A live-bearing reproductive strategy can induce large morphological changes in the mother during pregnancy. The evolution of the placenta in swimming animals involves a shift in the timing of maternal provisioning from pre-fertilization (females supply their eggs with sufficient yolk reserves prior to fertilization) to post-fertilization (females provide all nutrients via a placenta during the pregnancy). It has been hypothesised that this shift, associated with the evolution of the placenta, should confer a morphological advantage to the females leading to a more slender body shape during the early stages of pregnancy. We tested this hypothesis by quantifying three-dimensional shape and volume changes during pregnancy and in full-grown virgin controls of two species within the live-bearing fish family Poeciliidae: Poeciliopsis gracilis (non-placental) and Poeciliopsis turneri (placental). We show that P. turneri is more slender than P. gracilis at the beginning of the interbrood interval and in virgins, and that these differences diminish towards the end of pregnancy. This study provides the first evidence for an adaptive morphological advantage of the placenta in live-bearing fish. A similar morphological benefit could drive the evolution of placentas in other live-bearing (swimming) animal lineages.
Citation: Fleuren M, Quicazan-Rubio EM, van Leeuwen JL, Pollux BJA (2018) Why do placentas evolve? Evidence for a morphological advantage during pregnancy in live-bearing fish. PLoS ONE 13(4): e0195976. https://doi.org/10.1371/journal.pone.0195976
Editor: Iman Borazjani, Texas A&M University System, UNITED STATES
Received: June 17, 2017; Accepted: April 3, 2018; Published: April 16, 2018
Copyright: © 2018 Fleuren et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: This study was supported by ALW Open Programme grant 821.02.024 to J.L.V.L. and VIDI grant 864.14.008 of the Netherlands Organisation for Scientific Research to B.J.A.P. and a Francisco Jose de Caldas doctorate grant from the Administrative Department of Science, Technology and Innovation (COLCIENCIAS) call 512-2010 to E.M.Q. The authors declare no conflict of interest. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
The placenta, defined as an intimate apposition or fusion of maternal and foetal tissues for physiological exchange , has evolved many times independently throughout the animal kingdom (e.g. in invertebrates, fish, amphibians, reptiles and mammals; [2–6]), including at least eight times within the live-bearing fish family Poeciliidae [7–10]. Despite the repeated emergence of placentas among widely diverged animal lineages, it is still unclear what selective forces drive the evolution of placental organs. Three non-mutually exclusive adaptive hypotheses have been proposed to explain why the placenta may have evolved in Poeciliid fish: the resource availability hypothesis, the life history facilitation hypothesis and the locomotor cost hypothesis.
The resource availability hypothesis suggests that the evolution of the placenta and associated reduction in egg size at fertilization might allow females to attain a higher fitness through increased litter sizes. A critical assumption of this hypothesis is that females must be able to abort embryos when facing adverse food conditions . Recent empirical studies in Poeciliidae, however, show that they are not able to do this, suggesting that the conditions under which the placenta might be favoured by natural selection are, at least in this taxonomic group, restricted to environments characterized by high and stable resource conditions [12–15].
The life history facilitation hypothesis states that the placenta might evolve to facilitate the evolution of other life history traits, for example to enable organisms to mature at an earlier age or to produce more or larger offspring that have a higher early-life survivorship [16–21]. However, recent studies in Poeciliidae show that there are no consistent associations between placentation and life history traits, arguing against this hypothesis as a likely explanation for the evolution of the placenta in this taxonomic group [22–24].
Finally, the locomotor cost hypothesis argues that the placenta might evolve to offset some of the locomotor cost associated with a live-bearing mode of reproduction. The physical and physiological burden of a pregnancy negatively affects a female’s locomotor performance in a broad range of live-bearing animals (e.g. scorpions, ; fishes, [26–29]; reptiles, [30,31]; and mammals, [32,33]). In aquatic animals an increase in abdominal volume may locally limit axial bending and, furthermore, enlarge frontal surface area thereby increasing the drag forces on the body [26,33–35]. An increase in body mass during pregnancy could reduce the ability to rapidly accelerate . It has been postulated that the evolution of the placenta reduces a female’s mean reproductive allotment (RA, the proportion of female mass allocated to developing offspring) during gestation, thereby reducing the distention of the female’s abdomen during the pregnancy without sacrificing her reproductive output [9,36,37]. The argument is that the evolution of the placenta coincides with a shift in the timing of maternal provisioning from pre-fertilization nutrient allocation by building up large amounts of yolk reserves in the eggs prior to fertilization, to the allocation of nutrients after fertilization (via a placental organ throughout the pregnancy). Livebearing species that allocate nutrients prior to fertilization will start with a high RA (and hence a high burden) at the beginning of their pregnancy, because they produce large fully-yolked eggs. Placental species on the other hand species will start with a low RA, because they produce smaller eggs that contain little to no resources and instead rely on nutrient provisioning during gestation. Theory thus predicts that placental females should have a lower reproductive burden (e.g. lower total volume and frontal surface area) at the start of the pregnancy that diminishes over the course of gestation (Fig 1; [9,36,37]). This morphological advantage may improve their locomotor performance (e.g. predator evasion ability; ) and hence survival  without sacrificing reproductive output, i.e. the locomotor cost hypothesis [9,36,37]. While it is known that Poeciliidae increase in body mass and frontal surface area during pregnancy , it is still unclear if, when, and to what extent, the evolution of a placenta alleviates this reproductive burden during pregnancy (Fig 1).
