sequencing and targeted qPCR data (Figure S6) and, therefore, the relevance of the trend needs further investigation. The fact that ASIP is upregulated over a narrow window of 2–3 days during the hair growth cycle (Bultman, Michaud, & Woychik, 1992; Miller et al., 1993) may explain failure to detect the expression peak in some individuals.

Additionally, we detected early upregulation of four genes involved in pigmentation (EN1, SPARC, and EDAR) or melanocyte differentiation (MEF2C) (Figure 3b; Table S16). SPARC and EDAR play major roles in hair follicle and skin related processes (Adhikari et al., 2016; Fliniaux et al., 2008; Millar, 2002; Strandjord et al., 1999). MEF2C is involved in the differentiation of follicular melanocytes of the new hair cycle (Tobin, Slominski, Botchkarev, & Paus, 1999), and its loss results in decreased pigmentation (Agarwal et al., 2011). EN1 is involved in establishing the boundary of expression of dorsal genes (Candille et al., 2004; Cygan, Johnson, & McMahon, 1997; Davis, Holmyard, Millen, & Joyner, 1991) and limiting the migration and proliferation of melanocytes in the ventral limb (Cygan et al., 1997). EN1 typically shows ventral expression, and thus, the upregulation early in the molt in our dorsal samples appears surprising. It is possible that transient expression of EN1 contributes to winter whitening, by allowing dorsal expression of the ventral isoform of ASIP. However, the expression of ventral ASIP isoform in the dorsal skin was negligible (Figure S6c; Table S17). Thus, as with snowshoe hares (Jones et al., 2018), expression of ventral ASIP isoform does not appear to be involved in the evolution of dorsal winter whitening in mountain hares.

There was no overlap of pigmentation candidate genes differentially expressed in mountain hares and snowshoe hares (Figure 3b). This may be due to season or species‐specific differences in expression. However, since we found a general conservation of the genic regulation of the early molt across species, we hypothesize that this difference may be related with the mechanisms of production of brown or white hairs across seasons. For example, we found upregulation of genes related with melanocyte differentiation early in autumn molt, which were not detected in spring molt (Figure 3b; Table S16; Ferreira et al., 2017), suggesting that the production of winter white coats may involve an impact on melanocyte differentiation. Whether this hypothesis can indeed be generalized from mountain to snowshoe hare must await the study of the autumn molt in the latter species and a better understanding of the genetic basis of the evolution of winter white coats in Lepus.

5. CONCLUSIONS AND SIGNIFICANCE

Coat color change is a complex adaptation to environmental seasonality that involves the regulation of circadian response, hair shedding and regrowth, and pigmentation. The joint analysis of the transcriptional landscape of the coat color molt in snowshoe and mountain hares shows that hair color in molting animals broadly delimits the underlying developmental timeline of seasonal molts. Particularly, early molt is the most transcriptionally active stage, marking the activation of circadian clock genes and inducing hair growth and pigmentation. As this inductive wave recedes, color change becomes noticeable, and in late molt a second transcription wave occurs, marking the transition from anagen to catagen. We now show that, in autumn, the upregulation of genes related with integration of nutritional cues into the circadian clock late in molt may further adjust molt rate in response to nutritional condition. This comprehensive mechanistic baseline will be a useful tool to guide functional molecular studies on the control of seasonal color change in hares and other species and to interpret quantitative genotype‐to‐phenotype studies on the genetic basis of winter coat color determination.

CONFLICT OF INTERESTS

The authors declare no conflict of interests.

AUTHOR CONTRIBUTIONS

JM‐F, JMG, KH, LSM, and PCA conceived the study; JM‐F, JMG, and MSF designed the work; HJ and KH organized and performed field work; CMC and LF performed laboratory work; MSF analyzed data; MSF and IG analyzed qPCR data; MSF, JM‐F, and JMG wrote the manuscript with comments from the other authors; all authors read, revised, and approved the final manuscript.

Supporting information

 

Click here for additional data file. (500.2KB, pdf)

 

Click here for additional data file. (80.4KB, pdf)

 

Click here for additional data file. (816.4KB, xlsx)

 

Click here for additional data file. (109.9KB, pdf)

