Sperm specificity and potential paternal effects in gynogenesis in the Amazon Molly (Poecilia formosa)

Ethics statement

All fish care and protocol procedures were performed according to United States federal regulations and institutional policies and approved by the University of Oklahoma Norman Institutional Animal Care and Use Committee (Protocol R21-027). Fish were acquired from the International Stock Center for Livebearing Fishes, University of Oklahoma. Surviving fish were returned to the International Stock Center for Livebearing Fishes, University of Oklahoma. Sample sizes were based on fish availability. Humane endpoints due to spontaneous disease were weight loss or decrease in activity level or appetite, and animals meeting humane endpoints were euthanized using the method described below prior to planned experimental endpoints.

Environment and timeline

All fish were fed either flake food (TetraMin Tropical Flakes, Spectrum Brands Pet LLC, Blacksburg, VA, USA) or frozen live feed (Hikari Bio-Pure Blood Worms for adults and Hikari Bio-Pure Brine Shrimp for juveniles, Hikari Sales USA Inc., Hayward, CA, USA) once daily for 4 to 6 days per week. Water conductivity was measured at water change and adjusted with Instant Ocean Sea Salt (Spectrum Brands Pet LLC, Blacksburg, VA, USA) to be at least 800 μS/cm. Water temperature was maintained at or above 20 °C, with water temperature from a randomly selected tank sampled multiple times per week. Fish were either housed in 10-liter static tanks or 40-liter static tanks each with a sponge filter, polyvinyl chloride tunnel, pebbles, and live plants as enrichment. Ten-liter tanks had 30% water changes performed weekly and 40-liter tanks had 30% water changes performed every other week. All fish were housed in the same room and experienced a 12:12 h light: dark cycle. Fish health was monitored daily. Experiments were conducted between March and December 2022.

Breeding design

Experiment 1 utilized 24 P. formosa VI/17 nulliparous (virgin) females from sympatry with P. mexicana, each measuring greater than 3 cm standard length, four sympatric P. mexicana VI/17 males, and four allopatric P. mexicana Oxolotan males. Another 12 P. formosa Airport Ditch 2021 (AD21) nulliparous females from sympatry with P. latipinna, measuring at least 2.5 cm in standard length, were paired with two sympatric P. latipinna AD21 males, or two allopatric P. latipinna Steve’s Ditch males. Females were observed for pregnancy weekly starting at week 3 after pairing. When a female was visibly pregnant as evidenced by a barrel-shape body, she was isolated in a 10-liter tank until she produced offspring.

In Experiment 2, we paired a total of 24 P. formosa nulliparous females, measuring at least 2 centimeters in standard length, with males from several species. We always used two males per species as independent replicates. The species used were Gambusia affinis, Gambusia sexradiata, Girardinus metallicus, Heterandria formosa, Limia islai, Limia melanogaster, Poeciliopsis prolifica, Poecilia dominicensis, Poecilia sphenops from allopatry and Poecilia sphenops from sympatry, Xiphophorus hellerii, and Xiphophorus variatus. Each female was paired individually with one approximately size-matched male in a 10-liter tank until she produced offspring or until the experimental endpoint of 12 or 15 weeks.

No female or male was shared between experiments to minimize the effect of a specific individual’s reproductive performance. All experiments utilized two males and at least two females from each population to compensate for variation in individual fertility.

Reproductive data collection

Pregnant females were checked daily for offspring. On the day that offspring were observed, the mother was humanely euthanized by immersion in (250 mg/L) tricaine methanesulfonate (Tricaine-S, Syndel USA, Ferndale, WA, USA) buffered with (500 mg/L) sodium bicarbonate for 30 min past the cessation of opercular movement. The carcass was then preserved in 4% formaldehyde. One day prior to dissection, the carcass was transferred to tap water to minimize personnel exposure to formaldehyde. The carcass was dissected to count embryos and observe oocytes. The stage of the embryos was classified as early-stage if eyes had not developed and classified as late-stage pregnancy if eyes had developed. Compared to the developmental stage classification of poeciliid species standardized by Haynes (1995), early-stage in this study was equivalent to stages 4–6 and late-stage was equivalent to stages 7–11. The standard length of the female was also measured at time of dissection. Number of offspring was estimated visually on the day of birth, and the number of offspring was confirmed 2 weeks after birth when offspring could safely be moved with a net.