Similar plots could be constructed for frontal or wetted surface area. For heuristic purposes the temporal patterns are assumed linear, because the exact shape of the relationship between female volume and time is currently unknown .
Here, we set out to test the morphological predictions of the locomotor cost hypothesis by comparing body shape changes during gestation in two closely-related sister species within the live-bearing fish genus Poeciliopsis (Family Poeciliidae). These species differ markedly in the way they provision their developing embryos: Poeciliopsis gracilis lacks a placenta and instead allocates all resources necessary for embryo development to the eggs before fertilization (lecithotrophy), while Poeciliopsis turneri has a well-developed placenta (placentotrophy; [7,10]). Specifically, we test whether, compared to P. gracilis, the placental species P. turneri has (1) a lower body volume and frontal surface area at the beginning of the interbrood interval and in virgin controls, and (2) a stronger increase in volume and frontal surface area (i.e. have a steeper slope) when pregnancy progresses, indicating that the potentially beneficial reduction in body volume associated with placentation will be greatest at the beginning of the pregnancy and will gradually diminish towards the end of the interbrood interval (Fig 1).
Material & methods
A detailed description of the used species and their origins, fish rearing protocols and (pre-)experimental husbandry is provided in S1 Text. All procedures were approved by the Animal Ethics Committee of Wageningen University & Research (permit number 2013103). All efforts were made to minimize suffering.
Time schedule and sample size
We studied changes in body shape during the pregnancy of Poeciliopsis gracilis (lecithotrophic) and Poeciliopsis turneri (placentotrophic) by creating a series of 3D body reconstructions. For each female, these models were created at evenly spaced time points during one interbrood interval (IB), defined as the period between two parturitions starting the day after a female gave birth (hereafter referred to as IB = 0) and lasting until the next parturition (IB = 1). Poeciliopsis turneri was measured every second day and P. gracilis was measured every fourth day. This served to maintain an approximately equal number of measurements per individual, as the interbrood interval lengths varied between species due to differences in the level of superfetation. Superfetation refers to a reproductive strategy in which females carry multiple broods at different developmental stages [39–41]. Assuming an equal embryo development time, species with a higher level of superfetation (i.e. more simultaneous overlapping broods) will have shorter interbrood intervals (defined as the period between two parturition events) compared to species with lower levels of superfetation [9,42]. Due to the presence of superfetation, IB = 1 does not represent the birth of embryos which eggs were fertilized at IB = 0, but that of an antecedent brood.
To avoid an effect of feeding on body shape (i.e. abdominal extension), the feeding schedules of both species included a 16–24 h food deprivation period prior to the measurements (see S1 Text for further information regarding feeding). Our final dataset comprised 246 three-dimensional body models for 10 pregnant (plus 10 virgin control) P. gracilis and 14 pregnant (plus 14 virgin control) P. turneri. Of these 246 data points, six P. turneri models were omitted preceding analysis because these females were fed shortly before imaging.
Creation of three-dimensional body models
To create a single body model, a fish was first transferred to a small tank (8 × 8 × 8 cm) with scale bars for image calibration on all walls. Orientation of the female was limited by a separate movable divider. Three photos were taken simultaneously with three Nikon D3200 DSLR cameras (Nikon, Tokyo, Japan; sensor resolution 24 Mpix, equipped with Micro-Nikkor f = 55mm lenses), synchronized with a remote trigger (JinJiaCheng Photography Equipment Co., Ltd., Shenzhen, China) and with LED lights behind glass fibre cloths opposite to the cameras providing diffuse back lighting. The three orthogonally placed cameras yielded a lateral, ventral and rostral/caudal view of the fish. Multiple sets of photos were taken during one measurement session; for further analysis a set of three synchronized pictures was selected in which the fish was in a straight and minimally rotated position.
These photos were subsequently processed with an in-house developed program in MATLAB 2013a (MathWorks, Natick, MA, United States), adapted from a program previously described by . The longitudinal axis of the fish was defined by a straight line between the most anterior point of the snout and the most posterior part of the caudal peduncle (Standard Length, LSL; white lines in Fig 2A). The longitudinal axis of the fish consisted of on average 3122 (SE ± 14) pixels and 3193 (± 15) pixels in the lateral and ventral views respectively. Outlines of trunk and eyes were manually digitized (Fig 2A, blue and red lines respectively) as was the position of the abdomen of the fish (lateral view only; delimited by the dorsal edge of the vertebral column, swim bladder and the bottom of the abdomen; Fig 2A, orange line). After cubic spline interpolation of the outlines, the position of the outlines with respect to the longitudinal axis was measured at 251 equidistant points along the longitudinal axis. Using cubic spline interpolation, these points were subsequently converted into ellipse-like cross-sections, that differed in shape depending on whether the section was located in the abdominal region (Fig 2B). In the abdominal region, the minor axis is shifted to half-way the abdominal polygon at that section (default at centre of major axis). Cubic spline interpolations were also used to create a 3D-model of the eyes (with a cubic spline resembling a super-ellipse), which was then stitched to the trunk to create a full 3D-model (Fig 2C).