ACKNOWLEDGMENTS

Financial support was obtained from Fundação para a Ciência e a Tecnologia (FCT), project grant "CHANGE" (reference PTDC/BIA‐EVF/1624/2014, funded by Portuguese National Funds through FCT) to JM‐F, and from the National Science Foundation (grant OIA‐1736249) to JMG. MSF was supported by POPH‐QREN funds from the European Social Fund (ESF) and Portuguese MCTES/FCT (PD/BD/108131/2015 PhD grant in the scope of BIODIV PhD program at Faculty of Sciences, University of Porto). JM‐F was supported by FCT Investigator grant IF/00033/2014/CP1256/CT0005 (POPH‐QREN funds from ESF and Portuguese MCTES/FCT). Additional support was provided by the National Science Foundation (DEB‐1907022) to JMG and LSM, by the National Science Foundation (DEB‐1743871) to LSM, by NORTE2020, PORTUGAL2020, through the European Regional Development Fund—ERDF (NORTE‐01‐0145‐FEDER‐000007), by COMPETE2020, PORTUGAL2020, and ERDF (POCI‐01‐0145‐FEDER‐022184), and by Portugal—United States of America Research Networks Program funds from Fundação Luso‐Americana para o Desenvolvimento (FLAD) to PCA and MSF. Instrumentation, laboratory, and computational support was provided by CIBIO NEW‐GEN sequencing platform, supported by European Union's Seventh Framework Program for research, technological development and demonstration under grant agreement no. 286431, the University of Montana Genomics Core, supported by a grant from the M.J. Murdock Charitable Trust, and the Vincent J. Coates Genomics Sequencing Laboratory at UC Berkeley, supported by NIH S10 Instrumentation Grants S10RR029668 and S10RR027303. We thank Pierre Boursot for providing access to the Montpellier Bioinformatics Biodiversity platform supported by the LabEx CeMEB, an ANR "Investissements d'avenir" program (ANR‐10‐LABX‐04‐01) for the de novo assembly of the transcriptome, and João Pedro Marques for bioinformatic support. We thank Fabiana Neves, from CIBIO‐InBIO's IMED group for primers to amplify ACTB gene. We are grateful to René Gadient and local hunters in Grisons for their tremendous effort in collecting samples under Alpine conditions.

Ferreira MS, Alves PC, Callahan CM, et al. Transcriptomic regulation of seasonal coat color change in hares. Ecol Evol. 2020;10:1180–1192. 10.1002/ece3.5956

Funding information

Fundação para a Ciência e a Tecnologia (Portuguese National Funds), Grant/Award Number: PTDC/BIAEVF/1624/2014; National Science Foundation, Grant/Award Number: OIA‐1736249, DEB‐1907022 and DEB‐1743871; Fundação para a Ciência e a Tecnologia (POPH‐QREN funds from the European Social Fund and Portuguese MCTES), Grant/Award Number: PD/BD/108131/2015 and IF/00033/2014/CP1256/CT0005; NORTE2020 and PORTUGAL2020 through the European Regional Development Fund, Grant/Award Number: NORTE‐01‐0145‐FEDER‐000007; COMPETE2020 and PORTUGAL2020 through the European Regional Development Fund, Grant/Award Number: POCI‐01‐0145‐FEDER‐022184; Fundação Luso‐Americana para o Desenvolvimento; European Union's Seventh Framework Program, Grant/Award Number: 286431; M.J. Murdock Charitable Trust, Grant/Award: University of Montana Genomic Core; NIH Instrumentation, Grant/Award Number: S10RR029668 and S10RR027303; LabEx CeMEB MBB (ANR “Investissements d'avenir” program), Grant/Award Number: ANR‐10‐LABX‐04‐01.

DATA AVAILABILITY STATEMENT

Raw reads were deposited at the Sequence Read Archive (SRA) under BioProject PRJNA590529. Raw and filtered transcriptome assembly, the blast output annotations of the transcriptome to the ENSEMBL 92 rabbit (OryCun2.0) and mouse (GRCm38) references and gene and isoform level counts were deposited at Dryad (https://doi.org/10.5061/dryad.x0k6djhfh). The pipeline and custom scripts used to generate and annotate the assembly, generate bam files and do a Gene Ontology analysis with a custom annotation are available at https://github.com/evochange/coat_color_change_transcriptomics.

