The Evolutionary Ecology of Poecilia Mexicana in the Cueva Del Azufre System Effects of Abiotic and Biotic Environmental Conditions

Zurich Open Repository and Archive University of Zurich Main Library Strickhofstrasse 39 CH-8057 Zurich www.zora.uzh.ch

Year: 2008

The evolutionary ecology of poecilia mexicana in the cueva del azufre system effects of abiotic and biotic environmental conditions

Tobler, M

Abstract: Caves, with their divergent environmental conditions, provide some of the most unusual habitats on earth and harbor a diversity of highly adapted endemic organisms. Many aspects of the ecology and evolution of cave organisms, however, are poorly understood; probably because the inaccessibility of their habitats, the often small population sizes and their conservation status, as well as the lack of closely related epigean species that would allow for comparative studies. For my thesis, I explore how divergent abiotic conditions and correlated biotic conditions affect the ecology of a small livebearing fish occurring in cave as well as in surface habitats. Furthermore, I identify the evolutionary responses to selective pressures imposed by the environment, ultimately with the goal to contribute to the understanding of the processes that lead to ecological and phenotypic diversity and speciation. My research was conducted in Cueva del Azufre system in southern Mexico where the study species (Poecilia mexicana, Poeciliidae) has occurs in four habitat types: non-sulfidic surface, sulfidic surface, non-sulfidic cave, and sulfidic cave.I documented independent and partially heritable morphological variation as well as genetic differentiation along environmental gradients, providing evidence for parapatric adaptive divergence. I also investigated aspects of the trophic ecology, host-parasite interactions, and predator-prey interactions in the system.

Posted at the Zurich Open Repository and Archive, University of Zurich ZORA URL: https://doi.org/10.5167/uzh-5335 Dissertation Published Version

Originally published at: Tobler, M. The evolutionary ecology of poecilia mexicana in the cueva del azufre system effects of abiotic and biotic environmental conditions. 2008, University of Zurich, Faculty of Science.

THE EVOLUTIONARY ECOLOGY OF POECILIA

MEXICANA IN THE CUEVA DEL AZUFRE SYSTEM

EFFECTS OF ABIOTIC AND BIOTIC ENVIRONMENTAL

CONDITIONS

Dissertation

zur

Erlangung der naturwissenschaftlichen Doktorwürde (Dr. sc. nat.)

vorgelegt der

Mathematisch-naturwissenschaftlichen Fakultät

der

Universität Zürich

von

Michael Tobler aus Thal SG

Promotionskomitee

Prof. Dr. Heinz-Ulrich Reyer (Vorsitz) Prof. Dr. Ingo Schlupp (Leitung der Dissertation) Prof. Dr. Lukas Keller Prof. Dr. Dustin Penn

Zürich, 2008

For Lea and Ally

Contents 1

CONTENTS Summary/ Zusammenfassung 3

Chapter 1 8 The evolutionary ecology of Poecilia mexicana in the Cueva del Azufre system – Effects of abiotic and biotic environmental conditions Introduction and synthesis

Chapter 2 21 Life on the edge: Hydrogen sulfide and the fish communities of a Mexican cave and surrounding waters. Extremophiles 10: 577-585 (2006)

Chapter 3 36 A new and morphologically distinct population of cavernicolous Poecilia mexicana (Poeciliidae: Teleostei) Environmental Biology of Fishes 82: 101-108 (2008)

Chapter 4 47 Phenotypic and genetic divergence across two abiotic environmental gradients in Poecilia mexicana Evolution (in press)

Chapter 5 78 Divergence in trophic ecology characterizes colonization of extreme habitats Biological Journal of the Linnean Society (in press)

Chapter 6 100

Predation of cave fish (Poecilia mexicana, Poeciliidae) by a giant water-bug (Belostoma, Belostomatidae) in a Mexican sulfur cave Ecological Entomology 32: 492-495 (2007)

2 Contents

Chapter 7 107 Extreme habitats as refuge from parasite infections? Evidence from an extremophile fish Acta Oecologica 31: 270-275 (2007)

Chapter 8 117 Perspectives

References 119

Acknowledgements 145

Curriculum vitae 147

Summary/ Zusammenfassung 3

SUMMARY Caves, with their divergent environmental conditions, provide some of the most unusual habitats on earth and harbor a diversity of highly adapted endemic organisms. Many aspects of the ecology and evolution of cave organisms, however, are poorly understood; probably because the inaccessibility of their habitats, the often small population sizes and their conservation status, as well as the lack of closely related epigean species that would allow for comparative studies. For my thesis, I explore how divergent abiotic conditions and correlated biotic conditions affect the ecology of a small livebearing fish occurring in cave as well as in surface habitats. Furthermore, I identify the evolutionary responses to selective pressures imposed by the environment, ultimately with the goal to contribute to the understanding of the processes that lead to ecological and phenotypic diversity and speciation. Chapter 1 provides a short introduction to pertinent concepts in cave biology and a synthesis of my major research questions and results. My research was conducted in Cueva del Azufre system in southern Mexico where the study species (Poecilia mexicana, Poeciliidae) has occurs in four habitat types: non-sulfidic surface, sulfidic surface, non-sulfidic cave, and sulfidic cave. Chapters 2 & 3 provide an introduction and characterization the abiotic and biotic environmental factors in the study system. Specifically, the distribution of toxic hydrogen sulfide (H2S) is investigated in chapter 2, and a newly discovered cave population of P. mexicana is described in chapter 3. The Cueva del Azufre system provides an unparalleled ‘natural experiment’ with two strong selective pressures (the presence or absence of light and H2S) occurring in a fully 2x2 factorial design. For chapter 4, I investigated the genetic and phenotypic differentiation of P. mexicana in different habitat types using molecular genetic and morphological analyses. I document independent and partially heritable morphological variation along each environmental gradient. Molecular genetic analyses using microsatellites as well as cytochrome b gene sequences indicate high population differentiation and very low rates of gene flow among populations from different habitat types despite the spatial proximity and the lack of physical barriers. Altogether, the study provides evidence for parapatric adaptive divergence in response to divergent natural selection by abiotic environmental conditions. 4 Summary/ Zusammenfassung

For chapter 5, I investigated differences in the tropic ecology of P. mexicana in the different habitat types. Resource use in different habitat types was investigated using gut content analysis. A shift in resource use, from algivory/ detrivory to the incorporation of invertebrate food items, was detected upon colonization of the divergent habitats. P. mexicana in cave habitats further exhibited a higher dietary niche width than conspecifics from surface habitats. Condition of P. mexicana was analyzed using storage lipid extractions, and fish from sulfidic and cave habitats exhibited a very poor condition hinting towards resource limitation or high costs of coping with extreme conditions. Finally, the shift in resource use was accompanied by divergence in viscerocranial morphology. Although the divergent morphological traits investigated were phenotypically plastic to some extent, they appear to have a genetic basis. It is suggested that the morphological diversification is an adaptation to the differential use of resources among populations. Caves are often assumed to be predator-free environment for cave fishes. This has been proposed to be a potential benefit of colonizing these otherwise relatively hostile environments. In chapters 6, I tested this hypothesis by investigating the predator-prey interaction of a belostomatid water-bug (predator) and P. mexicana. I determined feeding rates and size-specific prey preferences of the predator, and estimated the population density of Belostoma using a mark-recapture analysis. Belostomatids were found to heavily prey on cave mollies and to exhibit a prey preference for large bodied fish. The mark-recapture analysis revealed a high population density of the heteropterans in the cave. Although the absence of predators is not a general habitat feature of cavernicolous P. mexicana, this study highlights the fundamental differences in predatory regimes between epigean and cave habitats. In chapter 7, I suggest that extreme environments in general, and cave habitats in particular, may function as refuge from parasite infections, since parasites can become locally extinct either directly, through selection by an extreme environmental parameter on free-living parasite stages, or indirectly, through selection on other host species involved in its life cycle. Populations from such sulfidic and cave habitats are significantly less parasitized by the trematode Uvulifer sp. than populations from a non-sulfidic habitat and it is suggested that reduced parasite prevalence may be a benefit of colonizing otherwise inhospitable habitats. Finally, in chapter 8, I provide some conclusion of my thesis and perspectives for future research. Summary/ Zusammenfassung 5

ZUSAMMENFASSUNG Höhlen gehören mit ihren abweichenden Umweltbedingungen zu den aussergewöhnlichsten Habitaten der Erde und werden von einer Vielfalt von hoch angepassten Endemiten bewohnt. Viele Aspekte der Ökologie und Evolution von Höhlenorganismen werden bisher allerdings nur unzureichend verstanden, was unter anderem daran liegt, dass die Habitate nur schwer zugänglich sind, dass Höhlenorganismen nur kleine Populationsgrössen haben und entsprechend gefährdet sind, und dass meist keine nah verwandten Arten aus Oberflächenhabitaten für vergleichende Studien vorhanden sind. Für meine Dissertation habe ich untersucht, wie abweichende abiotische Umweltfaktoren und korrelierte biotische Bedingungen die Ökologie einer kleinen lebendgebärenden Fischart beeinflusst, die in Höhlen- wie auch Oberflächenhabitaten vorkommt. Zudem habe ich erforscht, welche Anpassungen die Fische an die unterschiedlichen Umweltbedingungen evolviert haben, um so zum Verständnis der Prozesse beizutragen, die zu ökologischer und phänotypischer Diversität und Artbildung führen. In Kapitel 1 werden die relevanten höhlenbiologischen Konzepte eingeführt und die wichtigsten Fragen und Resultate dieser Arbeit zusammengefasst. Meine Forschungsarbeit wurde im Cueva del Azufre System in Südmexiko durchgeführt, wo die studierte Art (Poecilia mexicana, Poeciliidae) in vier Habitaten vorkommt: nicht-schweflige Oberflächenhabitate, schweflige Oberflächenhabitate, nicht-schweflige Höhlenhabitate sowie schweflige Höhlenhabitate. Die Kapitel 2 und 3 bieten eine Einführung in die unterschiedlichen Habitate und charakterisieren die abiotischen und biotischen Umweltbedingungen. Im Speziellen wird die Verbreitung von giftigem Schwefelwasserstoff (H2S) in Kapitel 2 untersucht, und eine neu entdeckte Höhlenpopulation von P. mexicana in Kapitel 3 beschrieben. Das Cueva del Azufre System bietet demnach ein einzigartiges ‚natürliches Experiment’ mit zwei starken Selektionsdrücken (das Vorhandensein und Fehlen von Licht und H2S), die in einem vollständigen faktoriellen Design vorkommen. Für Kapitel 4 habe ich die genetische und phänotypische Differenzierung von P. mexicana in den unterschiedlichen Habitattypen mit molekulargenetischen und morphologischen Analysen untersucht. Ich dokumentiere unabhängige und teilweise erbliche morphologische Variation entlang beider Umweltgradienten. Die molekulargenetischen Analysen mittels Mikrosatelliten und Cyochrom b 6 Summary/ Zusammenfassung

Gensequenzen zeigen eine hohe genetische Differenzierung und wenig Genfluss zwischen Populationen trotz der geringen Distanzen und dem Fehlen von Migrationsbarrieren. Diese Ergebnisse geben Hinweise auf eine parapatrische, adaptive Divergenz von P. mexicana durch die unterschiedliche natürliche Selektion von abiotischen Umweltfaktoren in den verschiedenen Habitaten. Für Kapitel 5 habe ich Unterschiede in der Nahrungsökologie von P. mexicana untersucht. Die Ressourcennutzung in den verschiedenen Habitaten wurde durch Mageninhaltsuntersuchungen ermittelt. Ein Wechsel in der Ressourcennutzung von Algivorie/ Detrivorie hin zum Konsum von Wirbellosen hat während der Kolonisation von Schwefel- und Höhlenhabitaten stattgefunden. Zudem nutzen P. mexicana aus Höhlenhabitaten eine grössere Vielfalt von Nahrungsquellen als Tiere an der Oberfläche. Die Kondition der Fische wurde mittels Fettextraktionen untersucht. Tiere aus Schwefel- und Höhlenhabitaten hatten eine tiefe Kondition, was ein Indiz für Energielimitierung oder hohe Kosten für die Toleranz der Extrembedingungen sein könnte. Letztendlich war der Wechsel in der Ressourcennutzung von Änderungen in der Kopf- und Darmmorphologie begleitet. Obwohl die Kopfmorphologie zu einem gewissen Grad phänotypisch plastisch ist, scheinen die Unterschiede zwischen den Populationen eine genetische Basis zu haben. Die morphologische Differenzierung zwischen Population ist wahrscheinlich eine Anpassung an die unterschiedliche Ressourcennutzung. Es wird oft angenommen, dass Höhlenfische keine natürlichen Räuber haben, was als ein möglicher Vorteil für das Leben in einem sonst unwirtlichen Habitat angesehen wird. In Kapitel 6 habe ich diese Hypothese getestet, in dem ich die Räuber-Beute-Beziehung zwischen einer Wasserwanze (Räuber) und P. mexicana untersucht habe. Ich habe Fressraten der Wanzen und ihr grössenspezifisches Beutewahlverhalten untersucht, sowie die Populationsdichten in der Höhle bestimmt. Die Wanzen haben sich als starke Fischprädatoren herausgestellt, die eine Beutepräferenz für grosse Fische haben. Auch kommen sie in der Höhle in grossen Dichten vor. Obwohl das Habitat von höhlenbewohnenden P. mexicana also nicht räuberfrei ist, hebt diese Studie trotzdem die Unterschiede in den Prädationregimen von Oberflächen- und Höhlenhabitaten hervor. In Kapitel 7 schlage ich vor, dass Extremhabitate im generellen – und Höhlenhabitate im speziellen – eine Art Rückzugsgebiet vor Parasiteninfektionen sein können, da Parasiten lokal aussterben können. Lokales Aussterben kann direkt Summary/ Zusammenfassung 7 erfolgen, wenn Parasiten mit den abweichenden Umweltbedingungen nicht zurecht kommen, oder indirekt, wenn die Selektion durch die abweichenden Umweltbedingungen andere im Lebenszyklus notwendige Wirte ausmerzt. Tatsächlich sind P. mexicana in Schwefel- und Höhlenhabitaten deutlich weniger vom Trematoden Uvulifer sp. parasitisiert als Populationen aus normalen Oberflächengewässern. Ich schlage vor, dass eine reduzierte Prävalenz von Parasiten ein möglicher Vorteil einer Kolonisation von sonst unwirtlichen Habitaten ist. In Kapitel 8 sind letztlich einige Schlussfolgerungen und Perspektiven für zukünftige Arbeiten zu finden.

8 Chapter 1: Introduction and synthesis

Chapter 1

THE EVOLUTIONARY ECOLOGY OF POECILIA MEXICANA IN

THE CUEVA DEL AZUFRE SYSTEM:

EFFECTS OF ABIOTIC AND BIOTIC ENVIRONMENTAL FACTORS

INTRODUCTION AND SYNTHESIS

In my thesis, I am interested in how species adapt to their environment; for this purpose I examine Poecilia mexicana (the Atlantic molly) living in surface and cave habitats. In this chapter, I first review the pertinent concepts and empirical data on the ecology and evolution of cave organisms highlighting the current gaps of knowledge. Subsequently, I provide a synthesis of my thesis research by outlining the major questions addressed in the subsequent chapters and briefly reviewing the main results in the context of other recent advances in the understanding of the model system.

The cave environment Caves provide us with exceptional ‘evolutionary laboratories’. Cave-dwelling organisms have long captured the interest of evolutionary biologists, probably because of their bizarre morphology and the exotic nature of their habitats (Darwin, 1859). Although there is a huge body of theoretical work, the ecology and evolution of many cave organisms are still poorly understood. Firstly, I address the biotic and abiotic environmental conditions in caves, most of which differ strikingly from most other habitats on earth.

Darkness Light generally can be viewed as transport of energy and information, and its effects on various levels of biological organization – from cells to ecosystems – in surface habitats can be classified in the following categories: (1) The UV-spectrum of the light matches the absorption spectrum of biomolecules and may thus act as a mutagen. (2) Light allows for vision. (3) Periodical fluctuations of solar radiation regulate daily and seasonal light cycles and accordingly fluctuations in temperatures rendering light Chapter 1: Introduction and synthesis 9 an important regulator of circadian and circannual rhythms. (4) Light allows for photoautotrophic primary production and thus serves as the fuel for ecosystem productivity. Consequently, the absence of light has profound effects on the inhabitants of subterranean ecosystems. On the one hand, certain selective pressures imposed by light are relaxed, which make adaptations to the presence of light obsolete (e.g., traits related to vision). On the other hand, darkness imposes a suite of novel selection pressures to which epigean organisms are generally not exposed (Langecker, 2000).

Other abiotic environmental factors Besides these direct and indirect effects of permanent darkness on the ecology and evolution of cave-dwellers, other specific divergent environmental factors characterize caves. Howarth (1993) emphasizes the importance of the complex maze- like habitat structure and barren rocky substrates. Furthermore, carbon dioxide, oxygen, and other gas concentrations may be at suboptimal or even lethal levels (Howarth and Stone, 1990). For example, the water and air of some caves contain high concentrations of hydrogen sulfide (H2S) (Sarbu et al., 1996; Macalady et al., 2006). Sulfide is an inhibitor of the cytochrome c oxidase, blocking the electron transport in aerobic respiration, thereby hampering the function of mitochondria and the production of ATP (Lovatt Evans, 1967; Nicholls, 1975; Petersen, 1977; National

Research Council, 1979). H2S is also able to modify oxygen transport proteins (Carrico et al., 1978; Park et al., 1986) and inhibit about 20 other enzymes

(Bagarinao, 1992; Reiffenstein et al., 1992). Due to its biochemical effects, H2S is highly toxic for aerobic organisms even in micromolar amounts (Torrans and

Clemens, 1982; Bagarinao, 1992; Grieshaber and Völkel, 1998). H2S is highly reactive at room temperature and spontaneously oxidizes in water (Cline and Richards, 1969; Chen and Morris, 1972), which also leads to and aggravates hypoxia in aquatic systems (Bagarinao, 1992). Although potentially important as selective factors, the effects of suboptimal gas concentrations and other abiotic environmental factors on subterranean organisms has rarely been evaluated; and research to date is primarily concerned with direct and indirect effects of continuous darkness.