(A) Lateral and ventral photographs in which the trunk (green), abdomen (orange) and eyes (red) are outlined by manually indicated polygons. The longitudinal axis is depicted by white lines. (B) At 251 equidistant points along the longitudinal axis, the width and height of the polygons are converted into ellipse-like cross-sections; in the abdominal area, the vertical position of the horizontal axis is shifted. (C) Stitching the cross-sections of trunk and eyes results in a 3D model from which volume, wetted surface area and frontal surface area (projection at the right) can be calculated. For illustrative purposes these examples only consist of one-fourth of the number of cross-sections.
From these 3D models maximum height, maximum width, frontal surface area (area of frontal view projection, Fig 2C), wetted surface area (total body surface area), and volume were calculated. Maximum height was determined for the whole body, while maximum width was determined for the abdominal region only as the level of opercular distention (the moment in the breathing cycle) caused the position of the widest point to fluctuate in the slender-most fish. These fluctuations only had a minimal effect on the measured total body volume (on average 0.53% with respect to the instance with the least distention). To correct for the effect of intra- and interspecies differences in body size, maximum values of all one-dimensional parameters were normalized by dividing the values by LSL, surface areas by dividing by LSL2 and volume was normalized by dividing by LSL3.
Litter wet mass
To get an estimate of the partial reproductive allotment at IB = 1, all new-borns from the litter were caught on the day of delivery and euthanized with a lethal dose of MS-222 (Tricaine-S; Western Chemical Inc., Ferndale, WA, United States). Total litter wet mass was measured after carefully removing excess liquid with a paper towel on a Mettler AE200 analytic balance (scale accuracy 0.0001 g; Mettler-Toledo B.V., Tiel, The Netherlands). Litter wet mass provides a better approximation of reproductive burden than litter dry mass, because the water content of the embryos contributes to the total volume of the brood. Not all P. gracilis litters could be weighed; however, since offspring size did not differ between females (Mixed model, F9,19 = 1.31, P = 0.2953) total litter wet mass was instead estimated using a linear fit between offspring number and measured litter wet mass (wet mass (g) = 0.0078 ∙ nnew-borns; R2 = 0.9469).
The change in morphological parameters was modelled as a two-level longitudinal growth model , using the Mixed procedure in SAS version 9.3 (SAS Institute, Cary, NC, United States) under restricted maximum likelihood (REML). This multi-level modelling (MLM) method compares individual growth trajectories between species, allows time to be processed as a continuous variable and is able to handle unbalanced and missing data [44,45]. The model consists of two levels, the level-one model (Eq 1) that represents individual change trajectories, and the level-two model (Eqs 2 and 3) that provides intercept and slope term for the sample average. For each individual (i) and time point (j), the measured parameter is a function of the individual intercept (αi), the individual growth trajectory (βi ∙ Tj) and a random error term for that specific individual and time point (εij). Litter wet mass (wi) at IB = 1 was added as a covariate in the level-two model for slope (Eq 3), as arguably larger broods result in increasingly larger morphological parameters due to a higher growth rate. This also allows comparison of the morphologies without the effect of offspring wet mass. The common intercept (ɤ11), linear slope (ɤ21) an covariate (ɤ23) terms in Eqs 2 and 3 represent the values for P. gracilis while the ɤ12, ɤ22 and ɤ24 terms represent the added difference for P. turneri for intercept, linear slope and covariate values respectively; ζ1i and ζ2i factor individual random error terms. ‘Variance components’ was used as covariance structure (default in SAS Proc Mixed), denominator degrees of freedom were calculated with Kenward-Roger and significance level alpha was set to 0.05 (default in SAS Proc Mixed). To compare model parameter estimates, post-hoc tests were performed using ‘contrast’ and ‘lsmeans’ statements. Virgin data were analysed using a similar MLM method, albeit with a simpler model. Because we did not expect any time-dependent effects, and the virgin controls did not have litter wet mass to use as a covariate, the model consisted solely of an effect of species.
Transformation of the data occasionally resulted in slightly better fits as indicated by marginally higher R2-values from linear fits (using Proc GLM in SAS version 9.3), but using transformed data did not change the outcomes of the previously mentioned statistical models. Furthermore, we did not have any a priori expectations for the curve of the line. Therefore, we opted to use the original untransformed data and a linear depiction of change.(1)(2)(3)
Type 3 tests for Fixed Effects for both the pregnant and the virgin MLM model can be found in S1 Table. All fixed effects in the model were significant, for all measured morphological parameters.
Morphological changes during pregnancy
At the beginning of the interbrood interval (IB = 0), Poeciliopsis gracilis females have an overall larger normalized body size than females of Poeciliopsis turneri. Except for maximum width (MLM contrast of intercepts: F1,21.2 = 2.52, P = 0.1275; Fig 3A), females of P. gracilis have a higher maximum height (F1,21.4 = 15.46, P = 0.0007; Fig 3B), frontal surface area (F1,21.6 = 12.49, P = 0.0019; Fig 3C), wetted surface area (F1,24.8 = 18.17, P = 0.0003; Fig 3D) and volume (F1,24.3 = 12.10, P = 0.0019; Fig 3E) than females of P. turneri.