If a female had not produced offspring by the experimental endpoint, the female was humanely euthanized and dissected using the methods described above. The experimental endpoint for Experiment 1 was 9 to 15 weeks based on at least twice the minimum time from pairing to parturition typically observed. The experimental endpoint for Experiment 2 was 12 to 15 weeks because no females had produced offspring by 9 weeks, but some did between 10 and 12 weeks. Fish were size matched between treatment and control groups, but randomization and blinding were not performed.

Variable definitions

The groups of interest for Experiment 1 were sympatry (sympatric vs allopatric) and female population (VI/17 vs AD21). Outcomes for statistical comparison were relative reproductive output, relative embryo output, and combined relative reproductive output. Relative reproductive output was the number of live offspring divided by female standard length in millimeters (Hinz & Schlupp, 2011). Relative embryo output was the number of embryos recovered on dissection divided by female standard length in millimeters. Combined relative reproductive output was the sum of relative reproductive output and relative embryo output.

Statistical analysis

Statistical software used was Microsoft Excel (2016, Microsoft Corporation, Redmond, WA, USA). All statistical tests were two-tailed Mann-Whitney U (MWU) tests, which were performed for each paired combination of variable of interest and outcome in Experiment 1. Statistical significance was set at α = 0.05.

Experiment 1

For logistical reasons we performed Experiment 1 as two runs that were analyzed separately due to unforeseen differences in fish health. In Experiment 1A, there was no significant difference in female size between the females used with allopatric and sympatric males (U-test, U = 18, n = 12, p = 1.0). All sympatric females had offspring or were pregnant by 9 weeks after breeding began, but only one allopatric female had offspring and no allopatric females were pregnant. The average brood size for sympatric females was 13.7 offspring/late-stage embryos, and the one allopatric female which produced offspring had a brood size of 13 offspring. There was no significant difference in relative reproductive output (U-test, U = 30.6, n = 12, p = 1.96) or relative embryo output (U-test, U = 12.75, n = 12, p = 0.4) between sympatric and allopatric females. However, there was a significant difference in combined relative reproductive output between sympatric and allopatric females (U-test, U = 2, n = 12, p = 0.01) (Fig. 1).

In Experiment 1A, one of the non-pregnant allopatric females died and two more allopatric non-pregnant females were moribund (meeting humane endpoints for decrease in activity level or appetite) and humanely euthanized 4 days before the experimental endpoint at 9 weeks.

Experiment 1B repeated the groupings of Experiment 1A with different fish from the same populations and added groupings of AD21 P. formosa females with sympatric AD21 P. latipinna males and allopatric Steve’s Ditch P. latipinna males. In Experiment 1B, there was no significant difference in female size between the sympatric VI/17 females and allopatric VI/17 females (U-test, U = 14, n = 12, p = 0.522). Three allopatric females had offspring with an average brood size of five offspring. Two sympatric females had offspring and embryos with an average brood size of 8.5 offspring/late-stage embryos. There was no significant difference in relative reproductive output (U-test, U = 23.6, n = 12, p = 1.63), relative embryo output (U-test, U = 39.5, n = 12, p = 2.0), or combined relative reproductive output (U-test, U = 25.1, n = 12, p = 1.74). All VI/17 females in Experiment 1B remained healthy for the duration of the experiment. One allopatric male died in the week 2 of the 15-week experiment and was replaced.

Comparing all VI/17 females between Experiments 1A and 1B, there was no significant difference in relative reproductive output (U-test, U = 119.6, n = 24, p = 1.99), relative embryo output (U-test, U = 125.6, n = 24, p = 2.0), or combined relative reproductive output in a two-sided MWU (U-test, U = 72.6, n = 24, p = 1.03). Comparing sympatric VI/17 females between Experiments 1A and 1B, there was no significant difference in relative reproductive output (U-test, U = 29.5, n = 12, p = 1.93) or relative embryo output (U-test, U = 13.7, n = 12, p = 0.493). In Experiment 1A, sympatric VI/17 females had a significantly greater combined relative reproductive output (U-test, U = 2, n = 12, p = 0.01) than sympatric VI/17 females in Experiment 1B in a two-sided MWU. Comparing allopatric VI/17 females between Experiments 1A and 1B, there was no significant difference in relative reproductive output (U-test, U = 23.6, n = 12, p = 1.63), relative embryo output (U-test, U = 39.5, n = 12, p = 2.0), or combined relative reproductive output (U-test, U = 23.6, n = 12, p = 1.63) in a two-sided MWU. There was a significant difference in VI/17 female size between Experiments 1A and 1B (U-test, U = 5, n = 24, p < 0.001) in a two-sided MWU. The VI/17 females in Experiment 1A had an average female standard length of 41.6 mm, while the VI/17 females in Experiment 1B had an average female standard length of 33.5 mm.