REFERENCES

  1. Adhikari, K. , Fontanil, T. , Cal, S. , Mendoza‐Revilla, J. , Fuentes‐Guajardo, M. , Chacón‐Duque, J.‐C. , … Ruiz‐Linares, A. (2016). A genome‐wide association scan in admixed Latin Americans identifies loci influencing facial and scalp hair features. Nature Communications, 7(10815), 1–12. 10.1038/ncomms10815 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Agarwal, P. , Verzi, M. P. , Nguyen, T. , Hu, J. , Ehlers, M. L. , McCulley, D. J. , … Black, B. L. (2011). The MADS box transcription factor MEF2C regulates melanocyte development and is a direct transcriptional target and partner of SOX10. Development, 138(12), 2555–2565. 10.1242/dev.056804. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Ahmed, M. , & Kim, D. R. (2018). pcr: An R package for quality assessment, analysis and testing of qPCR data. PeerJ, 6, e4473 10.7717/peerj.4473 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Allain, D. , & Rougeot, J. (1980). Induction of autumn molt in mink (Mustela vison Peale and Beauvois) with melatonin. Reproduction Nutrition Development, 20(1A), 192–201. [DOI] [PubMed] [Google Scholar]
  5. Alvarez, M. , Schrey, A. W. , & Richards, C. L. (2014). Ten years of transcriptomics in wild populations: What have we learned about their ecology and evolution? Molecular Ecology, 24, 710–725. 10.1111/mec.13055 [DOI] [PubMed] [Google Scholar]
  6. Angerbjörn, A. et al (2018). Lepus timidus, Mountain hare In Smith A. T. (Ed.), Lagomorphs: Pikas, rabbits, and hares of the world (pp. 212–216). Baltimore: Johns Hopkins University Press. [Google Scholar]
  7. Atmeh, K. , Andruszkiewicz, A. , & Zub, K. (2018). Climate change is affecting mortality of weasels due to camouflage mismatch. Scientific Reports, 8(1), 7648 10.1038/s41598-018-26057-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Balsalobre, A. (2002). Clock genes in mammalian peripheral tissues. Cell and Tissue Research, 309(1), 193–199. 10.1007/s00441-002-0585-0 [DOI] [PubMed] [Google Scholar]
  9. Barrett, R. D. H. , & Hoekstra, H. E. (2011). Molecular spandrels: Tests of adaptation at the genetic level. Nature Review Genetics, 12(11), 767–780. 10.1038/nrg3015 [DOI] [PubMed] [Google Scholar]
  10. Bauer, S. , Grossmann, S. , Vingron, M. , & Robinson, P. N. (2008). Ontologizer 2.0 – a multifunctional tool for GO term enrichment analysis and data exploration. Bioinformatics, 24(14), 1650–1651. 10.1093/bioinformatics/btn250 [DOI] [PubMed] [Google Scholar]
  11. Bergengren, A. (1969). On genetics, evolution and history of distribution of the heath‐hare, a distinct population of the Arctic hare, Lepus timidus. Lin. Viltrevy: Swedish Wildlife, 6(5), 381–460. [Google Scholar]
  12. Bissonnette, T. , & Bailey, E. (1944). Experimental modification and control of molts and changes of coat‐color in weasels by controlled lighting. Annals of the New York Academy of Sciences, 45(6), 221–260. 10.1111/j.1749-6632.1944.tb47953.x. [DOI] [Google Scholar]
  13. Botchkarev, V. A. , & Kishimoto, J. (2003). Molecular control of epithelial‐mesenchymal interactions during hair follicle cycling. Journal of Investigative Dermatology Symposium Proceedings, 8(1), 46–55. 10.1046/j.1523-1747.2003.12171.x [DOI] [PubMed] [Google Scholar]
  14. Bradshaw, W. E. , & Holzapfel, C. M. (2007). Evolution of animal photoperiodism. Annual Review of Ecology, Evolution, and Systematics, 38, 1–25. 10.1146/annurev.ecolsys.37.091305.110115 [DOI] [Google Scholar]
  15. Bultman, S. J. , Michaud, E. J. , & Woychik, R. P. (1992). Molecular characterization of the mouse agouti locus. Cell, 71(7), 1195–1204. 10.1016/S0092-8674(05)80067-4 [DOI] [PubMed] [Google Scholar]
  16. Candille, S. I. , Van Raamsdonk, C. D. , Chen, C. , Kuijper, S. , Chen‐Tsai, Y. , Russ, A. , … Barsh, G. S. (2004). Dorsoventral patterning of the mouse coat by Tbx15. PLoS Biology, 2(1), 0030–0042. 10.1371/journal.pbio.0020003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Crocker, M. , & Cattanach, B. M. (1979). The genetics of Sleek: A possible regulatory mutation of the tabby‐crinkled‐downless syndrome. Genetics Research, 34, 231–238. [DOI] [PubMed] [Google Scholar]
