Genomic Erosion Through the Lens of Comparative Genomics

24 Loss of genetic diversity threatens species survival, yet the dynamics of such loss and species' 25 responses thereof can vary widely depending on their evolutionary histories, life-history traits 26 and demographic trajectories. Comparative genomics offers a powerful framework to explore 27 the dynamics of genomic erosion across species. Here, we analysed the genomes of three 28 species — the Mauritius parakeet, the Mauritius kestrel, and the pink pigeon — that 29 experienced extreme and well-documented population bottlenecks. We compared them to 36 30 species spanning the avian phylogeny, with varied IUCN Red List statuses to investigate the 31 genomic consequences of their demographic collapses. For each species, we assessed 32 nucleotide diversity, genetic load, and runs of homozygosity (ROH), alongside genome 33 synteny and transposable elements. We found a negative correlation between nucleotide 34 diversity and ROH, but neither metric was a good predictor of the species’ Red List status. 35 Rather, the population effective to census size ratio showed a strong correlation to Red List 36 status. Moreover, species with larger historical effective population sizes showed greater 37 heterozygosity but carried a higher heterozygous load, highlighting the importance of historical 38 demography to assess species vulnerability to genomic erosion. We found significant 39 differences in homozygous load between taxonomic groups of our target species, possibly 40 due to differences in life-history traits and demographic histories. Genome structure analyses 41 revealed differences in transposable elements and genomic rearrangements between groups, 42 suggesting their potential role in shaping genome architecture and adaptive potential across 43 species. Our findings underscore the value of multispecies comparisons in understanding the 44 evolutionary dynamics of genomic erosion and its relevance for biodiversity conservation. 45 46 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 3

47 Genetic diversity is an essential component of a species' ability to adapt and persist under 48 changing environmental conditions (Spielman et al. 2004; Allendorf et al. 2013; Kardos et al. 49 2021). For small or isolated populations, maintaining genetic diversity is particularly 50 challenging, as reduced effective population size (Ne) diminishes the efficacy of natural 51 selection, intensifies genetic drift, and leads to increased inbreeding. These processes 52 ultimately lead to genomic erosion, often characterised by the loss of genetic diversity and an 53 increased accumulation or fixation of deleterious mutations. As inbreeding becomes more 54 frequent, the exposure of recessive deleterious mutations exacerbates these effects, 55 culminating in inbreeding depression (Charlesworth and Willis 2009; Blomqvist et al. 2010; 56 Hasselgren et al. 2021). These processes collectively diminish fitness and adaptive potential, 57 exacerbating vulnerability to environmental changes and threatening the long -term 58 persistence of the population (Blomqvist et al. 2010; Hasselgren et al. 2021; Jackson et al. 59 2022; van Oosterhout et al. 2022; Jeon et al. 2024). Additionally, even populations that have 60 partially recovered demographically bear the genetic legacy of past bottlenecks, known as 61 "drift debt", which manifests as a time lag between demographic change and loss of genome-62 wide variation (Dussex et al. 2023; Gilroy et al. 2017; Pinto et al. 2024). Recent analyses have 63 shown that species are losing genetic diversity worldwide (Exposito-Alonso et al. 2022; Shaw 64 et al. 2025) , highlighting the critical need for understanding genomic erosion in biodiversity 65 conservation. 66 With the continuous and rapid production of genomic data for wild species worldwide, 67 conservation genomics can now take advantage of high -resolution tools to assess genetic 68 diversity, genetic load and structural variation (Lewin et al. 2018; Wright et al. 2020; van 69 Oosterhout et al. 2022; Theissinger et al. 2023). For species at risk of extinction, such insights 70 are critical for guiding conservation interventions aimed at reducing genetic load and 71 enhancing population viability (e.g., vaquita (Morin et al. 2021), kākāpō (Dussex et al. 2021), 72 pink pigeon (Speak et al. 2024)). Additionally, the use of genomic resources across multiple 73 species with diverse evolutionary histories, demographic trajectories, and conservation status 74 within a comparative genomic framework offers valuable insights (Grueber 2015) . This 75 approach can elucidate how these factors interact with genomic traits such as genetic 76 diversity, genetic load, or structural variations to influence species' long -term viability and 77 extinction risk. 78 The echo parakeet (Alexandrinus [Psittacula] eques), Mauritius kestrel (Falco punctatus), and 79 pink pigeon (Nesoenas mayeri) exemplify how species can recover demographically from near 80 extinction but remain genetically imperilled. These birds, endemic to Mauritius—a biodiversity 81 hotspot that has witnessed over 100 species extinctions in recent centuries (Florens 2013)—82 experienced some of the most extreme population bottlenecks ever recorded in wild 83 populations but recovered thanks to targeted conservation efforts (Jones and Swinnerton 84 1997). Only four Mauritius kestrels remained by 1974, the pink pigeon declined to 10 85 individuals by 1990, and the Mauritius parakeet to just 20 individuals by 1986 (Jones and 86 Swinnerton 1997; Jones 2013; Jones et al. 2013) . Conservation programs, including captive 87 breeding, supplementary feeding, habitat restoration, and predator control, facilitated their 88 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 4 demographic recovery to current free-living adult population sizes of approximately 500 pink 89 pigeons, 250 Mauritius kestrels and 650 echo parakeets (Jones and Swinnerton 1997; Jones 90 2010; Nicoll et al. 2021) (Figure 1D). Despite successful demographic recoveries, the legacy 91 of the extreme historical population collapses can jeopardise their long -term viability, as 92 genetic diversity in these species continues to decline due to the accrued drift debt (Tollington 93 et al. 2013; Jackson et al. 2022). These species are also at risk of accumulating an increased 94 genetic load of deleterious mutations, as has been shown in the pink pigeon (Jackson et al. 95 2022). Beyond genetic challenges, ecological pressures such as habitat loss and degradation 96 persist, compounded by threats like emerging infectious diseases, such as the Psittacine Beak 97 and Feather Disease (PBFD) in the Mauritius parakeet, can affect individuals fitness and 98 population viability(Tollington et al. 2015). 99 This study aims to leverage the power of comparative genomics to explore the interplay 100 between genome-wide diversity, genetic load, demographic history, and conservation status 101 across a diverse set of avian species. Using recently generated high -quality, chromosome-102 level reference genomes for these three Mauritian birds, we compared them to a set of bird 103 species spanning the avian phylogeny with distinct evolutionary histories and conservation 104 statuses. By narrowing our focus to species within the same order as the Mauritius ones, we 105 further aim to investigate potential group -specific differences in genomic metrics. This 106 comparative framework aims to uncover evolutionary mechanisms influencing the 107 maintenance or erosion of genetic diversity, providing insights to inform conservation priorities 108 for these and other vulnerable species. 109 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 5

110 Dataset 111 The reference genomes from three bird species endemic to Mauritius —the echo parakeet 112 (Morales, Groombridge, et al. 2024), the pink pigeon (Morales, Van Oosterhout, et al. 2024), 113 and the Mauritius kestrel (Morales, Norris, et al. 2024) —were recently sequenced and 114 reported. To build a comparative dataset, 36 additional bird species were selected based on 115 the availability of high-quality reference genomes (i.e., based on N50, average scaffold size, 116 and total scaffold count) and ensuring a comprehensive representation across the avian 117 phylogeny (Figure 1B). Metadata for all 39 species, including current census population sizes 118 and IUCN conservation statuses, are compiled in Table S1. 119 Mapping and variant calling 120 Raw reads used for assembling the reference genomes were downloaded from NCBI (see 121 Table S1 for Assembly IDs) and aligned to the corresponding genomes. NGS short reads were 122 mapped using BWA (v0.7.17) mem (Li and Durbin 2009) with default parameters. Read 123 duplicates were marked with GATK (4.4.0.0) MarkDuplicates (DePristo et al. 2011) . PacBio 124 HiFi reads were mapped and sorted using pbmm2 v1.5.0 125 (https://github.com/PacificBiosciences/pbmm2) with the parameter “ --preset HIFI”. GATK 126 HaplotypeCaller was used to call variants for each alignment. Only SNPs were kept for further 127 analyses. 128 Coverage 129 After mapping the raw reads to each reference genome, we estimated the average genome -130 wide coverage per scaffold in each genome using MosDepth v0.3.3 (Pedersen and Quinlan 131 2018). 132 Sex chromosome removal 133 Each reference genome was mapped to the chicken genome (assembly GRCg6a) with 134 minimap2 v2.1 (Li 2018). Any scaffold mapped to chicken sex chromosomes for more than 135 20% was treated as potential regions from sex chromosomes and removed for further 136 analyses. Any additional scaffolds annotated in the reference genomes as sex chromosomes 137 were also removed. 