The multi-level longitudinal growth models (MLM) indicate changes in normalized (A) maximum width, (B) maximum height, (C) frontal surface area, (D) wetted surface area and (E) volume during one interbrood interval for pregnant P. turneri (red, N = 14) and P. gracilis (blue, N = 10). To account for individual variation in body size, one-dimensional parameters (A and B) were normalized by dividing the values by standard length (LSL), the surface areas (C and D) by dividing by LSL2 and volume (E) by dividing by LSL3. Connected circles represent individual female growth trajectories, solid lines are plotted from the MLM estimates for intercept and slope with equal litter wet mass (NS = P > 0.05, * = 0.01 < P < 0.05, ** = 0.001 < P < 0.01, *** = P < 0.001). Projections show examples of the respective model projections (A–C) or the complete model (D,E).
We found that P. turneri increases in body size faster than females of P. gracilis, as indicated by the steeper slopes of the former species in maximum width (MLM contrast of slopes: F1,23 = 11.42, P = 0.0026; Fig 3A), maximum height (F1,17.9 = 13.01, P = 0.0020; Fig 3B), frontal surface area (F1,21.6 = 11.84, P = 0.0024; Fig 3C), wetted surface area (F1,21.1 = 6.76, P = 0.0167; Fig 3D) and volume (F1,22.2 = 5.92, P = 0.0235; Fig 3E). As a consequence, the measured differences at IB = 0 diminished towards the end of pregnancy (IB = 1) (post hoc comparison, maximum height: P = 0.6531; frontal surface area: P = 0.0866; wetted surface area: P = 0.9541; volume: P = 0.4696; Fig 3B–3E respectively), while at this point in time females of P. turneri have a larger maximum width (P = 0.0127; Fig 3A). The steeper slopes in P. turneri are also reflected in the relative increase for maximum width (for P. gracilis maximum width at IB = 1 is 112% of its value at IB = 0 compared to P. turneri whose maximum width at IB = 1 is 122% of the value at IB = 0), maximum height (P. gracilis: 107%, P. turneri: 113%), frontal surface area (P. gracilis: 121%, P. turneri: 140%), wetted surface area (P. gracilis: 104%, P. turneri: 107%), and volume (P. gracilis: 111%, P. turneri: 120%).
Morphological differences between virgins
In line with the measured differences at IB = 0 for their pregnant conspecifics, virgin fish of P. gracilis have a larger overall normalized body size than virgins of P. turneri. We found significant effects of species on maximum width (MLM contrast: F1,20.9 = 39.62, P < 0.0001; MLM estimate ± SE: P.g. 0.1800 ± 0.0021, P.t. 0.1624 ± 0.0018), maximum height (F1,21.2 = 84.02, P < 0.0001; P.g. 0.2652 ± 0.0025, P.t. 0.2354 ± 0.0021), frontal surface area (F1,20.6 = 77.98, P < 0.0001; P.g. 0.0367 ± 0.0006, P.t. 0.0298 ± 0.0005), wetted surface area (F1,19.5 = 35.28, P < 0.0001; P.g. 0.4972 ± 0.0032, P.t. 0.4726 ± 0.0027), and volume (F1,19.3 = 32.62, P < 0.0001; P.g. 0.0191 ± 0.0003, P.t. 0.0171 ± 0.0002).
A key aspect of this study is that we compare two different reproductive states (pregnant and virgin fish) in two phylogenetically closely related ‘sister’ species that differ in the way that they provision their embryos : Poeciliopsis gracilis is a lecithotrophic species that lacks a placenta, while Poeciliopsis turneri represents one of three independent origins of extensive placentation in the genus Poeciliopsis [7,10]. Poeciliopsis gracilis and P. turneri have moderate levels of superfetation [23,36]. Due to smaller litters per parturition, higher levels of superfetation could result in reduced litter wet mass (S2 Table). Since the level of superfetation also affects the length of the interbrood interval [9,23,42], time was normalized in our MLM models to account for variation in interbrood interval both within and between species. In absolute terms, the difference in growth rate between the two species would be even more pronounced than is currently shown by our results. Finally, we measured the morphology of virgin controls, because they offer a morphological ‘base-line’ similar to the start of the first pregnancy: all virgin females carried unfertilized, fully yolk-provisioned eggs but were not yet affected by having to carry overlapping broods (superfetation). We showed that virgins of P. turneri have a more slender body shape than those of P. gracilis, in line with the observed differences at the beginning of the interbrood interval of pregnant females. At IB = 0, pregnant P. gracilis are morphologically more alike their virgin conspecifics than pregnant P. turneri, probably because the subsequent brood is already further developed in the latter species due to its higher level of superfetation (S3 Table).