In Experiment 1B for AD21 females, none of the females produced offspring and only one allopatric female had a possible early-stage pregnancy with three large, yellow oocytes. Both groups had frequent mortality of the male fish, which were replaced when a male was noted to be sick or dead. Due to the high mortality rate, there were no statistical comparisons performed on relative reproductive output, relative embryo output, or combined relative reproductive output. There was a significant difference in size between the AD21 females and VI/17 females in Experiment 1B (U-test, U = 29.7, n = 24, p = 0.015), but there was no significant size difference between the sympatric and allopatric AD21 females (U-test, U = 10.3, n = 12, p = 0.22). The average standard length of an AD21 female in Experiment 1B was 28.58 mm, and the average standard length of a VI/17 female in Experiment 1B was 33.5 mm. Table 1 includes the data for VI/17 females in Experiment 1.

Experiment 2

Experiment 2 examined reproductive output of pairings between P. formosa females and males from species within the genus Poecilia and related genera (i.e. Gambusia, Girardinus, Heterandria, Limia, Poeciliopsis, and Xiphophorus). All pairings with males from the genera Poecilia (sympatric and allopatric P. sphenops, P. dominicensis) and Limia (L. melanogaster, L. islai) had at least one pair that produced offspring or late-stage embryos. Two follow-up 15-week pairings between P. formosa females and Poecilia wingei males from February 2023 to May 2023 in a greenhouse setting did not result in offspring or embryos but did have large, yellow, well-developed oocytes present on dissection of the females. None of the males from other genera (Gambusia, Girardinus, Heterandria, Poeciliopsis, Xiphophorus) produced offspring or late-stage embryos. Poeciliopsis, Girardinus, and Xiphophorus males were seen attempting to mate with the P. formosa females. On dissection, one female from each pairing with P. prolifica, G. metallicus, and X. hellerii appeared to have large, yellow, well-developed oocytes, but the oocytes did not appear to be fertilized because they lacked a visible blastodisc and blood vessels (Fig. 2). In one G. metallicus pairing, the male died in the first and fourth week of the 15-week experiment. In one L. islai pairing, the male died in week 12 of the 15-week experiment. These two pairs in Experiment 2 that had male mortality did not produce offspring or late-stage embryos despite prompt replacement of the males within a week of the deaths.

A female paired with P. dominicensis (combined relative reproductive output = 0.2) and a female paired with L. melanogaster (combined relative reproductive output = 0.2) had the greatest combined relative reproductive output (Table 2). Females paired with sympatric P. sphenops were the only species cross to have both females have offspring or late-stage embryos. Comparing females who had offspring or embryos from Experiment 2 to females who had offspring or embryos in Experiment 1A, the combined relative reproductive output of females from Experiment 2 was significantly lower (U-test, U = 0, n = 11, p = 0.008) and the female standard lengths were significantly shorter (U-test, U = 2.3, n = 11, p = 0.026). Comparing females who had offspring or embryos from Experiment 2 to females who had offspring or embryos in Experiment 1B, there was no significant difference in combined relative reproductive output (U-test, U = 8, n = 9, p = 0.624), and the female standard lengths did not differ significantly (U-test, U = 2, n = 9, p = 0.05).

Are there differences in male capability to trigger embryogenesis in Amazon Mollies?