138 Heterozygosity 139 We estimated genome-wide heterozygosity using ANGSD (Korneliussen et al. 2014). We first 140 obtained genotype likelihoods on scaffolds larger than 500 kb and only considered sites with 141 a depth of coverage between ⅓ ( -setMinDepth) and two times ( -setMaxDepth) the average 142 coverage for each sample. We assumed that the reference and ancestral states were the 143 same. We applied the following parameters: -uniqueOnly 1 -remove_bads 1 -144 only_proper_pairs 1 -C 50 -baq 0 -minMapQ 30 -minQ 20 -setMinDepth $minDP -145 setMaxDepth $maxDP -doCounts 1 -GL 2 -doSaf 1. Next, we calculated the folded site 146 frequency spectrum (SFS) with realSFS. 147 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 6 To compare different estimations of heterozygosity, we also estimated genome -wide 148 heterozygosity directly from the VCF files with a custom pipeline. First, we divided scaffolds 149 into sliding windows of size 100 kb (with a slide of 50 kb) using bedtools makewindows v2.30.0 150 (Quinlan and Hall 2010) . Next, we obtained the total number of callable heterozygous and 151 total genotypes per window using bcftools v1.20 (Danecek et al. 2021), tabix v1.14 (Li 2011), 152 and vcfhetcount from vcflib (Garrison et al. 2022). Genotypes were considered callable if their 153 read depth was between ⅓ and 2 times the average coverage per sample and had a minimum 154 genotype quality (GQ) of 30 or reference genome quality (RGQ) of 10. Indels and windows 155 with less than 50% of callable sites were excluded from the analysis. 156 Runs of Homozygosity (ROH) 157 As some reference genomes were assembled with only Pacbio long reads, and no short-read 158 data are available for the same individual, commonly used methods like ROHan (Renaud et 159 al. 2019) could not be applied to identify ROH. To address this limitation, we developed a 160 custom method for this analysis. Given the varying fragmentation levels of the assemblies 161 (Table S1), for each species, we retained only scaffolds with a minimum size of 5 Mb. As a 162 result, two species (Red -faced mousebird, Urocolius indicus and Wilson's storm petrel, 163 Oceanites oceanicus) were excluded, as neither had scaffolds larger than 5Mb. We identified 164 ROH based on per -window heterozygosity estimates (see above). Using the R package 165 bedtoolsr (Patwardhan et al. 2019) , we concatenated windows with a heterozygosity lower 166 than 5e -4 bp-1 (Figure S1), except for two genomes with exceptionally low average 167 heterozygosity (the flightless cormorant, Nannopterum harrisi, and Mauritius kestrel), for which 168 a lower threshold of 1e -4 bp-1 was used. Adjacent homozygous regions were merged if 169 separated by a gap shorter than 100 kb, and only ROHs with a minimum size of 500 kb were 170 retained. The inbreeding coefficient (F ROH) was calculated as the ratio of the total length of 171 ROH segments to the total length of the analyzed genome (scaffolds > 5 Mb). We also 172 estimated heterozygosity for the analyzed scaffolds per species, both including and excluding 173 ROHs. 174 To validate our custom method, we compared our ROH estimates with those from ROHan for 175 five species for which short-read data were available. ROHan was employed using a window 176 size of 100Kb and a rohmu of 5e-4 to mimic the parameters used in the custom method. 177 Genetic load 178 To compare genetic load across species, we used chicken CADD score annotations as a 179 proxy for substitution deleteriousness at ultraconserved elements (UCEs) in our target 180 species. We extracted UCE regions and corresponding flanking regions from each reference 181 genome and from the chicken genome (GRCg6a), and performed a multi -species alignment 182 for each UCE, using the recommended pipeline of Phyluce v1.7.3 (Faircloth 2016)see Figure 183 S2 for detailed pipeline). Each genome was converted from fasta to 2bit format using UCSC 184 FaToTwoBit (Casper et al. 2018) . The UCE probe file was downloaded from 185 https://github.com/faircloth-lab/uce-probe-sets/tree/master/uce-5k-probe-set, and used to 186 extract and validate UCE regions per species. The CADD score file of the chicken genome 187 was downloaded from https://osf.io/c97ez. For each species, CADD scores were lifted from 188 the chicken genome for homozygous (relative to chicken) and heterozygous sites across UCE 189 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 7 regions with customised scripts partially adapted from LoadLift (Speak et al. 2024) . 190 Heterozygous sites were extracted from VCF files and filtered using bcftools with the same 191 parameters used for heterozygosity estimations (see above). Only heterozygous sites with 192 one allele equal to the corresponding site in the chicken genome were kept for further 193 analyses. 194 To quantify genetic load in each genome, we counted the number of heterozygous sites and 195 homozygous substitutions with CADD scores ⩾ 20 in UCE regions, representing the top 1% 196 most deleterious sites in the chicken genome. To control for the the evolutionary distance to 197 the chicken among species we rescaled the counts of sites with CADD scores ⩾ 20 with the 198 counts of substitutions with CADD scores < 3, the latter representing nearly-neutral sites. To 199 account for potential lineage -specific adaptive substitutions that have occurred since the 200 divergence with chicken, homozygous substitutions that were shared by 20 or more species 201 were excluded, as these are more likely to reflect long -term adaptive changes rather than 202 harmful mutations.To further confirm that we were correctly retaining sites that have not 203 changed since the divergence with chicken, we compared three species to their closest 204 reconstructed ancestral nodes in the phylogenetic tree from Stiller et al. (2024). We extracted 205 the reconstructed full sequences from the closest ancestral nodes and cut them into 200 -bp 206 short sequences with 20 -bp step sliding windows. The reads were mapped to the chicken 207 genome with BWA. For each species, we confirmed that homozygous substitutions had at 208 least one ancestral read mapped to the chicken genome, and the ancestral state matched the 209 chicken sequence. Homozygous substitutions with CADD scores above 20 are concentrated 210 at the terminal branches of the phylogenetic tree of birds (Figure S3), indicating that they are 211 more likely to be deleterious substitutions than lineage -specific adaptations. The rescaled 212 relative ratio of heterozygous sites was used as a proxy of masked load, which we refer as 213 “heterozygous load” in the following sections. The relative ratio of homozygous sites, being 214 referred to as “homozygous load”, was used as a proxy realized load. 215 Demographic history 216 We inferred historical fluctuations of effective population size (Ne) for each species using 217 PSMC (Li and Durbin 2011) with the parameters “ -N30 -t5 -r5 -p 1+1+1+1+30*2+4+6+10” 218 (Hilgers et al. 2025) and estimated the harmonic Ne mean from 10 kya to 100 kya for further 219 analyses. 220 Statistical and phylogenetic comparative analyses 221 To explore the relationships among genome-wide heterozygosity, FROH, Ne, and genetic load 222 across species while accounting for phylogenetic signals, we reconstructed the phylogenetic 223 relationships among the studied species and incorporated this inference into two statistical 224 frameworks. The phylogeny was reconstructed from a subset of the UCE dataset described 225 above (see Genetic Load section). Using the software AMAS (Borowiec 2016), we selected 226 and concatenated 1,526 UCE sequences, retaining only loci with ≤2% missing data and >30% 227 parsimony-informative sites. We used IQ -TREE (Minh et al. 2020) to infer the maximum -228 likelihood tree using the edge-linked partition model (Chernomor et al. 2016) and constraining 229 the topology to reflect higher-order taxonomic relationships following Stiller et al. (2024). Using 230 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 8 this inferred phylogenetic tree, we assessed different univariate models with a phylogenetic 231 generalised least squares (PGLS) approach (Freckleton et al. 2002) . These analyses were 232 conducted using the R packages ape (Paradis et al. 2004; Paradis and Schliep 2019), caper 233 (Orme et al. 2023) and nlme (Pinheiro et al. 2014). When PGLS indicated a non-zero lambda 234 (λ) value, suggesting a significant phylogenetic signal, we further examined these effects using 235 a Bayesian phylogenetic generalised linear mixed model (pGLMM) (Hadfield and Nakagawa 236 2010). In this framework, the phylogenetic relationship among species was modelled as a 237 random effect. pGLMM analyses were performed using the R packages ape and MCMCglmm 238 (Hadfield 2010). 239 Genomic synteny 240 We inferred multi -genome synteny for chromosome -level reference genomes for pigeons 241 (three species in Columbidae), parrots (five species in Psittaciformes), and falcons (six 242 species in Falco) separately using ntSynt v1.0.2 (Coombe et al. 2024) with the divergence 243 range (-d) set to 10. Synteny results were visualised using scripts from ntSynt based on the R 244 package gggnomes (Hackl et al. 2024). 245 Identification of Transposable Elements 246 To annotate repetitive elements (RE) in the genomes of the pink pigeon, Mauritius kestrel, and 247 Mauritius parakeet, we produced de novo libraries of RE for each species using 248 RepeatModeler2 (Flynn et al. 2020) . We combined the de novo libraries with previously 249 published manually curated libraries of RE from the Collared flycatcher ( Ficedula albicollis) 250 and Blue-capped cordon-bleu (Uraeginthus cyanocephalus) from Storer et al. (2021), and 251 from the Emu (Dromaius novaehollandiae), Anna's hummingbird (Calypte anna), and Kākāpō 252 (Strigops habroptilus ) from Peona et al. (2021). Using the resulting custom libraries, we 253 annotated the RE from the genomes using RepeatMasker version 4.0.8 (Smit et al. 2015) . 254 We repeated this process for three, five and four additional species of Columbiformes, 255 Falconiformes and Psittaciformes, respectively, to enable comparisons of proportions of RE 256 across species. 257

