1 2 3 4 The functional genetic architecture of egg-laying and live-bearing 5 reproduction 6 7 8 H. Recknagel1, M. Carruthers1,3, A. Yurchenko1,4, M. Nokhbatolfoghahai1, N. A. Kamenos2, 9 M.M. Bain1, K.R. Elmer1*. 10 11 Affiliations: 12 1Institute of Biodiversity, Animal Health and Comparative Medicine, College of Medical, 13 Veterinary & Life Sciences, University of Glasgow, Glasgow, G12 8QQ, UK. 14 2School of Geographical and Earth Sciences, University of Glasgow, Glasgow, G12 8QQ, 15 UK. 16 3current address: School of Biological Sciences, University of Bristol, Bristol, UK. 17 4current address: Inserm U981, Gustave Roussy Cancer Campus, Université Paris Saclay, 18 Villejuif, France. 19 20 Abstract: 21 All amniotes reproduce either by egg-laying (oviparity), which is ancestral to vertebrates, or 22 by live-bearing (viviparity), which has evolved many times independently. However, the 23 genetic basis of these parity modes has never been resolved and consequently the homology 24 across evolutionary scales is currently unknown. Here we describe the functional genes and 25 genetic architecture of parity mode and its key traits, eggshell and gestation length, by 26 leveraging natural hybridisations between oviparous and viviparous lizards, and comparing 27 our findings across vertebrates. In lizards, parity trait genes were associated with 28 progesterone-binding functions and enriched for tissue remodelling and immune system 29 pathways. Viviparity involved more genes and complex gene networks than did oviparity. 30 Angiogenesis, vascular endothelial growth, and adrenoreceptor pathways were enriched in 31 the viviparous female reproductive tissue, while pathways for transforming growth factor 32 were enriched in the oviparous. Natural selection on these parity mode genes was evident 33 genome-wide. Our analysis of seven independent origins of viviparity in mammals, 34 squamates and fish, showed that genes active in pregnancy related to immunity, tissue 35 remodelling, and blood vessel generation. Therefore, our results demonstrate that pre- 36 established regulatory networks are repeatedly recruited for viviparity and that these are 37 shared at deep evolutionary scales. 38 2 39 Whether offspring hatch from calcified eggs (oviparity) or emerge from the mother via live 40 birth (viviparity) is a fundamental and dichotomous reproductive strategy with wide-ranging 41 consequences for animal ecology, development, physiology and evolution1. The difference 42 between these parity modes is complex; however, major transitions from oviparity to 43 viviparity evolved repeatedly across amniotes. All birds, most reptiles and a few mammals 44 retain the ancestral state of oviparity, while nearly all mammals and ~20% of reptiles are 45 viviparous. Mammals transitioned to viviparity once, while squamate reptiles have 46 independently transitioned perhaps more than 100 times1,2. Nonetheless, the reproductive 47 physiology, neuroendocrine, and developmental pathways regulating viviparous 48 reproduction are thought to be highly analogous among amniotes3–7. Gathering evidence to 49 test this hypothesis has been impeded by the deep evolutionary divergences between 50 oviparous–viviparous sister taxa, which obscure inference and preclude crossing experiments. 51 Consequently, the genetic bases of the parity modes have not been resolved for any 52 animal3,4,8. 53 Major adjustments to embryo development are necessary for the transition from 54 oviparity to viviparity including i) an increase in the internal gestation time, to the point 55 where the embryo is fully developed prior to parturition, and ii) the loss of a calcified 56 eggshell9. Simultaneously, changes are necessary for the mother and for the healthy 57 development of the embryo3,4, such as an active physiological exchange of gas, water and 58 calcium and an adjusted immune response for maternal–foetal communication4,10, which is 59 facilitated by uterine angiogenesis and membrane vascularisation3,4. To prevent abortion, the 60 mother’s immune response is suppressed, for example, via the down-regulation of pro- 61 inflammatory cytokines4. These developmental and physiological changes necessitate 62 coordinated integration, making the transition a step change; intermediate phenotypes are 63 considered a fitness valley, and most species are fixed for oviparity or viviparity11. 3 64 Here, we identify the functional genomic architecture of alterative parity modes— 65 oviparity and viviparity—using an exceptional natural model. The Eurasian common lizard 66 (Zootoca vivipara) is viviparous across the majority of its distribution; however, some 67 southern populations are oviparous. In this reproductively bimodal species, alternative parity 68 modes are genetically fixed but can interbreed12,13. These lizards are therefore uniquely 69 informative to definitively genetically map the bases of the parity modes, demonstrate the 70 functional genetic mechanisms and targets of natural selection and thereby infer the 71 molecular underpinnings of the egg-laying and live-bearing reproduction modes. 