  18. Cygan, J. A. , Johnson, R. L. , & McMahon, A. P. (1997). Novel regulatory interactions revealed by studies of murine limb pattern in Wnt‐7a and En‐1 mutants. Development, 124(24), 5021–5032. [DOI] [PubMed] [Google Scholar]
  19. Dalziel, A. C. , Rogers, S. M. , & Schulte, P. M. (2009). Linking genotypes to phenotypes and fitness: How mechanistic biology can inform molecular ecology. Molecular Ecology, 18(24), 4997–5017. 10.1111/j.1365-294X.2009.04427.x [DOI] [PubMed] [Google Scholar]
  20. Davis, C. A. , Holmyard, D. P. , Millen, K. J. , & Joyner, A. L. (1991). Examining pattern formation in mouse, chicken and frog embryos with an En‐specific antiserum. Development, 111, 287–298. Available at: http://dev.biologists.org/content/develop/111/2/287.full.pdf (Accessed: 13 December 2018). [DOI] [PubMed] [Google Scholar]
  21. DiTommaso, T. , Jones, L. K. , Cottle, D. L. , Gerdin, A.‐K. , Vancollie, V. E. , Watt, F. M. , … Smyth, I. M. (2014). Identification of genes important for cutaneous function revealed by a large scale reverse genetic screen in the mouse. PLoS Genetics, 10(10), 1004705 10.1371/journal.pgen.1004705 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Duncan, M. , & Goldman, B. (1984). Hormonal regulation of the annual pelage color cycle in the Djungarian hamster, Phodopus sungorus. II. Role of prolactin. Journal of Experimental Zoology, 230(1), 97–103. 10.1002/jez.1402300113 [DOI] [PubMed] [Google Scholar]
  23. Everts, H. B. (2012). Endogenous retinoids in the hair follicle and sebaceous gland, Biochimica et Biophysica Acta (BBA) – Molecular and Cell Biology of Lipids . Elsevier, 1821(1), 222–229. 10.1016/J.BBALIP.2011.08.017 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Ferreira, M. S. , Alves, P. C. , Callahan, C. M. , Marques, J. P. , Mills, L. S. , Good, J. M. , & Melo‐Ferreira, J. (2017). The transcriptional landscape of seasonal coat color molt in the snowshoe hare. Molecular Ecology, 26(May), 4173–4185. 10.1111/mec.14177 [DOI] [PubMed] [Google Scholar]
  25. Fliniaux, I. , Mikkola, M. L. , Lefebvre, S. , & Thesleff, I. (2008). Identification of dkk4 as a target of Eda‐A1/Edar pathway reveals an unexpected role of ectodysplasin as inhibitor of Wnt signalling in ectodermal placodes. Developmental Biology, 320(1), 60–71. 10.1016/j.ydbio.2008.04.023 [DOI] [PubMed] [Google Scholar]
  26. Foitzik, K. , Spexard, T. , Nakamura, M. , Halsner, U. , & Paus, R. (2005). Towards dissecting the pathogenesis of retinoid‐induced hair loss: All‐trans retinoic acid induces premature hair follicle regression (Catagen) by upregulation of transforming growth factor‐β2 in the dermal papilla. Journal of Investigative Dermatology. Elsevier, 124(6), 1119–1126. 10.1111/J.0022-202X.2005.23686.X [DOI] [PubMed] [Google Scholar]
  27. Geyfman, M. , Gordon, W. , Paus, R. , & Andersen, B. (2012). Identification of telogen markers underscores that telogen is far from a quiescent hair cycle phase. The Journal of Investigative Dermatology, 132, 721–724. 10.1038/jid.2011.367 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Geyfman, M. , Plikus, M. V. , Treffeisen, E. , Andersen, B. , & Paus, R. (2015). Resting no more: Re‐defining telogen, the maintenance stage of the hair growth cycle. Biological Reviews, 90(4), 1179–1196. 10.1111/brv.12151 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Girardet, C. , Begriche, K. , Ptitsyn, A. , Koza, R. A. , & Butler, A. A. (2014). Unravelling the mysterious roles of melanocortin‐3 receptors in metabolic homeostasis and obesity using mouse genetics. International Journal of Obesity Supplements, 4(S1), S37–S44. 10.1038/ijosup.2014.10 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Giska, I. , Farelo, L. , Pimenta, J. , Seixas, F. A. , Ferreira, M. S. , Marques, J. P. , … Melo‐Ferreira, J. (2019) Introgression drives repeated evolution of winter coat color polymorphism in hares. Proceedings of the National Academy of Sciences, USA, 116(48), 24150–24156. 10.1073/pnas.1910471116 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Gordon, A. (2011). Illumina CASAVA‐1.8 FASTQ Filter. Available at: http://cancan.cshl.edu/labmembers/gordon/fastq_illumina_filter/.