258 Genome-wide diversity and inbreeding 259 The sequencing depth across the compiled dataset ranged between 15x and 96x (mean = 46, 260 SD = 19). We estimated genome -wide heterozygosity with both genotype likelihoods in 261 ANGSD and by SNP-calling, resulting in very similar estimates (adjusted R2 = 0.78; Table S1, 262 S2, Figure S4). We used heterozygosity estimates from ANGSD for all subsequent analyses, 263 except for the estimation of ROHs (see Material and Methods). Neither heterozygosity or FROH 264 estimates showed significant correlation with the quality of the genomes (e.g., N50) or depth 265 (Figure S5). Our in -house method produced consistent ROH results to those from ROHan 266 (Figure S6), validating our approach. 267 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 9 268 Figure 1. Demographic trajectory of the three Mauritius species and phylogenetic, genetic diversity, and 269 Ne/Nc ratio distribution of the 39 species used in this study. A Demographic trajectory of the wild population 270 derived from field monitoring of adult individuals over time. Note that the three species have been monitored in 271 different ways so presented trends are approximations of their total numbers. B Phylogenetic tree topology, 272 adapted from Stiller et al. (2024). Each circle represents a sampled species within its respective order. The colour 273 and initials indicate the IUCN Red List category of each species. C Genome-wide heterozygosity (circles) and runs 274 of homozygosity-based inbreeding coefficient (F ROH; diamonds) for each species. Colour coding corresponds to 275 IUCN Red List categories. Domestic species are highlighted with a thicker vertical line. D Correlation between the 276 log-transformed Ne/Nc ratio (Log(Ne)/Log(Nc)) and IUCN Red List categories. The effective population size (Ne) 277 was estimated as the harmonic mean of PSMC values from 10 kya to 100 kya, whereas the census population size 278 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 10 (Nc) was obtained from IUCN Red List data. Photo credits: Samantha Cartwright for the Mauritius kestrel ( Falco 279 punctatus), Jacques de Speville for the Mauritius parakeet ( Alexandrinus eques), and Gregory Guida for the pink 280 pigeon (Nesoenas mayeri). 281 Genome-wide heterozygosity showed a strong negative correlation with inbreeding coefficient 282 FROH (Figure 1C; PGLS: λ = 0, R2 = 0.46, F1-28 = 26.56, p < 0.001). As expected, samples with 283 domestic or pet origins (Table S1) showed higher FROH compared to wild samples with similar 284 levels of heterozygosity. Thus, we excluded the former samples from further analyses. IUCN 285 Red List status was not a good predictor of genome -wide heterozygosity nor of F ROH (Figure 286 S7) or genetic load (Figure S8). The log -transformed Ne/Nc ratio (Log(Ne)/Log(Nc)) was 287 significantly lower (Wilcoxon two -sample test p < 0.001) in non -threatened species (Least 288 Concern, mean = 0.91, SD = 0.23) compared to threatened species (remaining IUCN status 289 categories, mean = 1.49), yet more varied (SD = 0.52). When numbers are assigned to the 290 status categories (0 to Critically Endangered, 1 to Endangered, …, 4 to Least Concern), Ne/Nc 291 ratios had a linear correlation with IUCN status (Figure 1D, GLM: R2 = 0.43, F1-24 = 12.49, p < 292 0.001). The Ne value (estimated with PSMC) represents an estimate of ancestral population 293 size deep-in-time, while the Nc value represents current census population size. Elevated 294 Ne/Nc ratios are indicative of recent population declines, with higher values reflecting more 295 abrupt changes. This result illustrates the discrepancy between IUCN Red List conservation 296 status relying only on demographic estimates and the shallow correlation to the genetic 297 diversity estimates. 298 We found a significant positive correlation between historical Ne and genome -wide 299 heterozygosity (Figure 2A; PGLS: λ = 0, R2 = 0.47, F1-22 = 19.65, p = 2.1e-4), highlighting the 300 predominant effect of long -term demographic history on genetic diversity. In contrast, F ROH 301 showed no correlation with historical Ne (Figure 2D; PGLS: λ = 0, R2 = 0.11, F1-22 = 2.82, p = 302 0.11). This lack of correlation is not surprising, given that our analysis focused only on long 303 ROHs (> 500 Kb), which reflect population history within tens to hundreds of generations ago. 304 Species with higher genetic diversity or a lower inbreeding coefficient tended to be burdened 305 by a higher heterozygous load, measured as the corrected ratio of heterozygous sites with a 306 CADD score above 20 (Figure 2B; PGLS: λ = 0, R2 = 0.61, F1-30 = 45.99, p = 1.6e-7, and Figure 307 2E; PGLS: λ = 0, R2 = 0.91, F1-28 = 6.74, p = 0.015). In contrast, homozygous load, measured 308 as the corrected ratio of homozygous substitutions with a CADD score above 20, showed a 309 statistically significant but weak association with heterozygosity (Figure 2C; PGLS: λ = 0.81, 310 R2 = 0.16, F1-30 = 5.61, p = 0.0245), and no association with FROH (Figure 2F; PGLS: λ = 0.86, 311 R2 = 0.02, F1-28 = 0.61, p = 0.4424). However, given the detected phylogenetic signal in the 312 homozygous load comparisons ( λ ≈ 0.8 ), we further explored these relationships using a 313 PGLMM approach. The results supported the previous described associations and revealed 314 that phylogenetic relatedness does not play a predominant role in homozygous load variation 315 (see Table S3). 316 317 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 11 318 Figure 2. Comparison between genetic diversity metrics, genetic load, and effective population sizes. The 319 dashed line represents the linear correlation when p<0.05. A Correlation between genome -wide heterozygosity 320 and effective population size (Ne), with Ne estimated as the harmonic mean of PSMC values between 10 kya and 321 100 kya. B Correlation between heterozygous load and genome-wide heterozygosity. Heterozygous load is defined 322 as the ratio of heterozygous substitutions with CADD scores above 20 to homozygous substitutions with CADD 323 scores below three. C Correlation between putatively homozygous load and genome -wide heterozygosity. 324 Homozygous load is defined as the ratio of counts of filtered homozygous substitutions with CADD scores above 325 20 and the number of filtered homozygous substitutions with CADD scores below three. D Correlation between 326 inbreeding coefficient (based on runs of homozygosity; F ROH) and Ne. E Correlation between heterozygous load 327 and inbreeding coefficient. F Correlation between putatively homozygous load and inbreeding coefficient. 328 329 Inbreeding, genetic diversity and load across taxonomic groups 330 Despite similar distribution ranges and histories of population decline in the past decades 331 (Figure 1A), the three Mauritius species showed contrasting levels of genetic heterozygosity, 332 inbreeding coefficients (Figure 1C, 3AB), and homozygous load (Figure 3C). The Mauritius 333 kestrel had the lowest heterozygosity of all samples (8.33 x 10 ⁻⁵ het x bp -1) and the second 334 highest FROH (0.71) among the wild species included in this study, with 50% of its genome in 335 very long ROHs (>10 Mb), as evidence of sustained recent inbreeding after recovering 336 demographically from a bottleneck of only four individuals. The echo parakeet had the second 337 lowest heterozygosity (8.07x10 ⁻⁴ het x bp -1) among parrots, closely following another 338 extremely bottlenecked species, the critically endangered Kākāpō ( Strigops habroptila ). 339 However, the echo parakeet's FROH, while high (0.40), is lower than that of the Kākāpō (0.69), 340 with 24.7% of their genome in ROHs longer than 10 Mb, evidence of their extreme bottleneck 341 of ~12 individuals. The pink pigeon exhibits a heterozygosity of 2.38 × 10 ⁻³ het x bp -1, the 342 lowest among the analysed pigeons, but higher than that of more than half of the species 343 included in the study and nearly 30 times greater than that of the Mauritius kestrel. Additionally, 344 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 12 the pink pigeon showed an FROH of 0.26, with 12.3% of its genome in ROHs of a length longer 345 than 10 Mb (Figure S9). 346 347 Figure 3. Genetic diversity, genetic load, and demographic history of Falconiformes, Psitaciformes, and 348 Columbiformes. A Genome-wide heterozygosity in het x bp -1, B inbreeding coefficient, and C homozygous load 349 distribution across the orders of the three target species. The inbreeding coefficient (F ROH) was estimated using 350 runs of homozygosity, and homozygous load was based on the ratio of homozygous substitutions with CADD 351 scores above 20 to those with CADD scores below 3. D Demographic histories are shown as variation in Ne 352 (effective population size) inferred with PSMC. Thick lines refer to Mauritius species. Only species with 353 chromosomal-level assemblies were included. 354 The differences in genetic diversity of Mauritius species were associated with the differences 355 between their taxonomic groups (Figure 3). Falcons exhibit the lowest heterozygosity and 356 homozygous load among the three taxonomic groups. Likewise, falcons carry 23.7% less 357 homozygous load than the parrots and 29.7% less than the pigeons. Within their respective 358 taxonomic groups, the three Mauritius species showed the lowest genome -wide 359 heterozygosity (Figure 3A) and the highest FROH (Figure 3B). Genetic diversity estimates carry 360 the signal of ancestral population size, as the three Mauritius species had relatively low 361 population sizes within their respective taxonomic groups (Figure 3D). However, the pink 362 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 13 pigeon had a larger historical population than most studied species, including falcons and 363 parrots, which is reflected in its higher heterozygosity compared to the average levels in the 364 other taxonomic groups (Figure 3A). This reveals the importance of considering genetic 365 diversity within the context of a species’ long-term evolutionary history and taxonomic group. 