10 Chapter 1: Introduction and synthesis

Biotic environmental factors Compared to surface habitats, caves are characterized by harboring low biodiversity and truncated food webs as both primary producers and specialized predators are rare (Gibert and Deharveng, 2002). Due to the absence of photosynthetic primary production, most cave ecosystems rely on organic matter brought in from epigean habitats as a source of carbon. Thus, caves are usually viewed as energy-poor habitats, in which resources are patchily distributed (Hüppop, 2000; Poulson and Lavoie, 2000). However, allochthonous input in caves may be substantial, especially in the tropics or if roosting bats are present (Poulson and Lavoie, 2000). In some caves, primary production through chemoautotrophic bacteria has been reported (Sarbu et al., 1996; Opsahl and Chanton, 2006). In summary, caves provide a vastly different abiotic and biotic environment to organisms colonizing them, and cave colonizers are subject to a set of divergent environmental selection pressures. What are the origins of diversity in cave faunas, and how did organisms adapt to the unusual abiotic and biotic environmental conditions?

Evolution of cave organisms Colonization of caves Originally, all cave organisms invariantly descended from epigean ancestors that colonized subterranean habitats and subsequently diverged into distinct evolutionary lineages. Cave colonization has long been viewed as a passive process, where organisms were accidentally washed into the underground and got trapped (Wilkens, 1979; Langecker, 1989). According to Romero and Green (2005), this notion predominantly derived from the – faulty – logic that caves provide a harsh environment that organisms would try to avoid if possible. In fact, most cave habitats are continuous with adjacent epigean habitats, and no permanent physical barriers prevent organisms from returning to their original habitats (e.g., Romero et al., 2002; Reis et al., 2006). Alternatively, cave colonization by epigean organisms may be active and advantageous, and potential advantages include environmental stability, exploitation of unoccupied niches, as well as a reduction in competition and predation (Romero and Green, 2005). To date, however, there are very few empirical studies investigating potential benefits of cave colonization, and it is unclear why certain evolutionary lineages are more successful than others in colonizing cave habitats. Chapter 1: Introduction and synthesis 11

Evolutionary divergence of troglobites A predominant view in the literature is that obligate cave-dwellers evolved from facultative cave colonizers that became isolated through extinction of the surface ancestor (Barr, 1968). Fluctuating environmental conditions during periods of climate change are thought to drive surface populations to extinction while cave populations sustain in suitable habitats (e.g., in terms of temperature or humidity); hence this hypothesis is known as the climate-relict model (Barr, 1968; Wilkens, 1973; Humphreys, 1993; Trajano, 1995; Allegrucci et al., 2005). Essentially, this is a two- step model of cave evolution, where caves were first colonized without restriction of gene flow with surface populations followed by allopatric genetic and phenotypic differentiation in geographically isolated caves after the local extinction of surface- dwelling ancestors. The climate-relict model is inherently difficult to test in specific cases, since it fails to provide exclusive testable predictions. Even if phylogenetic and biogeographic patterns are congruent with the reconstruction of climatic changes in the past, it is impossible to exclude the possibility that troglobites diverged parapatrically from epigean ancestors before the respective surface populations went extinct. So, although many caves with their geological age and their environmental stability may indeed act as climatic refugia for phylogenetic lineages long extinct in adjacent surface habitats, the climate-relict hypothesis only offers limited insight to the ultimate mechanisms of the origin of troglobites. An alternative hypothesis prevalent in the biospeleological literature is the ‘adaptive shift’ model, where epigean species invade caves to exploit novel resources and subsequently adapt to the divergent environmental conditions present (Howarth, 1987). Divergence between epigean ancestors and cave inhabitants is usually assumed to occur parapatrically along the steep environmental gradient. Additional, sometimes very specific, models have been proposed to explain the origins of troglobites (see Danielopol and Rouch, 2005 for an overview). Generally the investigation of the origins of subterranean organisms has been plagued by the lack of appropriate study systems. Many lineages of troglobites are phylogenetically old, making the elucidation of the mechanisms of initial colonization and divergence an improbable task. Future work will have to put more emphasis on phylogenetically young systems, where cave colonization occurred recently or is still in progress (e.g., Schilthuizen et al., 2005). Upon the identification of suitable 12 Chapter 1: Introduction and synthesis systems, questions about the ecology of cave colonization as well as the mechanisms of divergence and the evolution of reproductive barriers could be addressed more vigorously.

Cave adaptation and regressive evolution Trait evolution in cave organisms is governed by two themes: the regression of traits that became obsolete in permanent darkness and the evolution of novel adaptations that allow for coping with the divergent environmental conditions. Convergent evolution has led to similar modifications of traits across taxonomic groups (e.g., Porter and Crandall, 2003; Verovnik et al., 2004; Xiao et al., 2005; Moller et al., 2006). Troglobites are characterized by the regression of traits, the function of which is impaired or obsolete in the cave environment. For example, troglobites exhibit reduced pigmentation, photoreceptors, and brain centers associated with vision (all due to the absence of light) as well as a reduction of overall body size, muscles, ossification, and scales (in fishes). The latter traits are thought to have evolved in response to food scarcity and the lack of predators (Poulson, 1963; Poulson and White, 1969; Culver, 1982; Culver et al., 1995; Langecker, 2000). The evolutionary mechanisms that lead to the regression of traits, especially the role of natural selection, have been continually debated over the past decades (e.g., Sket, 1985; Culver and Fong, 1986). Regressive evolution of traits has been thought to be a purely neutral process, during which deleterious mutations in the underlying genetic architecture are not counter-selected in a lightless environment (Culver, 1982; Wilkens, 1988). Alternatively, it has been suggested that trait regression may be adaptive and caused by costs of making and maintaining certain structures (Culver, 1982) or pleiotropic effects, where organs beneficial to survival in the cave environment are enhanced at the expense of others (Barr, 1968). Recent, genetic and developmental studies in cave-dwelling Astyanax mexicanus indeed suggest that eye reductions are likely adaptive (Jeffery, 2001; Jeffery et al., 2003; Jeffery, 2005; Protas et al., 2007), whereas body pigment reduction in the same species apparently did not evolve in response to natural selection but rather due to genetic drift (Baker and Montgomery, 1999; Protas et al., 2006; Protas et al., 2007). Cave dwellers also exhibit a number of novel traits that were shaped by natural selection and unambiguously are adaptive in the cave environment. Along with the Chapter 1: Introduction and synthesis 13 reduction of visual senses, many cave organisms have elaborated non-visual senses such as tactile and chemical receptors as well as respective brain areas (Poulson, 1963; Langecker and Longley, 1993; Montgomery et al., 2001; Miller, 2005a). Cave organisms also evolved astonishing adaptations to the food scarcity in subterranean habitats (see Hüppop, 2000 for a review). Increased starvation resistance and reduced metabolic rates allow troglobites to withstand prolonged periods without food (Gannon et al., 1999; Adams and Johnson, 2001; Poulson, 2001). Adaptations to food scarcity also include shifts in life history strategies, i.e. different strategies of resource allocation, and cave organisms usually exhibit slow growth rates, increased longevity, delayed reproduction, as well as a reduced fecundity and larger individual offspring size (Christiansen, 1965; Hüppop and Wilkens, 1991; Winemiller, 1992).

Synthesis of the dissertation For my theses, I investigated the ecology and evolution of a small livebearing fish (Poecilia mexicana, Poeciliidae, Teleostei). Not only is this species widespread in surface habitats along the Atlantic versant of Mexico and Central America, but it has also colonized at least two caves in the Cueva del Azufre system. My goal was not only to address some of the major open questions in the evolutionary ecology of cave organisms outlined above and test some of the relevant concepts with empirical data. I also tried to link ecologically based divergent natural selection to evolutionary processes, thus providing insights to the origin of ecological and phenotypic diversity, which are generally relevant in ecological and evolutionary research. Specifically, I characterized the abiotic and biotic environmental conditions of cave and adjacent surface habitats, and I investigated the evolutionary origin of the cave populations as well as the differentiation of P. mexicana in different habitats. Furthermore, I compared the trophic ecology of populations inhabiting the different habitats. Below, I present my major research questions and briefly review my findings in the context of other studies on the model system.

Characterization of the model system The Cueva del Azufre is a cave in the southern Mexican state of Tabasco, which was first described by Gordon and Rosen in 1962, and was reported to be inhabited by a unique population of cavernicolous P. mexicana (also known as the cave molly). Subsequently, the cave molly served primarily as a model system to study adaptations 14 Chapter 1: Introduction and synthesis

– particularly on the behavioral level – to life in darkness. Differences in traits between cave and surface mollies were generally interpreted as consequential to the absence of light (see Parzefall, 2001 for a review). The extensive body of work on the cave molly was primarily based on laboratory studies. This led to a somewhat reductive view in two directions: (1) The predominant view of the Cueva del Azufre system was that it consists of two distinct habitat types, the Cueva del Azufre and normal surface habitats. (2) The absence of light was assumed to be the most important selective factor. Chapters 2 and 3 of this dissertation redefine and characterize the Cueva del Azufre system to amend these views by including additional habitats in comparative analyses and by recognizing the importance of hydrogen sulfide (H2S), which is present in some habitats of this system, as a selective factor. The discovery of a non-sulfidic cave (Cueva Luna Azufre; chapter 3) and independent sulfidic springs in surface habitats (El Azufre) in 2006 allowed me to extend the comparative approach of studying the ecology and evolution of P. mexicana under divergent environmental conditions. Consequently, four habitat types are recognized (non-sulfidic surface, sulfidic surface, non-sulfidic cave, and sulfidic cave) for most comparative analyses (see also Figures 1.1 to 1.3). The Cueva del Azufre system thus provides an unparalleled ‘natural experiment’ with two abiotic environmental factors occurring in a fully 2x2 factorial design.

For chapter 2, H2S concentrations were measured for the first time in the Cueva del Azufre system. The water in sulfidic cave and surface habitats contains up to 320 µM of H2S. Generally, the concentrations are not only considered toxic, but also high compared to other systems with naturally occurring H2S, such as deep sea hydrothermal vents (see Van Dover, 2000; Nybakken, 2001). Although the water in sulfidic habitats of the Cueva del Azufre system still contains low concentrations of oxygen, the conditions are considered hypoxic.

Genetic differentiation and the parapatry of divergent lineages As outlined above, questions about the ecology of cave colonization as well as the mechanisms of divergence among cave colonizers and epigean ancestors are ideally addressed in systems where cave colonization occurred recently or is ongoing. To test whether P. mexicana in the Cueva del Azufre system is fulfilling this prerequisite, the phylogenetic relationships as well as migration patterns and genetic Chapter 1: Introduction and synthesis 15 differentiation among populations of P. mexicana from different habitat types were investigated using molecular genetic analyses (chapter 4). Cytochrome b sequence data as well as microsatellite analyses in P. mexicana from the Cueva del Azufre system suggest that fish from the different habitat types are closely related but clearly distinct from each other. Each habitat type harbors a genetically highly differentiated population of P. mexicana, and gene flow among habitat types is low despite their spatial proximity and the lack of physical barriers. These results highlight that P. mexicana in the Cueva del Azufre system is one of the few known examples where cave colonization has occurred only recently, and cave populations still coexist parapatrically with their recent ancestors, to which they are connected by low rates of gene flow. The system thus provides a unique and rare opportunity not only to study ecological and evolutionary processes associated with cave colonization, but it is an emergent model system to investigate mechanisms leading to phenotypic and ecological diversification and speciation.

Phenotypic differentiation along abiotic gradients Cave organisms evolved a suite of traits due to the divergent selection in their habitats. I tested for phenotypic differentiation in P. mexicana not only in cave and surface habitats, but also in sulfidic and non-sulfidic habitats using a geometric morphometric analysis of body shape and a morphometric analysis of the gills (chapter 4). I also tested whether the divergent morphological traits have a heritable basis by comparing the shape variation of wild-caught fish and laboratory stocks housed under identical conditions. The results indicate that P. mexicana from different habitat types diverged phenotypically in terms of their morphology. Each habitat type in the Cueva del Azufre system harbors a morphologically distinct population. Most importantly, variation along the surface-cave axis is independent of that along the non-sulfide- sulfide gradient. The observed morphological differences among P. mexicana from different habitat types are not caused by environmentally induced phenotypic variation alone, since laboratory stocks retained the distinct morphology of wild- caught fish from the respective habitat types, indicating that morphological traits are partially heritable. Compared to fish from surface habitats, P. mexicana from the two caves are characterized by a reduction in eye size and more slender bodies (chapter 4) as well as 16 Chapter 1: Introduction and synthesis a reduction in body pigmentation (chapter 3). Individuals from sulfidic habitats are characterized by an increase in head size, which is correlated with an increase in gill filament length (chapter 4). An increase in gill size facilitates oxygen uptake in hypoxic environments (Graham, 2005). This highlights the importance of respiratory adaptations facilitating efficient oxygen acquisition for survival in sulfidic habitats. Survival of P. mexicana in sulfidic water is critically dependent on the possibility to perform aquatic surface respiration (ASR), where fish exploit the oxygen-rich air- water interface using their gills (Plath et al., 2007c).

Shifts in resource use Divergent habitat types differ not only in abiotic conditions, but these are correlated with differences in the biotic environment. For example, habitats may differ in productivity and species diversity, which may lead to differences in resource availability and quality as well as changes in competitive interactions within and between species. To study the ecological consequences of living in habitats differing in abiotic conditions, I investigated whether the colonization of divergent habitats was accompanied by a shift in resource use (chapter 5). A gut content analysis indicated such a shift. Unlike conspecifics from non- sulfidic surface habitats that foraged exclusively on algae and detritus, P. mexicana in the divergent habitats included invertebrate prey into their diet. Individuals from cave habitats further exploited a higher diversity of resources than those from surface habitats. These changes in resource use are probably driven by differences in resource availability among habitats as well as by the reduction or even lack of interspecific competition in sulfidic and cave habitats.

Energy limitation Two different hypotheses predict differences in energy availability and/ or demand in different habitat types, thus I examined the condition of P. mexicana from different habitat types using extraction of storage lipids (chapter 5). Survival in stressful environments is considered to be costly, and short-term survival of P. mexicana in sulfidic water critically depends on energy availability (Plath et al., 2007c). Likewise, caves are usually considered to be energy-poor (Hüppop, 2000), but the sulfidic cave investigated here has been reported to be energy-rich (even Chapter 1: Introduction and synthesis 17 compared to surface habitats) due to the chemoautotrophic primary production and the colony of roosting bats (Langecker et al., 1996). P. mexicana from the divergent habitats consistently exhibited poor body condition when compared to individuals from normal surface habitats (chapter 5).

Consequently, neither the cave environment nor the presence of H2S per se affect body condition negatively, suggesting that fish from different habitat types may have low body condition for different reasons. P. mexicana from the non-sulfidic cave exhibit low amounts of storage fats, probably because resources are scarce like in other caves. The body condition of fish in – apparently resource-rich – sulfidic habitats in turn may be low because coping with the toxic environment is energetically costly. Availability of energy and the possibility to perform ASR directly affect survival in P. mexicana (Plath et al., 2007c); and although ASR is necessary, it is also physiologically costly and constrains an individual’s energy budget, leaving less time for foraging (Kramer, 1983; Weber and Kramer, 1983).

Phenotypic differentiation in response to the divergent ecology Due to the pronounced differences of resource use among habitat types, I tested whether differences in the ecology among habitats – just like the abiotic conditions – led to phenotypic differentiation in trophic morphology (chapter 5). Indeed, populations foraging on invertebrates had wider mouths, stronger jaws, and shorter intestinal tracts. The differences in viscerocranial morphology did not mirror the differentiation found along the two abiotic gradients (where each population turns out to have a unique morphology with respect to overall body shape), but rather all insectivorous populations were relatively similar with respect to their viscerocranium and are clearly distinguishable from populations that do not consume invertebrates. Thus, differences in viscerocranial morphology are not simply explained by a correlated response to selection on other characteristics of the head, but may actually be adaptive to differential resource use among populations. Other ecological differences among P. mexicana from different habitats documented in this thesis offer potential explanations for other divergent traits documented in the past. Energy limitation may have selected for the genetically based reduction of costly behavioral traits. Shoaling behavior (Plath and Schlupp, 2008) as well as male aggression (Parzefall, 1974) and mating activity (Plath et al., 2003a; Plath et al., 2007b; Plath, 2008) are substantially reduced in P. mexicana from sulfidic 18 Chapter 1: Introduction and synthesis and cave habitats. Likewise, females of the sulfidic cave population have evolved a mating preference for males with a high nutritional state, a trait that is thought to have indicator value for male fitness in energy-poor but not necessarily in energy-rich habitats (Plath et al., 2005). P. mexicana from different habitat types also differ in their life-history strategies, and populations in sulfidic and cave habitats have reduced fecundity and larger offspring size at birth compared to fish from non-sulfidic surface habitats (Riesch et al., in press and unpublished data).

