Life-history evolution in livebearing fishes (Teleostei: Poeciliidae)

Abstract

Life histories lie at the heart of biology, because the tradeoffs each organism faces concerning the distribution of limited resources into either reproduction or maintenance and growth determine that organism's fitness. Generally speaking, an organism can choose to either invest in traits reducing age-specific mortality or to invest in traits increasing age-specific fecundity. Each decision in this regard leads to tradeoffs between current versus future reproduction, somatic maintenance or growth versus reproduction, or number, size and sex of offspring. Because of this, basic life history data provides the backbone for any organism-oriented research, effectively linking behavior, ecology, morphology, and physiology together.

One family of teleosts fishes, the livebearing fishes (Poeciliidae), has proven to be excellent models for studying life history adaptations. Within Poeciliidae, at least three characters evolved to give rise to the present diversity within the family: (1) internal fertilization using a transformed anal fin, referred to as the gonopodium, for sperm transfer, (2) livebearing, and (3) different degrees of maternal provisioning for the developing young.

Over the years, certain aspects of poeciliid life-history evolution have received particular attention. First and foremost, different predatory regimes have been demonstrated to drive rapid microevolutionary change in guppies, although evidence for macroevolutionary processes is as yet missing. Second, placental evolution and maternal provisioning strategies have been extensively studied, with the main focus of these studies on interspecific rather than intraspecific differences.

The research detailed in my dissertation investigates several specific aspects of life-history evolution in livebearing fishes (Poeciliidae) that have so far been largely neglected. The first three chapters broaden our understanding of the importance of life-history evolution in the colonization of and ecological speciation in extreme habitats. Taken together, they highlight the relevance of life-history evolution in maintaining and even driving ecological speciation processes. The fourth chapter emphasizes the significance of environmental affects and intraspecific population differences on maternal provisioning strategies in livebearing fishes. Thus, cautioning against the traditional approach of investigating few ecologically similar populations and ascribing the identified patterns as rigid strategies for that particular species across its natural distribution. Finally, my last chapter opens the door for research on life histories as regulatory mechanisms for stability in the unisexual/bisexual mating system of the Amazon molly.

Extremophile poeciliids

Although mechanisms that can lead to speciation are of fundamental importance to evolutionary biology, the actual process of speciation is still one of the least understood aspects of evolution. Recently, the idea that reproductive isolation and ultimately speciation can be the result of ecologically-based divergent selection (i.e., ecological speciation) has received a lot of attention. Two main components of ecological speciation are currently recognized: different sources of divergent selection, and different forms of reproductive isolation.

Basically, sources of divergent selection can stem from differences between environments, ecological interactions, and/or sexual selection. Under the first scenario, populations inhabit environments that, for example, differ in resource availability or habitat structure, and as populations adapt to their specific environment, the may begin to diverge from one another. Under the second scenario, divergent selection between populations arises due to ecological interactions, for example via competition for shared resources. Finally, under the third scenario, mate preferences differ between populations from ecologically different environments, thus leading to divergent sexual selection.

Different forms of reproductive isolation are usually distinguished based on whether they occur before (premating isolation) or after mating (postmating isolation). Premating isolation can for example evolve if populations are separated in time or space, or if populations are sexually isolated, i.e., populations simply differ in their mating signals and preferences. Postmating isolation on the other hand arises due to reduced hybrid fitness, which can originate from genetic incompatibilities, a mismatch between hybrid phenotypes and the environment, or a mismatch between hybrid phenotypes and the mating preferences of conspecifics.

Poeciliids inhabit a wide variety of different habitats, ranging from small creeks to large streams, freshwater lakes to coastal (brackish) lagoons, and even subterranean to toxic environments. In particular, poeciliid species of at least three genera (i.e., Gambusia, Limia, and Poecilia) are known to have independently colonized various toxic habitats in the Dominican Republic, Mexico, and the United States. Some species, like Gambusia eurystoma, Limia sulphurophila, and Poecilia sulphuraria are even endemic to sulfidic waters, and via ecological speciation, are actually derived from poeciliids that can be found in sulfidic and nonsulfidic waters of adjacent habitats. In these systems, toxicity stems from naturally occurring hydrogen sulfide, which is acutely toxic to most metazoans, because it competes with oxygen in the respiratory chain. Nonetheless, extremophile poeciliids thrive in these habitats and often occur in high local population densities.