366 Note that all species that we have included from Falconiformes are from the same genus, 367 Falco, with a divergence time of 12 million years (Kumar et al. 2022), which could explain the 368 small deviations for heterozygosity and homozygous load, whereas the divergence time of the 369 study Columbiformes and Psittaciformes species was roughly 16 and 55.6 milion years, 370 respectively (Figure 4). 371 Genome synteny and repetitive elements 372 The three taxonomic groups differed in degree of genome conservation, with Falconiformes 373 and Columbidae having relatively stable genomes, while Psittaciformes showed highly 374 complex rearrangements between most chromosomes and species (Figure 4). However, also 375 within Falconiformes and Columbidae, individual chromosomes showed evidence of 376 rearrangement. For example, Chr1 of the peregrine falcon (Falco peregrinus) was homologous 377 to Chr7 and Chr9 of the other Falco species (indicating a fusion or a fission), and the lesser 378 kestrel (Falco naumanni) showed evidence of intra-chromosomal inversions on both Chr2 and 379 Chr4. In general, closely related species are expected to show maintained synteny, and in line 380 with this reasoning, the three Mauritius species showed well-maintained synteny to the closest 381 related species included in the comparison (Figure 4). 382 383 Figure 4. Pattern of chromosome synteny within Falconiformes, Columbiformes and Psittaciformes. For 384 each species, chromosomes (grey horizontal bars) are ordered according to their respective genome assemblies. 385 Syntenic relationships (i.e., syntenic blocks identified by ntSynt) between macro- and medium-sized chromosomes 386 are marked with differently coloured vertical lines, while synteny between sex chromosomes and micro -387 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 14 chromosomes is marked with grey vertical lines. On the left, the phylogenetic relationship among species is 388 provided (cladogram; branch lengths not scaled to time), with the time to the most recent common ancestor (millions 389 of years ago) indicated for each group of species. The names of the three Mauritius species are in orange. 390 Repetitive Element Annotation 391 The average percentage of repetitive elements (RE) was 13.25% for the Columbiformes 392 (14.92% for the pink pigeon, highest among the studied pigeons), 12.85% for the 393 Falconiformes (13.67% for the Mauritius kestrel), and 14.37% for the Psittaciformes (17.63% 394 for the Mauritius parakeet). These results align with previous studies suggesting a lower 395 proportion of RE in avian genomes compared to those in mammals, with most avian species 396 presenting around 15 -20% of RE (Hughes and Piontkivska 2005) . All Columbiformes 397 presented the transposon Tc1.IS630, which was not present in Psittaciformes and 398 Falconiformes. This transposon element has been involved in structural rearrangement and 399 has a gene regulatory function (Shen et al. 2021; Wang et al. 2021). 400 In the falcons, we found a sharp difference of almost 5% in RE between the sequenced 401 species using PacBio Hi-Fi (N = 13) and Illumina short-read sequencing (N = 1, Falco cherrug) 402 techniques (Table S1). It has been shown that short -read sequences are not ideal for 403 annotating RE due to their inherent fragmentation (Mann et al. 2024) . However, the three 404 Mauritius species presented in this study have been long -read sequenced. The Mauritius 405 kestrel presented the highest percentage (0.48%) of the transposon hobo activator compared 406 to all species analyzed. The hobo transposon has been associated with developmental 407 regulatory genes, suggesting a role in the evolution of developmental gene networks (Deprá 408 et al. 2009) . It can induce transposition through the “cut and paste” mechanism, which can 409 impact developmental processes (Kim et al. 2011). 410 In the parrots, the Mauritius parakeet presented the second highest percentage of RE 411 (17.63%) compared to the rest of the Psittaciformes studied, highlighting potential lineage -412 specific retention or amplification of repetitive elements. In contrast, the blue-fronted amazon 413 (Amazona aestiva ) and kākāpō ( Strigops habroptila ) exhibited markedly lower RE 414 percentages relative to other species. Given the functional implications of RE in genome 415 organization and regulatory evolution, these differences may reflect distinct evolutionary 416 pressures shaping the genomic architecture of these taxa. 417 418 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 15 419 Figure 5. Annotated repeated elements (RE) in Columbiformes, Falconiformes and Psittaciformes. The 420 proportion of different classes of RE in the genome (different colours) and the total proportion of RE (value on top 421 of each bar) are shown for each species. 422 423 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 16

424 By analysing 39 genomes from a diverse range of avian species, we demonstrate that deep 425 demographic history exerts a lasting influence on genetic diversity and, consequently, on 426 masked genetic load. Species from different taxonomic groups exhibit different levels of 427 heterozygosity, and while their demographic history can account for some of these differences, 428 this also suggests that there might be taxonomic -specific differences in life-history traits that 429 exert an influence on how genetic diversity and inbreeding respond to demographic decline 430 and recovery. Overall, our results highlight the importance of accounting for demographic 431 history and taxonomy in interspecies comparisons of genetic metrics. While we identified some 432 challenges for standardizing genomic metrics comparisons across diverse taxonomic groups, 433 our study demonstrates that a comparative genomics framework offers powerful insights for 434 addressing key questions in biodiversity conservation. 435 Understanding conservation genetics with comparative methods 436 Leveraging the rapidly expanding availability of high-quality avian genomes (Feng et al. 2020; 437 Stiller et al. 2024) , we are now able to employ comparative phylogenetic methods to gain 438 deeper insights into genetic traits that play important roles in conservation biology (Supple and 439 Shapiro 2018; Wright et al. 2020) . We observed a negative correlation between genetic 440 diversity and FROH (Figure 1C), consistent with previous studies (Brüniche-Olsen et al. 2018; 441 Grossen et al. 2020), showing a higher risk of genetic drift and inbreeding in small populations. 442

genomes sequenced from domestic or pet samples had elevated F ROH, indicating 443 the importance of checking the resources of the samples when using public genomic data 444 (Figure S9). 445 Data on modern population sizes are undoubtedly crucial in conservation assessments 446 (Shaffer 1981; Lande 1988; Willi et al. 2006; Frankham et al. 2014). The rate of decline is one 447 of the major factors considered by the IUCN Red List rating (Frankham et al. 2014; McNeely 448 et al. 1990). The Ne/Nc (with modern Ne) ratio reflects the balance between genetic diversity 449 and current population size (Frankham 1995; Kalinowski and Waples 2002), with an increased 450 Ne/Nc ratio suggesting a rapid population decline that has not yet resulted in significant 451 genetic diversity loss measured as genome-wide heterozygosity. This highlights the prevalent 452 time-lag between population decline and genetic diversity loss (Gargiulo et al. 2024) resulting 453 from the drift debt (Gilroy et al. 2017; Dussex et al. 2023; Pinto et al. 2024; Liu et al. 2025) , 454 serving as an early warning sign of an imminent population collapse (Amos and Balmford 455 2001; Wilder et al. 2023) . Our findings indicate that the Ne/Nc ratio, even with historical Ne, 456 serves as an indicator of a species' conservation status (Figure 1D), showing the importance 457 of understanding the long-term demographic history of the species in conservation. 458 Genetic diversity has been considered a classical indicator for population resilience and risk 459 of extinction (Breed et al. 2019; DeWoody et al. 2021; Teixeira and Huber 2021; Jeon et al. 460 2024). We showed that genome-wide heterozygosity, as a measurement of genetic diversity, 461 strongly correlates with historical population size (Figure 2A) spanning 10,000 to 100,000 462 years ago, with the former corresponding to at least 500 to 5,000 generations in birds given a 463 generation time of roughly 2 -20 years (Bird et al. 2020) . This indicates a species' "genetic 464 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 17 vulnerability" to future challenges is associated with its long -term evolutionary history, even 465 prior to the accelerated environmental changes induced by human activity (Tan et al. 2023). 466 Thus, monitoring both population demography and environmental threats could be important 467 to the conservation of species and populations with low genetic diversity, even if the population 468 sizes remain stable, as they are initially more likely to be vulnerable to environmental changes 469 (Ellstrand and Elam 1993; van der Valk, de Manuel, et al. 2019; Brüniche -Olsen et al. 2021; 470 Liu et al. 2025; Willi et al. 2006) . FROH, on the other hand, did not show a strong correlation 471 with the historical population size, as long runs of homozygosity reflect recent demographic 472 history within tens of generations (Ceballos et al. 2018) , highlighting the importance of 473 combining different genomic metrics to evaluate the genetic health of a species and their likely 474 short- and long-term risks. 