1. Viviparity (live-bearing) has independently evolved from oviparity (egg-laying) in more than 100 lineages of squamates (lizards and snakes).
2. We might expect consequent shifts in selective forces to affect per-brood reproductive investment (RI = total mass of offspring relative to maternal mass) and in the way in which that output is partitioned (number versus size of offspring per brood). Based on the assumption that newly-born offspring are heavier than eggs, we predicted that live-bearing must entail either increased reproductive investment or a reduction in offspring size and/or fecundity.
3. However, our phylogenetically-controlled analysis of data on 1,259 squamate species revealed no significant differences in mean offspring size, clutch size or RI between oviparous and viviparous squamates.
4. We attribute this paradoxical result to (1) strong selection on optimal offspring sizes, unaffected by parity mode, (2) the lack of a larval stage in amniotes, favouring large eggs even in the ancestral oviparous mode, and (3) the ability of viviparous females to decrease the mass of uterine embryos by reducing extra-embryonic water stores.
5. Our analysis shows that squamate eggs (when laid) weigh about the same as the hatchlings that emerge from them (despite a many-fold increase in embryo mass during incubation). Most of the egg mass is due to components (such as water stores and the eggshell) not required for oviductal incubation.  That repackaging enables live-born offspring to be accommodated within the mother’s body without increasing total litter mass.
6. The consequent stasis in reproductive burden during the evolutionary transition from oviparity to viviparity may have facilitated frequent shifts in parity modes.

To compare oviparous and viviparous species with respect to fecundity (clutch vs. litter size), offspring size (hatchling vs. neonate mass), and reproductive investment (RI), we collected data on squamate life history traits, body sizes, and geographic ranges from the literature, museum databases and our own observations in the field, of captive reptiles and in museum collections.

We define all live-bearing species (including ovoviviparous taxa) as “viviparous” because our interest lies not in the physiology of gas and material transfer between the embryo and the air or uterus, but with the effects of carrying developing embryos inside the body cavity of the mother.  We excluded from the dataset all records of squamates that lay fixed clutch sizes of one or two eggs (Schwarz & Meiri, 2017). These are defined here by phylogenetic criteria, as members of the Gekkota (geckos and pygopodids), Dactyloidae (anoles) and Gymnophthalmidae. Because the number of eggs per clutch is tightly constrained in these taxa, their responses to the evolution of viviparity may differ from those of other squamates. In practice, excluding these taxa has little impact because, as far as is known, all anoles and gymnophthalmids and ~98% of gekkotans lay eggs (Meiri, 2018). Members of other lineages with very small brood sizes (e.g., 1 or, less often, 2 in the Solomon Island skink, Corucia zebrata; Honegger, 1985) were included because their small broods presumably reflect adaptations rather than constraints.

Data on clutch and litter (henceforth: brood) sizes and on body sizes of hatchlings, neonates and adult females are means. In cases when multiple mean values were available, we used the average between the largest and smallest reported means. If data on means were unavailable, we used the average between the largest and smallest reported numbers or sizes. To compare among animals that vary greatly in shape (e.g., heavyset lizards vs slender snakes) we converted data on snout vent lengths (SVL) (for all lizards and some snakes) or total lengths (TL, for most snakes) to masses using clade-specific equations, taking lizard leg reduction into account (Meiri, 2010, Feldman & Meiri, 2013, Feldman, Sabath, Pyron, Mayrose & Meiri 2016). For four species (Uromastyx thomasi, Lerista arenicola, Varanus timorensis and V. dumerilii) we could only obtain data on female mass, but not length, so we used mass data (of non-gravid females) directly. To calculate RIs we multiplied the mean brood size by mean hatchling/neonate mass and divided this by female mass. We only used species for which we had data for all parameters (mode of reproduction, hatchling/neonate size, brood size, female mass) as well as phylogenetic information.