Evidence for active cave colonization The traditional view in cave biology is that cave colonization is primarily passive, where organisms accidentally got trapped in subterranean habitats. More recently, cave colonization has been hypothesized to be active and advantageous (Romero and Green, 2005), but comprehensive data testing this idea were lacking. Chapters 6 and 7 specifically address potential benefits of cave colonization in terms of predation by a giant water-bug and a novel parasite refuge hypothesis, respectively, but data from other chapters (especially 2 and 5) also give insights into the question. The biotic environmental data of the different habitat types as well as the feeding ecology of P. mexicana provide the first comprehensive evidence that advantages of cave colonization may exist. Firstly, the colonization of divergent habitat types was accompanied by a shift in resource use pointing towards the exploitation of thus far unused resources (chapter 5). The shift in resource use appears to be at least partially driven by the reduction or even lack of interspecific competition (chapter 2). Secondly, predators differ vastly among habitat types. Avian and piscine predators are absent in the cave habitats, but are common at least in non-sulfidic surface habitats (chapters 2 and 7). However, the sulfidic cave is not a predator free refuge for P. mexicana as it harbors a specialized predator (a giant belostomatid water-bug) in high densities (chapter 6 and Tobler et al., 2008a). Lastly, I hypothesized that cave habitats may offer an advantage in terms of reduced parasite exposure (chapter 7). Some parasites (e.g., the trematode Uvulifer sp.) that are highly abundant in non-sulfidic surface habitats have reduced prevalence in sulfidic surface habitats and are even completely absent in the sulfidic cave. Chapter 1: Introduction and synthesis 19

Figures chapter 1

Figure 1.1. Sample sites used in this thesis. Exact locations can be found in Table 4.1. CA depicts the entrance of the sulfidic caves, LA the entrance of the non-sulfidic cave, and EAI & II are the sulfidic surface sites. All other sites are non-sulfidic surface habitats used for comparison. The star in the inset indicates the location of the study system in Mexico.

20 Chapter 1: Introduction and synthesis

Figure 1.2. Map of the Cueva del Azufre.

Figure 1.3. Map of the Cueva Luna Azufre. Chapter 2: Hydrogen sulfide 21

Chapter 2

LIFE ON THE EDGE: HYDROGEN SULFIDE AND THE FISH

COMMUNITIES OF A MEXICAN CAVE AND SURROUNDING 1 WATERS

Michael Tobler, Ingo Schlupp, Katja U. Heubel, Rüdiger Riesch, Francisco J. García

de León, Olav Giere and Martin Plath

Abstract Most eukaryotic organisms classified as living in an extreme habitat are invertebrates. Here we report of a fish living in a Mexican cave (Cueva del Azufre) that is rich in highly toxic H2S. We compared the water chemistry and fish communities of the cave and several nearby surface streams. Our study revealed high concentrations of H2S in the cave and its outflow (El Azufre). The concentrations of H2S reach more than 300 µM inside the cave, which is immediately deadly for most fishes. In both sulfidic habitats, the diversity of fishes was heavily reduced, and Poecilia mexicana was the dominant species indicating that the presence of H2S has an all-or-none effect, permitting only few species to survive in sulfidic habitats. Compared to habitats without H2S, P. mexicana from the cave and the outflow have a significantly lower body condition. Although there are microhabitats with varying concentrations of H2S within the cave, we could not find a higher fish density in parts of the cave with lower concentrations H2S. We discuss that P. mexicana is one of the few extremophile vertebrates. Our study supports the idea that extreme habitats lead to impoverished species diversity.

1 Published as: M. Tobler, I. Schlupp, K. U. Heubel, R. Riesch, F. J. García de León, O. Giere & M. Plath (2006): Life on the edge: Hydrogen sulfide and the fish communities of a Mexican cave and surrounding waters. Extremophiles 10: 577-585. 22 Chapter 2: Introduction and synthesis

Introduction Townsend et al. (2003) defined an extreme environmental condition as one that requires, of any organism tolerating it, costly adaptations absent in most related species. It is often claimed that habitats with extreme environmental parameters have a reduced species richness (Begon et al., 1996; Townsend et al., 2003). For example, plant diversity is reduced on plots with low pH in the Alaskan tundra (Gough et al., 2000), and deep-sea hydrothermal vents possess a low species diversity due to extremes in temperature, hypoxia, sulfide, and heavy metals (McMullin et al., 2000; Price, 2002; Tsurumi, 2003). However, not all habitats with reduced species richness are harsh. For example, low species richness in an apparently extreme habitat may also be explained by its limited size (MacArthur and Wilson, 1967), low productivity, or low spatial heterogeneity (Begon et al., 1996). Hence, it remains an open question if harsh environments are in fact low in species diversity because of the abiotic stressors themselves (Townsend et al., 2003). Physiochemical stressors like toxic chemicals are thought to directly influence the composition of ecological communities from the zoogeographical to local scale

(Begon et al., 1996; Matthews, 1998; Townsend et al., 2003). Hydrogen sulfide (H2S) can clearly be considered an extreme environmental factor for all animal life, because it is acutely toxic (Evans, 1967; Theede, 1973; Smith et al., 1977; Bagarinao and Vetter, 1989; Grieshaber and Völkel, 1998). Most known animals from sulfidic habitats are invertebrates, which cope with naturally occurring H2S by (1) avoiding microhabitats with high sulfide concentrations, (2) switching to anaerobic metabolism, (3) excluding sulfide from sensitive tissues, or (4) oxidizing sulfide to more benign forms (see Grieshaber and Völkel, 1998; McMullin et al., 2000 for reviews). Previous research has focused on the impact of H2S on species assemblages in deep-sea hydrothermal vents (Peek et al., 1998; Sarrazin and Juniper, 1999; Van

Dover, 2000), or on H2S as a chemical pollutant for invertebrates (Oseid and Smith Jr., 1974) and fishes (Colby and Smith Jr., 1967; Adelman and Smith Jr., 1970; Smith Jr. et al., 1976; Abel et al., 1987; Bagarinao and Vetter, 1989; Bagarinao and Vetter, 1990; Geiger et al., 2000). Very little is known about the effects of naturally occurring

H2S and its influence on the composition of freshwater species communities (Dare et al., 2001). Chapter 2: Hydrogen sulfide 23

Natural H2S is present in a cave, the Cueva del Azufre, and its outflow in tropical Mexico (Gordon and Rosen, 1962). The dominant species in the cave is a cavernicolous form of a live-bearing fish, the Atlantic molly, Poecilia mexicana (the cave molly, Parzefall, 2001). Although P. mexicana from the Cueva del Azufre and adjacent waters are used as a model system to study the evolution of cave adaptations (Parzefall, 1969, 1993, 2001; Plath et al., 2003a; Plath et al., 2003b; Plath et al., 2004; Plath et al., 2005; Plath et al., 2006), so far little is known about the environmental characteristics of their habitat. Here we used the Cueva del Azufre and adjacent waters to study the effects of H2S on the diversity of fish communities. The reduction of species diversity in caves is usually attributed to the lack of light and the associated lack of photoautotrophy (Barr and Holsinger, 1985; Hüppop, 2000). Many cave ecosystems completely rely on organic matter washed in from the surface (Poulson and White, 1969; Parzefall, 1993; Poulson and Lavoie, 2000), and only specialized cave dwellers are thought to be able to cope with these conditions. The Cueva del Azufre is thought to be different from most other caves in that the food web appears to be energy rich even compared to surface habitats and to rely mainly on in situ chemoautotrophic bacterial primary production and the input of guano by bats (Langecker et al., 1996). Within the cave, mollies were reported to feed on bacterial detritus and bat guano (Langecker et al., 1996), and mosquito larvae were found in their guts (Tobler personal observation). The major objective of our study was an analysis of the abiotic environmental conditions in different habitats in and around the Cueva del Azufre to estimate the effects of these parameters on the composition of fish communities. To account for possible interactive effects of darkness and the presence of H2S, we also examined a sulfidic creek outside the cave, where the absence of light can be ruled out to have an influence on the fish communities, and contrasted it with nearby non-sulfidic, but otherwise similar habitats. For a comparison at a between-species level, we examined the fish communities of the different habitat types. For the only species that occurs in all habitats examined, the Atlantic molly (Poecilia mexicana), data on body conditions across habitat types as well as population densities in two different cave chambers were determined, which allowed us to estimate how different environmental conditions might act as limiting factors on a within-species level.

24 Chapter 2: Introduction and synthesis

Methods Study sites All study sites are located near the village of Tapijulapa in the state of Tabasco, south Mexico. All creeks studied eventually drain into the Río Oxolotan, which is part of the Río Grijalva drainage system. We included several habitats in the immediate vicinity (within a perimeter of about 2 km) of the Cueva del Azufre (17°26.5'N,

92°46.5'W), where H2S and darkness occur in varying combinations and/or intensities (Figure 1.1; Table 2.1): 1) The cave itself is sulfidic and the front chambers obtain some dim light, whereas the rearmost cave chambers are completely dark. Nomenclature of the cave chambers follows Gordon and Rosen (1962, Figure 1.2). Chambers III, IV, V, X and XIII were sampled. 2) The creek flowing out of the cave (El Azufre) is sulfidic but is exposed to sunlight. Other surface waters lack any sulfidic components and have also normal exposure to light: 3-5) Three creeks of similar size and structure to the El Azufre were used for a direct comparison: Arroyo Cristal, Arroyo Bonita and Arroyo Tres (the latter two were only sampled in January 2006). 6) A small freshwater tributary of the El Azufre (Clear Creek) and 7) the Río Oxolotan were also included in our analysis.

Water chemistry Water parameters were measured in September 2002, August 2004 and January 2006 using a Hydrolab Multiprobe 4A, which measures several variables at the same time. Measurements and calibration of probes were conducted according to the manufacturers recommendations. Specific conductance was measured in mS/cm, dissolved oxygen in mg/l and % saturation, temperature in °C and turbidity using a shuttered turbidity probe in nephelometric turbidity units [NTU]. Measurements of temperature and light over 24 h were conducted using Onset Stow Away loggers. Data presented are means of several measurements (2 to 4), which were collected at several sites within the mentioned habitats (Table 2.2).

To determine H2S contents, represented by the total concentration of sulfide, samples were collected in August 2004 and January 2006 on site. One milliliter of water was injected into a vial containing 2 ml of zinc acetate (0.12 M with 0.5 ml NaOH 1.5 M) using a syringe. The vials were stored at room temperature and photometric measurements were conducted according to Cline (1969). The data presented in Table 2.2 are means of 1 to 4 measurements. Chapter 2: Hydrogen sulfide 25

Comparison of fish communities In order to compare the fish communities, fish were collected in August 2004 and January 2006 at one to three sites within each habitat, and data from each site were pooled (Figure 1.1; Table 2.3). Within the cave, chambers III, IV, V, X and XIII, which include all major microhabitat types, were sampled. In the El Azufre, data were collected right outside the cave exit, around the mouth of the Clear Creek and 250 m downstream, which includes several riffle and pool areas. In the Arroyo Cristal and the Arroyo Bonita, fishes were collected in a 150 m long stretch including riffles and pools. In the Arroyo Tres, the sampling area included a 100 m stretch of riffles and pools. In the Clear Creek, data were collected in a 200 m long stretch from the mouth upstream, which also included a sequence of several riffles and pools. In the Río Oxolotan fishes were caught downstream of the mouth of the Arroyo Cristal as well as at two boat ramps downstream of the mouth of the El Azufre and in the village of Tapijulapa. Because habitat structures differed strongly between sampling sites, various sampling methods were employed. In the cave, where the water is very shallow, fishes were caught with dip nets (13x14 cm, 1 mm mesh-width). In the El Azufre, the Arroyo Cristal, the Arroyo Bonita and the Arroyo Tres, fishes were caught with a seine (4 m long, 4 mm mesh-width) and a cast net (2.5 m in diameter, 6 mm mesh- width). In the Clear Creek, both dip nets and the seine were employed. In the Río Oxolotan, the seine, the cast net and dip nets were used, and catches of local fishermen were qualitatively surveyed in January 2006. Fishes were counted and photographed using a Nikon D70 digital camera. Species identity was determined ad hoc or using the photographs following Miller (2005b). After identification, the fishes were released at the collection site. Abundance of fish species was classified in the following categories: rare: 1-5 individuals; common: 5-50 individuals; abundant: >50 individuals. Nomenclature is in accordance with Miller (2005b). For the comparison of the species diversity of each habitat type, the Shannon-Wiener diversity index (H) and the evenness index (J) were calculated with the combined data from 2004 and 2006 (Begon et al., 1996).

Population densities in the cave We compared population densities in two sub-populations of the cave molly from two cave chambers (X and XIII). Both chambers are essentially dark. This comparison 26 Chapter 2: Introduction and synthesis

was especially interesting because the two chambers differ in the presence of H2S and oxygen concentrations (Table 2.2). A small cascade (1.5 m high) separates both chambers, so that migration is likely mostly unidirectional from chambers XIII to X. Population sizes were estimated by using mark-recapture analyses and by calculating the Lincoln index (Mühlenberg, 1993). Fish were caught with dip nets for 45 minutes by two persons. We marked the fish by clipping their dorsal fin. Observations from laboratory-reared mollies have shown that this procedure does not harm the fish, and the removed fin tissue usually regenerates within approximately one week. No dead fish were observed in the cave after releasing the handled fish. After 24 hours, sampling was repeated. We counted the total number of mollies caught and the number of marked (recaptured) individuals. Densities were calculated by dividing the mean values of the estimated population sizes by the area of the respective cave chamber. The area of chamber XIII was estimated as 10 m2, that of chamber X as 85 m2.

Condition factor of P. mexicana Another factor we considered in this study was the general body condition of the only fish species present in all habitat types, P. mexicana. We determined the body condition factor for male and female P. mexicana living in the different habitats. The condition factor (1000 * mass [g] / standard length [mm]3) was determined in P. mexicana larger than 20 mm from cave chambers III, IV, V, XI (N=265) and XIII (N=144), from the El Azufre (N=100) and the Arroyo Cristal (N=100). Standard lengths were measured to the closest millimeter using scale paper. Mass was measured to the closest 0.1 gram using a Pesola scale. Based on the results of the water parameter analyses, P. mexicana from cave chamber XIII were treated as a separate population compared to the rest of the cave. This distinction is further justified due to the barrier between chamber XIII and the other parts of the cave (see above). Data were analyzed using ‘population’ as between factor and ‘sex’ as within factor in a two-way ANOVA. Since the interaction term was not significant (F3,

601=0.19, P=0.90), only the main effects were analyzed. For post hoc contrasts, Fisher’s protected least significant difference (PLSD) was employed.

Chapter 2: Hydrogen sulfide 27

Results Water chemistry

We found variation in H2S and oxygen concentrations within the apparently homogenous cave, whereby H2S and oxygen concentrations differed even within very short distances. The typical inverse relationship of oxygen and H2S was found (Table

2.2). Furthermore, H2S concentrations apparently vary over time. The values measured in 2006 were generally lower than those from 2004, possibly due to heavy rainfalls before and during sampling. In the innermost cave chamber (XIII), the least extreme conditions within the cave were found. The water entering this chamber through cracks in the wall has relatively high amounts of oxygen and very low sulfide concentrations. By contrast, a small springhead only a few meters away in chamber X has almost no oxygen, but is very rich in H2S with concentrations reaching 300 µM (Table 2.2). Downstream areas of the cave (chambers III, V) were richer in oxygen and had less H2S. Turbidity within the cave appears to coincide with amounts of colloidal sulfur in the water, which is produced by the oxidation of H2S. While the springs are clear and rich in H2S, parts of the cave with mixing appear milky. Specific conductivity is uniform throughout the cave and pH is well buffered, probably due to the limestone of the cave, except for the actual springs where it is lower than 7, which may reflect an interaction with the H2S (Table 2.2). Typical for the cave habitat is a nearly constant water temperature of 28.3°C. Continuous temperature measurements over 24 hours in February of 1998 and August of 2004 in chamber XIII revealed no variability in temperature. Continuous measurements over 24 hours (1998) of the light intensities in chambers X and XIII read 0 Lux and confirmed complete darkness for these parts of the cave.

In the El Azufre, H2S concentrations are lower than in the cave and decrease with increasing distance from the cave exit. Although clear and most likely without

H2S, the Clear Creek also has low oxygen and lowered specific conductance (Table 2.2). This contrasts with high values for oxygen in the two other surface habitats, the Arroyo Cristal and the Río Oxolotan (Table 2.2).

Comparison of fish communities Extensive sampling in different chambers inside the Cueva del Azufre revealed only one species of fish: the cavernicolous form of P. mexicana (Table 2.3). Juveniles and adults were caught. An estimation of the species richness revealed a pattern of low 28 Chapter 2: Introduction and synthesis

numbers of species in habitats containing H2S and higher species richness in habitats without sulfur components (Table 2.3). A direct comparison of the El Azufre and the Arroyo Cristal, two streams of similar size and structure, revealed a considerably higher number of species in the non-sulfurous habitat (Table 2.3). The low value of the evenness index in the El Azufre compared to the benign surface habitats reflects the over-dominance of one species: P. mexicana. Only one further fish species, the predatory cichlid 'Cichlasoma' salvini occurs, but only in small numbers. The Río Oxolotan harbors a fish community comparable in species composition and diversity to that of the Arroyo Cristal. In the Clear Creek, a small stream that is directly connected to the El Azufre, P. mexicana was not dominant, and Heterandria bimaculata occurred at high abundance. Xiphophorus hellerii was recorded in small numbers near the mouth of the Clear Creek into the El Azufre, but always in clear water.