475 As whole-genome data become more accessible, the integration of genomic-derived metrics 476 (e.g., demographic reconstructions, heterozygosity, F ROH, Ne/Nc) with demographic, 477 ecological and environmental metrics can substantially improve conservation assessments 478 and planning. Combining genomic insights with ecological data will enable more precise 479 predictions of population collapse and help prioritise efforts for species most at risk. 480 A comparative perspective on genetic load 481 Population decline often leads to the expression of masked genetic load, driven by genetic 482 drift and reduced purging (van der Valk, de Manuel, et al. 2019; Dussex et al. 2023). Here, we 483 found a strong correlation between heterozygous deleterious sites, as a proxy for masked 484 load, and genome -wide heterozygosity (Figure 2B). This suggests that species with higher 485 genetic diversity face a different type of threat under demographic and environmental changes 486 than those with lower genetic diversity. With habitat loss predicted to intensify in the near 487 future, resulting in accelerated population declines and loss of genetic diversity (Exposito-488 Alonso et al. 2022), species with currently higher diversity may face rapid exposure of these 489 deleterious mutations before effective purging can occur, increasing the risk of fitness 490 reduction and jeopardising population viability (van Oosterhout et al. 2022). 491 Conversely, species with lower genetic diversity exhibit reduced homozygous load (Figure 492 2C). Despite that one sample cannot reflect the whole picture of a species, this pattern across 493 a wide range of species likely reflects long -term purging of deleterious alleles in populations 494 with small Ne. Although we cannot rule out lineage -specific adaptive substitutions when 495 measuring homozygous load, focusing on sites with CADD > 20 within the most conserved 496 regions provides a reasonable control (Rentzsch et al. 2019; Fontsere, Speak, Caven, 497 Rodriguez, et al. 2024; Speak et al. 2024) (Figure S3). However, such purging does not 498 necessarily translate to improved fitness (Kardos et al. 2021). If this interpretation is correct, it 499 implies that the signal we measured primarily captures mildly deleterious sites, as strongly 500 deleterious alleles are typically purged quickly (Robinson et al. 2016; Dussex et al. 2021; 501 Dussex et al. 2023; Fontsere et al. 2024), and our estimation included substitutions reflecting 502 the accumulation of putatively deleterious homozygous alleles during a long evolutionary 503 process, rather than a potential negative fitness effect. This also suggests that genetic load 504 analyses tend to detect mainly mildly deleterious alleles as a proxy of the realized load 505 (Grossen et al. 2020; Dussex et al. 2023; Kardos et al. 2023; Wang et al. 2023) . Addressing 506 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 18 this challenge requires improving ancestral state inference, which currently relies on species 507 that diverged tens of thousands of generations ago. 508 Moreover, of key importance is advancing our understanding of the fitness effects of putatively 509 deleterious alleles to better integrate genetic load estimation into conservation strategies 510 (Kardos et al. 2021; Dussex et al. 2023) . This could involve incorporating direct fitness data, 511 leveraging temporal genomic data to trace load dynamics, and refining methods for detecting 512 strongly deleterious alleles (Bosse et al. 2019; van der Valk, de Manuel, et al. 2019; Bertorelle 513 et al. 2022; Kyriazis et al. 2023). Understanding the interplay between genetic load, structural 514 variation, and demographic history will be crucial for predicting species' resilience to 515 environmental change and guiding effective management. 516 Comparative genomics reveals patterns across taxonomic groups 517 The three Mauritius species exhibit low genetic diversity and high F ROH within their species 518 groups, as expected from their recent histories of population collapse. However, substantial 519 genetic differences between their taxonomic groups highlight the interplay of evolutionary 520 history, demographic processes, life-history traits and genomic architecture. For instance, the 521 pink pigeon exhibits higher heterozygosity than most Falco species (Figure 3A), whereas 522 Falco species show significantly lower homozygous load compared to parrots and pigeons 523 (Figure 3C). These patterns underlie the effect of long-term effective population sizes (Figure 524 3D). On the other hand, F ROH values based on long ROHs have similar values across 525 taxonomic groups (Figure 3B), as these do not reflect the effect of long -term demographic 526 history. 527 Life-history traits may help further explain these patterns. Parrots, such as the Mauritius 528 parakeet, have long generation times, low reproductive rates, and high parental investment 529 (Jones and Swinnerton 1997; Jones 2010; Jones et al. 2013) , making them particularly 530 vulnerable to genomic erosion. These traits slow the recovery of genetic diversity after 531 bottlenecks and exacerbate the accumulation of homozygous load. In contrast, pigeons, such 532 as the pink pigeon, have shorter generation times and higher reproductive rates (Jones 2013), 533 facilitating faster recovery and preserving higher genetic diversity despite similar population 534 collapses. Falcons, including the Mauritius kestrel, exhibit intermediate traits, with low 535 reproductive rates but shorter generation times and the ability to disperse to new environments 536 (Jones et al. 1995; Cartwright et al. 2014; Nicoll et al. 2021), which can limit genetic drift and 537 inbreeding but may not fully mitigate the effects of historically small population sizes. However, 538 the connection between life -history traits and genetic traits remains to be better studied 539 (Germain et al. 2023). 540 Distinct patterns in genomic architecture further illustrate the importance of considering 541 phylogenetic context in conservation genetics. Falcons exhibit reduced levels of Long 542 Interspersed Nuclear Elements (LINEs), while parrots have higher proportions of Long 543 Terminal Repeats (LTRs), suggesting taxonomic variation in transposable element 544 composition (Kapusta and Suh 2017; Benham et al. 2024) . These differences may influence 545 key genomic processes, including gene regulation and alternative splicing, potentially shaping 546 species' adaptive capacities (Lin et al. 2009; Schmitz and Brosius 2011; Chénais et al. 2012; 547 Casacuberta and González 2013). Similarly, parrots are known to experience more frequent 548 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 19 chromosome reshuffling (Z. Huang et al. 2022) , a pattern confirmed in our study (Figure 4). 549 Chromosome rearrangements, such as inversions, may impact the evolution of genetic load 550 by reducing recombination and preserving deleterious alleles (Jay et al. 2021; K. Huang et al. 551 2022). Beyond traditional metrics of genetic diversity, chromosome rearrangements and 552 transposable elements may become more important in our understanding of genomic erosion, 553 although their links to conservation outcomes remain underexplored. Future studies 554 incorporating high-resolution genomic analyses across diverse taxa could help clarify the role 555 of structural variation in shaping species' evolutionary trajectories and their responses to 556 ongoing environmental challenges (Brüniche-Olsen et al. 2021). 557 Our findings illustrate the importance of accounting for taxonomy, genomic architecture, and 558 life-history traits when comparing species of conservation concern to closely related taxa (e.g., 559 Robinson et al. 2018, 2019; Grossen et al. 2020) . While comparative methods provide 560 valuable insights, future research could investigate how life -history traits and genomic 561 features, such as transposable elements and chromosome rearrangements, interact with 562 genomic erosion and demographic change. 563 Future directions to benefit conservation genomics 564 Our study underscores the value of expanding the availability of reference genomes to 565 enhance the utility of genomic resources for conservation biology (Grueber 2015; Supple and 566 Shapiro 2018) . Given the substantial genetic differences observed between groups, it is 567 advantageous, even in the absence of a species -specific reference genome, to identify a 568 closely related reference genome to achieve better accuracy for certain analyses and 569 inferences (Prasad et al. 2022). However, it is important to recognise that population genetic 570 data are crucial, as a single individual cannot fully represent the genetic diversity of an entire 571 species. Intra -species genetic variation and structure should be considered (Gutiérrez-572 Espeleta et al. 2000; Bowen et al. 2005; Turchetto et al. 2016), and relying on data from one 573 or a few individuals may inadvertently capture outliers (Figure S9). Incorporating population -574 level data also allows for more robust estimates of realised genetic load by leveraging site 575 frequency distributions (Grossen et al. 2020; Bertorelle et al. 2022). 