Comparative Genomics Reveals Convergent Evolution Between Avivorous Bats (Ia io and Nyctalus aviator)

ABSTRACT

Investigating the genetic basis of dietary specialization can provide insights into the evolution of niche breadth. In this study, we employed comparative genomics to investigate the adaptive mechanisms enabling two bat species (Nyctalus aviator and Ia io) to shift from insectivory to seasonal bird consumption (avivorous bats). Our findings revealed adaptation related to immune response and lipid metabolism in avivorous bat species. Avivorous bats exhibit strong positive selection and convergent evolution in immune‐related genes, which are under heightened selective pressure compared to those of non‐avivorous bats. These species also display significantly fewer endogenous retroviral elements. These findings emphasized the significance of immune‐driven adaptive evolution in avivory. Additionally, our results showed that the dietary evolution of avivorous bats is accompanied by convergent evolution associated with the lipid metabolism. Notably, CEPT1, the upstream gene required for the activation of the PPAR pathway, underwent positive selection and convergence, which may have affected lipid metabolism. These adaptations may enable avivorous bat species to face the challenge of immune response and nutrition during dietary niche expansion. These findings not only provide comprehensive insights into the adaptive evolution driving the unique diet of avivorous bats but also offered novel perspectives on the molecular mechanisms underlying ecological niche evolution in a dietary context.

In this study, we employed comparative genomics to investigate the adaptive mechanisms enabling two bat species (Nyctalus aviator and Ia io) to shift from insectivory to seasonal bird consumption (avivorous bats). Our findings revealed adaptation related to immune response and lipid metabolism in avivorous bat species.

1. Introduction

Insights into the evolution of niche width are important for understanding the ecological and evolutionary responses of organisms to changes in the environment (Sexton et al. 2017). The dietary niche, a key component of the multidimensional ecological niche, provides insights into the evolutionary mechanisms of niche. Phenotypic variations may influence the evolution of the niche breadth of organisms. Dietary shifts may induce phenotypic variations in animals, including physiological (i.e., nutrient assimilation and energy metabolism), morphological, and behavioral variations (Palm and Thompson 2017), providing new ecological opportunities (Yoder et al. 2010). Molecular adaptations linked to dietary niche expansion help uncover the genetic mechanisms of niche breadth evolution. However, the genomic architecture underlying dietary niche expansion in distantly related taxa remains poorly characterized.

Chiropterans (bats) have evolved diverse phenotypic traits to utilize various dietary resources via adaptive radiation. Dietary diversification, including insectivory, carnivory, piscivory, frugivory, nectarivory, and sanguivory, is associated with physiological, biochemical, and morphological adaptations in bats (Altringham 2011). This dietary diversity makes bats ideal for studying the evolutionary mechanism of dietary niches. For example, in Desmodus rotundus, the loss of the RAB15 effector protein (REP15) gene is associated with an iron‐rich diet (Blumer et al. 2022), while a reduction in bitter taste receptor genes likely reflects adaptations to its exclusive blood‐feeding niche (Lu et al. 2021). Similarly, an increase in the activity of the trehalase (Treh) gene may contribute to the enhanced ability of insectivorous bat species to metabolize insects (Jiao et al. 2019). Investigating carnivory in bats could provide valuable insights into dietary niche evolution. Carnivorous bat species have evolved more elaborate traits associated with carnivory, compared to insectivorous bats; these traits include higher body mass, greater bite force, and distinctive cranial shapes (Gual‐Suárez and Medellín 2021). No carnivorous bat species feeds exclusively on terrestrial vertebrates (Norberg and Rayner 1987). Elucidating molecular adaptations underlying carnivory‐associated traits might offer novel insights into the evolution of dietary niches. However, research on the genetic basis of carnivory, a relatively uncommon dietary strategy among bats, remains limited. Although several studies have characterized the dietary, morphological, and behavioral adaptations of carnivorous bat species, more studies are needed to fully understand the genetic basis of this diet (Gual‐Suárez and Medellín 2021; Yue et al. 2024).

Three carnivorous bat species (Ia io, Nyctalus lasiopterus, and Nyctalus aviator) are specialized in preying on seasonally migratory birds and are referred to as “avivorous bats.” These avivorous bats are relatively large‐bodied, with convergent wing morphologies and echolocation call characteristics adapted for avian detection (Thabah et al. 2007). Ecologically, they primarily feed on insects in summer, while employing an aerial‐hawking strategy to prey on nocturnally migratory birds in spring and autumn. Compared to their closely related insectivorous bat relatives, their dietary niches have expanded from insect‐based to bird‐based diets. N. lasiopterus preys on birds from 8 families, encompassing 31 species, predominantly nocturnally migratory birds (Ibáñez et al. 2001). N. aviator can prey on birds from 7 families and 14 species, with an average prey mass of 13.3 ± 5.4 g (Ibáñez et al. 2021). DNA metabarcoding analysis on fecal sample revealed that I. io can prey on birds from 7 families and 22 species, primarily small (6 to 19 g) migratory birds (Gong, Shi, et al. 2021). Behavioral experiments showed that nocturnally active birds fail to detect I. io via effective visual and acoustic cues (Gong et al. 2024). Phylogenetically, the three species belong to two distinct genera (Ia and Nyctalus) within the family Vespertilionidae, representing independent evolutionary lineages. This phylogenetic divergence, combined with their shared avivorous traits, indicates that this specialized foraging behavior evolved via convergent evolution (Thabah et al. 2007). Avian predation requires predators to develop physiological adaptations related to energy demands and nutrient metabolism. We previously reported I. io has underwent molecular adaptations for dietary niche expansion, including energy metabolism, cardiovascular system, and nutrient metabolism (Gong et al. 2022). However, whether avivorous bats have undergone convergent evolution because of their shared aerial‐hawking strategy and similar avian prey types needs to be determined.