Population densities in the cave Population sizes were repeatedly analyzed in cave chamber XIII. The estimated population sizes were highly consistent between years (Table 2.4). Population densities were similar between the two cave chambers examined, with 12.47±0.35 (mean±SD) individuals per m2 in chamber XIII and 19.58±11.12 individuals per m2 in cave chamber X.

Condition factors of P. mexicana A comparison of condition factors between P. mexicana from different habitats revealed pronounced differences between populations (ANOVA: F3, 604=64.54, P<0.0001; Figure 2.1). Surface fish from the Arroyo Cristal showed higher condition factors than cave fish and fish from the El Azufre, which in turn had higher condition factors than fish from the cave. Furthermore, P. mexicana from chamber XIII showed a slightly, but significantly worse body condition compared to fish from the other cave chambers. Sex had no significant effect on the condition factors (F1, 604=1.58, P=0.21).

Discussion

Our study revealed high concentrations of H2S in the Cueva del Azufre and its outflow. In both habitats, the diversity of fishes was heavily reduced and Poecilia Chapter 2: Hydrogen sulfide 29

mexicana was the dominating species. Compared to habitats without H2S, P. mexicana from the cave and the El Azufre have a significantly lower body condition.

Although there are microhabitats with different concentrations of H2S within the cave, we could not find a higher fish density in habitats with lower concentrations.

Environmental conditions

Except for the presence of H2S in the Cueva del Azufre and the El Azufre, the water parameters reported here are in agreement with those reported for the wider area (Mayland, 1984; Stawikowski and Werner, 1998). Outside the cave, differences between habitats are best explained by the fundamental differences between rivers and small creeks. Within the cave, there was considerable variation in H2S and oxygen concentrations within short distances. This patchiness was previously unrecognized. Sulfide concentrations seem to vary to some extent over time and are likely dependent on the discharge of the springs relative to the precipitation in the area. Thus, the toxicity of the water might peak during the dry season in February to April. Further studies are needed to estimate the degree and relevance of temporal variation in sulfide concentrations in the cave. The presence of H2S indicates a chemically reduced environment. Potentially, other toxic substances, such as elevated concentrations of dissolved metals, coincide with H2S, but this remains to be studied. Two competing mechanisms apparently influence the oxygen content of the water. Oxygen concentrations clearly rise below areas of turbulence (e.g., chamber IX), but become low again downstream (chamber V). While mixing with air leads to increased oxygen, bacterial metabolism likely leads to decreased oxygen content. This can explain the relatively low oxygen values towards the exit and outside the cave. The close match of the readings from 2002, 2004 and 2006 indicates very high constancy of the abiotic conditions in the cave.

H2S and fish communities The fish communities documented here are typical for Central American fish communities in that cichlids and poeciliids were the dominant species (Miller, 1976; Miller, 2005b). A potential criticism of our study could be that the catch per unit effort was not identical across habitats and that we may have underestimated the fish diversity of some habitats, especially that of the Río Oxolotan, where water levels were high during our visits. However, catching efforts were lower in non-sulfidic 30 Chapter 2: Introduction and synthesis surface habitats so that the reported high species diversity is a rather conservative estimate of the actual diversity. In contrast, catching efforts were very high within the cave and the El Azufre during several expeditions. Therefore, our data are suitable for a comparison of the fish communities between the different habitats studied. Fish communities were most diverse in benign habitats, but were impoverished in both sulfidic habitats. In comparison to the El Azufre, the Arroyo Cristal is of similar size and structure, but harbors a more diverse fish community.

Our data therefore suggest that the presence of H2S strongly influences the composition of the fish communities, which leads to pronounced differences even within short distances. We did not find any evidence for fine-scale changes of fish community compositions along the gradient of H2S concentration in the Cueva del

Azufre and the El Azufre, but the presence of H2S rather seems to have an all-or-none effect, permitting only few species to survive in these habitats. The same pattern was found in other sulfidic freshwater habitats in southern Mexico, such as the Baños del Azufre (Tobler et al., 2008c) and in metazoan communities in deep sea habitats containing H2S (McMullin et al., 2000; Price, 2002; Tsurumi, 2003). This supports the idea that extreme conditions directly translate into low species diversity (Townsend et al., 2003). The small-scale distribution of oxygen-rich areas in the cave makes it likely that cave mollies can choose more favorable microhabitats, avoiding areas with extreme conditions. However, our estimations of fish densities in cave chamber X

(high concentrations of H2S) and XIII (low concentrations of H2S) did not show pronounced differences. Possibly, food in chamber XIII is especially scarce, since primary production relies on H2S, which is low in this chamber. Food shortage in chamber XIII is reflected by the eminently low condition factor of its inhabitants. Fish densities appear to be highest downstream towards the cave exit coinciding with intermediate H2S and oxygen values, but a systematic survey there remains to be done. Given that the major differences of fish community composition are evident between the Clear Creek and the El Azufre, not between the cave and the El Azufre, light seems to play a subordinate or no role on the between-species level. On a within- species level, the presence or absence of light appears to have a strong effect on the distribution of the two phenotypically distinct forms of P. mexicana: the cave form (Parzefall, 2001) and the surface form living in the El Azufre. It remains to be studied Chapter 2: Hydrogen sulfide 31 if – and to what extent – the phenotypic differences between the surface form and the cave form have a genetic basis, and how environmental effects, namely the presence of light and the availability of food, influence the ontogeny of this species.

Adaptations to H2S

The most plausible explanation for how the presence of H2S causes the observed reduction of species diversity is its toxic nature. High concentrations of H2S are acutely toxic for most eukaryotic organisms, because H2S binds at the iron of the heme to replace O2 and also at respiratory chain proteins (e.g., the cytochrome c oxidase), where it prohibits electron transport in aerobic respiration (Evans, 1967; Theede, 1973; Nicholls, 1975; Smith et al., 1977; Carrico et al., 1978; National Research Council, 1979; Bagarinao, 1992; Grieshaber and Völkel, 1998). Lower concentrations of H2S are chronically toxic (National Research Council, 1979). H2S is known to potentially play a role in natural fish kills (Bagarinao and Lantin-Olaguer, 1999; Luther et al., 2004) and sulfide toxicity is known to aggravate mortality due to hypoxia and low pH (Bagarinao and Lantin-Olaguer, 1999). Sulfide resistance in fishes varies with habitat and lifestyle, and species that encounter naturally occurring sulfide usually show increased resistance [e.g. Megalopus atlanticus tolerate concentrations up to 230 µM (Geiger et al., 2000); Hoplosternum littorale up to 87

µM (Affonso and Rantin, 2005)]. In fishes from shallow-water marine habitats, LD50 concentrations of H2S range from 1 µM in open-coast inhabitants up to 700 µM in species inhabiting tidal-marshes with naturally occurring H2S (Bagarinao and Vetter, 1989).

Because of the typical inverse relationship between oxygen and H2S concentrations, the reduction of species diversity may also be influenced by hypoxia. However, the Clear Creek also has very low oxygen concentrations, but harbors a completely different fish community compared to the H2S-containing habitats. Although migration between Clear Creek and El Azufre is possible without having to overcome physical barriers or strong current, the composition of fish communities changes dramatically within about 3 meters. Previous experiments have shown that P. mexicana from non-sulfidic habitats lack any sulfide tolerance and that the short-term survival of cave mollies in water containing hydrogen sulfide depends on the possibility to perform aquatic surface respiration (ASR) and sufficient food availability (Plath et al., 2007c). During ASR, 32 Chapter 2: Introduction and synthesis cave mollies exploit the air water interface, which is relatively oxygen-rich compared to the rest of the water column (Kramer and Mehegan, 1981; Kramer, 1987). The fact that energy availability is crucial for survival (individuals supplemented with a high energy food source have a higher chance of surviving in toxic water; Plath et al. 2007b) highlights the biological significance of the low body condition of cave mollies and may point to costly physiological adaptations to detoxify H2S. In fishes, detoxification of sulfide is known through its oxidation to thiosulfate in liver mitochondria (Bagarinao and Vetter, 1990) and its binding to ferrous and ferric hemoglobin and other blood compounds (Torrans and Clemens, 1982; Bagarinao and Vetter, 1989). So far it is not clear how cave mollies cope with the long-term toxic effects of H2S.

Our results suggest that the presence of H2S reduces the diversity of fish communities. Furthermore, they contradict the hypothesis that food (sulfur bacteria and bat guano) is abundant in this cave ecosystem (Langecker et al., 1996), or at least they suggest that the food base is of poor quality for the fish. Cave mollies have reduced a number of energetically costly behaviors like aggression, shoaling (Parzefall, 1993) and male sexual activity (Plath et al., 2003a). Reduction of shoaling and aggression were previously interpreted as adaptations to the absence of light (Parzefall, 1993, 2001). The results of this study, however, suggest that the presence of H2S and restricted food availability are more likely explanations. Future research will have to answer which of the adaptations reported for this species are really adaptations to cave life per se (darkness), which are driven by the extreme water chemistry or low food availability, and how exactly these components interact. Chapter 2: Hydrogen sulfide 33

Figures and tables chapter 2

Figure 2.1. Condition factors of P. mexicana from different habitats. Clear water surface habitat (Arroyo Cristal), sulfidic surface creek (El Azufre), cave chambers with sulfidic components (cave with sulfur), and rearmost cave chamber XIII without sulfide (cave without sulfur). A post hoc analysis revealed that condition factors differed significantly between all populations (Fisher’s protected least significant difference, P<0.01 in all cases).

Table 2.1. Brief comparison of the sampling sites. Cave El Azufre Arroyo Cristal Clear Creek Rio Oxolotan Light absent present present present present H2S present present absent absent absent Width [m] 1-7 2-5 3-5 1-2 40 Mean depth [m] 0.5 1.2 1 0.3 ? Current low - moderate moderate moderate low high Predominant ground bedrock, silty gravel cobble, gravel gravel ? sediments Surrounding secondary tropical rainforest, pastures, agriculture Sampling effort high high moderate low low

34 Chapter 2: Introduction and synthesis

Table 2.2. Water chemistry of several chambers of the Cueva del Azufre from 2002 and 2004. Nomenclature of cave chambers follows Gordon & Rosen (1962). For sites outside the cave refer to main text.

1 M)

pH Date rbidity(NTU) DO(mg/L) Sulfide( Tu DOSaturation (%) Watertemperature (°C)

Specific conductivity (mS/cm) Site Cave (III hole) Aug 2004 28 7.0 3.95 95.4 0.84 13.2 Cave (III stream) Aug 2004 28 7.0 3.95 81.4 0.85 13.6 Cave (IV hole) Aug 2004 28 7.0 3.99 67.9 0.72 11.4 Cave (IV stream) Aug 2004 28 7.0 3.95 76.8 0.91 14.2 Cave (V) Aug 2004 28 7.0 3.96 85.0 1.35 21.3 50 Jan 2006 28 7.1 4.23 45.3 1.07 14.1 162 Cave (IX below cascades) Aug 2004 28 7.1 3.94 82.1 2.04 32.1 45 Jan 2006 28 7.2 4.19 39.0 1.79 23.3 404 Cave (X clear area) Sept 2002 28 6.8 4.25 6.9 1.23 15.6 Aug 2004 28 6.8 3.95 43.9 1.44 23.1 199 Jan 2006 28 7.0 4.20 9.0 2.45 32.0 Cave (X spring I) Aug 2004 28 6.7 4.18 10.1 0.32 4.8 234 1554 Cave (X spring II) Sept 2002 28 6.7 4.20 18.4 0.29 3.8 Aug 2004 28 6.7 3.99 9.6 0.35 5.7 3093 Jan 2006 28 6.7 4.32 8.0 0.90 3.3 1364 Cave (X turbid area) Sept 2002 28 7.1 4.22 70.0 2.06 26.8 Aug 2004 28 7.1 3.93 74.6 1.95 30.6 137 Jan 2006 28 6.9 4.25 39.0 1.65 21.6 254 Cave (XIII) Sept 2002 28 7.2 4.06 37.3 3.10 41.0 Aug 2004 28 7.2 3.94 33.0 4.01 64.1 14 Jan 2006 28 7.6 4.23 12.5 3.11 40.4 14 El Azufre (cave exit) Sept 2002 28 7.1 4.18 67.0 1.10 14.7 Aug 2004 28 7.1 3.94 68.4 1.18 18.9 37 Jan 2006 28 7.1 4.22 104.3 1.05 13.8 324 El Azufre (bridge2) Aug 2004 28 7.3 3.91 50.5 1.82 28.9 11 Jan 2006 28 7.2 4.09 45.3 1.35 17.7 224

Clear Creek Sept 2002 27 7.3 2.17 55.0 2.7 34.1 Aug 2004 25 7.5 2.16 13.2 4.06 60.0 5 Arroyo Cristal Aug 2004 28 7.8 0.38 28.6 5.75 82.8 Jan 2006 23 8.2 0.38 0.0 4.34 50.8 02 Arroyo Bonita Jan 2006 23 8.3 0.34 0.0 4.7 54.1 04 Arroyo Tres Jan 2006 22 7.6 0.29 80.5 3.1 36.3 04 Rio Oxolotan Sept 2002 23 8.3 0.27 87.6 8.3 97.9 Aug 2004 26 8.3 0.39 155.6 6.25 95.0 5 1 2 3 Water contains 6,4ml/l 02 at 20°C at sea-level, Average of 4 measurements, Average of 3 measurements, 4Average of 2 measurements, 5Samples were taken, but produced no reliable readings Chapter 2: Hydrogen sulfide 35

Table 2.3. Summary of the fish communities in the different habitats sampled in and around the Cueva del Azufre in August 2004 and January 2006. The total number of species, the Shannon-Wiener diversity index (H) and the evenness index (J) is given for each habitat. Additionally, the abundance for each species within habitats is presented for 2004/ 2006. a = abundant, c = common, r = rare.

1 Cave AzufreEl ClearCreek Arroyo Cristal Arroyo Bonita Arroyo Tres Río Oxolotan Number of species 1 2 3 11 9 6 14 H 0.00 0.05 0.82 1.79 1.70 1.59 1.92 J - 0.06 0.77 0.81 0.76 0.80 0.78 Characidae Astyanax aeneus (Günther 1860) a / a a a a / a Brycon guatemalensis Regan 1908 Ariidae Potamarius nelsoni (Evermann & Goldsborough 1902) 2 Pimelodidae Rhamdia guatemalensis (Günther 1864) r Batrachoididae Batrachoides goldmani Evermann & Goldsborough 1902 r / - Atherinopsidae Atherinella alvarezi (Díaz Pardo 1972) - / r c c / c Poeciliidae Heterophallus milleri Radda 1987 c a / a Heterandria bimaculata (Heckel 1848) a / c c / c c r / r Poecilia mexicana Steindachner 1863 a /a a / a c / c a / c c c c / c Priapella chamulae Schartl, Meyer & Wilde 2006 c / a c Xiphophorus hellerii Heckel 1848 r / c r / c r c Centropomidae Centropomus undecimalis (Bloch 1792) 2 Cichlidae 'Cichlasoma' salvini (Günther 1862) r / - - / r r / - Oreochromis cf. aureus (Steindachner 1864) 2,3 Paraneetroplus gibbiceps (Steindachner 1864) r / r Thorichthys helleri (Steindachner 1864) c / c r c / c Vieja bifasciata (Steindachner 1864) c / c r r c / r Vieja intermedia (Günther 1862) - / r Eleotridae Gobiomorus dormitor Lacépède 1800 r 1Only sampled in 2006; 2Recorded in catches of local fishermen 2006; 3Introduced species

Table 2.4. Cave molly population size estimates (mean ± SD) from two cave chambers (X and XIII). Population Year Marked (day 1) Caught (day 2) Recaptured Estimated population size Cave (XIII) 1998 53 54 28 102 ± 13 adults 1999 31 18 4 139 ± 62 adults 2002 39 34 16 309 ± 91 adults + juveniles 2004 83 55 12 380 ± 97 adults + juveniles Cave (X) 1998 52 96 3 1664 ± 954 adults

36 Chapter 3: A new cave population

Chapter 3

A NEW AND MORPHOLOGICALLY DISTINCT POPULATION OF

CAVERNICOLOUS POECILIA MEXICANA (POECILIIDAE: 2 TELEOSTEI)

Michael Tobler, Rüdiger Riesch, Francisco J. García de León, Ingo Schlupp and Martin Plath

Abstract The cave molly (Poecilia mexicana) from the Cueva del Azufre, a sulfur cave in Tabasco, Mexico, ranks among the best-studied cave fishes worldwide, despite being known from a single population only. Here we describe a newly discovered second population of cave-dwelling P. mexicana from a nearby, but mostly non-sulfidic cave (Luna Azufre). Despite apparent similarities between the two populations (such as reduced eye diameter and reduced pigmentation), a geometric morphometric analysis revealed pronounced morphological differentiation between the two cave forms.

2 Published as: M. Tobler, R. Riesch, F. J. García de León, I. Schlupp & M. Plath (2008): A new and morphologically distinct population of cavernicolous Poecilia mexicana (Poeciliidae: Teleostei). Environmental Biology of Fishes 82: 101-108.