576 Our results also emphasise the importance of integrating demographic history at multiple 577 temporal scales to correctly interpret genetic diversity trends. Historical genetic data plays a 578 critical role in accurately assessing trends in population size and genetic diversity (van der 579 Valk et al. 2019; Femerling et al. 2023; Cavill et al. 2024; Dehasque et al. 2024; Silver et al. 580 2024; Fontsere et al. 2024) . Such data are invaluable for identifying rapid declines in 581 population size and genetic health, which may pose significant conservation risks of genetic 582 erosion (Díez-del-Molino et al. 2018). 583 Comparative genomics offers a powerful framework for understanding how evolutionary 584 history, demographic processes, and life-history traits shape genetic diversity across species 585 (Bertorelle et al. 2022; van Oosterhout et al. 2022) . By identifying commonalities and 586 differences among taxonomic groups, this approach can inform targeted strategies, such as 587 selecting species for genetic rescue or identifying populations most vulnerable to 588 environmental change. Future research should explore how genomic features, such as genetic 589 load, genome synteny and transposable elements, interact with population collapse and 590 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 20 environmental change to influence species’ resilience (Díez-del-Molino et al. 2018; van 591 Oosterhout et al. 2022; Germain et al. 2023) . Moreover, integrating genomic data with 592 ecological models and leveraging emerging tools like AI can provide a more holistic 593 understanding of biodiversity conservation (van Oosterhout 2024). 594 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 21 Acknowledgments 595 We are grateful to Anna Brüniche -Olsen and Roberto Biello for providing comments on an 596 early draft version of the manuscript. This work was supported by the European Research 597 Council (101078303); and the Swedish Research Council for Sustainable Development (2022-598 00536). Further support was obtained from the Royal Society International Collaboration 599 Awards 2020 (ICA/R1/201194), the Earth and Life Systems Alliance (ELSA), the Swedish 600 Research Council (621 -4996), the Erik Philp -Sörensen’s foundation, Science for Life 601 Laboratory (SciLifeLab), and Biodiversity and Ecosystem services in a Changing Climate 602 (BECC). Views and opinions expressed are however those of the authors only and do not 603 necessarily reflect those of the European Union or the European Research Council. Neither 604 the European Union nor the granting authority can be held responsible for them. 605 Data Availability 606 The scripts used in this study are available on GitHub: PachecoMC/CompConGen. 607 608

609 610 Allendorf FW, Luikart G, Aitken SN, Antunes A. 2013. Conservation and the genetics of populations. 611 2nd ed. Hoboken: John Wiley & Sons 612 Amos W, Balmford A. 2001. When does conservation genetics matter? Heredity 87:257–265. 613 Benham PM, Cicero C, Escalona M, Beraut E, Fairbairn C, Marimuthu MPA, Nguyen O, 614 Sahasrabudhe R, King BL, Thomas WK, et al. 2024. Remarkably High Repeat Content in the 615 Genomes of Sparrows: The Importance of Genome Assembly Completeness for Transposable 616 Element Discovery. Genome Biol. Evol. 16:evae067. 617 Bertorelle G, Raffini F, Bosse M, Bortoluzzi C, Iannucci A, Trucchi E, Morales HE, van Oosterhout 618 C. 2022. Genetic load: genomic estimates and applications in non-model animals. Nat. Rev. 619 Genet. 23:492–503. 620 Bird JP, Martin R, Akçakaya HR, Gilroy J, Burfield IJ, Garnett ST, Symes A, Taylor J, Şekercioğlu 621 ÇH, Butchart SHM. 2020. Generation lengths of the world’s birds and their implications for 622 extinction risk. Conserv. Biol. 34:1252–1261. 623 Blomqvist D, Pauliny A, Larsson M, Flodin L-Å. 2010. Trapped in the extinction vortex? Strong 624 genetic effects in a declining vertebrate population. BMC Evol. Biol. 10:33. 625 Borowiec ML. 2016. AMAS: a fast tool for alignment manipulation and computing of summary 626 statistics. PeerJ 4:e1660. 627 Bosse M, Megens H-J, Derks MFL, de Cara ÁMR, Groenen MAM. 2019. Deleterious alleles in the 628 context of domestication, inbreeding, and selection. Evol. Appl. 12:6–17. 629 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 22 Bowen BW, Bass AL, Soares L, Toonen RJ. 2005. Conservation implications of complex population 630 structure: lessons from the loggerhead turtle (Caretta caretta). Mol. Ecol. 14:2389–2402. 631 Breed MF, Harrison PA, Blyth C, Byrne M, Gaget V, Gellie NJC, Groom SVC, Hodgson R, Mills JG, 632 Prowse TAA, et al. 2019. The potential of genomics for restoring ecosystems and 633 biodiversity. Nat. Rev. Genet. 20:615–628. 634 Brüniche-Olsen A, Kellner KF, Anderson CJ, DeWoody JA. 2018. Runs of homozygosity have utility 635 in mammalian conservation and evolutionary studies. Conserv. Genet. 19:1295–1307. 636 Brüniche-Olsen A, Kellner KF, Belant JL, DeWoody JA. 2021. Life-history traits and habitat 637 availability shape genomic diversity in birds: implications for conservation. Proc. R. Soc. B 638 Biol. Sci. 288:20211441. 639 Cartwright SJ, Nicoll MAC, Jones CG, Tatayah V, Norris K. 2014. Anthropogenic Natal 640 Environmental Effects on Life Histories in a Wild Bird Population. Curr. Biol. 24:536–540. 641 Casacuberta E, González J. 2013. The impact of transposable elements in environmental adaptation. 642 Mol. Ecol. 22:1503–1517. 643 Casper J, Zweig AS, Villarreal C, Tyner C, Speir ML, Rosenbloom KR, Raney BJ, Lee CM, Lee BT, 644 Karolchik D, et al. 2018. The UCSC Genome Browser database: 2018 update. Nucleic Acids 645 Res. 46:D762–D769. 646 Cavill EL, Morales HE, Sun X, Westbury MV, van Oosterhout C, Accouche W, Zora A, Schulze MJ, 647 Shah N, Adam P-A, et al. 2024. When birds of a feather flock together: Severe genomic 648 erosion and the implications for genetic rescue in an endangered island passerine. Evol. Appl. 649 17:e13739. 650 Ceballos FC, Joshi PK, Clark DW, Ramsay M, Wilson JF. 2018. Runs of homozygosity: windows 651 into population history and trait architecture. Nat. Rev. Genet. 19:220–234. 652 Charlesworth D, Willis JH. 2009. The genetics of inbreeding depression. Nat. Rev. Genet. 10:783–653 796. 654 Chénais B, Caruso A, Hiard S, Casse N. 2012. The impact of transposable elements on eukaryotic 655 genomes: From genome size increase to genetic adaptation to stressful environments. Gene 656 509:7–15. 657 Chernomor O, von Haeseler A, Minh BQ. 2016. Terrace Aware Data Structure for Phylogenomic 658 Inference from Supermatrices. Syst. Biol. 65:997–1008. 659 Coombe L, Kazemi P, Wong J, Birol I, Warren RL. 2024. Multi-genome synteny detection using 660 minimizer graph mappings. :2024.02.07.579356. Available from: 661 https://www.biorxiv.org/content/10.1101/2024.02.07.579356v1 662 Danecek P, Bonfield JK, Liddle J, Marshall J, Ohan V, Pollard MO, Whitwham A, Keane T, 663 McCarthy SA, Davies RM. 2021. Twelve years of SAMtools and BCFtools. GigaScience 664 [Internet] 10. Available from: https://pubmed.ncbi.nlm.nih.gov/33590861/ 665 Dehasque M, Morales HE, Díez-del-Molino D, Pečnerová P, Chacón-Duque JC, Kanellidou F, Muller 666 H, Plotnikov V, Protopopov A, Tikhonov A, et al. 2024. Temporal dynamics of woolly 667 mammoth genome erosion prior to extinction. Cell 187:3531-3540.e13. 668 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 23 Deprá M, da Silva Valente VL, Margis R, Loreto ELS. 2009. The hobo transposon and hobo-related 669 elements are expressed as developmental genes in Drosophila. Gene 448:57–63. 670 DePristo MA, Banks E, Poplin R, Garimella KV, Maguire JR, Hartl C, Philippakis AA, del Angel G, 671 Rivas MA, Hanna M, et al. 2011. A framework for variation discovery and genotyping using 672 next-generation DNA sequencing data. Nat. Genet. 43:491–498. 673 DeWoody JA, Harder AM, Mathur S, Willoughby JR. 2021. The long‐standing significance of 674 genetic diversity in conservation. Mol. Ecol. 30:4147–4154. 675 Díez-del-Molino D, Sánchez-Barreiro F, Barnes I, Gilbert MTP, Dalén L. 2018. Quantifying 676 Temporal Genomic Erosion in Endangered Species. Trends Ecol. Evol. 33:176–185. 677 Dussex N, Morales HE, Grossen C, Dalén L, Oosterhout C van. 2023. Purging and accumulation of 678 genetic load in conservation. Trends Ecol. Evol. 38:961–969. 679 Dussex N, van der Valk T, Morales HE, Wheat CW, Díez-del-Molino D, von Seth J, Foster Y, 680 Kutschera VE, Guschanski K, Rhie A, et al. 2021. Population genomics of the critically 681 endangered kākāpō. Cell Genomics 1:100002. 682 Ellstrand NC, Elam DR. 1993. Population Genetic Consequences of Small Population Size: 683 Implications for Plant Conservation. Annu. Rev. Ecol. Evol. Syst. 24:217–242. 684 Exposito-Alonso M, Booker TR, Czech L, Gillespie L, Hateley S, Kyriazis CC, Lang PLM, Leventhal 685 L, Nogues-Bravo D, Pagowski V, et al. 2022. Genetic diversity loss in the Anthropocene. 686 Science 377:1431–1435. 687 Faircloth BC. 2016. PHYLUCE is a software package for the analysis of conserved genomic loci. 688 Bioinformatics 32:786–788. 689 Femerling G, van Oosterhout C, Feng S, Bristol RM, Zhang G, Groombridge J, P. Gilbert MT, 690 Morales HE. 2023. Genetic Load and Adaptive Potential of a Recovered Avian Species that 691 Narrowly Avoided Extinction.Teeling E, editor. Mol. Biol. Evol. [Internet] 40. Available 692 from: https://dx.doi.org/10.1093/molbev/msad256 693 Feng S, Stiller J, Deng Y, Armstrong J, Fang Q, Reeve AH, Xie D, Chen G, Guo C, Faircloth BC, et 694 al. 2020. Dense sampling of bird diversity increases power of comparative genomics. Nature 695 587:252–257. 696 Florens FBV. 2013. Conservation in Mauritius and Rodrigues: Challenges and Achievements from 697 Two Ecologically Devastated Oceanic Islands. In: Raven PH, Sodhi NS, Gibson L, editors. 698 Conservation Biology. 1st ed. Wiley. p. 40–50. Available from: 699 https://onlinelibrary.wiley.com/doi/10.1002/9781118679838.ch6 700 Flynn JM, Hubley R, Goubert C, Rosen J, Clark AG, Feschotte C, Smit AF. 2020. RepeatModeler2 701 for automated genomic discovery of transposable element families. Proc. Natl. Acad. Sci. U. 702 S. A. 117:9451–9457. 