  32. Grabherr, M. G. , Haas, B. J. , Yassour, M. , Levin, J. Z. , Thompson, D. A. , Amit, I. , … Regev, A. (2011). Full‐length transcriptome assembly from RNA‐Seq data without a reference genome. Nature Biotechnology, 29(7), 644–652. 10.1038/nbt.1883 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Hansen, K. D. , Brenner, S. E. , & Dudoit, S. (2010). Biases in Illumina transcriptome sequencing caused by random hexamer priming. Nucleic Acids Research, 38(12), e131 10.1093/nar/gkq224 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Hart, J. S. (1956). Seasonal changes in insulation of the fur. Canadian Journal of Zoology, 53–57. 10.1139/z56-007 [DOI] [Google Scholar]
  35. Hart, S. N. , Therneau, T. M. , Zhang, Y. , Poland, G. A. , & Kocher, J.‐P. (2013). Calculating sample size estimates for RNA sequencing data. Journal of Computational Biology, 20(12), 970–978. 10.1089/cmb.2012.0283 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Helfer, G. , Ross, A. W. , Russell, L. , Thomson, L. M. , Shearer, K. D. , Goodman, T. H. , … Morgan, P. J. (2012). Photoperiod regulates vitamin A and Wnt/β‐Catenin signaling in F344 rats. Endocrinology, 153(2), 815–824. [DOI] [PubMed] [Google Scholar]
  37. Helm, B. , Ben‐Shlomo, R. , Sheriff, M. J. , Hut, R. A. , Foster, R. , Barnes, B. M. , & Dominoni, D. (2013). Annual rhythms that underlie phenology: Biological time‐keeping meets environmental change. Proceedings of the Royal Society B: Biological Sciences, 280(20130016), 20130016 10.1098/rspb.2013.0016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Hewson, R. (1958). Moults and winter whitening in the mountain hare lepus timidus scoticus, hilzheimer. Proceedings of the Zoological Society of London, 131(1), 99–108. 10.1111/j.1096-3642.1958.tb00635.x [DOI] [Google Scholar]
  39. Hoffmann, K. (1978). Effects of short photoperiods on puberty, growth and moult in the Djungarian hamster (Phodopus sungorus). Journal of Reproduction and Fertility, 54, 29–35. Retrieved from: http://www.reproduction-online.org/content/54/1/29.short (Accessed: 14 August 2014). [DOI] [PubMed] [Google Scholar]
  40. Jones, M. R. , Mills, L. S. , Alves, P. C. , Callahan, C. M. , Alves, J. M. , Lafferty, D. J. R. , … Good, J. M. (2018). Adaptive introgression underlies polymorphic seasonal camouflage in snowshoe hares. Science, 360(6395), 1355–1358. 10.1126/science.aar5273 [DOI] [PubMed] [Google Scholar]
  41. Lin, K. K. , Chudova, D. , Hatfield, G. W. , Smyth, P. , & Andersen, B. (2004). Identification of hair cycle‐associated genes from time‐course gene expression profile data by using replicate variance. Proceedings of the National Academy of Sciences, USA, 101(45), 15955–15960. 10.1073/pnas.0407114101s [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Lin, K. K. , Kumar, V. , Geyfman, M. , Chudova, D. , Ihler, A. T. , Smyth, P. , … Andersen, B. (2009). Circadian clock genes contribute to the regulation of hair follicle cycling. PLoS Genetics, 5(7), e1000573 10.1371/journal.pgen.1000573 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Lincoln, G. A. , Clarke, I. J. , Hut, R. A. , & Hazlerigg, D. G. (2006). Characterizing a mammalian circannual pacemaker. Science, 314(5807), 1941–1944. 10.1126/science.1132009 [DOI] [PubMed] [Google Scholar]
  44. Ling, J. K. (1972). Adaptive functions of vertebrate molting cycles. American Zoologist, 12(1), 77–93. Retrieved from: http://www.jstor.org/stable/3881733. [Google Scholar]
  45. Logan, A. , & Weatherhead, B. (1980). Post‐tyrosinase inhibition of melanogenesis by melatonin in hair follicles in vitro. The Journal of Investigative Dermatology, 74(1), 47–50. Retrieved from: http://www.ncbi.nlm.nih.gov/pubmed/6766170 [DOI] [PubMed] [Google Scholar]
  46. Loomis, C. A. , Kimmel, R. A. , Tong, C. X. , Michaud, J. , & Joyner, A. L. (1998). Analysis of the genetic pathway leading to formation of ectopic apical ectodermal ridges in mouse Engrailed‐1 mutant limbs. Development, 125(6), 1137–1148. Retrieved from: http://dev.biologists.org/content/125/6/1137.short (Accessed: 13 December 2018). [DOI] [PubMed] [Google Scholar]