Chapter 3: A new cave population 37

Introduction More than 100 species of teleost fishes permanently live in subterranean habitats (Proudlove, 2006). Although the understanding of the diversity of hypogean fishes has increased steadily, the ecology and evolution of most species are still poorly examined, and detailed information is restricted to just a few systems (e.g., Poulson, 1963; Strecker, 2003; Wilkens and Strecker, 2003; Jeffery, 2005). One of the best studied cave fishes is the cave molly (Gordon and Rosen, 1962), a distinct population of the Atlantic molly, Poecilia mexicana. The Atlantic molly is widespread in freshwater surface habitats along the Atlantic versant of Central America (Miller, 2005b). Compared to conspecifics from surface habitats, fish from the cave population have reduced albeit functional eyes and reduced pigmentation (Gordon and Rosen, 1962; Walters and Walters, 1965; Peters et al., 1973; Körner et al., 2006). Furthermore, cave mollies have reduced a set of behavioral traits, such as shoaling, aggression (Parzefall, 1993), and the intensity of male sexual activity (Plath et al., 2003a). On the other hand, cave mollies show a series of traits, most of which seem to improve communication and orientation in darkness and are absent in mollies from surface habitats. Additionally, female cave mollies exhibit a distinct genital pad that is absent in epigean fish. Supposedly, this pad secrets chemical cues that play a role during reproduction (Walters and Walters, 1965; Zeiske, 1968; Parzefall, 1970, 1973) and are perceived by males during pre-mating behavior (nipping) utilizing the increased number of taste buds on their heads (Parzefall, 1970). Furthermore, cave mollies exhibit a hyper-developed cephalic lateral line system (Walters and Walters, 1965; Parzefall, 2001), which has been documented to mediate spatial orientation in other cave fishes (Abdel-Latif et al., 1990; Burt de Perera, 2004). On a behavioral level, cave mollies have evolved the ability to assess mate quality (such as the size or the nutritional state of mates) in darkness, whereas surface fish rely mainly on visual cues and are unable to assess mate quality in the absence of light (Plath et al., 2004; Plath et al., 2005). The cave molly has only been reported from one cave, the Cueva del Azufre, in the state of Tabasco, Mexico (Gordon and Rosen, 1962; Parzefall, 2001). Its habitat is characterized not only by complete darkness (although some of the front cave chambers receive some light through openings in the ceiling), but also by high concentrations of hydrogen sulfide (H2S) and hypoxic conditions (Gordon and Rosen, 38 Chapter 3: A new cave population

1962; Tobler et al., 2006). Hydrogen sulfide is a potent respiratory toxicant and is lethal for most metazoans even in micro-molar amounts (Bagarinao, 1992; Grieshaber and Völkel, 1998). Parzefall (2001) notes that the cave molly may have an increased

H2S tolerance, but generally the effects of H2S on the evolutionary ecology of the cave molly are only poorly understood. A behavioral adaptation (aquatic surface respiration), by which fish exploit the oxygen-rich air-water-interface, seems to mediate the short-term survival of cave mollies in the toxic environment (Plath et al., 2007c). Recently, Pisarowicz (2005) reported the discovery of a new cave, the Luna Azufre, in the vicinity of the Cueva del Azufre that is also inhabited by a molly population. Mollies from both caves share characteristics such as reduced eyes and pigmentation (Figure 3.1). Here we provide the first comparison of the newly discovered cave-dwelling population with fish from the Cueva del Azufre using geometric morphometrics. Furthermore, we present data from the first environmental survey comparing the two different habitats.

Materials and Methods Study sites The caves studied are located near the village of Tapijulapa in Tabasco, Mexico. Cave mollies (males N=6; females N=26) were collected in the newly discovered Luna Azufre just south of the Entrada Marabunda in January 2006 [Figure 1.3; Pisarowicz (2005)]. For a comparison, cave mollies (males N=10; females N=10) also were collected in chamber V of the Cueva del Azufre during the same expedition [see Gordon and Rosen (1962) for a map of the Cueva del Azufre]. Fish were sacrificed by overdose of MS222 and preserved on site in 10% formalin for later investigation.

Geometric morphometric analysis Each specimen was photographed on a millimeter grid using a Nikon D70 digital camera. Because of distinct sexual dimorphism (male poeciliids have a modified anal fin used for sperm transfer) males and females were analyzed separately. Thirteen landmarks were digitized on each specimen using tpsDig (Rohlf, 2004): the tip of the upper jaw (1); the anterior (2) and posterior (3) margin of the eye; the anterior (4) and posterior (5) junction of the dorsal fin with the dorsal midline; the junction of the caudal fin with the dorsal (6) and ventral (7) midline; the anterior (8) and posterior (9)

Chapter 3: A new cave population 39 junction of the anal fin with the ventral midline; the anterior junction of the pelvic fins and the ventral midline (10); the bottom of the head where the operculum breaks away from the body outline (11); the upper end of the operculum where it connects to the body (12); and the dorsal junction of the left pectoral fin with the body (13). Data were translated to NTS format using tpsUtil (Rohlf, 2006). Subsequent analyses were performed using the thin-plate spline software tpsRegr (Rohlf, 2005). Landmark coordinates were aligned using least-squares superimposition to remove effects of translation, rotation, and scale, and a consensus configuration was calculated. Cartesian transformation grids were generated illustrating the relative shape differences among populations. Furthermore, superimposed coordinates were subjected to a principal components analysis (PCA). Ten principal components, which accounted for 91.7% and 97.4% of the total shape variation in females and males, respectively, were included as dependent variables in a MANCOVA, in which ‘population’ was used as independent variable and ‘centroid size’ as a covariate to control for the effect of body size. The assumptions of normal distribution and homogeneities of variances and covariances were met for these analyses. Furthermore, discriminant function analysis was used to test whether individuals were correctly assigned to the population of origin.

Color polymorphism Besides differences in shape we also investigated the distinct color polymorphism known from cave mollies in the Cueva del Azufre. Whereas the majority of the fish are pale, some show yellow coloration (Gordon and Rosen, 1962; Parzefall, 2001). We recorded the presence of both color morphs in the Luna Azufre and compared their relative frequencies to the Cueva del Azufre population using a Χ2-test. Data for the Cueva del Azufre were collected during a survey in cave chambers III, IV, V, XI and XIII in August 2004.

Length/weight relationship Cave mollies from the Cueva del Azufre also are reported to have an eminently lower body condition factor (i.e. a lower weight per given standard length) than populations from surface habitats (Plath et al., 2005; Tobler et al., 2006), and their short-term survival is critically dependent on energy availability (Plath et al., 2007c). We compared the length/weight relationship between the two cave population using an 40 Chapter 3: A new cave population

ANCOVA with ‘ln(weight)’ as dependent variable, ‘population’ and ‘sex’ as independent variables, and ‘ln(standard length)’ as covariate. The assumptions of normal distribution and homogeneity of variances were met.

Environmental conditions Water parameters were measured using a Hydrolab Multisonde 4A (Hach Environmental). Measurements and calibration of probes were conducted according to the manufacturer’s recommendations. Specific conductance was measured in mS/cm, dissolved oxygen in mg/l and % saturation, and turbidity using a shuttered turbidity probe in nephelometric turbidity units [NTU].

For the determination of H2S concentrations in the Luna Azufre, 1 ml of water was injected into a vial containing 1 ml of zinc acetate (0.12 M with 0.5 ml NaOH 1.5

M in a N2-atmosphere) using a syringe. The vials were stored at room temperature and photometric measurements were conducted later in the laboratory according to

Cline (1969). To further check for the presence of H2S in the Luna Azufre, H2S was also measured in the air using an Industrial Scientific Gas Badge Pro®.

Results Geometric morphometric analyses Both allometry (centroid size) and population identity had a significant effect on the shape variation among females of the two populations (Table 3.1A). The discriminant function analysis correctly assigned 100% of the specimens to the population of origin (see Table 3.2 for equality test of means). The consensus configuration and the Cartesian transformation grids for females are presented in Figure 3.2A-C. By inspection of these grids, pronounced differences in body shape between females from the Luna Azufre and the Cueva del Azufre were found with a smaller head and a higher caudal peduncle in females from Luna Azufre. The MANCOVA could not detect significant effects of allometry and population among males (Table 3.1B; likely this is an effect of the small sample size of N=16); the discriminant function analysis correctly assigned 87.5% of males to the population of origin, whereas two males from the Cueva del Azufre were misclassified to the Luna Azufre population (Table 3.2). The consensus configuration and the Cartesian transformation grids for males are shown in Figure 3.2D-F. Although no significant differences could be detected between males from the

Chapter 3: A new cave population 41 different population, there is a trend of males from the Luna Azufre having smaller heads and a higher caudal peduncle as it was found in the females.

Color polymorphism The distinct color polymorphism with pale and yellow morphs is not only present in cave mollies from the Cueva del Azufre, but also in the Luna Azufre population. The frequency of the yellow morph was significantly higher in the Luna Azufre population compared to the Cueva del Azufre population (Luna Azufre: 18 of 49 individuals yellow (37 %); Cueva del Azufre: 21 of 391 individuals yellow (0.5 %; Χ2: 53.03, P<0.001).

Length/weight relationship As expected, there was a strong positive relationship between standard length and body weight (Figure 3.3; F=371.29, P<0.001), but there were no differences between the two cave populations (F=0.31, P=0.580). Males were significantly lighter than females (F=12.87, P=0.001). The interaction term of ‘population’ and ‘sex’ was not significant (F=0.52, P=0.474).

Environmental conditions Measurements of the abiotic habitat characteristics in five different pools inside the Luna Azufre yielded the following results (mean ± SD): Temperature 27.5 ± 4.3 ºC, specific conductivity 3.613 ± 1.255 mS, pH 7.1 ± 0.0, oxygen content 1.65 ± 0.54 mg -1 l and relative oxygen saturation 22.0 ± 7.7 %. The two H2S samples taken revealed concentrations of 4 µM H2S in a spring and 0 µM in a stagnant pool south of the Entrada Marabunda. No hydrogen sulfide was detected in the air.

Discussion The new population of cave mollies from the Luna Azufre described here is only the second cavernicolous population of a poeciliid fish known to date. Similarities in appearance between cave mollies from the Luna Azufre and the nearby Cueva del Azufre, are evident. Compared to surface populations, both cave populations have reduced eyes and reduced pigmentation (Gordon and Rosen, 1962; Parzefall, 2001). However, the geometric morphometric analysis revealed that the females of the new population are morphologically distinct from the population of the Cueva del Azufre 42 Chapter 3: A new cave population in that individuals from the Luna Azufre have shorter heads and a higher caudal peduncle. The same trend, albeit not significant, was observed for the males. The biological significance of these traits is as yet unknown. The differences among the populations may be the result of phenotypic plasticity as suggested for other cave- dwelling organisms (see Romero and Green, 2005 for a review). The cave environment in the Luna Azufre differs from the Cueva del Azufre by the generally much lower H2S concentrations. Inside the Cueva del Azufre, H2S concentrations can reach more than 300 µM (Tobler et al., 2006). This is also reflected by the low abundance of sulfur oxidizing bacteria in the Luna Azufre, which cover almost all wet surfaces in the Cueva del Azufre (Pisarowicz, 2005). Caves are usually considered to be energy-poor habitats (Poulson and Lavoie, 2000), but because of the presence of bacterial primary production and the input of bat guano, the Cueva el Azufre was suggested to be energy-rich even compared to surface habitats (Langecker et al., 1996). Bats and accumulations of bat guano can also be found in the Luna Azufre and likely provide the energy basis for the cave ecosystem. Although the Cueva del Azufre is considered to be an energy-rich habitat, the mollies inhabiting the cave appear to be malnourished (Plath et al., 2005; Tobler et al., 2006) and energy availability affects the short-term survival of the fish (Plath et al., 2007c). It has been hypothesized that the low body condition is caused by costly adaptations necessary to cope with the high concentration of H2S (Tobler et al., 2006; Plath et al., 2007c). This idea is not supported by data on mollies from the Luna

Azufre, where very low concentrations of H2S were detected. Here, length/weight relationships did not differ significantly from that at Cueva del Azufre. This points to low energy availability in the Luna Azufre where chemoautotrophic primary production by sulfide-oxidizing bacteria likely plays an insignificant role. The effects of energy availability and the presence of H2S on the body condition of cave mollies certainly warrant further investigations. Both cave populations contain yellow morphs, but the frequency of the yellow morph differs significantly between the caves. This and the morphological differences observed indicate that the populations may be isolated. Although the caves are in close proximity, they are located within different hills that are separated by a surface valley. Previous research has shown that mollies from the Cueva del Azufre are reproductively isolated from the adjacent surface populations and that there is genetic structure even among mollies from different cave chambers within the Cueva del

Chapter 3: A new cave population 43

Azufre (Plath et al., 2007a). Therefore it is unlikely that gene-flow between the Luna Azufre and the Cueva del Azufre populations exists; however population genetic analyses are needed to determine the extent of genetic isolation. The cave molly system includes not only the two cave populations, but also closely related populations in adjacent surface habitats (sulfidic and non-sulfidic). This cluster of populations living under vastly different environmental conditions provides a unique system to study ecological diversification. The discovery of an additional cave population living in low-sulfide conditions is a keystone for future research on local adaptation and adaptive diversification in Poecilia mexicana. 44 Chapter 3: A new cave population

Figures and tables chapter 3

Figure 3.1. Cave mollies from the Luna Azufre (A: female, 38 mm SL; B: male, 26 mm SL) and the Cueva del Azufre (cave chamber V; C: female: 38 mm SL; D: male 28 mm SL).

Figure 3.2. Cartesian transformation grids showing the consensus configuration of all specimens examined (A: females; D: males), and the relative shape differences between the consensus and the two populations (B: Luna Azufre females; C: Cueva del Azufre females; E: Luna Azufre males; F: Cueva del Azufre males). Deformations are exaggerated three times.

Chapter 3: A new cave population 45

Figure 3.3. Length/weight relationship in mollies from the Luna Azufre (males: ; females: ) and the Cueva del Azufre (males: ; females: ).

Table 3.1. Results of the MANCOVA on principal components depicting the shape variance of cave mollies of the Cueva del Azufre and the Luna Azufre. A. Females (N=36) Effect F dfnum dfdenom P Estimated effect size Centroid size 6.909 10 24 <0.001 0.742 Population 6.158 10 24 <0.001 0.720

B. Males (N=16) Effect F dfnum dfdenom P Estimated effect size Centroid size 2.901 10 4 0.158 0.879 Population 1.575 10 4 0.351 0.798

46 Chapter 3: A new cave population

Table 3.2. Equality test of group means from the discriminant function analysis for the different coordinates (see materials and methods for a definition of coordinates). Females Males Wilks' Lambda F P Wilks' Lambda F P X1 0.769 10.208 0.003 0.942 0.860 0.369 Y1 0.811 7.918 0.008 0.954 0.678 0.424 X2 0.681 15.955 0.000 0.993 0.092 0.766 Y2 0.927 2.682 0.111 0.917 1.272 0.278 X3 1.000 0.016 0.901 1.000 0.007 0.936 Y3 0.963 1.319 0.259 0.993 0.094 0.764 X4 0.964 1.253 0.271 0.982 0.255 0.622 Y4 0.882 4.570 0.040 0.753 4.596 0.050 X5 0.882 4.560 0.040 0.963 0.533 0.478 Y5 0.944 2.018 0.165 0.822 3.031 0.104 X6 0.975 0.865 0.359 0.953 0.687 0.421 Y6 0.994 0.196 0.661 0.983 0.243 0.630 X7 0.942 2.074 0.159 0.851 2.460 0.139 Y7 0.745 11.630 0.002 0.807 3.356 0.088 X8 0.966 1.190 0.283 0.999 0.007 0.934 Y8 0.999 0.040 0.842 0.973 0.390 0.543 X9 0.980 0.687 0.413 1.000 0.001 0.982 Y9 0.971 1.006 0.323 0.980 0.283 0.603 X10 0.943 2.052 0.161 0.817 3.137 0.098 Y10 0.990 0.328 0.571 0.984 0.227 0.641 X11 0.701 14.519 0.001 0.964 0.518 0.484 Y11 0.921 2.924 0.096 0.996 0.054 0.820 X12 0.649 18.358 0.000 0.853 2.421 0.142 Y12 0.993 0.256 0.616 1.000 0.003 0.961 X13 0.882 4.529 0.041 0.998 0.024 0.879 Y13 0.942 2.079 0.158 0.998 0.032 0.861

Chapter 4: Adaptive divergence 47

Chapter 4

TOXIC HYDROGEN SULFIDE AND DARK CAVES: PHENOTYPIC

AND GENETIC DIVERGENCE ACROSS TWO ABIOTIC

3 ENVIRONMENTAL GRADIENTS IN POECILIA MEXICANA

Michael Tobler, Thomas J. DeWitt, Ingo Schlupp, F. J. García de León, Roger Herrmann, Philine G. D. Feulner, Ralph Tiedemann and Martin Plath

Abstract Divergent natural selection drives evolutionary diversification. It creates phenotypic diversity by favoring developmental plasticity within populations or genetic differentiation and local adaptation among populations. We investigated phenotypic and genetic divergence in the livebearing fish Poecilia mexicana along two abiotic environmental gradients. These fish typically inhabit non-sulfidic surface rivers, but also colonized sulfidic and cave habitats. We assessed phenotypic variation among a factorial combination of habitat types using geometric and traditional morphometrics, and genetic divergence using quantitative and molecular genetic analyses. Fish in caves (sulfidic or not) exhibited reduced eyes and slender bodies. Fish from sulfidic habitats (surface or cave) exhibited larger heads and longer gill filaments. Common- garden rearing suggested that these morphological differences are partly heritable. Population genetic analyses using microsatellites as well as cytochrome b gene sequences indicate high population differentiation over small spatial scale and very low rates of gene flow, especially among different habitat types. This suggests that divergent environmental conditions constitute barriers to gene flow. Strong molecular divergence over short distances as well as phenotypic and quantitative genetic divergence across habitats in directions classic to fish eco-morphology suggest that divergent selection is structuring phenotypic variation in this system.