72 73 Egg-laying and live-bearing lizards hybridise 74 We intensively sampled a contact zone between oviparous and viviparous common lizards 75 with a divergence time of ~4 Mya14 where hybridisation had been speculated13,15. We 76 quantified the parity mode traits for pregnant females and their clutches (N = 606) (Extended 77 Data Fig. 1): i) embryonic stage at oviposition/parturition, ii) number of days a clutch was 78 externally incubated, iii) calcium content (% Ca) within eggshells/membranes and iv) 79 eggshell/membrane thickness. In addition, we genotyped adults (ddRADSeq12), resulting in 80 80,696 genome-mapped16 polymorphic loci (N = 831, including the phenotyped females). 81 As expected, oviparous and viviparous clutches differed substantially. Eggs from 82 oviparous females had a thick shell (49.8 ± 0.8 μm) covered with a dense layer of calcium 83 crystals (25.4 ± 1.3% Ca), encircling embryos laid at an early developmental stage (stage 84 31.1 ± 0.1) requiring extended external incubation (35 ± 0.2 days) (Figs. 1A–1D) before they 85 hatched at full development (stage 41). Conversely, offspring from viviparous females were 86 delivered fully developed (stage 41), surrounded by a thin uncalcified membrane (4.2 ± 0.1 4 87 μm; 1.6 ± 0.6% Ca), which they emerged from within a short time (1.9 ± 0.1 days) (Figs. 1R– 88 1U) (Extended Data Fig. 1). 89 Despite these dramatic differences in reproduction traits, phenotypic and genotypic 90 data revealed oviparous–viviparous hybrids. We found that 14.0% of adults inherited at least 91 10% of their genome from the alternative parity mode and that 6.1% of adults were first- 92 generation (F1) hybrids (Fig. 1V). F1 hybrid females (45–55% genomic admixture) were 93 intermediate between viviparous and oviparous characteristics with respect to eggshell 94 thickness (mean: 26.7 ± 2.3 μm; 13.0 ± 2.9% Ca), gestation time (mean: 25.8 ± 1.4 days) and 95 developmental stage at oviposition (mean: stage 33.9 ± 0.4) (Figs. 1J–1M and 1V). To the 96 best of our knowledge15,17, this is the first definitive evidence of contemporary natural 97 hybridisation between oviparous and viviparous lizards. 98 99 The genetic bases of the parity modes 100 This striking hybridisation between parity modes presents a natural genetic crossing 101 experiment for determining the genetic architectures of oviparity and viviparity. Using 102 genotypic and phenotypic data from females (N = 458; 80,696 single nucleotide 103 polymorphisms (SNPs)), we admixture-mapped two essential reproductive traits, i) the 104 gestation time (a score combining the number of external incubation days and the embryonic 105 stage at oviposition/parturition per clutch) and ii) the eggshell traits (a score calculated from 106 the eggshell/membrane thickness and calcium content), using linear mixed models (Extended 107 Data Figs. 2, 3). To identify relevant genes and biological processes, we estimated the 108 chromosome-wide linkage disequilibrium (LD) decay from a whole genome resequencing 109 dataset (N = 65 admixed females; mean genome coverage 4.1× each) (Supplementary Table 5 110 1). Genes within the same chromosome-specific LD-window as the SNPs associated with 111 gestation time and/or eggshell traits (Extended Data Fig. 4) were considered candidate genes. 112 We found that eggshell traits and gestation time were both highly genetically 113 determined and that our experiment had the ability to resolve these traits with great fidelity, 114 even in this complex natural population. The phenotypic variance explained by the genotype 115 data was very high: gestation time = 0.97 (95% CI = 0.96–0.98) and eggshell traits = 0.98 116 (95% CI = 0.96–0.98) (Supplementary Table 2). Genome-wide, 221 genetic variants (488 117 genes) were significantly associated with reproductive traits (adjusted P < 0.01). More 118 genetic variants were associated with the gestation time (210 SNPs, 439 candidate genes) 119 than with the eggshell traits (17 SNPs, 38 genes), indicating that the genetic basis of embryo 120 retention is more complex than that of the eggshell traits (Supplementary Table 3). From the 121 17 SNPs associated with eggshell traits, an excess was shared with gestation time (6 when the 122 expected shared SNPs by chance is 0.05; P < 0.0001; 11 genes). These SNPs were located on 123 a locus on chromosome 7 (two SNPs within 100 bp) and on a locus on chromosome 3 (two 124 SNPs within 100 bp), and two SNPs were located on unplaced scaffolds. The candidate genes 125 were significantly enriched for eight pathways, including the immune system response (Fas 126 signalling and T-cell activation) and tissue remodelling (the cadherin signalling pathway) 127 (Supplementary Table 4). 128 For gestation time, significant associations were found on most chromosomes (total 129 accumulated span of 41.2 Mb; Supplementary Table 3; Fig. 2a; Extended Data Fig. 5), with 130 an excess on chromosome 8 (64 significant SNPs; total span of 19.4 Mb) (Fig. 2a). The 131 candidate genes were significantly enriched (P < 0.01) for processes associated with cell 132 growth, proliferation and death (Supplementary Table 5), indicative of tissue remodelling.