Together these findings provide the first evidence in support of two key predictions of the locomotor cost hypothesis that the evolution of post-fertilization maternal provisioning by means of a placenta leads to a more slender body shape at the beginning of the pregnancy, and that this ‘morphological benefit’ diminishes over the course of the pregnancy (Fig 1; [9,36,37]). Our results further show that maximum width is higher in females of P. turneri during late pregnancy, which could lead to a reduction in abdominal flexibility during this period, more than that experienced by P. gracilis.
Whether, and to what extent, the measured morphological differences translate directly into differences in swimming performance requires further investigation, as other parameters that determine performance (e.g. physiology, flexural stiffness) could also be affected by pregnancy. Body shape, however, is known to affect the drag forces a fish experiences during swimming, such that more slender animals have a better continuous swimming performance and lower metabolic costs of swimming (e.g. [46–50]). There are two main types of drag that act on swimming fish of this size: pressure drag and friction drag; the former is related to the frontal surface area and the latter to the wetted surface area [51,52]. In P. turneri, both surface areas are lower at the beginning of pregnancy, and increase more rapidly over time than in P. gracilis (Fig 3C and 3D), implying an absolute benefit during the early stages and a mean benefit over the whole interbrood interval. During undulatory swimming, the experienced drag is highly complex due to the changing pressure and shear stress distribution on the deforming body . It is, however, apparent that the survival value of optimizing swimming speed and cost of transport (and thus reducing drag forces on the body) could be a driving force behind the evolution of fish morphology .
To study female morphology in three dimensions, we used a novel method for collecting longitudinal data, adapted from an in-house developed program originally designed to create 3D-models for zebrafish (Danio rerio) larvae . Our method offers two advantages over conventional two-dimensional geometric morphometric approaches. First, we took pictures while the female is in the water, thereby avoiding the need to kill, anesthetize and/or handle the fish with a paintbrush or tweezers [26,55–59]. Minimizing stress, potential physical damage and risk of death is particularly important in longitudinal studies where a single individual is measured repeatedly over a period of time. Second, we use information from 251 equidistant cross-sections and two orthogonal planes to reconstruct 3D body models. Even in a single 2D-plane (e.g. lateral view), our method produces a more accurate approximation of female body shape than landmark-based geometric morphometric approaches, which are often based on a limited number of (semi-)landmarks (typically between 12 to 17) and are hampered by a lack of clear landmarks in the abdominal region of pregnant females [55–59]. Perhaps more importantly, we show that different planes can yield different patterns of shape change through time (e.g. compare the effects in maximum width and height; Fig 3A and 3B), suggesting that information from one plane cannot be readily used to make inferences about temporal changes during the pregnancy in the other planes nor overall streamlining (e.g. volume or frontal surface area).
We studied only one of eight independent origins of the evolution of the placenta in the family Poeciliidae [8,10]. To test the generality of our results, a wider comparative survey is required that includes other independent evolutionary origins of the placenta. Recent studies in the family Poeciliidae have revealed three independent origins of placentation in the genus Poeciliopsis (of which this study examined one; ) and two independent origins of placentation in the genus Poecilia (in the subgenera Micropoecilia and Pamphorichthys, respectively; [60,61]). These (sub)genera contain closely related species that differ in whether they have a placenta and are eminently suitable for further comparative experimental studies [13,14,23,24,37] to test whether the morphological benefits we found in this study are repeated in other placental lineages.
Moreover, it is possible that drag reduction is one of the driving forces behind the evolution of a placenta in other families of live-bearing bony fish (e.g. Anablepidae, Goodeidae, Zenarchopteridae) [62–64] and of a placenta or other ways of post-fertilization nutrient allocation (matrotrophy) in live-bearing cartilaginous fishes [5,65,66]. Furthermore, it would be worthwhile to study the effects of a placenta on morphology and relevant performance parameters in mobile animals that otherwise try to maximize slenderness, for instance live-bearing Squamate reptiles with a predominantly burrowing or ‘sand-swimming’ mode of locomotion (e.g. viviparous skinks)  or animals that are girth-restricted, for instance by the crevices they inhabit . Finally, the placenta evolved many times independently throughout the animal kingdom, in livebearing animal lineages with a large diversity of lifestyles [2–6]; the broader applicability of the locomotor cost hypothesis requires further study.
To conclude, in this study we compared changes in volume and frontal surface area during gestation between a lecithotrophic and a placental fish species using a new 3D-modelling approach. Our results provide the first empirical evidence in support of the locomotor cost hypothesis, which states that the evolution of a placenta can lead to a more slender body shape at the start of the pregnancy and that this effect disappears towards the end of the pregnancy (Figs 1 and 3). To test the generality of our findings, future research should focus on additional independent placental lineages. The biomechanical importance of drag reduction for locomotion, however, suggests that the locomotor cost hypothesis could potentially be applicable to the evolution of placentas in other swimming live-bearing lineages.
S1 Table. Multi-level modelling output for fixed effects in the pregnant and virgin control models, for all measured morphological parameters.
LWM: Litter wet mass.
S2 Table. General and reproductive parameters of the experimental fish used in this study.
S3 Table. Relative values of morphological parameters of Poeciliopsis gracilis and Poeciliopsis turneri, compared to their virgin conspecifics.