3 Published as: M. Tobler, T. J. DeWitt, I. Schlupp, F. J. García de León, R. Herrmann, P. G. D. Feulner, R. Tiedemann & M. Plath (in press): Toxic hydrogen sulfide and dark caves: Phenotypic and genetic divergence across two environmental gradients in Poecilia mexicana. Evolution. 48 Chapter 4: Adaptive divergence

A fundamental question in evolutionary biology is how populations adapt to heterogeneous environments (Levins, 1968; Bohonak, 1999; Schluter, 2000). When populations are exposed to spatially divergent selection there are three typical evolutionary outcomes (Kawecki and Ebert, 2004). These scenarios are not mutually exclusive but rather constitute the extremes of a spectrum: (1) A single specialist optimally adapted to one habitat (usually the more common or productive one) and poorly adapted to others may evolve. In this case, source-sink dynamics are expected, whereby persistence in habitats to which the species is not adapted depends on migration from habitats where the species is better adapted (Dias, 1996; Dias and Blondel, 1996; Day, 2000; Holt et al., 2004). (2) Generalists adapted to tolerate multiple habitat types may evolve. Generalists can be phenotypically uniform intermediates (Van Tienderen, 1991; Palaima, 2007) or express alternate phenotypes under different environmental conditions (i.e. phenotypic plasticity, West-Eberhard, 1989; Pigliucci, 1996; DeWitt and Scheiner, 2004). In the case of a generalist, bi- directional migration between habitat types may occur (Wilson and Yoshimura, 1994; Sultan and Spencer, 2002). (3) Multiple specialists may be locally adapted to alternative habitat types (Levene, 1953). Hence, one expects divergent specialized phenotypes that maximize fitness in a given habitat and do not migrate between habitat types, which results in limited gene flow among populations (Kawecki and Ebert, 2004; Hays, 2007). Depending on the pattern of environmental heterogeneity, populations of organisms respond evolutionarily by evolving intermediate generalist phenotypes, phenotypic plasticity, or local adaptation, and either exhibit considerable or minimal migration. Local adaptation is hindered by gene flow because gene flow homogenizes allele frequencies among populations and prevents divergent selection from creating genetic divergence (Storfer and Sih, 1998; Lenormand, 2002; Moore et al., 2007). However, if divergent selection is sufficiently strong it can maintain population differentiation even when gene flow is present and can cause local adaptation on small spatial scales (Jimenez-Ambriz et al., 2006; Hays, 2007; Manier et al., 2007). If the response to environmental heterogeneity is heritable, local adaptation can proceed to speciation and adaptive radiation from a single ancestor (Schluter, 2000; Streelman and Danley, 2003). Ecological speciation occurs when divergent selection, in addition to driving trait divergence among populations, also leads to evolution of reproductive isolation. In traditional models of ecological speciation,

Chapter 4: Adaptive divergence 49 reproductive isolation evolves incidentally as a by-product (Schluter, 2000, 2001; Dieckmann et al., 2004; Rundle and Nosil, 2005); but whenever divergent natural selection occurs among populations, there may be direct selection for premating isolation (i.e. reinforcement, Schluter, 2001; Rodriguez et al., 2004). Evidence for ecological speciation in the wild is mounting (Funk, 1998; McPeek and Wellborn, 1998; Rundle et al., 2000; Jiggins et al., 2001; Nosil et al., 2002; McKinnon et al., 2004; Boughman et al., 2005; Langerhans et al., 2007b). In animals, a variety of— mostly biotic—selective agents have been documented to lead to reproductive isolation, including reproductive interference (Servedio and Noor, 2003), resource use (Funk, 1998; Ryan et al., 2007), interspecific resource competition (Pfennig and Rice, 2007; Tyerman et al., 2008), predation (Nosil and Crespi, 2006; Langerhans et al., 2007b), and parasitism (Blais et al., 2007). Adaptive divergence in response to divergent abiotic conditions is predominantly known from plants exposed to different soil types or elevation gradients (Macnair and Christie, 1983; Wang et al., 1997; Rajakaruna et al., 2003; Silvertown et al., 2005; Antonovics, 2006). In the present study, we examined phenotypic and genetic divergence in the livebearing fish Poecilia mexicana (Atlantic molly, Poeciliidae). This species has colonized habitats differing in abiotic conditions in the Cueva del Azufre system in southern Mexico. Habitats in this system are characterized by the presence or absence of naturally occurring hydrogen sulfide (H2S) and/or light (i.e. cave versus surface habitats), providing a natural 2×2 factorial design of these two environmental conditions (Tobler et al., 2006; Tobler et al., 2008b). Both the presence of H2S and the absence of light are potential sources of divergent natural selection. H2S is correlated with extreme hypoxia in aquatic environments and is a potent respiratory toxicant lethal for most metazoans even in micromolar amounts (Evans, 1967;

Bagarinao, 1992; Grieshaber and Völkel, 1998). In the Cueva del Azufre system, H2S is present in acutely toxic concentrations up to 300 µM (Tobler et al., 2006). Similarly, the absence of light in caves inhibits the use of visual senses, and cave- dwellers are under selection to cope with darkness, especially if they evolved from a diurnal surface-dwelling form like in P. mexicana (Poulson and White, 1969; Howarth, 1993; Culver et al., 1995; Langecker, 2000; Plath et al., 2004). Therefore these environmental axes should provide two divergent natural selection gradients along which to test for phenotypic divergence and patterns of gene flow in the absence of vicariance. 50 Chapter 4: Adaptive divergence

We asked four major questions: (1) Is phenotypic differentiation in P. mexicana populations evident across two environmental gradients? (2) Does divergence correspond to eco-morphological expectations? (3) Do divergent traits have a heritable component? (4) Is gene flow limited among P. mexicana from different habitats? We address these questions by surveying morphology across environmentally diverse sites, comparing diversification with previous studies on eco- morphology, analyzing phenotypes of fish from alternative populations when raised in a common garden, and analyzing marker genetics among sites.

Materials and Methods Populations Poecilia mexicana is common in freshwater habitats on the Atlantic side of Central America from northern Mexico to Costa Rica (Miller, 2005b). Our study sites are located near the village of Tapijulapa in the southern Mexican state of Tabasco (Figure 1.1 for an overview and Table 4.1 for exact locations). We sampled four different habitat types that are characterized by the presence or absence of H2S and/or light. All sites are within 10 km of each other (river distance), and the average distance between sites is about 3.5 km (Table 4.2). Sites sampled include normal (non-sulfidic, surface) rivers (N=6 sites), sulfidic surface rivers (N=2 sites), a non- sulfidic cave (N=1 site), and a sulfidic cave in which we sampled from multiple (N=5) cave chambers. The two caves investigated are the only known subterranean habitats inhabited by P. mexicana. • The Cueva del Azufre is a sulfidic cave. The cave is structured into different chambers (Figure 1.2), the nomenclature of which follows Gordon and Rosen (1962). The front chambers obtain some dim light, whereas the rearmost cave chambers are completely dark. The cave is drained by a creek fed by a number of

springs throughout the cave, most of which contain high levels of dissolved H2S (Tobler et al., 2006). Poecilia mexicana occur throughout the cave, and for this study they were collected in chambers II, V, X, XI and XIII. • Despite its name, the Cueva Luna Azufre is a non-sulfidic cave (Tobler et al., 2008b). It is smaller than the Cueva del Azufre, and P. mexicana occur at lower densities. Although the two caves are in close proximity, they are located within different hills that are separated by a surface valley. The creek in the Cueva Luna

Azufre is also fed by springs, however, these do not contain H2S (Tobler et al.,

Chapter 4: Adaptive divergence 51

2008b). Poecilia mexicana were collected south of the Entrada Marabunda (Figure 1.3). • The El Azufre is a sulfidic surface habitat. It is fed by multiple independent sulfidic as well as non-sulfidic springs and flows through the valley that separates the two caves. Both caves drain into the El Azufre, which eventually joins the Río Oxolotan. Hydrogen sulfide concentrations are comparable to those in the Cueva del Azufre. Poecilia mexicana were collected upstream around some sulfidic springs as well as downstream near the resurgence of the Cueva del Azufre. • Six non-sulfidic surface habitats were sampled. These habitats include large rivers like the Río Oxolotan (most proximate to the other habitat types; Figure 1.1) and the Río Amatan, as well as some of their tributaries that are similar in size and structure to the El Azufre. Fish were collected in January 2006 and May 2007. Because habitat structures differed between sampling sites, different methods were employed. In the caves, where the water is very shallow and low ceilings preclude seining, fish were caught with dip nets (13 x 14 cm, 1 mm mesh-width). In the other habitats, fish were caught using a seine (4 m long, 4 mm mesh-width). All specimens were euthanized using MS222 immediately after capture and fixed in a 10% formaldehyde solution. Fin clips for extraction of DNA were stored in 96% ethanol at 4°C. Table 4.1 summarizes the material collected and examined in the different analyses. Heritability of traits was estimated by analysis of a population-level common garden rearing experiment (Weir, 1996). Laboratory stocks of fish were available from three populations: the sulfidic cave, the non-sulfidic cave, and a non-sulfidic surface habitat (Rio Oxolotan). Fish from sulfidic surface habitats were not available in the laboratory. All stocks were founded in January 2006 and maintained as randomly out-bred populations in 1000-liter tanks in a greenhouse at the Aquatic Research Facility of the University of Oklahoma. All stocks were exposed to identical environmental conditions (i.e., natural light cycle and no H2S). Algae, detritus, and invertebrates were present in the stock tanks, and the diet was supplemented with commercial flake food twice a week. Random samples of fish from these stocks were collected in May 2007 (Table 4.1). At this point the stocks were established in the laboratory for multiple generations. As for the wild-caught fish, specimens were euthanized using MS222 and fixed in a 10% formaldehyde solution.

52 Chapter 4: Adaptive divergence

Morphometrics We investigated divergence in P. mexicana morphology across habitat types as well as similarity of laboratory-reared to wild-caught fish using a geometric morphometric analysis of body shape. Due to the hypoxic nature of sulfidic habitats, we further analyzed gill morphology of wild-caught fish from different habitats.

Geometric morphometics For geometric morphometric analysis, lateral radiographs were taken with a Hewlett- Packard (Palo Alto, CA) Faxitron cabinet X-ray system. We digitized 13 landmark points on each image using the software program tpsDig (Rohlf, 2004). Landmarks included the tip of the upper jaw (1); the center of the orbital (2); the posterodorsal tip of the skull (3); the anterior (4) and posterior (5) junction of the dorsal fin with the dorsal midline; the junction of the caudal fin with the dorsal (6) and ventral (7) midline; the anterior (8) and posterior (9) junction of the anal fin with the ventral midline; the anterior junction of the pelvic fins and the ventral midline (10); the bottom of the head where the operculum breaks away from the body outline (11); the center of the first vertebra (12); and the center of the third vertebra with a hemal arch (13). Based on the coordinates of the digitized landmarks, we performed a geometric morphometric analysis (e.g., Zelditch et al., 2004). Data were translated to NTS format using tpsUtil (Rohlf, 2006). Landmark coordinates were aligned using least-squares superimposition as implemented in the program tpsRelw (Rohlf, 2007) to remove effects of translation, rotation, and scale. Eye diameter was measured to the nearest 0.01mm with calipers. This distance was halved, and used to position two reference points anterior (14) and posterior (15) to the orbit landmark (with the same y-value). The aligned coordinates plus reference points were subjected to eigendecomposition (principal component analysis) to reduce the data to true dimensionality. The last seven eigenvalues were null; four due to superimposition (two for translation, one for rotation, one for scaling) and three due to deficiency (sensu Bookstein, 1991) of the two reference points. Null dimensions were dropped from the analysis and the remaining principle axes were retained as shape variables. Body shape variation (23 principle components) was analyzed using multivariate analyses of covariance (MANCOVA). Assumptions of multivariate normal error and

Chapter 4: Adaptive divergence 53 homogeneity of variances and covariances were met for all analyses performed. Effect strengths were estimated using partial eta squared (ηp²). For wild-caught fish, we tested for effects of centroid size to control for multivariate allometry, and sex as well as presence or absence of H2S and light as independent variables. Shape variation along the two environmental gradients was visualized using thin-plate spline transformation grids (Zelditch et al., 2004; Rohlf, 2005). To provide a quantitative basis for the nature of shape effects, we calculated correlations of superimposed landmark coordinates with the shape gradients. This is done by creating a score for each specimen on the focal shape axis. To wit, we multiplied the eigenvector of the effect SSCP matrix by the principle components block to yield a column of scores. Correlation is then calculated between these scores and superimposed coordinate values. For the comparison of wild-caught and laboratory fish, we used centroid size as a covariate, and sex, habitat type, as well as treatment (i.e., wild-caught or laboratory-reared) as independent variables. If morphological variation were entirely caused by environmentally induced phenotypic plasticity, differences among fish from alternative habitat types should disappear in laboratory stocks housed under identical conditions. Likewise, if morphological differences were principally heritable, no differences between laboratory raised and wild caught individuals would be expected. An intermediate result would suggest that the traits under investigation have a heritable basis, but phenotypic plasticity also plays a role. To provide another intuitive measure of effect strength, we conducted heuristic discriminant function analyses (DFA) to determine the percentage of specimens that could be correctly classified to the population of origin based on body shape. To facilitate the DFAs we first removed the effects of sex and allometry by using the residuals of preparatory MANCOVAs. In these MANCOVAs, the 23 principle components were used as dependent variables, centroid size as a covariate, and sex as an independent variable. DFA on the pooled laboratory and wild-caught fish also allowed us to test whether laboratory reared individuals clustered with wild- caught specimens from their original habitat type. All statistical analyses were performed using SPSS 16 (SPSS, Inc., 2007).

Gill morphometrics Total gill filament length (TGFL) was measured as a proxy for the gill surface area in a sub-sample of individuals. TGFL is correlated with gill surface area in the closely 54 Chapter 4: Adaptive divergence related Poecilia latipinna (Timmerman and Chapman, 2004) and other fishes (Chapman et al., 2000; Langerhans et al., 2007a). To determine TGFL, each of the four gill arches from the left branchial basket was removed in a random sub-sample of individuals from each habitat type. Arches were placed on a microscope slide, and a picture was taken from both sides using a Spot Insight digital camera mounted on an Olympus stereomicroscope. For each hemibranch, the length of every fifth filament was measured using an image analysis program (Spot Advanced 4.5, Diagnostic Instruments, 2005). The mean of successive measurements was calculated to estimate the length of intermediate filaments. Then, filament lengths were summed for the eight hemibranches and multiplied by two to produce an estimate of TGFL. Variation in TGFL among habitats differing in abiotic environmental conditions was examined using an analysis of covariance (ANCOVA), in which TGFL (log-transformed) was used as a dependent variable, body mass of the individual (log-transformed) as a covariate (to control for allometry, see Timmerman and Chapman, 2004; Graham,

2005), and presence of H2S, as well as presence of light as independent variables. Homogeneity of slopes was observed for this analysis.

Genetic analyses We used a population genetic approach to distinguish between the evolutionary scenarios outlined in the introduction. If a single specialist adapted to non-sulfidic surface habitats was present, we would expect little genetic differentiation between different habitat types and primarily unidirectional migration patterns from non- sulfidic surface habitats to sink populations residing in the other habitats. If P. mexicana is a generalist equally adapted to multiple habitat types, genetic differentiation among population is also expected to be low if not absent, but bi- directional migration across gradients should occur. Alternatively, P. mexicana in each habitat type may be locally adapted to the respective abiotic condition. In this case genetic differentiation among populations from different habitat types would be expected, and migration events may be more common between sites with similar abiotic conditions. To test these alternative hypotheses, we performed a population genetic study using microsatellite markers and cytochrome b gene sequence variation.

Chapter 4: Adaptive divergence 55

Microsatellite analysis DNA was extracted from tissue samples using the DNeasy DNA Extraction kit (QIAGEN, Hilden, Germany) according to the manufacturer’s recommendations. Ten microsatellite loci were amplified according to previously described cycling parameters using approximately 400 ng of genomic DNA as a template (Tiedemann et al., 2005; Plath et al., 2007a). Fragment sizes were determined on an ABI 3100 automatic sequencer using GENESCAN 2.1 and an internal size standard (GeneScan- 500 LIZ, Applied Biosystems). Data for N=99 individuals from a previous study were re-analyzed (Plath et al., 2007a). We checked for the independent inheritance of all loci (linkage disequilibrium) with a likelihood ratio test using GENEPOP on the internet (http://wbiomed.curtin.edu.au/genepop/). FSTAT (Goudet, 2002) was used to calculate allelic richness. GenAlEx (Peakall and Smouse, 2001) was employed to calculate observed (HO) and expected heterozygosity (HE). GENEPOP was also used to conduct a probability test for deviation from Hardy-Weinberg equilibrium (HWE). For all tests, we used 1,000 dememorization steps and 100 batches with 10,000 iterations each.