  47. Lu, D. , Willard, D. , Patel, I. R. , Kadwell, S. , Overton, L. , Kost, T. , … Cone, R. D. (1994). Agouti protein is an antagonist of the melanocyte‐stimulating‐hormone receptor. Nature, 371(6500), 799–802. 10.1038/371799a0 [DOI] [PubMed] [Google Scholar]
  48. Manceau, M. , Domingues, V. S. , Linnen, C. R. , Rosenblum, E. B. , & Hoekstra, H. E. (2010). Convergence in pigmentation at multiple levels: mutations, genes and function. Philosophical Transactions of the Royal Society B: Biological Sciences, 365(1552), 2439–2450. 10.1098/rstb.2010.0104 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Manceau, M. , Domingues, V. S. , Mallarino, R. , & Hoekstra, H. E. (2011). The developmental role of agouti in color pattern evolution. Science, 331(6020), 1062–1065. 10.1126/science.1200684 [DOI] [PubMed] [Google Scholar]
  50. Matthee, C. , Van Vuuren, B. , Bell, D. , & Robinson, T. (2004). A molecular supermatrix of the rabbits and hares (Leporidae) allows for the identification of five intercontinental exchanges during the miocene. Systematic Biology, 53(3), 433–447. 10.1080/10635150490445715 [DOI] [PubMed] [Google Scholar]
  51. Mayer, U. , Nischt, R. , Pöschl, E. , Mann, K. , Fukuda, K. , Gerl, M. , … Fässler, R. (1993). A single EGF‐like motif of laminin is responsible for high affinity nidogen binding. The EMBO Journal, 12(5), 1879–1885. 10.1093/emboj/19.15.3990 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. McCarthy, D. J. , Chen, Y. , & Smyth, G. K. (2012). Differential expression analysis of multifactor RNA‐Seq experiments with respect to biological variation. Nucleic Acids Research, 40(10), 4288–4297. 10.1093/nar/gks042 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Melo‐Ferreira, J. , Boursot, P. , Carneiro, M. , Esteves, P. J. , Farelo, L. , & Alves, P. C. (2012). Recurrent introgression of mitochondrial DNA among hares (Lepus spp.) revealed by species‐tree inference and coalescent simulations. Systematic Biologyology, 61(3), 367–381. 10.1093/sysbio/syr114 [DOI] [PubMed] [Google Scholar]
  54. Millar, S. E. (2002). Molecular mechanisms regulating hair follicle development. Journal of Investigative Dermatology, 118(2), 216–225. 10.1046/j.0022-202x.2001.01670.x [DOI] [PubMed] [Google Scholar]
  55. Miller, M. W. , Duhl, D. M. J. , Vrieling, H. , Cordes, S. P. , Ollmann, M. M. , Winkes, B. M. , & Barsh, G. S. (1993). Cloning of the mouse agouti gene predicts a secreted protein ubiquitously expressed in mice carrying the lethal yellow mutation. Genes & Development, 7(3), 454–467. 10.1101/gad.7.3.454 [DOI] [PubMed] [Google Scholar]
  56. Mills, L. S. , Bragina, E. , Kumar, A. V. , Zimova, M. , Feltner, J. , Davis, B. , … Fay, K. (2018). Winter coat color polymorphisms identify global hotspots for evolutionary rescue from climate change. Science (New York, N.Y.), 359(6379), 1033–1036. [DOI] [PubMed] [Google Scholar]
  57. Mills, L. S. , Zimova, M. , Oyler, J. , Running, S. , Abatzoglou, J. T. , & Lukacs, P. M. (2013). Camouflage mismatch in seasonal coat color due to decreased snow duration. Proceedings of the National Academy of Sciences of the USA, 110(28), 7360–7365. 10.1073/pnas.1311224110 [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Helm, B. , & Stevenson, T. J. (2014). Circannual rhythms: History, present, challenges, future directions In Numata H., & Helm B. (eds) Annual, lunar, and tidal clocks (pp. 203–225) Tokyo Heidelberg New York Dordrecht London: Springer. [Google Scholar]
  59. Oh, E. , Park, J. , Lee, J. , Myung, C. , Kim, S. , Chang, S. , … Hwang, J. S. (2016) A Novel Role of Serotonin Receptor 2B Agonist as an Anti‐Melanogenesis Agent. International Journal of Molecular Sciences, 17(4), 546– 10.3390/ijms17040546 [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Parakkal, P. F. (1969). Role of macrophages in collagen resorption during hair growth cycle. Journal of Ultrastructure Research, 29(3), 210–217. 10.1016/S0022-5320(69)90101-4 [DOI] [PubMed] [Google Scholar]