The unique evolutionary adaptations in bats, including dietary specializations, may also influence their immune responses. The outbreak of the coronavirus disease 2019 (COVID‐19) attracted public attention to the transmission of viruses among bat species. As natural reservoirs of numerous viruses, bats exhibit a unique immune tolerance that supports long‐term viral coexistence (Wu et al. 2023), and the direct and indirect interactions between bats and birds may contribute to the transmission of viruses (Nabi et al. 2021). For example, the influenza A virus (IAV) can infect bat cells, and H9 IAV has been identified in bats (Kandeil et al. 2019). Recently, the hemagglutinin (HA) gene of H19 IAV, isolated from a wild duck, exhibited characteristics of avian and bat influenza viruses (Fereidouni et al. 2023). Such cross‐species viral exposure also leaves traces in the host genome (Zheng et al. 2022). Endogenous retroviruses (ERVs) are genomic relics of ancient retroviral infections that integrate into the host genome during evolution (Lamba et al. 2024), and their abundance and diversity can reflect dynamics of long‐term viral exposure and host immune adaptation (Da Silva et al. 2024). For avivorous bats, which predate on birds, these unique ecological interactions increases the risk of cross‐species viral transmission (Nabi et al. 2021), potentially driving distinct adaptive changes in their immune systems and genomic stability. Investigating immune response in avivorous bats, characterized by their predation on birds, may help understand viral transmission mechanisms. However, research on immune adaptations of avivorous bats remains limited.

Addressing these questions clarifies the genetic basis of dietary niche evolution, particularly in distantly related taxa with convergent dietary niche expansion. In this study, we performed comparative genome analysis on the genomes of bat species with various diets, including two avivorous bats, I. io and N. aviator, as well as other bats representing various dietary types. A recent study identified convergent evolution in lipid metabolism‐related genes in two carnivorous bat species, Vampyrum spectrum and Lyroderma lyra (Yue et al. 2024). This evolutionary adaptation underpins their enhanced lipid metabolism, which enables these bat species to efficiently process high‐fat diets. Given that birds have a higher fat content than insects (Payne et al. 2016), we hypothesized that convergent evolution in lipid metabolism would be observed in these two avivorous bat species. Furthermore, given that both birds and bats are natural viral reservoirs, the immune response of these two avivorous bat species may have undergone convergent evolution.

2. Materials and Methods

2.1. Data Collection and Identification of Orthologous Genes

In this study, we used the high‐quality chromosome‐level genome assemblies of two avivorous bats (I. io and N. aviator) from our prior publication for subsequent analyses (Gong et al. 2022; Geng et al. 2024). The genomes of I. io and N. aviator have been characterized at the chromosome level. We also analyzed the genomes of other non‐avivorous bat species to enable comparisons across bats with diverse diets strategies. To ensure that the results were robust, we only selected chromosome‐level genome assemblies. Finally, 13 representative bat species formed the diverse set, including four Yinpterochiropterans (Rousettus aegyptiacus, Pteropus alecto, Rhinolophus ferrumequinum, and Hipposideros armiger) and nine Yangochiropterans (Phyllostomus discolor, D. rotundus, Artibeus jamaicensis, Molossus molossus, Miniopterus natalensis, Myotis myotis, Myotis davidii, Pipistrellus kuhlii, and Eptesicus fuscus). In addition, four outgroup species (Homo sapiens, Mus musculus, Felis catus, and Equus caballus) were incorporated into this study. All genomes were downloaded from NCBI, and detailed information on genome accession numbers and diets is listed in Table S1. The longest transcript isoforms for each gene were selected and translated to proteins. OrthoFinder v2.5.4 (Emms and Kelly 2019) were used to identify orthologous relationships among all protein‐coding genes across 19 species. From these results, we extracted two complementary datasets: (i) 9132 single‐copy orthologous genes—defined as genes present in exactly one copy in all 19 species (1:1:1 orthologs)—used for phylogenetic tree construction and selection analysis; and (ii) all gene families with variable copy numbers across species, used for gene family expansion/contraction analysis.

2.2. Phylogenetic Tree Construction

The phylogenetic relationship was determined using CDS sequences of orthologous genes. The single‐copy orthologous genes shared across 19 species were aligned using PRANK v.170703 with the parameter “+F −codon −termgap” (Löytynoja 2014) and refined by using Gblock v0.91b with the parameter “−b4 = 5 −b5 = h” (Talavera and Castresana 2007). Then, the gene trees of orthologous genes were constructed using IQ‐TREE v 2.2.2.6 (Minh et al. 2020) with default parameters. The species tree for 19 species was estimated using ASTRALIII under a multispecies coalescent model (Zhang et al. 2018). As ASTRALIII cannot estimate the length of the terminal branch, the ASTRALIII branch length information could not be used directly for subsequent analysis (Rabiee et al. 2019). The supergene matrix was constructed by concatenating the conserved alignments of all orthologous genes. Finally, the branch lengths of the species tree were re‐estimated based on the supergene matrix using IQ‐TREE with the parameter “−te.” The best model was identified using ModelFinder.

The divergence times of species were estimated using MCMCTree in PAML (Yang 2007), with the approximated likelihood determined using the GTR model (model = 7). The third codons extracted from the supergene matrix were used to estimate divergence times. First, the initial mean substitution rate was estimated by BaseML (Yang 2007). Based on this rate, the overall substitution rate to G (1, 20.83) and the rate‐drift parameter to G (1, 4.5) were set. The analysis was performed with 5 000 000 iterations as burn‐in and sampling at every 500 iterations, until 1 000 000 samples were collected. The root was set to 100 million years ago (MYA). Divergence time estimation incorporated four fossil calibration points, derived from TimeTree (Kumar et al. 2022) and study of Hao et al. (2024), as constrains, including Yinpterochiropterans and Yangochiropterans (57.9–64.0 MYA), R. ferrumequinum and H. armiger (39.0–47.5 MYA), horse and cat (65.8–74.5 MYA), and human and mouse (81.3–91.0 MYA). The MCMCTree software (version 4.9j) was used to perform the analysis twice to check for the convergence of results.