A plateau in Tabasco, southern Mexico, provides an even more extreme suit of habitats. Here, divergent selection stems from drastic differences between the physio-chemical characteristics of environments inhabited by populations of the Atlantic molly (Poecilia mexicana), and habitats can be grouped into four distinct classes: toxic/cave, nontoxic/cave, toxic/surface, and nontoxic/surface. All habitat types are interconnected (i.e., no physical barriers prevent movement between environments) and merely several hundred meters apart; nonetheless, locally adapted populations are characterized by profound genetic differentiation coupled with strong behavioral, morphological, and physiological divergence. Premating isolation seems to be largely driven by natural selection in the cave molly system, because translocation experiments revealed high mortalities in fish transferred to waters of a different adjacent habitat type. But even a weak form of sexual isolation has been demonstrated: females prefer males from their own populations over males from adjacent environments. However, the role of life histories in population divergence in this and similar systems has so far remained unstudied.

In my first chapter, I therefore began to address this issue by conducting a preliminary study that investigated differences in life-history evolution (i.e., fecundity) between Poecilia mexicana from the sulfidic Cueva del Azufre and from a benign surface habitat. Using a combination of data from field-collected and laboratory-reared animals, I was able to demonstrate that cave mollies not only exhibit reduced fecundity, but also that fecundity in cave mollies is less plastic than in their surface-dwelling counterparts. Although this clearly suggests a heritable component to fecundity divergence in cave mollies, the extent of total life history divergence between cave and surface mollies remained unexplored. Furthermore, these results led me to ask the following question: Was reduced fecundity a response to permanent darkness, to toxicity, or to the combination of both selective forces?

In my second chapter, I attempted to address these new questions. To do so, I conducted a full life-history analysis on females collected from four types of habitats, which provided me with a natural 2x2 design, where the same species inhabits environments characterized by all possible combinations of the two selective forces darkness and toxicity (i.e., dark/toxic vs. dark/nontoxic vs. light/toxic vs. light/nontoxic). The data demonstrate a habitat-specific divergence in P. mexicana life histories, and both darkness and toxicity seem to select for the same trait dynamic: low fecundity and large offspring size. This particular trait dynamic most likely arose as a strategy to avoid cannibalism in the extreme habitats, and to create a more efficient body-volume-to-body-surface-area ratio with regards to the amount of body surface exposed to toxicity. Thus, my third chapter adds further evidence to the notion of ecological speciation driving population differentiation between populations from these different habitat types. However, are any of these life history patterns more generally applicable to life-history evolution and ecological speciation processes in poeciliids and other organisms, or is this type of divergence unique to the cave molly system?

In order for me to focus on this, I needed to take a broader approach. Hence, for my third chapter, I turned to another system of poeciliids in toxic waters: the Banos del Azufre system, in Tabasco, southern Mexico. Here, two species (Gambusia eurystoma and Poecilia sulphuraria) have diverged, and eventually speciated, in sulfidic waters. Both of these sulfide-endemics derived from two more widespread poeciliids that inhabit surrounding nontoxic and toxic habitats (Gambusia sexradiata and Poecilia mexicana, respectively). In chapter 3, I report on a comparative analysis, in which I contrast life histories of all four species from different habitat types. Even though I also found evidence for genus-specific responses to toxicity, the most pronounced pattern was the same as in the cave molly: hydrogen sulfide induced low fecundity and large offspring size, and the higher the toxicity, the lower the fecundity and the larger the offspring. Overall, my first three chapters therefore provide me with evidence that poeciliid fishes colonizing sulfidic habitats exhibit convergent life history evolution, with the potential to eventually result in complete speciation (as is the case with P. sulphuraria and G. eurystoma). I therefore propose that after an initial period of trait divergence, life-history evolution may provide an additional mechanism for further divergence, because dispersers between habitats will suffer from reduced fitness due to the wrong life history strategy in the new environment. Hence, life histories can be an important mechanism for further divergence in the advanced stages of ecological speciation processes.