Amazon Mollies are one of a small number of sperm-dependent asexual fishes. They produce clonal offspring without using male genetic contributions. Nonetheless, they require sperm from heterospecific males to trigger embryogenesis. This sperm dependency raises a number of important questions, two of which we addressed in the present study. On the reproductive biological level, we asked if males from allopatric or sympatric populations differ in effectiveness in triggering embryogenesis. We found that there is a non-significant trend for sympatric males to sire more offspring. Based on our design, we cannot distinguish whether this may be a direct effect of the sperm or males, or indirectly a female response. Females may influence the success of fertilization either by behaviorally rejecting mating or through sperm selection by cryptic female choice in the female reproductive tract. When presented with animated P. mexicana male models, P. formosa preferred models with lateral projection areas similar to those of sympatric males, but it is unknown if the behavioral preference would result in more mating or more offspring (Kim et al., 2014). Cryptic female choice has not been studied in P. formosa. However, in a recent zebrafish study, female reproductive fluid preferentially attracted spermatozoa with longevity greater than 20 s and better DNA integrity (Cattelan et al., 2023). Our study was limited by small sample sizes, due to logistical reasons and unexpected morbidity and mortality. Nonetheless, if correct, our finding of differences between the sympatric and allopatric pairings would point to coevolution of Amazon Mollies and their host males by suggesting that sympatric donor sperm or the receiving Amazon Mollies have adapted to activate embryogenesis more successfully during mating. Something similar has been reported by Gabor, Ryan & Morizot (2005), who found reproductive character displacement in Sailfin Mollies depending on sympatry or allopatry with Amazon Mollies. Based on these findings one would predict very high specificity of the sperm-egg interaction in Amazon Mollies.

Our study did not confirm this prediction, but our findings had several limitations. We ran the experiments in two independent rounds, which may have introduced effects of seasonality and the females used differed in size. Based on the data collected, larger P. formosa (measuring 40 mm in female standard length) may be more suitable for detecting subtle differences in reproductive success as they appear more likely to successfully produce offspring and embryos, even though P. formosa as small as 25 mm in female standard length occasionally produced live offspring. Further studies of the relationship between female standard length and reproductive output could develop new coefficients or equations to refine comparisons of reproductive output between P. formosa of different sizes. Although seasonality of reproduction has not been studied thoroughly in P. formosa, Robinson et al. (2011) studied the mating season of a parental species P. latipinna in Texas. The percentages of gravid P. latipinna females were more than 4.5 times greater in the early mating season of March and April and the mid mating season of May through August than in the late mating season of September and October. Out of the 988 female P. latipinna sampled, 4.8% had embryos present in November through February, 69.3% had embryos present in March and April, 66.4% had embryos present in May through August, and 14.7% had embryos present in September and October (Robinson et al., 2011). To the contrary, the highest gonadosomatic index for females of the other parental species P. mexicana occurred between August and February, in a population near Veracruz, Mexico. However, this sample only included 100 P. mexicana females and gonadosomatic index was based on many criteria other than presence of embryos, such as weight, standard length, weight of ovary, number of mature oocytes, and number of embryos (Chávez-López, Rocha-Ramírez & Cortés-Garrido, 2015). Behavioral interactions may also be important in this context: Schlupp & Plath (2005) found that P. mexicana males did mate with P. mexicana females significantly more than P. formosa females over 30 min and that P. mexicana females received significantly more sperm. Makowicz, Muthurajah & Schlupp (2018) found that neither P. mexicana nor P. latipinna males showed a behavioral preference between sympatric and allopatric P. formosa, and P. formosa did not show a behavioral preference between P. mexicana and P. latipinna. These findings suggest that although P. mexicana males can differentially allocate sperm and choose conspecifics, P. mexicana males may not breed more effectively with sympatric P. formosa females than allopatric P. formosa females. If the significant difference in VI/17 P. formosa combined relative reproductive output favoring sympatric populations of P. mexicana is a true difference, then the difference is quite possibly at the epigenetic level rather than driven by male behavior, based on the findings of Makowicz, Muthurajah & Schlupp (2018).