2.3. Gene Family Evolution

The results of the phylogenetic hierarchical orthologous groups (HOGs) were considered to analyze the evolution of the gene family using CAFE5 v5.1.0 (Mendes et al. 2021). Based on the method described in another study (Jebb et al. 2020), we used the CAFE software under single lambda (birth–death rate, ω) and gamma model (multiple evolution rate category, k). To fit the model better, some gene families were filtered based on the following rules: (i) gene family with one or more species that had more than 100 gene copies; (ii) gene family with single‐copy genes; (iii) some species in the gene family had no gene copy (Wang et al. 2020). After filtering, 3965 gene families with variable copy numbers across species were retained for expansion/contraction analysis. To decrease the possibility of errors stemming from genome assembly, we first estimated the error model. Subsequently, the models with various gamma category counts (by setting the parameter “k” from 2 to 5) were tested, and the results that exhibited the highest likelihood were selected. The highest likelihood was found using k = 5 categories (−lnL = 41734.8), with λ = 0.00233 and α = 0.87. Gene family expansion or contraction was inferred based on significant changes in gene copy number, with p < 0.05 considered statistically significant. For gene families exhibiting significant changes, the symbol of the human gene was used as the name of the family.

2.4. Selection Analysis

Conserved alignment of single‐copy orthologous genes was used for selection analysis, which was conducted using CodeML in the PAML package. We investigated the positively selected genes (PSGs) and rapidly evolved genes (REGs) in two avivorous bats to assess the adaptive evolution associated with an avivorous diet (Figure S1). The branch‐site model was used to identify PSGs on two distinct foreground branches: (i) I. io was selected as the foreground branch, with N. aviator excluded from background branches; (ii) N. aviator was designated as the foreground branch, with I. io excluded from the background branches. The likelihood ratio test (LRT) between alternative models (model = 2, NSsites = 2, fix_omega = 0, omega = 1) and null models (model = 2, NSsites = 2, fix_omega = 1, omega = 1) was evaluated with chi2 in PAML. According to Bayes Empirical Bayes (BEB) analysis (Yang 2005), the threshold of positively selected sites was more than 0.95. Genes exhibiting significant differences (p < 0.05) were considered to be PSGs. The branch model was used to identify REGs by adopting a strategy that was similar to the approach used for identifying PSGs. The LRT was also performed between the two‐ratio model (model = 2, NSsites = 0) and the one‐ratio model (model = 0, NSsites = 0). Genes showing significant differences and with higher ω values for foreground than background branches were considered REGs.

2.5. Convergent Evolution in Avivorous Bats

The convergent evolution of I. io and N. aviator was identified with two distinct methods (Figure S1): (i) based on the gene evolutionary rate; (ii) focused on specific amino acid sites. As adaptively driven convergence at specific amino acid sites is very rare, changes in convergent amino acids may only represent a fraction of the total adaptive changes that have occurred (Partha et al. 2019). To systematically examine convergent molecular adaptations in avivorous bats, we initially used RERconverge v0.3.0 (Kowalczyk et al. 2019) to identify genes exhibiting shared evolutionary rate shifts between I. io and N. aviator. All single‐copy genes were pooled to construct a master tree. To correct non‐specific factors affecting divergence in the branches, the relative evolutionary rates (RERs) for all branches of the tree for each gene were calculated by normalizing the branch lengths of the master tree. These RER values were used to determine the correlation between the binary phenotype “avivorous” terminal branches (I. io and N. aviator). A gene was considered to have undergone convergent evolution when the correlation was significant (p < 0.05). We also identified convergent evolution at specific amino acid sites following the method described in another study (Zou and Zhang 2015). To elaborate, for determining a convergent amino acid substitution, the following rules had to be met: (i) the amino acid sites of I. io and N. aviator were identical; (ii) the amino acid sites in I. io were different from those in the most recent ancestral node (I. io and E. fuscus); (iii) the amino acid sites in N. aviator were different from those in the most recent ancestral node (N. aviator and P. kuhlii). The expected number of convergent sites for each gene due to random substitution was calculated under the JTT‐f_site model. The Poisson test was conducted to determine if the observed number of convergent sites was significantly higher than the expected number. The p‐values were adjusted using the Benjamini–Hochberg method. Genes with FDR < 0.05 were considered to be convergent. The combined results of the two methods were used for subsequent analysis.

2.6. Functional Enrichment Analysis

The functional enrichment analyses of PSGs, REGs, evolutionary gene families, and convergent genes were performed using Metascape v3.5.20240101 (Zhou et al. 2019), with a significant threshold of p < 0.01. In this study, five databases were used to conduct the enrichment analysis in Metascape, including Gene Ontology Biological Processes, Reactome Gene Sets, KEGG Pathway, WikiPathways, and Canonical Pathways. Specific keywords can also be used to extract genes of interest from these databases. The protein–protein interaction networks of convergent genes were constructed using the STRING v11 database. The cutoff for the interaction score was 0.6. The genes in the networks were further clustered using the MCL method, and the inflation parameter was 1.5.

2.7. Annotation of Endogenous Retroviruses

The ERVs were identified from the genomes of all bat species using ERVin (https://github.com/strongles/ervin). The protein sequences of 15 representative retroviruses were downloaded from NCBI, including alpha‐, beta‐, delta‐, epsilon‐, gamma‐, lentivirus, and spumavirus. The complete gag, env, and pol protein sequences for each retrovirus were used in the analysis. Detailed information on the retroviruses used in the analysis is presented in Table S2. The ERVs in bat genomes were identified using reciprocal BLAST Hits between bat genomes and viral proteins (Camacho et al. 2009). The BLAST hits that were >200 amino acids long and had an E‐value <0.009 were selected for gag and env proteins, while those that were >400 amino acids long and had an E‐value <0.009 were selected for pol proteins.

2.8. Gene Loss Identification

Whole‐genome alignment between the bat genome and the human reference genome was performed using LASTZ v1.04.45 (Harris 2007). Gene loss events were systematically identified using TOGA v1.1.7 with alignment chain file (Kirilenko et al. 2023). A gene was classified as lost if it met the following criteria: (i) at least 80% of the gene's central region was fully aligned, and (ii) the gene contained inactivating mutations in at least two exons. For the identified lost genes, transcriptome‐wide expression levels across different tissues were analyzed using Salmon v1.5.2 to validate the findings (Patro et al. 2017).