In my fourth chapter, I am turning to another important life-history aspect of livebearing animals; namely the question of maternal strategies for embryo provisioning. Basically, a mother has two choices: she can either store all the resources required for successful embryo development within the yolk prior to fertilization (lecithotrophy), or she can supplement yolk-stored nutrients with additional provisions via direct transfer after fertilization (matrotrophy). Even though recent theoretical models and empirical studies have stressed that maternal provisioning strategies are resource-dependent, mainstream life history research still attempts to classify an organism as being either exclusively lecithotrophic or matrotrophic. I therefore compared maternal provisioning strategies between two populations of Poecilia mexicana that inhabit vastly different environments: a toxic, resource-poor limestone cave and a benign, resource-rich surface habitat. Furthermore, I directly compared two different techniques that are widely used to quantify maternal provisioning: the indirect matrotrophy index analysis and the direct radio-tracer assay of maternal provisioning. According to the matrotrophy index analysis, both populations of P. mexicana are purely lecithotrophic, while according to the radio-tracer assay, both populations provide similar levels of postfertilization nutrient transfer (i.e., matrotrophy). Together with results from chapter 2, this suggests that P. mexicana is at least capable of incipient matrotrophy and that, to avoid misclassification, both techniques of quantifying nutrient transfer should ideally be employed together. Finally, I propose that current theories on the evolution of matrotrophy in poeciliids need to be revised, since this chapter and other recent research suggest that most livebearing fishes are probably characterized by dual provisioning rather than distinct strategies of either lecithotrophy or matrotrophy.

Unisexual poeciliids

Unisexual vertebrates have long been used as model systems to study the maintenance of recombination and all species described to date originated as hybrids. Among unisexual vertebrates, fishes are of central interest because of their peculiar reproductive mechanisms. Several unisexual fishes reproduce via sperm-dependent parthenogenesis (gynogenesis), where sperm is required to initiate embryogenesis, but inheritance is strictly maternal.

In such mating systems involving a bisexual and a unisexual species we find two competing types of females, which rely on the same resource, sperm, but have fundamentally different population dynamics. According to theoretical models, the unisexuals should quickly outcompete the bisexuals, thus driving them to extinction. However, this would inevitably lead to their own subsequent demise. Hence, the stability of bisexual/unisexual mating systems and the factors underlying the maintenance of this stability are currently of great interest in evolutionary ecology.

The Amazon molly, Poecilia formosa, is a gynogenetic, all-female poeciliid. Poecilia formosa resulted from a single natural hybridization of the two sexual species Poecilia latipinna and Poecilia mexicana, and uses sperm from males of the two parental species for gynogenetic reproduction. Ecological differences and behavioral regulation have been proposed as two possible mechanisms explaining the stability in complexes of gynogens and sexuals, because at certain stages in their life history the gynogen could suffer a significant reduction of fitness. Since P. latipinna and P. mexicana males are not related to the resulting offspring from matings with Amazon mollies, they should be under selection to avoid them. Accordingly, both P. latipinna and P. mexicana males have been shown to (a) discriminate between heterospecific and conspecific females, and (b) to prefer to mate with conspecific females under most circumstances. However, these mating preferences are only of importance if they translate into reduced mating success of gynogens relative to conspecific females.

For my fifth and final chapter, I therefore investigated the mate choice:life history interface by documenting the presence of sperm in five natural populations. I extracted sperm from female fish of five syntopic populations of P. formosa and P. latipinna in Texas by flushing out the genital tract. A higher proportion of bisexual females had sperm than unisexuals. Also, among those females that had sperm, bisexuals had more sperm than unisexuals. Even though the results gained from this analysis cannot be used to infer the amount of sperm transferred per copulation or the number of heterospecific matings, they nonetheless show that Amazon mollies receive less sperm in the wild than Sailfin molly females. This represents the first study to investigate the ultimate effects of male mate choice as a stabilizing factor in natural populations, and my results suggest that P. formosa may in fact be sperm limited; however, future studies on life-history differences will have to determine whether this actually results in a fitness reduction.