Heterospecific matings with males throughout the family Poeciliidae

Considering which other species are able to provide sperm provides a different perspective on sperm specificity. Based on our data and previous studies (summarized in Table 3), sperm-egg interactions are not very specific, as a large number of species can trigger embryogenesis in Amazon Mollies. It appears that every species of the genus Poecilia that was tested so far is capable of triggering embryogenesis with the possible exception of Poecilia wingei. Poecilia wingei is closely related to Poecilia reticulata, which does successfully produce offspring with Amazon Mollies (Hubbs & Hubbs, 1946a; Schlupp, Parzefall & Schartl, 2001). The taxonomy of this group is unfortunately not fully resolved, but the major groups are well defined. If one breaks down the genus Poecilia into Acanthocephalus (the guppy and relatives), Allopoecilia, Mollienesia (Mollies sensu stricto), Limia, Pamphorichthys, and Micropoecilia, only Pamphorichthys has not been tested yet. The genus Limia, which is the sister group to Mollienesia, is also clearly capable of triggering embryogenesis. Specifically, in the present study we tested the species that is at the base of the genus, L. melanogaster and obtained offspring, as with all other species of Limia that were tested so far. When consulting the phylogeny by Reznick et al. (2017), the genus Cnestorodon would be a good next choice to further investigate the role of phylogenetic distance in sperm specificity in the Amazon Molly. While many species of livebearing fishes can successfully provide sperm, this is not true for all species or genera in the family (Fig. 3). Our study and previous ones also considered species that show geographical overlap with Amazon Mollies that are only distantly related. This includes for example the speciose genera Xiphophorus and Poeciliopsis, both of which seem unable to trigger embryogenesis in Amazon Mollies. Our work specifically tested a representative of the genus Girardinus, which is basal to the clades that are unable to trigger embryogenesis. A notable exception is reported in the article by Hubbs & Hubbs (1946a), where they claim to have found reproduction with an unspecified species of Gambusia. In a meeting abstract, Hubbs & Hubbs (1946b) clarified that offspring from crosses with Gambusia were principally obtained through use of pituitary injections and artificial insemination. Our study, which examined natural mating using G. affinis and G. sexradiata, was unable to confirm the previous finding, but additional work on this would be useful. Overall, we confirmed that P. formosa is not compatible in natural mating with X. hellerii or X. variatus. Additional new species tested belonged to the genera Gambusia, Girardinus, Heterandria, and Poeciliopsis, but none of the pairings produced live offspring or late-stage embryos.

It should also be noted that the taxonomy of the mollies is in flux and that a recent phylogeny by Palacios et al. (2023) classified the sympatric Poecilia species from Baños de San Ignacio as P. mexicana rather than P. sphenops. A similar problem is associated with the article by Hubbs & Hubbs (1946a), where we carefully tried to align old taxonomy with the currently accepted. Gambusia species were previously reported by Hubbs & Hubbs (1946a, 1946b) to produce live offspring with P. formosa, likely with G. affinis using artificial insemination and pituitary hormone treatment of the female. In the same abstract, Hubbs & Hubbs (1946b) claim “fecundity was roughly proportional to degree of relatedness”. It is not clear what direction that relationship has had (presumably more offspring with closer related taxa), but we were not able to corroborate this finding. Nonetheless, the notion that artificial insemination may reveal even more species as potential sperm donors, is of critical importance as it points to prezygotic isolating mechanisms to be potentially behind the currently known pattern. Prezygotic factors could include behavioral or physical barriers to intercourse, incompatibility of the gonopodium with the gonopore of the female, spermatozoa unable to travel to the oocyte, failure of fertilization due to lack of capacitation of the spermatozoon or binding to the vitelline membrane, or failure of embryogenesis by inappropriate epigenetic or other biochemical signaling for successful cell divisions. Based on Hubbs & Hubbs (1946b) and our study, an experiment that would test the role of pre- and postzygotic isolation using artificial insemination would be useful.

Connection to infertility and future research

Our experiments confirm that fertilization in P. formosa is fairly non-specific as species from the genera of Poecilia and Limia, including multiple allopatric species and species that diverged more than two million years ago (Palacios et al., 2016), can produce offspring with P. formosa through mating in a laboratory setting. If techniques such as artificial insemination could successfully bypass prezygotic barriers, fertilization may prove to be even less specific. With removal of the outer zona pellucida of the oocyte, sperm-egg fusion between species as disparate as humans and Syrian hamsters (Mesocrecitus auratus) supports that sperm-egg interactions during fertilization may not be highly species specific (Yanagimachi, Yanagimachi & Rogers, 1976; Berros et al., 1978; Aitken, 2006). Unlike in most sexually reproducing species, sperm-egg fusions in P. formosa with sperm from other species can result in embryo development and live viable offspring. Further study of the differences between sperm that result in live offspring compared to sperm that does not result in embryo development can help identify non-genetic components of sperm, such as messenger RNA, long non-coding RNA, or short non-coding RNA, associated with embryo development. Some sperm RNAs are differentially expressed between fertile and infertile men, but it is difficult to extrapolate association to causation in humans (Santiago et al., 2022). If important sperm RNAs are identified using a P. formosa model, the identified RNAs would be a good target for experiments microinjecting identified RNAs into a mouse zygote to confirm epigenetically driven differences in fertility and offspring phenotype (Grandjean et al., 2015). Because many RNAs are highly evolutionarily conserved between fish and mammals (Ord et al., 2020), P. formosa may be a valuable model for translational research to human infertility.