S1 Text. Fish rearing, feeding and husbandry.
Detailed description of the study species used, husbandry and feeding.
S1 Data. Spreadsheet containing pregnant fish morphology data.
We thank David Reznick (UC Riverside) for supplying the original stocks of our study species and Cees J. Voesenek for his help with the custom-made MATLAB program. We further thank David Reznick, Andrew Furness (UC Irvine) and the Experimental Zoology Journal Club for commenting on an earlier version of the manuscript.
- 1. Mossman HW. Comparative morphogenesis of the fetal membranes and accessory uterine structures. Contrib Embryol. 1937;26: 129–246.
- 2. Wourms JP. Viviparity: the maternal-fetal relationships in fishes. Am Zool. 1981;21: 473–515.
- 3. Wourms JP, Grove BD, Lombardi J. The maternal-embryonic relationship in viviparous fishes. In: Hoar WS, Randal DJ, editors. Fish Physiology. New York: Academic Press; 1988. pp. 1–134.
- 4. Wake MH. Evolutionary Scenarios, Homology and Convergence of Structural Specializations for Vertebrate Viviparity. Am Zool. 1992;32: 256–263.
- 5. Blackburn DG. Evolution of vertebrate viviparity and specializations for fetal nutrition: A quantitative and qualitative analysis. J Morphol. 2015;276: 961–990. pmid:24652663
- 6. Ostrovsky AN, Lidgard S, Gordon DP, Schwaha T, Genikhovich G, Ereskovsky A V. Matrotrophy and placentation in invertebrates: a new paradigm. Biol Rev. 2016;91: 673–711. pmid:25925633
- 7. Reznick DN, Mateos M, Springer MS. Independent origins and rapid evolution of the placenta in the fish genus Poeciliopsis. Science. 2002;298: 1018–1020. pmid:12411703
- 8. Hrbek T, Seckinger J, Meyer A. A phylogenetic and biogeographic perspective on the evolution of poeciliid fishes. Mol Phylogenet Evol. 2007;43: 986–998. pmid:17185005
- 9. Pollux BJA, Pires MN, Banet AI, Reznick DN. Evolution of Placentas in the Fish Family Poeciliidae: An Empirical Study of Macroevolution. Annu Rev Ecol Evol Syst. 2009;40: 271–289.
- 10. Pollux BJA, Meredith RW, Springer MS, Garland T, Reznick DN. The evolution of the placenta drives a shift in sexual selection in livebearing fish. Nature. 2014;513: 233–236. pmid:25043015
- 11. Trexler JC, DeAngelis DL. Resource allocation in offspring provisioning: An evaluation of the conditions favoring the evolution of matrotrophy. Am Nat. 2003;162: 574–585. pmid:14618536
- 12. Reznick DN, Yang AP. The influence of fluctuating resources on life history: patterns of allocation and plasticity in female guppies. Ecology. 1993;74: 2011–2019.
- 13. Banet AI, Reznick DN. Do placental species abort offspring? Testing an assumption of the Trexler-DeAngelis model. Funct Ecol. 2008;22: 323–331.
- 14. Banet AI, Au AG, Reznick DN. Is mom in charge? Implications of resource provisioning on the evolution of the placenta. Evolution (N Y). 2010;64: 3172–3182. pmid:20649814
- 15. Pollux BJA, Reznick DN. Matrotrophy limits a female’s ability to adaptively adjust offspring size and fecundity in fluctuating environments. Funct Ecol. 2011;25: 747–756.
- 16. Blackburn DG, Vitt LJ, Beuchat CA. Eutherian-like reproductive specializations in a viviparous reptile. Proc Natl Acad Sci. 1984;81: 4860–4863. pmid:16593499
- 17. Wourms JP, Lombardi J. Reflections on the evolution of piscine viviparity. Am Zool. 1992;32: 276–293. Available: http://icb.oxfordjournals.org/content/32/2/276.short
- 18. Trexler JC. Resource Availability and Plasticity in Offspring Provisioning: Embryo Nourishment in Sailfin Mollies. Ecology. 1997;78: 1370–1381.
- 19. Holbrook GL, Schal C. Maternal investment affects offspring phenotypic plasticity in a viviparous cockroach. Proc Natl Acad Sci. 2004;101: 5595–5597. pmid:15064397
- 20. Schrader M, Travis J. Population Differences in Pre- and Post-Fertilization Offspring Provisioning in the Least Killifish, Heterandria formosa. Copeia. 2005;2005: 649–656.
- 21. Wildman DE, Chen C, Erez O, Grossman LI, Goodman M, Romero R. Evolution of the mammalian placenta revealed by phylogenetic analysis. Proc Natl Acad Sci. 2006;103: 3203–3208. pmid:16492730
- 22. Reznick DN, Meredith R, Collette BB. Independent Evolution of Complex Life History Adaptations in Two Families of Fishes, Live-bearing Halfbeaks (Zenarchopteridae, Beloniformes) and Poeciliidae (Cyprinodontiformes). Evolution (N Y). 2007;61: 2570–2583.