2.9. In Silico Analysis of Protein Structures

To determine the effect of amino acid substitutions on gene function, we modeled the three‐dimensional structures of protein sequences that exhibited convergent amino acid substitutions. The protein sequence of humans was selected as the target sequence, and amino acids at the corresponding positions in the target sequence were replaced with amino acids that underwent convergent evolution. The protein structure was predicted using AlphaFold2 Colab v1.5.5 (Mirdita et al. 2022) with default parameters and visualized using PyMOL v3.0 (Schrödinger 2015). The effect of amino acid substitutions on protein stability was assessed using the DUET online server (Pires et al. 2014). The SIFT online server (2024.04.25) was used to predict whether the amino acid substitutions affected protein function (Sim et al. 2012).

3. Results

3.1. Genome‐Wide Signatures of Adaptive Evolution in Avivorous Bat Species

To resolve the phylogenetic relationships of two avivorous bat species, we reconstructed a genome‐wide phylogenetic tree (Figure 1) for bat species with diverse dietary strategies. The tree was constructed using a multispecies coalescent model and was based on 9132 single‐copy orthologous genes from 19 species. All nodes of the tree received 100% bootstrap support. The estimation of divergence times revealed that I. io originated in the Middle Miocene approximately 14.2 MYA, whereas N. aviator originated later at approximately 7.9 MYA (Figure 1).

FIGURE 1.

The phylogenetic tree of 19 species was reconstructed using ASTRAL‐III. The divergence time was estimated by MCMCTree. Geological timescale information was obtained from TimeTree (http://www.timetree.org/). The significantly expanded (red) and contracted (blue) gene families of each node are shown using a pie plot. The number of gene families that expanded and contracted in each species is highlighted with corresponding colors. The various diets are represented by different colors: frugivorous (purple), omnivorous (yellow), sanguivorous (red), insectivorous (blue), and avivorous (orange).

Based on the estimated divergence times, we analyzed the evolutionary dynamics (expansion and contraction) of 3965 gene families. We identified 32 gene families with significant expansion and 10 gene families that underwent significant contraction in I. io (Table S3). Similarly, in N. aviator, we detected 52 gene families with significant expansion and 35 gene families with significant contraction (Table S4). Functional enrichment analysis results (Tables S5 and S6) revealed that the evolutionary gene families of the two avivorous bat species were associated with several immune response terms. In N. aviator, the expanded gene families were significantly enriched in immune‐related terms, including leukocyte activation (GO:0046649), innate immune response (GO:0045087), leukocyte‐mediated immunity (GO:0002443), and T‐cell activation (GO:0042110). In contrast, contracted gene families were significantly enriched in interferon signaling (R‐HSA‐913531), humoral immune response (GO:0006959), and regulation of viral process (GO:0050792). In I. io, immune‐related enriched terms were relatively scarce, with only regulation of type I interferon (IFN‐I) production (GO:0032479) and innate immune response (GO:0045087) detected in expanded gene families. Further analysis of these terms revealed that several gene families were involved in the immune response induced by the major histocompatibility complex (MHC). Specifically, KLRC1 underwent an expansion in N. aviator, whereas HLA‐A was expanded in N. aviator but contracted in I. io.

To trace the molecular adaptations associated with the avivorous dietary strategy, we identified PSGs and REGs in I. io and N. aviator, respectively. When I. io was designated as the foreground branch, 245 PSGs and 464 REGs were identified (Table S7). When N. aviator was designated as the foreground branch, 246 PSGs and 588 REGs were identified (Table S8). Functional enrichment analysis of PSGs revealed that immune‐related terms were significantly enriched in both avivorous bat species (Table S9). In I. io, cell activation involved in immune response (GO:0002263) and positive regulation of immune response (GO:0050778) were enriched. The enrichment of N. aviator exhibited more extensive immune response terms, including inflammatory response (GO:0006954), positive regulation of immunoglobulin production (GO:0002639), leukocyte degranulation (GO:0043299), inflammatory response to antigenic stimulus (GO:0002437), and leukocyte degranulation (GO:0043299). Besides immune‐related terms, we also found the terms relevant to lipid metabolism, including cholesterol metabolic processes (GO:0008203), insulin processing (R‐HSA‐264876), and lipid catabolic processes (GO:0016042).

Given their shared predation strategies and diets, we explored convergent genes in avivorous bats. In total, 293 genes exhibited convergent evolution based on RERs (Table S10). The protein–protein interaction network consisted of convergent genes with 88 nodes and 93 edges (Figure 2a). MCL clustering identified two distinct clusters from this network, where the genes were found to be associated with immune response and lipid metabolism functions (Figure 2a). Functional enrichment results of these convergent genes yielded results consistent with the network, and multiple terms related to immune response and lipid metabolism were enriched significantly (Figure 2b, Table S11). For example, the terms related to immune response included the complement system (WP2806), positive regulation of leukocyte chemotaxis (GO:0002690), and positive regulation of interleukin‐8 production (GO:0032757). Lipid metabolism‐related terms included alcohol metabolic processes (GO:0006066), cellular lipid catabolic processes (GO:0044242), lipid localization (GO:0010876), and the PPAR signaling pathway (WP3942).

FIGURE 2.

Network analysis and functional enrichment of convergent genes in avivorous bat species. (a) The network was constructed using the convergent genes of I. io and N. aviator. The nodes were clustered by the MCL method. Each cluster was visually distinguished using a unique color scheme, with a special focus on convergent genes involved in immunity and lipid metabolism. (b) The functional enrichment results of the convergent genes.