703 Fontsere C, Speak SA, Caven AJ, Rodriguez JA, Wang X, Pacheco C, Cassatt-Johnstone M, 704 Femerling G, Maloney B, Balacco J, et al. 2024. Persistent genomic erosion in whooping 705 cranes despite demographic recovery. Available from: 706 https://www.biorxiv.org/content/10.1101/2024.12.12.628160v1 707 Fontsere C, Speak SA, Caven AJ, Rodríguez JA, Wang X, Pacheco C, Cassatt-Johnstone M, 708 Femerling G, Maloney B, Balacco J, et al. 2024. Persistent genomic erosion in whooping 709 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 24 cranes despite demographic recovery. :2024.12.12.628160. Available from: 710 https://www.biorxiv.org/content/10.1101/2024.12.12.628160v1 711 Frankham R. 1995. Effective population size/adult population size ratios in wildlife: a review. Genet. 712 Res. 66:95–107. 713 Frankham R, Bradshaw CJA, Brook BW. 2014. Genetics in conservation management: Revised 714 recommendations for the 50/500 rules, Red List criteria and population viability analyses. 715 Biol. Conserv. 170:56–63. 716 Freckleton RP, Harvey PH, Pagel M. 2002. Phylogenetic Analysis and Comparative Data: A Test and 717 Review of Evidence. Am. Nat. 160:712–726. 718 Gargiulo R, Budde KB, Heuertz M. 2024. Mind the lag: understanding genetic extinction debt for 719 conservation. Trends Ecol. Evol. [Internet] 0. Available from: 720 https://www.cell.com/trends/ecology-evolution/abstract/S0169-5347(24)00259-3 721 Garrison E, Kronenberg ZN, Dawson ET, Pedersen BS, Prins P. 2022. A spectrum of free software 722 tools for processing the VCF variant call format: vcflib, bio-vcf, cyvcf2, hts-nim and slivar. 723 PLOS Comput. Biol. 18:e1009123. 724 Germain RR, Feng S, Chen G, Graves GR, Tobias JA, Rahbek C, Lei F, Fjeldså J, Hosner PA, Gilbert 725 MTP, et al. 2023. Species-specific traits mediate avian demographic responses under past 726 climate change. Nat. Ecol. Evol. 2023 76 7:862–872. 727 Gilroy DL, Phillips KP, Richardson DS, van Oosterhout C. 2017. Toll‐like receptor variation in the 728 bottlenecked population of the Seychelles warbler: computer simulations see the ‘ghost of 729 selection past’ and quantify the ‘drift debt.’ J. Evol. Biol. 30:1276–1287. 730 Grossen C, Guillaume F, Keller LF, Croll D. 2020. Purging of highly deleterious mutations through 731 severe bottlenecks in Alpine ibex. Nat. Commun. 11:1–12. 732 Grueber CE. 2015. Comparative genomics for biodiversity conservation. Comput. Struct. Biotechnol. 733 J. 13:370–375. 734 Gutiérrez-Espeleta GA, Kalinowski ST, Boyce WM, Hedrick PW. 2000. Genetic variation and 735 population structure in desert bighorn sheep: implications for conservation. Conserv. Genet. 736 1:3–15. 737 Hackl T, Ankenbrand MJ, van Adrichem B. 2024. gggenomes: A Grammar of Graphics for 738 Comparative Genomics. R Package Version 101 [Internet]. Available from: 739 https://github.com/thackl/gggenomes 740 Hadfield JD. 2010. MCMC Methods for Multi-Response Generalized Linear Mixed Models: The 741 MCMCglmm R Package. J. Stat. Softw. 33:1–22. 742 Hadfield JD, Nakagawa S. 2010. General quantitative genetic methods for comparative biology: 743 phylogenies, taxonomies and multi-trait models for continuous and categorical characters. J. 744 Evol. Biol. 23:494–508. 745 Hasselgren M, Dussex N, Von Seth J, Angerbjörn A, Olsen R, Dalén L, Norén K. 2021. Genomic and 746 fitness consequences of inbreeding in an endangered carnivore. Mol. Ecol. 30:2790–2799. 747 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 25 Hilgers L, Liu S, Jensen A, Brown T, Cousins T, Schweiger R, Guschanski K, Hiller M. 2025. 748 Avoidable false PSMC population size peaks occur across numerous studies. Curr. Biol. 749 35:927-930.e3. 750 Huang K, Ostevik KL, Elphinstone C, Todesco M, Bercovich N, Owens GL, Rieseberg LH. 2022. 751 Mutation Load in Sunflower Inversions Is Negatively Correlated with Inversion 752 Heterozygosity. Mol. Biol. Evol. 39:msac101. 753 Huang Z, De O. Furo I, Liu J, Peona V, Gomes AJB, Cen W, Huang H, Zhang Y, Chen D, Xue T, et 754 al. 2022. Recurrent chromosome reshuffling and the evolution of neo-sex chromosomes in 755 parrots. Nat. Commun. 13:944. 756 Hughes AL, Piontkivska H. 2005. DNA repeat arrays in chicken and human genomes and the 757 adaptive evolution of avian genome size. BMC Evol. Biol. 5:12. 758 Jackson HA, Percival-Alwyn L, Ryan C, Albeshr MF, Venturi L, Morales HE, Mathers TC, Cocker J, 759 Speak SA, Accinelli GG, et al. 2022. Genomic erosion in a demographically recovered bird 760 species during conservation rescue. Conserv. Biol. 36:e13918. 761 Jay P, Chouteau M, Whibley A, Bastide H, Parrinello H, Llaurens V, Joron M. 2021. Mutation load at 762 a mimicry supergene sheds new light on the evolution of inversion polymorphisms. Nat. 763 Genet. 53:288–293. 764 Jeon JY, Black AN, Heenkenda EJ, Mularo AJ, Lamka GF, Janjua S, Brüniche-Olsen A, Bickham 765 JW, Willoughby JR, DeWoody JA. 2024. Genomic Diversity as a Key Conservation 766 Criterion: Proof-of-Concept From Mammalian Whole-Genome Resequencing Data. Evol. 767 Appl. 17:e70000. 768 Jones CG. 2010. Back from the brink: the echo parakeet story. PsittaScene 22:3–5. 769 Jones CG. 2013. Pink Pigeon Nesoenas mayeri. In: The birds of Africa: Volume VIII: The Malagasy 770 region: Madagascar, Seychelles, Comoros, Mascarenes. Safford, R. and Hawkins, F. 771 Christopher Helm. p. 484–489. 772 Jones CG, Heck W, Lewis RE, Mungroo Y, Slade G, Cade T. 1995. The restoration of the Mauritius 773 Kestrel Falco punctatus population. Ibis [Internet] 137. Available from: 774 https://onlinelibrary.wiley.com/doi/10.1111/j.1474-919X.1995.tb08439.x 775 Jones CG, Malham J, Reuleaux A, Richards H, Raisin C, Tollington S, Zuel N, Chowrimootoo A, 776 Tatayah V. 2013. Echo parakeet psittacula eques. In: The Birds of Africa. Vol. VIII: The 777 Malagasy Region. Safford, R. J. and Hawkins, A. F. A. London: Christopher Helm. p. 517–778 522. 779 Jones CG, Swinnerton KJ. 1997. A summary of conservation status and research for the Mauritius 780 kestrel Falco punctatus, Pink Pigeon Columba mayeri and echo parakeet Psittacula eques. 781 Dodo 33:72–75. 782 Kalinowski ST, Waples RS. 2002. Relationship of Effective to Census Size in Fluctuating 783 Populations. Conserv. Biol. 16:129–136. 784 Kapusta A, Suh A. 2017. Evolution of bird genomes—a transposon’s-eye view. Ann. N. Y. Acad. Sci. 785 1389:164–185. 786 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 26 Kardos M, Armstrong EE, Fitzpatrick SW, Hauser S, Hedrick PW, Miller JM, Tallmon DA, Funk 787 WC. 2021. The crucial role of genome-wide genetic variation in conservation. Proc. Natl. 788 Acad. Sci. 118:e2104642118. 789 Kim YJ, Hice RH, O’Brochta DA, Atkinson PW. 2011. DNA sequence requirements for hobo 790 transposable element transposition in Drosophila melanogaster. Genetica 139:985–997. 791 Korneliussen TS, Albrechtsen A, Nielsen R. 2014. ANGSD: Analysis of Next Generation Sequencing 792 Data. BMC Bioinformatics 15:356. 793 Kumar S, Suleski M, Craig JM, Kasprowicz AE, Sanderford M, Li M, Stecher G, Hedges SB. 2022. 794 TimeTree 5: An Expanded Resource for Species Divergence Times. Mol. Biol. Evol. 795 39:msac174. 796 Kyriazis CC, Robinson JA, Lohmueller KE. 2023. Using Computational Simulations to Model 797 Deleterious Variation and Genetic Load in Natural Populations. Am. Nat. 202:737–752. 798 Lande R. 1988. Genetics and Demography in Biological Conservation. Science 241:1455–1460. 799 Lewin HA, Robinson GE, Kress WJ, Baker WJ, Coddington J, Crandall KA, Durbin R, Edwards SV, 800 Forest F, Gilbert MTP, et al. 2018. Earth BioGenome Project: Sequencing life for the future 801 of life. Proc. Natl. Acad. Sci. 115:4325–4333. 802 Li H. 2011. Tabix: fast retrieval of sequence features from generic TAB-delimited files | 803 Bioinformatics | Oxford Academic. Bioinformatics 27:718–719. 804 Li H. 2018. Minimap2: pairwise alignment for nucleotide sequences. Bioinformatics 34:3094–3100. 805 Li H, Durbin R. 2009. Fast and accurate short read alignment with Burrows–Wheeler transform. 806 Bioinformatics 25:1754–1760. 807 Li H, Durbin R. 2011. Inference of human population history from individual whole-genome 808 sequences. Nature 475:493–496. 809 Lin L, Jiang P, Shen S, Sato S, Davidson BL, Xing Y. 2009. Large-scale analysis of exonized 810 mammalian-wide interspersed repeats in primate genomes. Hum. Mol. Genet. 18:2204–2214. 811 Liu X, Milesi E, Fontsere C, Owens HL, Heinsohn R, Gilbert MTP, Crates R, Nogués-Bravo D, 812 Morales HE. 2025. Time-lagged genomic erosion and future environmental risks in a bird on 813 the brink of extinction. Proc. R. Soc. B Biol. Sci. 292:20242480. 814 Mann L, Balasch K, Schmidt N, Heitkam T. 2024. High-fidelity (repeat) consensus sequences from 815 short reads using combined read clustering and assembly. BMC Genomics 25:109. 816 McNeely JA, Miller KR, Reid WV, Mittermeier RA, Werner TB. 1990. Conserving the world’s 817 biological diversity. 818 Minh BQ, Schmidt HA, Chernomor O, Schrempf D, Woodhams MD, von Haeseler A, Lanfear R. 819 2020. IQ-TREE 2: New Models and Efficient Methods for Phylogenetic Inference in the 820 Genomic Era. Mol. Biol. Evol. 37:1530–1534. 821 Morales HE, Groombridge JJ, Tollington S, Henshaw S, Tatayah V, Ruhomaun K, Van Oosterhout C, 822 Gilbert MTP, Wellcome Sanger Institute Tree of Life Management, Samples and Laboratory 823 team, Wellcome Sanger Institute Scientific Operations: Sequencing Operations, et al. 2024. 824 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 27 The genome sequence of the Mauritius parakeet, Alexandrinus eques (formerly Psittacula 825 eques) (A.Newton & E. Newton, 1876). Wellcome Open Res. 9:378. 826 Morales HE, Norris K, Henshaw S, Tatayah V, Ruhomaun K, Van Oosterhout C, Groombridge JJ, 827 Gilbert MTP, Wellcome Sanger Institute Tree of Life Management, Samples and Laboratory 828 team, Wellcome Sanger Institute Scientific Operations: Sequencing Operations, et al. 2024. 