  61. Pinkus, H. , Iwasaki, T. , & Mishima, Y. (1981). Outer root sheath keratinization in anagen and catagen of the mammalian hair follicle. A seventh distinct type of keratinization in the hair follicle: trichilemmal keratinization. Journal of Anatomy, 133(Pt 1), 19–35. Available at: http://www.ncbi.nlm.nih.gov/pubmed/6172407 (Accessed 12 December 2018). [PMC free article] [PubMed] [Google Scholar]
  62. Porter, R. M. , Corden, L. D. , Lunny, D. P. , Smith, F. J. D. , Lane, E. B. , & McLean, W. H. I. (2001). Keratin K6irs is specific to the inner root sheath of hair follicles in mice and humans. British Journal of Dermatology, 145(4), 558–568. 10.1046/j.1365-2133.2001.04463.x [DOI] [PubMed] [Google Scholar]
  63. Robinson, M. D. , McCarthy, D. J. , & Smyth, G. K. (2010). edgeR: A Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics, 26(1), 139–140. 10.1093/bioinformatics/btp616 [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Rosenblum, E. B. , Parent, C. E. , & Brandt, E. E. (2014). The molecular basis of phenotypic convergence. Annual Review of Ecology, Evolution, and Systematics, 45(1), 203–226. 10.1146/annurev-ecolsys-120213-091851 [DOI] [Google Scholar]
  65. Schai‐Braun, S. C. , & Hackländer, K. (2016). Mountain hare Lepus timidus In Wilson D. E., Lacher T. E., & Mittermeier R. A. (Eds.), Handbook of the mammals of the world – Lagomorphs and rodents I. Lynx, Vol. 6 (pp. 145–147). [Google Scholar]
  66. Schlake, T. , Beibel, M. , Weger, N. , & Boehm, T. (2004). Major shifts in genomic activity accompany progression through different stages of the hair cycle. Gene Expression Patterns: GEP, 4(2), 141–152. 10.1016/j.modgep.2003.09.009 [DOI] [PubMed] [Google Scholar]
  67. Schmidt, B. , & Horsley, V. (2012). Unravelling hair follicle‐adipocyte communication. Experimental Dermatology, 21(11), 827–830. 10.1111/exd.12001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  68. Schwartz, C. , & Andrews, M. T. (2014). Circannual transitions in gene expression: Lessons from seasonal adaptations. Current Topics in Development Biology, 105, 247–273. 10.1016/B978-0-12-396968-2.00009-9.Circannual [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. Shipman, M. , Chase, H. B. , & Montagna, W. (1955). Glycogen in skin of the mouse during cycles of hair growth. Experimental Biology and Medicine, 88(3), 449–451. 10.3181/00379727-88-21615 [DOI] [PubMed] [Google Scholar]
  70. Slominski, A. , & Paus, R. (1993). Melanogenesis is coupled to murine anagen: toward new concepts for the role of melanocytes and the regulation of melanogenesis in hair growth. Journal of Investigative Dermatology, 101(1), 90S–97S. Retrieved from: http://www.nature.com/jid/journal/v101/n1s/full/5611565a.html (Accessed: 14 August 2014). [DOI] [PubMed] [Google Scholar]
  71. Slominski, A. , Pisarchik, A. , Zbytek, B. , Tobin, D. J. , Kauser, S. , & Wortsman, J. (2003). Functional activity of serotoninergic and melatoninergic systems expressed in the skin. Journal of Cellular Physiology, 196(1), 144–153. 10.1002/jcp.10287 [DOI] [PubMed] [Google Scholar]
  72. Slominski, A. , Tobin, D. J. , Shibahara, S. , & Wortsman, J. (2004). Melanin pigmentation in mammalian skin and its hormonal regulation. Physiological Reviews, 84(4), 1155–1228. 10.1152/physrev.00044.2003 [DOI] [PubMed] [Google Scholar]
  73. Smith‐Unna, R. D. , Boursnell, C. , Patro, R. , Hibberd, J. M. , & Kelly, S. (2016). TransRate: Reference free quality assessment of de novo transcriptome assemblies. Genome Research, 26, 1134–1144. 10.1101/gr.196469.115 [DOI] [PMC free article] [PubMed] [Google Scholar]
  74. Stark, H. J. , Breitkreutz, D. , Limat, A. , Bowden, P. , & Fusenig, N. E. (1987). Keratins of the human hair follicle: ‘Hyperproliferative’ keratins consistently expressed in outer root sheath cells in vivo and in vitro. Differentiation, 35(3), 236–248. 10.1111/j.1432-0436.1987.tb00174.x [DOI] [PubMed] [Google Scholar]
  75. Stenn, K. S. , & Paus, R. (2001). Controls of hair follicle cycling. Physiological Reviews, 81(1), 449–494. Retrieved from: http://physrev.physiology.org/content/81/1/449.abstract [DOI] [PubMed] [Google Scholar]