3.2. Adaptive Evolution of Immunity in Avivorous Bat Species

3.2.1. Elevated Selection Pressure Related to Immunity in Avivorous Bat Species

The evolutionary gene families, positive selection, and convergence revealed potential adaptive evolution specifically tailored to the immune response in I. io and N. aviator. We performed a comprehensive investigation to characterize the adaptation to immune response in avivorous bats. Using “innate immune,” “adaptive immune,” and “virus” as keywords, we extracted gene sets corresponding to innate immunity, adaptive immunity, and antiviral defense, respectively, from Metascape. These gene sets comprised 392 associated with innate immunity, 225 with adaptive immunity, and 457 with antiviral defense. Compared to the background branches (bats with other diets), I. io and N. aviator (foreground branches) showed significantly higher ratios (ω) between the nonsynonymous substitution rate (dN) and the synonymous substitution rate (dS) (Wilcoxon test, p < 0.05) in genes for adaptive immunity and antiviral defense (Figure 3a). For genes linked to innate immunity, N. aviator showed a significantly higher ω value compared with the background branches. In contrast, while I. io exhibited a higher ω value; the statistical significance showed boundary effects.

FIGURE 3.

The adaptive evolution of immunity in avivorous bat species. (a) The kernel density plot shows the dN/dS ratios of genes involved in innate immune response, adaptive immune response, and antiviral defense in I. io (red), N. aviator (yellow), and other bat species (blue). The mosaic plot (b) and histogram plot (c) illustrate the composition and number of ERV elements in different bat species. (d) The synteny analysis of TLR10 and flanked genes among 19 species. (e) Visualizing the inactivating frameshift mutations of TLR10 in I. io. The frameshifting deletions are shown as gray vertical lines and frameshifting insertions are shown as red arrowheads. (f) The heat map illustrates the expression of TLR10 and flanked genes in I. io. The left subplot displays the normalized transcripts per million (TPM) for TLR10 and flanked genes, while the right subplot shows the read counts for these genes. The topmost bar chart reveals the expression patterns in the spleen of I. io.

3.2.2. Fewer Elements of ERVs in Avivorous Bats

As bats are natural viral reservoirs, the predation of birds by bats provides the opportunity for viral transmission, which facilitates infections in bats (Nabi et al. 2021). After invading the host, retroviruses integrate into the host genome by reverse transcription and integration (Skirmuntt et al. 2020). Across the 15 bat genomes (Table S12), most integrated ERV elements belonged to betaretrovirus and gammaretrovirus families, consistent with previous studies in mammals. Previous study identified alpharetrovirus‐like elements in three bat genera, including Rhinolophus, Rousettus, and Phyllostomus (Jebb et al. 2020). Our study contributed more information to previous findings by revealing the presence of alpharetrovirus‐like elements in expanded bat species (Figure 3b), including H. armiger, P. alecto, A. jamaicensis, D. rotundus, M. molossus, M. natalensis, M. davidii, M. myotis, E. fuscus, and I. io. Both I. io and N. aviator exhibited a lower abundance of ERV elements in their respective genomes (Figure 3c). The N. aviator genome contained 144 ERV elements. Although I. io had a higher total ERV count compared to N. aviator, its ERV abundance, particularly in the env and gag genes of retroviruses, was relatively lower compared to that of some insectivorous and frugivorous bat species.

3.2.3. The Adaptation of TLR Family in Avivorous Bats

Toll‐like receptor (TLR) family members, constituting the most prominent pattern recognition receptors, play a key role in detecting viruses, bacteria, fungi, and parasites (Barreiro et al. 2009). Our results revealed that many TLR members have undergone adaptive evolution. For example, TLR2 and TLR7 showed convergent evolution in two avivorous bat species (Figure 2a). In N. aviator, TLR9 was under positive selection (Table S8), with a positively selected site at position 643, where valine is replaced by leucine (p.V643L). Comparative analysis of other TLR family members across two avivorous bat species showed that TLR10 lost its function in I. io and N. aviator. No complete TLR10 sequence was identified in N. aviator genome. Synteny analysis of conserved TLR10 flanking genes (Figure 3d) confirmed N. aviator lacks TLR10, not due to assembly or annotation errors. In contrast, the TLR10 sequence in the I. io possesses a complete structure. However, multiple frameshift mutations in the coding region resulted in functional loss of TLR10 in I. io (Figure 3e). TLR10 in I. io shows low expression (only few aligned reads) in spleen, kidney, lung (10 tissues tested; Figure 3f). Selection pressure analysis confirmed relaxed purifying selection on TLR10 (p = 0.0002, K = 0.06), which is consistent with loss of functional constraint. These features are consistent with typical pseudogenization patterns, indicating that TLR10 in I. io is undergoing pseudogenization.

3.3. Convergent Evolution of Lipid Metabolism in Avivorous Bat Species

Functional enrichment analysis of convergent genes revealed that the two avivorous bat species have undergone convergent adaptation not only in the immune response but also in lipid metabolism. We identified 12 convergent genes related to lipid metabolism. These genes are involved in multiple key processes in lipid metabolism (Figure 4a). Notable, APOA1, APOF, and CEPT1 were under both positive selection and convergent evolution. APOA1 and APOF were involved in lipoprotein metabolism. APOA1 is the primary protein component of high‐density lipoproteins (HDLs), which mediate the transport of peripheral tissue cholesterol to the liver for excretion (Niesor 2015). Cholesteryl ester transfer protein (CETP) mediates the transfer of cholesteryl ester (CE) from HDLs to low‐density lipoproteins (LDLs) and very low‐density lipoproteins (VLDLs) (Morton et al. 2019). APOF modulates lipoprotein homeostasis by selectively inhibiting the CETP (Morton and Mihna 2022). CEPT1 contributes to the terminal phase of the Kennedy pathway, catalyzing the synthesis of phosphatidylcholine and phosphatidylethanolamine (Wang et al. 2023). Specifically, phosphatidylcholine plays a pivotal role in activating the PPARα pathway, which regulates lipid metabolism in hepatocytes (Chakravarthy et al. 2009).

FIGURE 4.