829 The genome sequence of the Mauritius kestrel, Falco punctatus (Temminck, 1821). Wellcome 830 Open Res. 9:312. 831 Morales HE, Van Oosterhout C, Whitford H, Tatayah V, Ruhomaun K, Groombridge JJ, Gilbert 832 MTP, Wellcome Sanger Institute Tree of Life Management, Samples and Laboratory team, 833 Wellcome Sanger Institute Scientific Operations: Sequencing Operations, Wellcome Sanger 834 Institute Tree of Life Core Informatics team, et al. 2024. The genome sequence of the Pink 835 Pigeon, Nesoenas mayeri (Prévost, 1843). Wellcome Open Res. 9:336. 836 Morin PA, Archer FI, Avila CD, Balacco JR, Bukhman YV, Chow W, Fedrigo O, Formenti G, 837 Fronczek JA, Fungtammasan A, et al. 2021. Reference genome and demographic history of 838 the most endangered marine mammal, the vaquita. Mol. Ecol. Resour. 21:1008–1020. 839 Nicoll MAC, Jones CG, Groombridge JG, Henshaw S, Ruhomaun K, Tatayah V, Zuel N, Norris K. 840 2021. Contrasting recovery trajectories of four reintroduced populations of the Endangered 841 Mauritius Kestrel (Falco punctatus). Ibis 163:1294–1309. 842 van Oosterhout C. 2024. AI-informed conservation genomics. Heredity 132:1–4. 843 van Oosterhout C, Speak SA, Birley T, Bortoluzzi C, Percival-Alwyn L, Urban LH, Groombridge JJ, 844 Segelbacher G, Morales HE. 2022. Genomic erosion in the assessment of species extinction 845 risk and recovery potential. :2022.09.13.507768. Available from: 846 https://www.biorxiv.org/content/10.1101/2022.09.13.507768v1 847 Orme D, Rob F, Gavin T, Thomas P, Susanne F, Nick I, Will P. 2023. caper: Comparative Analyses 848 of Phylogenetics and Evolution in R. Available from: https://github.com/davidorme/caper 849 Paradis E, Claude J, Strimmer K. 2004. APE: Analyses of Phylogenetics and Evolution in R language. 850 Bioinformatics 20:289–290. 851 Paradis E, Schliep K. 2019. ape 5.0: an environment for modern phylogenetics and evolutionary 852 analyses in R. Bioinformatics 35:526–528. 853 Patwardhan MN, Wenger CD, Davis ES, Phanstiel DH. 2019. Bedtoolsr: An R package for genomic 854 data analysis and manipulation. J. Open Source Softw. 4:1742. 855 Pedersen BS, Quinlan AR. 2018. Mosdepth: Quick coverage calculation for genomes and exomes. 856 Bioinformatics 34:867–868. 857 Peona V, Palacios-Gimenez OM, Blommaert J, Liu J, Haryoko T, Jønsson KA, Irestedt M, Zhou Q, 858 Jern P, Suh A. 2021. The avian W chromosome is a refugium for endogenous retroviruses 859 with likely effects on female-biased mutational load and genetic incompatibilities. Philos. 860 Trans. R. Soc. B Biol. Sci. 376:20200186. 861 Pinto AV, Hansson B, Patramanis I, Morales HE, van Oosterhout C. 2024. The impact of habitat loss 862 and population fragmentation on genomic erosion. Conserv. Genet. 25:49–57. 863 Prasad A, Lorenzen ED, Westbury MV. 2022. Evaluating the role of reference-genome phylogenetic 864 distance on evolutionary inference. Mol. Ecol. Resour. 22:45–55. 865 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 28 Quinlan AR, Hall IM. 2010. BEDTools: A flexible suite of utilities for comparing genomic features. 866 Bioinformatics 26:841–842. 867 Renaud G, Hanghøj K, Korneliussen TS, Willerslev E, Orlando L. 2019. Joint Estimates of 868 Heterozygosity and Runs of Homozygosity for Modern and Ancient Samples. Genetics 869 212:587–614. 870 Rentzsch P, Witten D, Cooper GM, Shendure J, Kircher M. 2019. CADD: predicting the 871 deleteriousness of variants throughout the human genome. Nucleic Acids Res. 47:D886–872 D894. 873 Robinson JA, Brown C, Kim BY, Lohmueller KE, Wayne RK. 2018. Purging of Strongly Deleterious 874 Mutations Explains Long-Term Persistence and Absence of Inbreeding Depression in Island 875 Foxes. Curr. Biol. 28:3487-3494.e4. 876 Robinson JA, Ortega-Del Vecchyo D, Fan Z, Kim BY, Vonholdt BM, Marsden CD, Lohmueller KE, 877 Wayne RK. 2016. Genomic Flatlining in the Endangered Island Fox. Curr. Biol. 26:1183–878 1189. 879 Robinson JA, R\" aikk\" onen J, Vucetich LM, Vucetich JA, Peterson RO, Lohmueller KE, Wayne1 880 RK. 2019. Genomic signatures of extensive inbreeding in Isle Royale wolves, a population on 881 the threshold of extinction. Sci. Adv. 5. 882 Schmitz J, Brosius J. 2011. Exonization of transposed elements: A challenge and opportunity for 883 evolution. Biochimie 93:1928–1934. 884 Shaffer ML. 1981. Minimum Population Sizes for Species Conservation. BioScience 31:131–134. 885 Shaw RE, Farquharson KA, Bruford MW, Coates DJ, Elliott CP, Mergeay J, Ottewell KM, 886 Segelbacher G, Hoban S, Hvilsom C, et al. 2025. Global meta-analysis shows action is 887 needed to halt genetic diversity loss. Nature 638:704–710. 888 Shen D, Song C, Miskey C, Chan S, Guan Z, Sang Y, Wang Y, Chen C, Wang X, Müller F, et al. 889 2021. A native, highly active Tc1/mariner transposon from zebrafish (ZB) offers an efficient 890 genetic manipulation tool for vertebrates. Nucleic Acids Res. 49:2126–2140. 891 Silver LW, Farquharson KA, Peel E, Gilbert MTP, Belov K, Morales HE, Hogg CJ. 2024. Temporal 892 loss of genome-wide and immunogenetic diversity in a near-extinct 893 parrot. :2024.11.10.622863. Available from: 894 https://www.biorxiv.org/content/10.1101/2024.11.10.622863v1 895 Smit AFA, Hubley R, Green P. 2015. RepeatMasker open-4.0. Available from: 896 http://www.repeatmasker.org 897 Speak SA, Birley T, Bortoluzzi C, Clark MD, Percival-Alwyn L, Morales HE, van Oosterhout C. 898 2024. Genomics-informed captive breeding can reduce inbreeding depression and the genetic 899 load in zoo populations. Mol. Ecol. Resour. 24:e13967. 900 Spielman D, Brook BW, Frankham R. 2004. Most species are not driven to extinction before genetic 901 factors impact them. Proc. Natl. Acad. Sci. 101:15261–15264. 902 Stiller J, Feng S, Chowdhury A-A, Rivas-González I, Duchêne DA, Fang Q, Deng Y, Kozlov A, 903 Stamatakis A, Claramunt S, et al. 2024. Complexity of avian evolution revealed by family-904 level genomes. Nature 629:851–860. 905 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 29 Storer J, Hubley R, Rosen J, Wheeler TJ, Smit AF. 2021. The Dfam community resource of 906 transposable element families, sequence models, and genome annotations. Mob. DNA 12:2. 907 Supple MA, Shapiro B. 2018. Conservation of biodiversity in the genomics era. Genome Biol. 19:131. 908 Tan HZ, Jansen JJ, Allport GA, Garg KM, Chattopadhyay B, Irestedt M, Pang SE, Chilton G, Gwee 909 CY, Rheindt FE. 2023. Megafaunal extinctions, not climate change, may explain Holocene 910 genetic diversity declines in Numenius shorebirds.Lovette I, Rutz C, editors. eLife 12:e85422. 911 Teixeira JC, Huber CD. 2021. The inflated significance of neutral genetic diversity in conservation 912 genetics. Proc. Natl. Acad. Sci. U. S. A. [Internet] 118. Available from: 913 https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.2015096118/-/DCSupplemental. 914 Theissinger K, Fernandes C, Formenti G, Bista I, Berg PR, Bleidorn C, Bombarely A, Crottini A, 915 Gallo GR, Godoy JA, et al. 2023. How genomics can help biodiversity conservation. Trends 916 Genet. TIG 39:545–559. 917 Tollington S, Greenwood A, Jones CG, Hoeck P, Chowrimootoo A, Smith D, Richards H, Tatayah V, 918 Groombridge JJ. 2015. Detailed monitoring of a small but recovering population reveals 919 sublethal effects of disease and unexpected interactions with supplemental feeding. J. Anim. 920 Ecol. 84:969–977. 921 Tollington S, Jones CG, Greenwood A, Tatayah V, Raisin C, Burke T, Dawson DA, Groombridge JJ. 922 2013. Long-term, fine-scale temporal patterns of genetic diversity in the restored Mauritius 923 parakeet reveal genetic impacts of management and associated demographic effects on 924 reintroduction programmes. Biol. Conserv. 161:28–38. 925 Turchetto C, Segatto ALA, Mäder G, Rodrigues DM, Bonatto SL, Freitas LB. 2016. High levels of 926 genetic diversity and population structure in an endemic and rare species: implications for 927 conservation. AoB PLANTS 8:plw002. 928 van der Valk T, Diez-del-Molino D, Marques-Bonet T, Guschanski K, Dalén L. 2019. Historical 929 Genomes Reveal the Genomic Consequences of Recent Population Decline in Eastern 930 Gorillas. Curr. Biol. 29:165-170.e6. 931 van der Valk T, Díez-del-Molino D, Marques-Bonet T, Guschanski K, Dalén L. 2019. Historical 932 Genomes Reveal the Genomic Consequences of Recent Population Decline in Eastern 933 Gorillas. Curr. Biol. 29:165-170.e6. 934 van der Valk T, de Manuel M, Marques-Bonet T, Guschanski K. 2019. Estimates of genetic load in 935 small populations suggest extensive purging of deleterious alleles. bioRxiv:696831. 936 Wang S, Diaby M, Puzakov M, Ullah N, Wang Y, Danley P, Chen C, Wang X, Gao B, Song C. 2021. 937 Divergent evolution profiles of DD37D and DD39D families of Tc1/mariner transposons in 938 eukaryotes. Mol. Phylogenet. Evol. 161:107143. 939 Wang X, Peischl S, Heckel G. 2023. Demographic history and genomic consequences of 10,000 940 generations of isolation in a wild mammal. Curr. Biol. 33:2051-2062.e4. 941 Wilder AP, Supple MA, Subramanian A, Mudide A, Swofford R, Serres-Armero A, Steiner C, 942 Koepfli K-P, Genereux DP, Karlsson EK, et al. 2023. The contribution of historical processes 943 to contemporary extinction risk in placental mammals. Science 380:eabn5856. 944 Willi Y, Buskirk JV, Hoffmann AA. 2006. Limits to the Adaptive Potential of Small Populations. 945 Annu. Rev. Ecol. Evol. Syst. 37:433–458. 946 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint 30 Wright BR, Farquharson KA, McLennan EA, Belov K, Hogg CJ, Grueber CE. 2020. A demonstration 947 of conservation genomics for threatened species management. Mol. Ecol. Resour. 20:1526–948 1541. 949 950 951 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 30, 2025. ; https://doi.org/10.1101/2025.03.28.645899doi: bioRxiv preprint