  76. Strandjord, T. P. , Madtes, D. K. , Weiss, D. J. , & Sage, E. H. (1999). Collagen accumulation is decreased in SPARC‐null mice with bleomycin‐induced pulmonary fibrosis. American Journal of Physiology‐Lung Cellular and Molecular Physiology, 277(3), L628–L635. 10.1152/ajplung.1999.277.3.L628 [DOI] [PubMed] [Google Scholar]
  77. Sutton, G. M. , Perez‐Tilve, D. , Nogueiras, R. , Fang, J. , Kim, J. K. , Cone, R. D. , … Butler, A. A. (2008). The Melanocortin‐3 receptor is required for entrainment to meal intake. Journal of Neuroscience, 28(48), 12946–12955. 10.1523/JNEUROSCI.3615-08.2008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  78. Tobin, D. J. , Slominski, A. , Botchkarev, V. , & Paus, R. (1999). The fate of hair follicle melanocytes during the hair growth cycle. Journal of Investigative Dermatology Symposium Proceedings, 4(3), 323–332. 10.1038/sj.jidsp.5640239 [DOI] [PubMed] [Google Scholar]
  79. Todd, E. V. , Black, M. A. , & Gemmell, N. J. (2016). The power and promise of RNA‐seq in ecology and evolution. Molecular Ecology, 25(6), 1224–1241. 10.1111/mec.13526 [DOI] [PubMed] [Google Scholar]
  80. Uebbing, S. , Künstner, A. , Mäkinen, H. , Backström, N. , Bolivar, P. , Burri, R. , … Ellegren, H. (2016). Divergence in gene expression within and between two closely related flycatcher species. Molecular Ecology, 25(9), 2015–2028. 10.1111/mec.13596 [DOI] [PMC free article] [PubMed] [Google Scholar]
  81. Visser, M. E. , Caro, S. P. , van Oers, K. , Schaper, S. V. , & Helm, B. (2010). Phenology, seasonal timing and circannual rhythms: Towards a unified framework. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences, 365(1555), 3113–3127. 10.1098/rstb.2010.0111 [DOI] [PMC free article] [PubMed] [Google Scholar]
  82. Voolstra, C. , Tautz, D. , Farbrother, P. , Eichinger, L. , & Harr, B. (2007). Contrasting evolution of expression differences in the testis between species and subspecies of the house mouse. Genome Research, 17(1), 42–49. 10.1101/gr.5683806 [DOI] [PMC free article] [PubMed] [Google Scholar]
  83. Xu, C. , Phd, C. L. , Hua, L. , Jin, H. , Yin, L. , Chen, S. , & Qian, R. (2009). Rhythm changes of clock genes, apoptosis‐related genes and atherosclerosis‐related genes in apolipoprotein E knockout mice. Canadian Journal of Cardiology, 25, 473 10.1016/S0828-282X(09)70122-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  84. Zimova, M. , Hackländer, K. , Good, J. M. , Melo‐Ferreira, J. , Alves, P. C. , & Mills, L. S. (2018). Function and underlying mechanisms of seasonal colour moulting in mammals and birds: what keeps them changing in a warming world? Biological Reviews, 93(3), 1478–1498. 10.1111/brv.12405 [DOI] [PubMed] [Google Scholar]
  85. Zimova, M. , Mills, L. , Lukacs, P. M. , & Mitchell, M. S. (2014). Snowshoe hares display limited phenotypic plasticity to mismatch in seasonal camouflage. Proceedings of the Royal Society B: Biological Sciences, 281(1782), 1–19. 10.1098/rspb.2014.0029 [DOI] [PMC free article] [PubMed] [Google Scholar]
  86. Zimova, M. , Mills, L. S. , & Nowak, J. J. (2016). High fitness costs of climate change induced camouflage mismatch in a seasonally colour moulting mammal. Ecology Letters, 19(3), 299–307. 10.1111/ele.12568 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

 

Click here for additional data file. (500.2KB, pdf)

 

Click here for additional data file. (80.4KB, pdf)

 

Click here for additional data file. (816.4KB, xlsx)

 

Click here for additional data file. (109.9KB, pdf)

Data Availability Statement

Raw reads were deposited at the Sequence Read Archive (SRA) under BioProject PRJNA590529. Raw and filtered transcriptome assembly, the blast output annotations of the transcriptome to the ENSEMBL 92 rabbit (OryCun2.0) and mouse (GRCm38) references and gene and isoform level counts were deposited at Dryad (https://doi.org/10.5061/dryad.x0k6djhfh). The pipeline and custom scripts used to generate and annotate the assembly, generate bam files and do a Gene Ontology analysis with a custom annotation are available at https://github.com/evochange/coat_color_change_transcriptomics.

Articles from Ecology and Evolution are provided here courtesy of Wiley

RESOURCES