The adaptive evolution linked to lipid metabolism in avivorous bats. (a) The convergent genes involved in lipid metabolism among I. io and N. aviator. These genes were highlighted with color and bolded in diagram. Genes in blue underwent convergent evolution, while genes in red underwent both convergent evolution and selection. (b) Convergent amino acid substitutions of CEPT1 were observed among two avivorous bat species. Four parallel amino acid substitutions underwent positive selection were highlighted with red. The ancestral state of these amino acids is shown in parentheses. “*” represents the significance of positive selection estimated by BEB. “*” means >0.95, “**” means >0.99. The membrane topology (top) of CEPT1 based on research of Wang et al. (c) The point plot illustrates the relative of evolution rates of CEPT1 in different species. Two avivorous bat species (red) have elevated evolution rates. (d) The predicted three‐dimensional structure of CEPT1. The four parallel amino acid substitutions within TM5 were highlighted. The red stick model depicts the four convergent amino acid substitutions, with their corresponding human amino acids indicated in blue.

We performed a comprehensive analyzed CEPT1, a key upstream activator of the PPARα pathway that underwent both positive selection and convergent evolution. Four identical positively selected sites were identified in both I. io and N. aviator (Figure 4b). Ancestral sequence comparison revealed that these CEPT1 sites have undergone parallel amino acid substitutions in the two avivorous bat species (Poisson test FDR = 4.77 × 10⁻⁹, Table S13). CEPT1 consists of 10 transmembrane segments (TMs), with TMs 1–6 constituting a conserved catalytic domain (Wang et al. 2023). Three (p.V216S, p.F219I, and p.M223C) of these four amino acid substitutions were in the fifth TM (Figure 4d). Previous studies have reported that the variant V216L led to a 40% loss of activity of CEPT1 (Wang et al. 2023). In the present study, we identified a valine (V) to serine (S) substitution at the 216th amino acid of CEPT1 in two avivorous bats. We also performed the SIFT and DUET analyses to evaluate the functional and structural implications of these parallel substitutions in two avivorous bat species. Our results showed that the substitution of the 223rd amino acid has harmful effects (SIFT score = 0.02) on function, while substitutions at position 216 and 219 amino acids might impair protein stability.

4. Discussion

Comparative genomics has advanced our understanding of genetic mechanisms underlying niche evolution. Similar foraging strategies and dietary expansion impose convergent selective pressures on avivorous bats, leading to convergent phenotypic adaptations. Despite belonging to two distinct genera within the Vespertilionidae family, three avivorous bat species share convergent wing morphologies and low‐frequency echolocation calls, which facilitate their aerial‐hawking strategy in open spaces (Thabah et al. 2007; Ibáñez et al. 2021). In the present study, we investigated molecular adaptations associated with the unique avivorous dietary by comparing genomes of two avivorous bats (I. io and N. aviator) with 13 bats with other diets (covering all major dietary niches). Our genome‐wide analyses uncovered molecular adaptations and convergent evolution linked to immunity and lipid metabolism in the two avivorous bats, shedding light on avivorous bats’ dietary niche‐specific molecular adaptations.

A key ecological challenge faced by avivorous bats is intensified exposure to avian‐derived viruses. As natural viral reservoirs (Nabi et al. 2021), birds and bats’ flying abilities facilitate cross‐species virus transmission through direct and indirect interactions, including predation, cohabitation, food competition, and spatiotemporal niche overlap (Ibáñez et al. 2001; Myczko et al. 2017; Perrella et al. 2020). For instance, the IAV H19 subtype identified in wild ducks exhibits characteristics of avian and chiropteran IAVs (Fereidouni et al. 2023), and a novel avian‐like IAV has been isolated from infected R. aegyptiacus (Kandeil et al. 2019). This heightened viral exposure exerts strong selective pressure on the immune systems of avivorous bats, which is supported by higher dN/dS ratios of immune‐related genes in I. io and N. aviator. Our genome‐wide selective pressure analyses provided insights into PSGs and REGs associated with immune response and antiviral response. Among these genes, several directly involved in antiviral defense underwent convergent evolution, including TLR2, TLR7, and CD8A. TLR2 mediates antiviral responses against various DNA viruses (Gaudreault et al. 2007; Sørensen et al. 2008; Zhou et al. 2008, 2021) and participates in defending against respiratory syncytial virus (RSV) infections (Murawski et al. 2009). TLR7 recognizes pathogen‐derived single‐stranded RNA and subsequently induces IFN‐I production, a cornerstone of antiviral immunity (Szabo et al. 2014). The CD8 antigen encoded by CD8A is essential for protecting hosts against retroviral infections through cytotoxic T lymphocyte‐mediated responses (Dittmer et al. 2004; Zelinskyy et al. 2009). These enhanced antiviral defense mechanisms, evolved in response to avian‐derived viral pressure, likely play a pivotal role in resisting retroviral colonization—an effect reflected in the lower ERVs abundance observed in avivorous bats compared to other dietary bats.

ERVs, as genomic “fossils” of ancient retroviral infections, accumulate in host genomes through the integration of exogenous retroviruses and subsequent vertical inheritance (Zheng et al. 2022). Reduced ERV load in avivorous bats does not indicate less historical viral exposure or infection. Instead, it likely reflects the enhanced efficiency of their immune adaptations in restricting retroviral integration and long‐term persistence. Dietary niche and foraging behavior are key drivers of retroviral exposure and endogenization in bats (Zhou et al. 2025). For instance, previous work has identified a higher abundance of ERVs in the genomes of carnivorous bats relative to herbivorous bats (Zhou et al. 2025). Yet unlike most carnivorous bats, avivorous bats do not show increased ERV accumulation. Instead, their low ERV abundance mirrors that of D. rotundus (Zepeda Mendoza et al. 2018). The unique immune adaptations of D. rotundus are associated with the low ERV abundance identified in genome (Zepeda Mendoza et al. 2018; Blumer et al. 2022). This parallel strongly suggests that avivorous bats, like D. rotundus, have evolved specialized immune adaptations. Specifically, the convergent evolution of TLR2, TLR7, and CD8A likely enhances the recognition and clearance of invading retroviruses, thereby reducing the likelihood of their integration into the host genome. These findings suggest that the specialized immune adaptations of avivorous bats may not only combat acute viral infections but also limit the long‐term genomic accumulation of retroviruses, with reduced ERV abundance potentially serving as a molecular signature of this defensive capacity.