We calculated pair-wise genetic distances (FST) using Arlequin (Schneider et al., 2000). P-values were based on 1,000 permutations. The same program was also used to test for overall differentiation among populations using analysis of molecular variance (AMOVA). We tested whether genetic differentiation would be greater among sites of a different habitat type compared to sites of the same habitat type by subjecting the pair-wise FST values to a partial Mantel test with 2,000 randomizations as implemented in FSTAT (Goudet, 2002). Predictor matrices were based on habitat type (same or different) and distance between sites as a covariate (to test for an effect of isolation by distance). STRUCTURE (version 2.1) (Pritchard et al., 2000) was used to identify the number of genetically distinct clusters (k) according to HWE and linkage equilibrium with the method presented by Evanno et al. (2005). For each value of k (k=1 through 12), three iterations were run using the admixture model with a burn-in period of 100,000 iterations followed by the same number of iterations for the collection phase. Each simulation was performed using an ancestry model incorporating admixture, a model of correlated allele frequencies, and the prior population information. 56 Chapter 4: Adaptive divergence

To estimate the number of first-generation migrants, we used GENECLASS2 (Piry et al., 2004). We used the L_home likelihood computation, the Bayesian method of classification (Rannala and Mountain, 1997), and a threshold P-value of 0.05. We used a partial Mantel test with 2,000 randomizations to compare the number of migrants (square root-transformed) between pair-wise sites (see Crispo et al., 2006). Predictor matrices were based on distance between sites and habitat type (same or different) as well as difference in habitat types with respect to the presence of H2S (-1: movement form a sulfidic to a non-sulfidic habitat; 0: no change; +1: from non- sulfidic to sulfidic) and the absence of light (-1: movement from a cave to a surface habitat; 0: no change; +1: from surface to cave).

Cytochrome b sequencing The complete mitochondrial cytochrome b gene was sequenced using the primers LA15058 and HA16249 (Schmidt et al., 1998). Approximately 800 ng of genomic

DNA was used as a template for each PCR. The annealing temperature was Ta=47ºC, otherwise PCRs were performed according to Feulner et al. (2005), but using GoTaq®Flexi (Promega, Mannheim, Germany) as polymerase. PCR products were purified using the QIA-quick PCR purification kit (Qiagen, Hilden, Germany). Cytochrome b was then sequenced in both directions with the primers used for amplification using the BigDye v3.1 Terminator Cycle-sequencing Kit (Applied Biosystems, Foster City, USA). The Multiscreen-HV (Millipore, Bedford, USA) purified products were analyzed on an AB 3100 multicapillary automatic sequencer (Applied Biosystems, Foster City, USA). All sequences are available on GenBank (Accession numbers: EU269039-EU269065). A network analysis was performed to estimate gene genealogies using the TCS program (Clement et al., 2000), which implements the Templeton et al. (1992) statistical parsimony. To summarize the degree of genetic differentiation, we calculated pair-wise FST values using F-statistics (Weir and Cockerham, 1984). The significance of FST was tested by permutation analysis, and AMOVA (Excoffier et al., 1992) was conducted as implemented in Arlequin (Schneider et al., 2000). We tested whether genetic differentiation would be greater among sites of a different habitat type compared to sites of the same habitat type by subjecting the pair-wise FST values to a partial Mantel test with 2,000 randomizations as implemented in FSTAT

Chapter 4: Adaptive divergence 57

(Goudet, 2002). Predictor matrices were based on habitat type (same or different) and distance between sites as a covariate (to test for an effect of isolation by distance).

Results Morphological analyses Geometric morphometrics A total of 497 wild-caught individuals were analyzed (Table 4.1). Body shape differed significantly and strongly (i.e. ηp² > 0.5) along both environmental gradients as well as between the sexes (Figure 4.1, Table 4.3A & 4.4). Fish from cave habitats had smaller eyes and were also more shallow bodied than fish from surface habitats, irrespective of whether the habitat of origin contained H2S or not. Fish from sulfidic habitats were characterized by an increase in head size, irrespective of whether the habitat was located in a cave or at the surface. Consequently, P. mexicana in non- sulfidic surface habitats were high bodied with large eyes but small heads; in sulfidic surface habitats, fish were high bodied with large eyes and large heads; in the non- sulfidic cave they were shallow bodied with small eyes and small heads; and in the sulfidic cave, they were shallow bodied with small eyes and large heads (Figure 4.1). The primary difference among sexes was in the position of the anal fin. In males, the anal fin is modified into a copulatory organ (the gonopodium, characteristic of the subfamily Poeciliinae), which is typically more anterior than the female anal fin (Rosen and Bailey, 1963). Although all effects in the model were significant, the interaction effects were generally weak (ηp² < 0.2), with the exception of the H2S × light interaction (ηp² ≈ 0.5). Over 91% of the specimens (compared to the expected 25% under a null hypothesis of no pattern) could be assigned to the habitat type of origin based on morphometric data (Figure 4.2A, Table 4.5A). Laboratory-raised fish differed significantly from wild-caught specimens, indicating that body shape was to some extent phenotypically plastic (Table 4.3B). This effect can be seen in Figure 4.2B as the lab-reared fish multivariate centroids do not superimpose directly on those for wild-caught specimens. Although laboratory fish were raised under identical conditions and never encountered H2S or permanent darkness, they arrayed geometrically like (clustered with) wild-caught individuals from their habitat type of origin (Figure 4.2B). Our result does not allow for an estimation of narrow sense heritability, but it shows that divergent body morphologies have a heritable basis. Though phenotypic plasticity (via tank effects) and maternal 58 Chapter 4: Adaptive divergence effects (if persistent for many generations) could have created some of the population difference, it is improbable those effects could completely replicate the geometry of difference in lab-reared fish in conformation with that observed for wild-caught fish. The DFA classified over 92% of the specimens to the correct habitat type (compared to the expected 33% under a null hypothesis of no pattern; Table 4.5B).

Gill morphometrics TGFL increased with increasing body mass, and fish from sulfidic habitats had a longer TGFL than fish from non-sulfidic habitats (Table 4.3C, Figure 4.3). This effect was not dependent on whether specimens were collected in a cave or a surface habitat.

Genetic analyses Microsatellite analysis Overall, we genotyped 269 specimens (Table 4.1). A total of 225 alleles were found within 10 loci, ranging from 4 to 48 per locus (for descriptive statistics see Table 4.6).

We observed strong genetic differentiation among populations (AMOVA: overall FST = 0.198; P <0.001), and 19.8% of variation was assigned to variability among sites.

The partial Mantel test explained 46.1% of the variance in pair-wise FST. Pair-wise

FST values were significantly lower between sites of the same habitat type (mean pair- wise FST ± SD: 0.074 ± 0.069) than between sites of a different habitat type (0.241 ± 0.088; Table 4.7; r = 0.678, P < 0.001). Distance between sites did not have a significant influence on genetic differentiation (r = 0.038, P = 0.75). The assignment test (STRUCTURE) found most support for k=5 clusters (Figure 4.4). Clusters corresponded with habitat types in all but one case. The P. mexicana collected in Arroyo Cristal were genetically distinct from their conspecifics in other non-sulfidic surface habitats, even though some of the individuals were genetically similar to conspecifics from other non-sulfidic surface habitats. The genetic differentiation among P. mexicana was also reflected in the analysis of first generation migrants. The partial Mantel test explained 35.2% of variance in the number of migrants between sites. Migration events were more common between sites of the same habitat type than between sites of different habitat types (Tables 4.8; r = -0.568, P < 0.001). There was no significant effect of distance between sites on the number of migrants (r = -0.154, P = 0.18). Furthermore, we

Chapter 4: Adaptive divergence 59 found no evidence that migration is more common from sulfidic to non-sulfidic (r=.005, P=0.70) or from cave to surface habitats (r = -0.076, P = 0.38).

Cytochrome b sequencing We sequenced the cytochrome b gene of 142 individuals. The minimum spanning network (Figure 4.5) showed a central haplotype, shared by most populations on the plateau on which El Azufre and the two caves are located (sulfur plateau), along with a set of haplotypes that formed a star-like topography around the central haplotype that were found throughout the various non-sulfidic surface sites. While both cave populations and the sulfidic surface habitats still share the central haplotype, each habitat type also harbored at least two private haplotypes. Altogether, this suggests a close relationship of the fish on the sulfur plateau. Both caves have likely been colonized from the El Azufre. The AMOVA assigned 39.5% of variation to variability among sites and indicated a high genetic differentiation among populations (Table 4.7; overall FST = 0.395; P < 0.001). The partial Mantel test explained 37.8% of variance in pair-wise

FST. Pair-wise genetic differentiation was slightly lower between sites of the same habitat type (mean pair-wise FST ± SD: 0.324 ± 0.202) than between sites of different habitat types (0.393 ± 0.130; r = 0.201. P = 0.059) and there was no isolation by distance (r = -0.581; P < 0.001). In fact, genetic distance decreased significantly with increasing distance between sites. This negative relationship was driven by high genetic differentiation in P. mexicana from sites on the sulfur plateau – both within as well as between habitat types – even at low spatial scales (<1km).

Discussion In southern Mexico, Poecilia mexicana has colonized four different habitat types characterized by the presence or absence of toxic H2S and/or light providing a ‘natural experiment’ with a fully factorial 2×2 design of the two abiotic selective agents. The results of our study indicate that fish diverged morphologically and genetically in each habitat despite the close spatial proximity of the sites, which is most consistent with a local adaptation scenario as outlined in the introduction. Patterns of heritable phenotypic differentiation and low gene flow among habitats excludes the single- specialist scenario, as there appear to be no sink populations maintained by migration. 60 Chapter 4: Adaptive divergence

Likewise the single generalist scenario is not supported by our results because each habitat has a clearly defined phenotype having a heritable basis.

Morphological differentiation Independent morphological variation occurs along each of the two environmental gradients. Poecilia mexicana from cave habitats are characterized by reduced eye size and more shallow bodies compared to fish from surface habitats, independent of whether or not H2S is present in their habitat. Reduced eye size has previously been reported for P. mexicana from the Cueva del Azufre (Walters and Walters, 1965; Peters et al., 1973; Plath et al., 2007a), but unlike in other cave organisms (Porter and Crandall, 2003), the eyes in cave-dwelling P. mexicana are still functional (Plath et al., 2004; Körner et al., 2006). The adaptive value of eye reduction in subterranean organisms is still under debate with two major opposing theories. The ‘neutral mutation hypothesis’ posits that eye regression is caused by the accumulation of mutations in eye-forming genes under relaxed selection in darkness (Culver, 1982; Wilkens, 1988). In contrast, the ‘adaptation hypothesis’ suggest that eye reduction provides fitness benefits in the cave environment (Poulson, 1963; Poulson and White, 1969). Different versions of the adaptation hypothesis attribute the regression of visual senses to energy economy, emphasizing the costs of making and maintaining an eye (Culver, 1982), or to pleiotropic effects, in which structures beneficial to survival in the cave environment are enhanced at the expense of eyes (Barr, 1968). Recent evidence for the adaptation hypothesis comes from studies investigating the genetic and developmental mechanisms of eye degeneration in cave-dwelling Astyanax mexicanus (Jeffery, 2005; Protas et al., 2007; but see Wilkens, 2007). It is argued that pleiotropic effects act on eye degeneration while enhancing traits that are adaptive in the cave environment, such as non-visual sensory structures (Jeffery, 2001, 2005). Poecilia mexicana from cave habitats have previously been shown to posses a hyper-developed cephalic lateral line system and an increased number of taste buds (Walters and Walters, 1965; Parzefall, 1970, 2001), but the developmental pathways of eye development/degeneration and their potential pleiotropic linkage to the development of non-visual sensory structures remain to be studied in this species. Poecilia mexicana from cave habitats are also more shallow bodied than fish from surface habitats, a trait commonly observed in cave organisms (Langecker, 2000). The reduction in body height is not due to poor nutritional condition of cave

Chapter 4: Adaptive divergence 61 populations because we find equally low storage lipid levels in both cave populations and surface fish from sulfidic habitats (Tobler, in press). Shallow bodies in caves could be driven by divergent predatory regimes, which have been shown to induce phenotypically plastic morphological changes in prey organisms (Spitze, 1992; DeWitt, 1998; Relyea, 2001) as well as heritable morphological differences among prey populations (McPeek, 1995; Nosil and Crespi, 2006; Johnson et al., 2007). In the two caves, piscine and avian predators are absent (Tobler et al., 2006; Tobler et al., 2007a), and the only predator in the Cueva del Azufre is a giant water-bug of the genus Belostoma (Tobler et al., 2007b; Tobler et al., 2008a). A lower body height (especially for the caudal peduncle) has been reported for other fish living in low predation environments (crucian carp: Brönmark and Miner, 1992; Western mosquitofish: Langerhans et al., 2004; guppy: Hendry et al., 2006; perch and roach: Eklöv and Jonsson, 2007; Bahamas mosquitofish: Langerhans et al., 2007b). Poecilia mexicana from non-sulfidic and sulfidic habitats diverged primarily in head size and total gill filament length (irrespective of whether the habitat was located at the surface or within a cave), which is consistent with findings in fishes (Chapman et al., 1999; Chapman et al., 2000; Chapman and Hulen, 2001; Timmerman and Chapman, 2004; Langerhans et al., 2007a), amphibians (Bond, 1960; Burggren and Mwalukoma, 1983), and invertebrates (Astall et al., 1997; Roast and Jones, 2003) living in other types of hypoxic environments. This highlights the importance of respiratory adaptations facilitating efficient oxygen acquisition for survival in sulfidic habitats (McMullin et al., 2000; Van Dover, 2000; Affonso and Rantin, 2005; Plath et al., 2007c). Sulfide detoxification in organisms capable of tolerating high and sustained concentrations of H2S is primarily achieved through its oxidation to less toxic sulfur species and subsequent excretion (Curtis et al., 1972; Bagarinao, 1992; Ip et al., 2004). Due to the hypoxic conditions in sulfidic habitats, however, oxygen available for respiration is generally limited, but at the same time oxygen is required for coping with the toxic effects of H2S. Some fish species rely on air-breathing to cope with low oxygen availability in sulfidic habitats (Bagarinao and Vetter, 1989; Brauner et al., 1995; Affonso and Rantin, 2005), but P. mexicana from the Cueva del Azufre rely on compensatory behavior (aquatic surface respiration), where the fish exploit the more oxygen-rich air-water interface using their gills (Plath et al., 2007c). 62 Chapter 4: Adaptive divergence

The significant H2S × light effect on body shape is characterized by a shift of eye position (and to a lesser degree a decrease in eye size) as head size increases. The interaction effect either indicates developmental constraints on morphological evolution or correlational selection (Brodie, 1992; Sinervo et al., 2001; DeWitt and Langerhans, 2003). Selection for an increase in gill size is generally thought to impose morphological trade-offs and to indirectly affect other head characteristics such as brain size (Chapman and Hulen, 2001) or trophic morphology (Chapman et al., 2000). Morphological differences among P. mexicana from different habitat types do not seem to be entirely caused by environmentally induced phenotypic variation, since laboratory stocks maintained under identical conditions clustered morphologically with wild-caught fish from the respective habitat types. This result indicates that these axes of morphological variation are at least partially heritable. Variation in body morphology (Greenfield et al., 1982; Greenfield and Wildrick, 1984; Ptacek, 2002; Langerhans et al., 2004; Langerhans et al., 2005) as well as gill morphometrics (Timmerman and Chapman, 2004) have been shown to have a heritable component in other poeciliid fishes. Likewise, other aspects of divergent morphological (Peters and Peters, 1968; Parzefall, 2001) as well as behavioral traits (Plath et al., 2004; Plath et al., 2006; Plath, 2008) of P. mexicana from different habitat types in the Cueva del Azufre system have a heritable basis. However, heritability of body shape in this study may have been overestimated, if epigenetic (e.g., maternal) effects influenced morphology (Holtmeier, 2001; Keller et al., 2001), or underestimated, if laboratory conditions (exposure to light and lack of H2S) exerted strong selection on body shape of fish naturally occurring in cave or sulfidic habitats leading to a rapid evolutionary change in the stock populations. Hence, future studies need to estimate narrow-sense heritability as well as the degree of phenotypic plasticity of morphological traits when fish are exposed to continuous darkness and/or H2S.

Genetic differentiation and Migration Genetic differentiation of P. mexicana in the Cueva del Azufre system parallels the observed morphological differentiation. Both marker systems used (microsatellites and cytochome b sequences) indicate that each habitat type harbors a distinct population, and genetic distance is lower among P. mexicana from sites of the same

Chapter 4: Adaptive divergence 63 than from different habitat types. No evidence for isolation by distance was uncovered. Likewise, contemporary dispersal between sites predominantly occurred within the same habitat type. This suggests that the divergent abiotic conditions indeed constitute strong (albeit not insurmountable) barriers to migration. Support for the hypothesis that migration events between different habitat types should be more common from harsh to benign environments than vice versa (see Railsback et al., 1999; Caskey et al., 2007) was not evident. Notably, there was no migration from surface to cave habitats, but low rates of migration were detected from non-sulfidic to sulfidic habitats. Thus, either the absence of light is a stronger selective agent than the presence of H2S, or the shallow passages with swift flow at the cave resurgences constitute stronger physical barriers for the movement of P. mexicana than previously thought. The latter hypothesis, however, seems unlikely, because bi-directional migration over potential physical barriers (waterfalls) were detected at least among populations from non-sulfidic surface habitats (e.g., between Arroyo Bonita and Río Oxolotan); and cave resurgences seem less likely as barriers than waterfalls. The mitochondrial haplotypes recorded in the Cueva del Azufre system differ by few mutation steps, suggesting that fish from different habitat types are closely related and have diverged only recently. The fish on the sulfur plateau share a common haplotype, which suggests common ancestry. The two caves were probably colonized independently from the sulfidic surface creek. Overall, evidence hints towards a parapatric divergence of P. mexicana populations in different habitat types, since physical separation of the divergent populations is not evident.