- 23. Pires MN, Bassar RD, McBride KE, Regus JU, Garland T, Reznick DN. Why do placentas evolve? An evaluation of the life-history facilitation hypothesis in the fish genus Poeciliopsis. Funct Ecol. 2011;25: 757–768.
- 24. Bassar RD, Auer SK, Reznick DN. Why do placentas evolve? A test of the life-history facilitation hypothesis in two clades in the genus Poeciliopsis representing two independent origins of placentas. Funct Ecol. 2014;28: 999–1010.
- 25. Shaffer LR, Formanowicz DR. A cost of viviparity and parental care in scorpions: Reduced sprint speed and behavioural compensation. Anim Behav. 1996;51: 1017–1023.
- 26. Plaut I. Does pregnancy affect swimming performance of female Mosquitofish, Gambusia affinis? Funct Ecol. 2002;16: 290–295.
- 27. Ghalambor CK, Reznick DN, Walker JA. Constraints on adaptive evolution: the functional trade-off between reproduction and fast-start swimming performance in the Trinidadian guppy (Poecilia reticulata). Am Nat. 2004;164: 38–50. pmid:15266369
- 28. Belk MC, Tuckfield RC. Changing costs of reproduction: age-based differences in reproductive allocation and escape performance in a livebearing fish. Oikos. 2010;119: 163–169.
- 29. Banet AI, Svendsen JC, Eng KJ, Reznick DN. Linking reproduction, locomotion, and habitat use in the Trinidadian guppy (Poecilia reticulata). Oecologia. 2016;181: 87–96. pmid:26809619
- 30. Shine R. “Costs” of reproduction in reptiles. Oecologia. 1980;46: 92–100. Available: http://link.springer.com/article/10.1007/BF00346972 pmid:28310632
- 31. Winne CT, Hopkins WA. Influence of sex and reproductive condition on terrestrial and aquatic locomotor performance in the semi-aquatic snake Seminatrix pygaea. Funct Ecol. 2006;20: 1054–1061.
- 32. Wu W, Meijer OG, Lamoth CJC, Uegaki K, van Dieën JH, Wuisman PIJM, et al. Gait coordination in pregnancy: transverse pelvic and thoracic rotations and their relative phase. Clin Biomech (Bristol, Avon). 2004;19: 480–488. pmid:15182983
- 33. Noren SR, Redfern J V, Edwards EF. Pregnancy is a drag: hydrodynamics, kinematics and performance in pre- and post-parturition bottlenose dolphins (Tursiops truncatus). J Exp Biol. 2011;214: 4151–9. pmid:22116757
- 34. Langerhans RB, Reznick DN. Ecology and Evolution of Swimming Performance in Fishes: Predicting Evolution with Biomechanics. In: Domenici P, Kapoor BG, editors. Fish Locomotion: An Eco Ethological Perspective. Science Publishers; 2010. pp. 200–248.
- 35. James R, Johnston I. Influence of spawning on swimming performance and muscle contractile properties in the short-horn sculpin. J Fish Biol. 1998;53: 485–501.
- 36. Thibault RE, Schultz RJ. Reproductive adaptations among viviparous fishes (Cyprinodontiformes: Poeciliidae). Evolution (N Y). 1978;32: 320–333. Available: http://www.jstor.org/stable/10.2307/2407600
- 37. Pires M, Arendt J, Reznick D. The evolution of placentas and superfetation in the fish genus Poecilia (Cyprinodontiformes: Poeciliidae: subgenera Micropoecilia and Acanthophacelus). Biol J Linn Soc. 2010;99: 784–796.
- 38. Walker JA, Ghalambor CK, Griset OL, McKenney D, Reznick DN. Do faster starts increase the probability of evading predators? Funct Ecol. 2005;19: 808–815.
- 39. Turner CL. Reproductive Cycles and Superfetation in Pœciliïd Fishes. Biol Bull. 1937;72: 145–164.
- 40. Turner CL. Superfetation in Viviparous Cyprinodont Fishes. Copeia. 1940; 88–91.
- 41. Scrimshaw N. Superfetation in Poeciliid Fishes. Copeia. 1944;3: 180–183.
- 42. Reznick DN, Miles DB. Review of life history patterns in Poeciliid fishes. In: Meffe GK, Snelson FF, editors. Ecology and Evolution of Livebearing Fishes (Poeciliidae). Prentice Hall, Englewood Cliffs, NJ; 1989. pp. 125–148.
- 43. Voesenek CJ, Pieters RPM, Van Leeuwen JL. Automated reconstruction of three-dimensional fish motion, forces, and torques. PLoS One. 2016;11: e0146682. pmid:26752597
- 44. Singer JD. Using SAS PROC MIXED to fit multilevel models, hierarchical models, and individual growth models. J Educ Behav Stat. 1998;23: 323–355.
- 45. Singer JD, Willett JB. Applied longitudinal data analysis: Modeling change and event occurrence. New York: Oxford University Press, Inc.; 2003.
- 46. Pettersson LB, Hedenström A. Energetics, Cost Reduction and Functional Consequences of Fish Morphology. Proc R Soc B. 2000;267: 759–764. pmid:10819144
- 47. Weihs D. Stability Versus Maneuverability in Aquatic Locomotion 1. Integr Comp Biol. 2002;42: 127–134. pmid:21708701
- 48. Law TC, Blake RW. Comparison of the Fast-Start Performances of Closely Related, Morphologically Distinct Threespine Sticklebacks (Gasterosteus spp.). J Fish Biol. 2005;67: 834–848.