The immune adaptations of avivorous bats extend beyond convergent antiviral genes to include specialized adjustments in immune‐related gene families. Previous studies have demonstrated that bats exhibit tolerance to viral infections through the expansion of MHC class I genes and the KLRC/KLRD family, which elevates the activation threshold of natural killer (NK) cells in R. aegyptiacus (Pavlovich et al. 2018). MHC class I molecules are central to antigen presentation and T‐cell activation, playing a key role in immune responses (Van De Weijer et al. 2015). KLRC1, a member of the NKG2 natural killer cell receptor family (Posch et al. 1998), encodes a protein that forms a complex with KLRD1, a transmembrane protein expressed on the surface of NK cells. This complex facilitates the recognition of MHC class I ligands, triggering the NK cells activation (Sullivan et al. 2007). In the present study, our gene family analyses detected avivorous bat species‐specific changes in the MHC I gene (HLA‐A expansion in N. aviator and contraction in I. io) and KLRC1 expansion in N. aviator, indicating enhanced viral infection tolerance in avivorous bats, particularly N. aviator. This tolerance, combined with the enhanced antiviral defenses mediated by convergent genes, creates a balanced immune strategy that minimizes both viral pathogenicity and immunopathology, could be a critical adaptation for bats given their unique physiological demands.

We also identified that TLR10 function is lost in both species of avivorous bats. The TLR10 gene is absent in N. aviator, while TLR10 is present in I. io but non‐functional due to multiple frameshift mutations in its coding region.

As a TLR1 subfamily member, TLR10 functions as a critical innate immune sensor during viral infections, contributing to early immune responses and inflammatory modulation (Lee et al. 2014; Oosting et al. 2014; Su et al. 2021). Given the functional overlap between TLR1 and TLR10, the polymorphism in their functions may modulate host responses in terms of direction and intensity (Su et al. 2021). In the absence of one of these genes, the other can compensate for host defense mechanisms, thereby modulating disease susceptibility. This evolutionary change may contribute to the fine‐tuning of immune responses in avivorous bats, preventing excessive inflammation that could disrupt physiological homeostasis. Notably, our study also identified convergent evolutionary events of immune genes (e.g., TLR2 and TLR7) in avivorous bats. This adaptive adjustment underscores the complexity of the immune adaptive landscape in avivorous bats, where multiple molecular changes converge to optimize antiviral defense. Similar evolutionary patterns in pattern recognition receptors have also been observed in studies on Carnivora species. Specifically, the pseudogenization of TLR10 in seven species is thought to be associated with immune adaptation to their unique living environments (Wu et al. 2022). For instance, most insectivorous and frugivorous bats retain a functional TLR10 gene under purifying selection, which is essential for recognizing diverse insect‐borne pathogens (Escalera‐Zamudio et al. 2015). In contrast, sanguivorous bats do not exhibit TLR10 pseudogenization, which represents a significant difference from the avivorous bats studied here (Zepeda Mendoza et al. 2018). These interspecific differences further confirm that dietary adaptation plays a key role in shaping the evolutionary pattern of TLR10 in bats.

It is worth noting, the two avivorous bat species exhibit distinct evolutionary trajectories of TLR10 loss: N. aviator has undergone complete structural deletion of TLR10, representing a more thorough pseudogenization process, while I. io has only experienced functional inactivation but retains the intact gene structure. TLR10 structural deletion is also observed in the close relatives of N. aviator. This further supports the inference that their common ancestor possessed an intact and functionally normal TLR10, with divergent TLR10 evolutionary trajectories in the two avivorous species occurring after speciation. The specific driving forces behind these different evolutionary paths require further investigation in future research.

Hibernation represents a critical physiological challenge for bats, as sufficient fat storage before hibernation and efficient lipid utilization during torpor are essential for survival (Popa‐Lisseanu et al. 2007; Fukui et al. 2013). Avivorous bats have evolved a specialized trophic strategy to address this demand: they predate on migratory birds primarily during pre‐ and post‐hibernation seasons (spring, autumn, and early winter), a period when insect resources are scarce (Fukui et al. 2013; Gong, Shi, et al. 2021). Migratory birds possess higher nutritional and energy value than insects, with notably higher fat content and a greater proportion of saturated fats (Payne et al. 2016). This dietary shift not only compensates for seasonal insect scarcity but also provides a concentrated lipid source for hibernation energy reserves.

However, high consumption of saturated fats imposes substantial metabolic burdens on the liver and cardiovascular system (Duval et al. 2007), creating strong selective pressure for adaptive evolution in lipid metabolism pathways. Consistent with this selective pressure, previous studies have documented adaptive evolution of lipid metabolism genes in carnivorous bats (Lyroderma lyra and Vampyrum spectrum) (Yue et al. 2024). Intriguingly, our analyses revealed that a similar adaptive pattern, lipid metabolism‐related genes of the two avivorous bat species underwent positive selection and convergent evolution. A core set of convergent genes are functionally linked to the PPARα signaling pathway, a key regulator of lipid homeostasis (Berger and Moller 2002): APOA1 and APOF contribute to lipoprotein balance (Niesor 2015; Morton et al. 2019), CPT1B and CPT2 mediate fatty acid β‐oxidation, and CYP7A1 is involved in bile acid synthesis (Chakravarthy et al. 2009). PPARα is predominantly expressed in the liver, where it promotes fatty acid oxidation, inhibits de novo fatty acid synthesis, and maintains normal blood triglyceride levels (Duval et al. 2007). Liver‐specific PPARα deficiency in mice leads to hepatic steatosis and inflammation, highlighting its critical role in lipid metabolism (Duval et al. 2007). For mammals, fat serves as the primary energy source during hibernation, and numerous lipid metabolism‐related genes are regulated by PPARα (Huang et al. 2025). Hibernating bats experience heightened selective pressure on PPARα, with distant bat lineages (e.g., Rhinolophidae, Hipposideridae, and Myotis species) showing increased PPARα expression at both the mRNA and protein levels during hibernation (Han et al. 2015). In contrast, brown bears (Ursus arctos) suppress PPARα expression to maintain a low fat consumption rate throughout hibernation (Vella et al. 2020).