- 49. Langerhans RB. Morphology, performance, fitness: functional insight into a post-Pleistocene radiation of mosquitofish. Biol Lett. 2009;5: 488–491. pmid:19411270
- 50. Van Wassenbergh S, Potes NZ, Adriaens D. Hydrodynamic drag constrains head enlargement for mouthbrooding in cichlids. J R Soc Interface. 2015;12: 20150461. pmid:26224567
- 51. Sfakiotakis M, Lane DM, Davies JBC. Review of fish swimming modes for aquatic locomotion. IEEE J Ocean Eng. 1999;24: 237–252.
- 52. Videler JJ. Fish Swimming. 1st ed. Videler JJ, editor. London: Chapman & Hall; 1993.
- 53. Tytell ED, Borazjani I, Sotiropoulos F, Baker TV, Anderson EJ, Lauder G V. Disentangling the functional roles of morphology and motion in the swimming of fish. Integr Comp Biol. 2010;50: 1140–1154. pmid:21082068
- 54. Tokic G, Yue DKP. Optimal shape and motion of undulatory swimming organisms. Proc R Soc B Biol Sci. 2012;279: 3065–3074. pmid:22456876
- 55. Zúñiga-Vega JJ, Reznick DN, Johnson JB. Habitat predicts reproductive superfetation and body shape in the livebearing fish Poeciliopsis turrubarensis. Oikos. 2007;116: 995–1005.
- 56. Zúñiga-Vega JJ, Suárez-Rodríguez M, Espinosa-Pérez H, Johnson JB. Morphological and reproductive variation among populations of the Pacific molly Poecilia butleri. J Fish Biol. 2011;79: 1029–1046. pmid:21967588
- 57. Wesner JS, Billman EJ, Meier A, Belk MC. Morphological convergence during pregnancy among predator and nonpredator populations of the livebearing fish Brachyrhaphis rhabdophora (Teleostei: Poeciliidae). Biol J Linn Soc. 2011;104: 386–392.
- 58. Ingley SJ, Billman EJ, Belk MC, Johnson JB. Morphological divergence driven by predation environment within and between species of Brachyrhaphis fishes. PLoS One. 2014;9: 31–36. pmid:24587309
- 59. Frías-Alvarez P, Zúñiga-Vega JJ. Superfetation in live-bearing fishes is not always the result of a morphological constraint. Oecologia. 2016;181: 645–658. pmid:26508151
- 60. Meredith RW, Pires MN, Reznick DN, Springer MS. Molecular phylogenetic relationships and the coevolution of placentotrophy and superfetation in Poecilia (Poeciliidae: Cyprinodontiformes). Mol Phylogenet Evol. 2011;59: 148–57. pmid:21292015
- 61. Meredith RW, Pires MN, Reznick DN, Springer MS. Molecular phylogenetic relationships and the evolution of the placenta in Poecilia (Micropoecilia) (Poeciliidae: Cyprinodontiformes). Mol Phylogenet Evol. 2010;55: 631–639. pmid:19922806
- 62. Uribe MC, Aguilar-Morales M, De la Rosa-Cruz G, García-Alarcón A, Campuzano-Caballero JC, Guerrero-Estévez SM. Ovarian Structure and Embryonic Traits Associated with Viviparity in Poeciliids and Goodeids. In: Uribe MC, Grier HJ, editors. Viviparous fishes II. New Life Publications, Homestead, Florida; 2010. pp. 211–229.
- 63. Uribe MC, De la Rosa-Cruz G, García-Alarcón A. Branchial placenta in the viviparous teleost Ilyodon whitei (Goodeidae). J Morphol. 2014;275: 1406–1417. pmid:25220154
- 64. Schindler JF. Structure and function of placental exchange surfaces in goodeid fishes (Teleostei: Atheriniformes). J Morphol. 2015;276: 991–1003. pmid:24797516
- 65. Dulvy NK, Reynolds JD. Evolutionary transitions among egg-laying, live-bearing and maternal inputs in sharks and rays. Proc R Soc B Biol Sci. 1997;264: 1309–1315.
- 66. Hamlett WC, Kormanik G, Storrie M, Stevens BS, Walker TI. Chondrichthyan Parity, Lecithotrophy and Matrotrophy. In: Hamlett WC, editor. Reproductive Biology and Phylogeny of Chondrichthyes. CRC Press; 2005. pp. 395–434.
- 67. Stewart JR, Blackburn DG. Viviparity and Placentation in Lizards. In: Rheubert JL, Siegel DS, Trauth SE, editors. Reproductive Biology and Phylogeny of Lizards and Tuatara. CRC Press; 2014. pp. 448–563.
- 68. Schwarzkopf L, Barnes M, Goodman B. Belly up: Reduced crevice accessibility as a cost of reproduction caused by increased girth in a rock-using lizard. Austral Ecol. 2010;35: 82–86.