Notably, an upstream gene essential for PPARα activation, CEPT1 (Wright and McMaster 2002), underwent both convergent evolution and positive selection in the two avivorous bats. Acting through the Kennedy pathway, CEPT1 catalyzes the final step in phosphatidylcholine (PC) synthesis, generating 1‐palmitoyl‐2‐oleoyl‐sn‐glycerol‐3‐phosphocholine (16:0/18:1‐GPC), an endogenous ligand that activates PPARα (Wright and McMaster 2002; Chakravarthy et al. 2009). In the present study, CEPT1 exhibits identical amino acid substitutions (defined as parallel convergent evolution) in the two avivorous bats. Specifically, substitutions at amino acid position 216 have been reported to reduce the catalytic activity of human CEPT1, which modulates activation and phosphorylation of PPARα (Wang et al. 2023). As an upstream regulator of PPARα, the reduced catalytic activity of CEPT1 in avivorous bats modulates the production of the endogenous PC ligand for PPARα (Wright and McMaster 2002), thereby regulating the fatty acid oxidation process mediated by the PPARα signaling pathway and promoting fat accumulation in preparation for hibernation—consistent with the fat accumulation phenotype observed in CEPT1‐knockdown mice (Chakravarthy et al. 2009).

Distinct from insectivorous bats, avivorous bats consume migratory avian prey characterized by a significantly higher content of saturated fatty acids (SFAs) (McWilliams et al. 2004). At low hibernation temperatures, an accumulation of SFAs in the lipid bilayer compromises membrane fluidity. Since the de novo synthesis pathway mediated by CEPT1 lacks stringent selectivity for the unsaturation of acyl chains, its downregulation likely functions as a regulatory filter to prevent the non‐selective incorporation of SFAs into the membrane (Shindou and Shimizu 2009). Such suppression of CEPT1 expression promotes a metabolic shift toward the more selective Lands’ cycle, which facilitates the precise enrichment of unsaturated fatty acids within the cellular membrane, thereby maintaining essential fluidity during extreme cold (Zhang et al. 2021). Therefore, we propose that the adaptive remodeling of the lipid metabolic landscape in avivorous bats involves a complex regulation of the PPAR𝛼 signaling pathway. By regulating PPARα activity to drive lipid metabolism while simultaneously downregulating CEPT1 to filter out saturated dietary lipids, avivorous bats may have evolved a convergent mechanism to balance the efficient utilization of high‐energy avian prey with the stringent requirements for membrane fluidity during deep hibernation.

Beyond hepatic lipid metabolism, convergent evolution was also detected in genes regulating intestinal lipid absorption, a key step in dietary lipid utilization (Ko et al. 2020). NPC1L1, predominantly expressed in gastrointestinal epithelial cells, binds to critical mediators of cholesterol absorption (Davis and Altmann 2009), while ABCG5 forms a complex with ABCG8 in the intestine to restrict cholesterol absorption and promote excretion (Bydlowski and Levy 2024). The convergent evolution of these genes likely optimizes the balance between lipid absorption and excretion, ensuring efficient extraction of energy from avian prey while avoiding excessive cholesterol accumulation. Additionally, studies on I. io’s gut microbiota showed increased abundances of Firmicutes and Bacteroidetes during bird‐consuming periods (Gong, Liu, et al. 2021), which may further enhance fat digestion and accumulation—suggesting a synergistic adaptation between host genetics and gut microbiota in lipid metabolism.

5. Conclusions

To summarize, our comparative genome analysis provided insights into the adaptive evolution of the unique diet of avivorous bat species. These adaptations were related to immunity and lipid metabolism. Our analyses revealed that convergent adaptations were associated with immunity and lipid metabolism in two phylogenetically distant avivorous bat species. These findings highlighted the significance of immune and lipid metabolism in the adaptation to an avivorous diet. Our results also revealed that immune‐related genes in avivorous bats are under greater selection pressure and harbor fewer endogenous retroviral elements in their genomes. Immune‐related adaptation may contribute to the defense of avivorous bats against viral infection while preying on birds. By investigating lipid metabolism adaptation, multiple genes involved in gut lipid absorption and PPAR signaling pathways exhibited convergent evolution. These adaptive changes in lipid metabolism might facilitate more efficient utilization of dietary lipids from birds. These findings not only provided greater insights into the adaptive evolution underlying the unique diet of avivorous bats but also offered novel perspectives on the molecular mechanisms underlying ecological niche evolution in a dietary context. Future studies should confirm the functions of the genes involved in immunity and lipid metabolism via cell culture and transfection.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1 The schematic of analytical workflow for positive selection and convergent evolution.

Table S1 The genomes used in comparative genomic analysis.

Table S2 The retrovirus sequences used in identification of ERVs elements in bats.

Table S3 The results of gene family evolution analysis in I. io.

Table S4 The results of gene family evolution analysis in N. aviator.

Table S5 The function enrichment analysis of significant changed gene families in I. io.

Table S6 The function enrichment analysis of significant changed gene families in N. aviator.

Table S7 The results of selective pressure analyses on single‐copy orthologous genes in I. io.

Table S8 The results of selective pressure analyses on single‐copy orthologous genes in N. aviator.

Table S9 The results of function enrichment analysis on positively selected genes in I. io and N. aviator, respectively.

Table S10: The convergent evolution analysis of I. io by using RERconverge.

Table S11 The results of function enrichment analysis on convergent genes among two avivorous bats. These enrichment analyses were performed on Metascape online server.

Table S12 The identification of endogenous retroviruses within genomes of bat species.

Table S13: The parallel amino acid substitutions identified by using Zou and Zhang's methods.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author, Tinglei Jiang, upon reasonable request. In this study, all analyses were conducted following the manuals and tutorials of software and pipeline. The detailed software versions are specified in the methods section. Unless specified otherwise, default or author‐recommended parameters were used for software and analysis pipeline. Data analysis commands and intermediate result files in this research can be obtained from github with https://github.com/life404/Avivorous_Bats_Comparative_Genomics.git.