Speciation along abiotic gradients? Divergent natural selection has been shown to shape the population genetic structure in several other studies (Turgeon et al., 1999; Steiniger et al., 2002; Martel et al., 2003; Dhuyvetter et al., 2007; Quesada et al., 2007). Abiotic gradients commonly structure phenotypic variation and perhaps facilitate speciation in plants (e.g. Donohue et al., 2001; McDonald et al., 2003; Swenson and Enquiest, 2007). However, abiotic gradients seem either to exert less influence, or perhaps just get less play in literature on animals. A recent study investigating possible genetic differentiation along a similar environmental gradient as addressed here (hypoxic versus normoxic habitats) did not find any effect of the oxygen regime on the 64 Chapter 4: Adaptive divergence population genetic structure in a cichlid fish (Crispo and Chapman, 2008). In that system, phenotypic plasticity is thought to play a central role for phenotypic differentiation of fish across habitat types (Crispo and Chapman 2008). It seems likely that the high levels of toxic hydrogen sulfide in our study system represent a stronger selection factor for aquatic animals than hypoxia alone. In our study we found strong phenotypic and genetic divergence across two abiotic gradients. The isolating mechanisms leading to genetic differentiation among populations of P. mexicana from different habitat types are unclear. Obvious physical barriers or significant distances among populations are lacking. It is unlikely that populations in the divergent habitats are genetically incompatible (i.e. not inter- fertile), since there is no intrinsic post-zygotic reproductive isolation known even in more distantly related poeciliid species (Hubbs, 1959; Schartl, 1995; Ptacek, 2002; Dries, 2003; Rosenthal et al., 2003; Alexander and Breden, 2004; Kittell et al., 2005). Likewise, isolation due to genetically-based preferences for separate habitat types (Rice and Salt, 1990; Johnson et al., 1996), which are common for radiations in phytophagous insects (Berlocher and Feder, 2002), is unlikely at least for the separation between surface- and cave-dwelling populations. Like surface-dwelling fish (El Azufre population), P. mexicana from the Cueva del Azufre exhibit photophilic behavior (Parzefall et al., 2007). We propose that divergent natural selection caused by the abiotic gradients, in combination with local adaptation in P. mexicana, limits gene flow across habitat types (Räsänen and Hendry, 2008). Correspondence between morphological variation along environmental gradients and eco-morphological expectations suggest local adaptation through divergent natural selection. The patterns of migration and genetic differentiation also point to abiotic conditions creating divergence among populations through divergent natural selection. Several isolating mechanisms, which may act in synchrony, seem possible in this system. (1) Selection could act directly on immigrants from divergent populations causing premating isolation (Nosil et al., 2005). For example, P. mexicana from non-sulfidic habitats are highly susceptible to the toxic effects of H2S (Tobler et al., 2008c). (2) Poecilia mexicana from different habitat types may be less attracted to conspecifics from divergent habitat types, which may cause prezygotic isolation (Schluter, 2000; Rundle and Nosil, 2005). (3) Divergent selection could act on hybrids of P. mexicana from different habitats

Chapter 4: Adaptive divergence 65

(Hatfield and Schluter, 1999; Schluter, 2000). To date no empirical evidence for the latter two mechanisms is available. Future studies will need to pay careful attention to the evolutionary forces causing the observed small-scale population differentiation in the Cueva del Azufre system in order to test whether parapatric ecological speciation is occurring. Regardless, the strong divergence observed along two abiotic gradients in the present study, and a potentially growing literature on divergence along abiotic gradients in animals (e.g., Schilthuizen et al., 2005; Fuller et al., 2007), suggests that abiotic factors may be potentially more important in animal diversification than is currently thought. These factors may be complex. We not only found effects of multiple abiotic gradients (see also Langerhans et al., 2007a), but also a significant interaction between gradients. It would be no stretch of imagination to expect interactions also between biotic and abiotic environmental factors. A key to understanding biological diversity will often be to embrace the complexity of nature and admit it conceptually and empirically into our investigations (DeWitt and Langerhans, 2003). 66 Chapter 4: Adaptive divergence

Figures and tables chapter 4

Figure 4.1. Morphological variation of P. mexicana along the two environmental gradients. Independent variation is explained along each environmental gradient (non- sulfidic to sulfidic, A; surface to cave, B), and there is also a significant interaction effect (H2S×light, C). This independent variation gives rise to unique phenotypes in each habitat type (non-sulfidic surface habitats, D; sulfidic surface habitats E; non- sulfidic cave, F; sulfidic cave, E).

Chapter 4: Adaptive divergence 67

Figure 4.2. Discriminant function plots (group centroids±SD for the first two discriminant functions) for the analyses presented in Table 4.5. (A) Analysis of wild- caught individuals from all four habitat types. (B) Comparison between wild-caught and laboratory-reared individuals. Non-sulfidic surface habitats (), sulfidic surface habitats (), non-sulfidic cave (), and sulfidic cave (). Closed symbols represent wild-caught individuals, open symbols laboratory-reared individuals.

68 Chapter 4: Adaptive divergence

Figure 4.3. Mean (± standard deviation) residual total gill filament length (TGFL) for P. mexicana of different habitat types. Residuals were obtained using a linear regression with log (TGFL) as dependent variable and log (body mass) as independent variable. N represents the sample size.

Figure 4.4. Population differentiation (STRUCTURE) for k=5 clusters. Abbreviations follow Table 4.1.

Chapter 4: Adaptive divergence 69

Figure 4.5. Minimum spanning network showing the relationship of different cytochrome b haplotypes. Different habitat types are coded using a gray-scale (white: surface, non-sulfidic; light gray: surface, sulfidic; dark gray: cave, non-sulfidic; black: cave, sulfidic). Abbreviations indicate field sites (see Table 4.1), numbers represent the number of individuals carrying the respective haplotype, whereby no number after the population code refers to only one individual showing this haplotype.

70 Chapter 4: Adaptive divergence

Table 4.1. List of the collection sites (abbreviations as used throughout the article in brackets behind the name), their location (latitude, longitude) and the number of individuals (males/ females for the morphological studies) examined from each site in the different parts of the study.

tochrome Site Location Geometric morphometrics (field) Geometric morphometrics (lab reared) Gills Microsatellites Cy

Non-sulfidic surface habitats Arroyo Bonita (AB) 17.42706, -92.75194 15/ 29 0/ 2 24 10 Arroyo Cristal (AC) 17.45063, -92.76369 4/ 3 20 10 Arroyo Tacubaya (AA) 17.45355, -92.78449 14/ 24 Arroyo Tres (AT) 17.48368, -92.77627 0/ 18 0/ 1 19 10 Río Amatan (RA) 17.43331, -92.79293 26/ 29 0/ 3 24 11 Río Oxolotan (RO) 17.44444, -92.76293 23/ 20 8/ 5 0/ 1 24 9

Sulfidic surface habitats El Azufre I (EAI) 17.44225, -92.77447 21/ 28 0/ 2 40 20 El Azufre II (EAII) 17.43852, -92.77475 19/ 34 0/ 4 20 11

Non-sulfidic cave habitat 1 Cueva Luna Azufre (LA) 17.44171, -92.77312 23/ 43 5/ 4 0/ 6 19 10

Sulfidic cave habitat 1 Cueva del Azufre, chamber II (II) 17.44234, -92.77542 10/ 14 0/ 3 1 Cueva del Azufre, chamber V (V) 17.44234, -92.77542 26/ 27 0/ 3 19 21 1 Cueva del Azufre, chamber X (X) 17.44234, -92.77542 15/ 20 5/ 6 0/ 2 20 10 1 Cueva del Azufre, chamber XI (XI) 17.44234, -92.77542 6/ 7 19 10 1 Cueva del Azufre, chamber XIII (XIII) 17.44234, -92.77542 21 10 1Location of cave entrance is provided

Table 4.2. Distance matrix between sites in km. Distances were estimated by plotting the coordinates of the collection sites into GoogleEarth, and then measuring the river- distance between sites with the measurement tool based on satellite images. AB AC AT RA RO EA1 EA2 LA V X XI AB AC 3.36 AT 8.50 5.78 RA 9.85 7.13 8.27 RO 2.52 0.82 5.98 7.33 EAI 4.10 1.77 6.72 8.07 1.58 EAII 4.60 2.27 7.22 8.57 2.08 0.50 LA 4.04 1.72 6.66 8.01 1.52 0.30 0.80 V 4.26 1.93 6.88 8.23 1.74 0.16 0.66 0.46 X 4.36 2.03 6.98 8.33 1.84 0.26 0.76 0.56 0.10 XI 4.40 2.07 7.02 8.37 1.88 0.30 0.80 0.60 0.14 0.04 XIII 4.41 2.08 7.03 8.38 1.89 0.31 0.81 0.61 0.15 0.05 0.01

Chapter 4: Adaptive divergence 71

Table 4.3. Results of multivariate analyses of covariance (MANCOVA) examining body shape variation of P. mexicana from field collections (A) and using both field- collected and laboratory-reared animals (B). F-ratios were approximated using Wilks’ Lambda. (C) Analysis of covariance (ANCOVA) results examining the total gill filament length (TGFL) of P. mexicana from different habitat types. Effect sizes were 2 2 estimated with partial Eta squared (ηp ). Significant effects and ηp ≥0.2 are given in boldface.

2 Effect F df P ηp A. Geometric morphometrics: wild-caught fish centroid size 36.99 23, 466 <0.001 0.646 sex 230.88 23, 466 <0.001 0.919 H2S 29.22 23, 466 <0.001 0.591 light 69.48 23, 466 <0.001 0.774 sex × H2S 2.83 23, 466 <0.001 0.123 sex × light 2.78 23, 466 <0.001 0.121 H2S × light 18.42 23, 466 <0.001 0.476 sex × H2S × light 3.23 23, 466 <0.001 0.137

B. Geometric morphometrics: wild-caught and lab-reared fish centroid size 28.31 23, 393 <0.001 0.624 sex 67.66 23, 393 <0.001 0.798 habitat 11.67 46, 786 <0.001 0.406 treatment 5.69 23, 393 <0.001 0.250 sex × habitat 1.61 46, 786 0.007 0.086 sex × treatment 3.78 23, 393 <0.001 0.181 habitat × treatment 5.90 46, 786 <0.001 0.257 sex × habitat × treatment 1.26 46, 786 0.117 0.069

C. Gill morphometrics log(mass) 27.74 1 <0.001 0.545 H2S 33.76 1 <0.001 0.611 light 0.50 1 0.488 0.013 H2S × light 0.91 1 0.350 0.040

72 Chapter 4: Adaptive divergence

Table 4.4. Correlations of superimposed landmark coordinates with the shape gradient between fish from sulfidic and non-sulfidic habitats, surface and cave habitats as well as the interaction between the two environmental factors. Correlations ≥ |0.5| are given in bold.

Trait H2S light H2S × light X1 -0.417 0.298 -0.413 Y1 0.066 0.304 0.213 X2 -0.064 0.300 -0.506 Y2 -0.049 0.175 -0.180 X3 0.438 -0.101 0.072 Y3 0.424 -0.135 0.538 X4 -0.094 0.089 -0.051 Y4 -0.118 0.605 0.159 X5 -0.222 0.035 0.067 Y5 -0.021 0.645 -0.036 X6 0.182 -0.190 -0.311 Y6 0.140 0.279 -0.171 X7 0.204 -0.081 -0.358 Y7 -0.073 -0.087 -0.326 X8 -0.129 -0.037 0.226 Y8 -0.044 -0.686 0.137 X9 -0.104 -0.008 0.173 Y9 -0.061 -0.588 0.110 X10 -0.052 0.161 0.147 Y10 0.119 -0.523 -0.014 X11 0.561 -0.192 0.232 Y11 -0.546 -0.066 -0.561 X12 0.401 -0.021 0.143 Y12 0.371 0.199 0.278 X13 -0.302 0.007 -0.067 Y13 -0.077 -0.047 0.226 X14 -0.034 0.560 -0.556 Y14 -0.049 0.175 -0.180 X15 -0.034 0.560 -0.556 Y15 -0.049 0.175 -0.180

Chapter 4: Adaptive divergence 73

Table 4.5. Discriminant function analyses (DFA) of the morphology in P. mexicana from different habitat types. (A) Analysis of wild-caught individuals from all four habitat types. (B) Comparison between wild-caught and laboratory-reared individuals.

A. Wild-caught fish B. Wild-caught and lab-reared fish Function 1 Function 2 Function 3 Function 1 Function 2 X1 0.413 -0.363 0.143 -0.324 0.126 Y1 0.102 0.148 -0.108 -0.106 0.053 X2 0.049 -0.245 0.412 0.067 -0.175 Y2 0.117 -0.034 0.176 -0.180 -0.207 X3 0.300 -0.121 0.420 -0.276 -0.292 Y3 0.117 -0.034 0.176 -0.182 -0.207 X4 0.506 0.027 0.357 -0.542 -0.328 Y4 0.117 -0.034 0.176 -0.184 -0.207 X5 -0.216 0.322 0.086 0.275 -0.193 Y6 0.329 0.160 -0.277 0.320 0.043 X6 0.074 -0.052 0.060 0.009 -0.006 Y6 -0.256 0.375 -0.297 -0.328 0.135 X7 0.071 -0.106 -0.184 -0.014 0.236 Y7 0.365 0.179 -0.060 -0.375 0.021 X8 -0.127 -0.056 0.383 0.140 -0.279 Y8 0.189 0.159 0.151 -0.162 -0.311 X9 -0.059 -0.020 0.530 0.048 -0.459 Y9 0.038 -0.190 0.247 -0.024 -0.214 X10 -0.078 -0.041 -0.460 0.033 0.398 Y10 -0.419 -0.230 -0.048 0.372 0.246 X11 -0.034 -0.008 -0.271 -0.009 0.228 Y11 -0.320 -0.186 -0.045 0.303 0.248 X12 0.087 0.083 -0.106 -0.106 0.096 Y12 -0.271 -0.079 0.098 0.272 0.020 X13 -0.329 0.419 -0.017 0.398 -0.156 Y13 0.187 -0.559 0.302 -0.065 0.067 X14 -0.138 0.301 -0.004 0.171 -0.059 Y14 -0.051 0.371 -0.107 0.114 -0.090 X15 0.125 -0.259 -0.193 -0.115 0.289 Y15 -0.031 0.004 -0.160 -0.016 0.282

Canonical correlation 0.906 0.772 0.623 0.991 0.664 Eigenvalue 4.565 1.479 0.635 4.871 0.787 % Variance 68.3 22.1 9.5 86.1 13.9 Chi-square 1503.51 675.29 237.23 973.01 240.23 df 69 44 21 46 22 P <0.001 <0.001 <0.001 <0.001 <0.001

74 Chapter 4: Adaptive divergence

Table 4.6. Genetic diversity in surface- and cave-dwelling P. mexicana. For each population and locus, observed (HO) and expected (HE) heterozygosity and allelic richness (A) are given. Zero-values indicate that the locus is monomorphic in this population. Locus No. of Range of Test RA RO AB AC AT alleles allele size N=24 N=24 N=24 N =20 N=19 GAI29B 16 217-255 HO 0.42 0.46 0.17* 0.35 0.53 HE 0.49 0.53 0.23 0.31 0.51 A 8.00 6.38 4.94 5.60 7.74

GAIV42 48 180-466 HO 0.74 0.79 0.92 0.70 0.95 HE 0.90 0.90 0.92 0.90 0.90 A 16.47 16.21 17.62 16.08 20.16

GTII33 14 167-231 HO 0.48 0.75 0.63 0.30 0.53 HE 0.63 0.69 0.62 0.60 0.65 A 5.35 5.50 4.69 5.70 5.84

GAII41 13 122-150 HO 0.88 0.83 0.67 0.65 0.72 HE 0.69 0.64 0.58 0.66 0.65 A 6.25 6.19 6.18 5.70 5.00

GAI29A 18 137-259 HO 0.83 0.83 0.67 0.50 0.72 HE 0.66 0.72 0.66 0.53 0.71 A 6.99 8.12 7.63 3.80 9.00

GAV18 18 114-154 HO 0.82 0.88 0.96 0.90 0.79 HE 0.86 0.85 0.88 0.88 0.89 A 9.58 10.12 11.42 12.67 11.94

GTI49 12 130-168 HO 0.55 0.46 0.57 0.50 0.32 HE 0.71 0.55 0.67 0.64 0.36 A 5.82 4.98 7.64 6.69 3.95

GAI26 45 167-295 HO 0.91 0.83 0.96 0.75 0.74 HE 0.89 0.92 0.89 0.94 0.86 A 14.68 16.55 17.25 21.75 13.58

GAIII28 33 193-265 HO 0.73 0.92 0.86* 0.30* 0.68 HE 0.91 0.88 0.82 0.71 0.78 A 16.27 13.15 12.76 8.78 9.79

GTI13B 4 217-237 HO 0.21* 0.29 0.21 0.10 0.00 HE 0.26 0.32 0.26 0.41 0.00 A 3.87 3.00 2.99 2.90 1.00

Mean across HO 0.66 0.70 0.66 0.50 0.60 loci HE 0.70 0.70 0.65 0.66 0.63 A 9.32 9.02 9.31 8.97 8.80

Chapter 4: Adaptive divergence 75

Table 4.6. continued. Locus Test EA I EA II V X XI XIII LA Mean N=40 N=20 N=19 N=20 N=19 N=21 N=19 across populations GAI29B HO 0.25 0.65 0.00 0.20 0.15 0.00 0.00 0.26 HE 0.29 0.51 0.00 0.18 0.10 0.00 0.00 0.26 A 2.00 2.90 1.00 2.90 2.90 1.00 1.00 3.86

GAIV42 HO 0.53 0.70 0.58 0.65 0.42 0.52 0.11 0.63 HE 0.70 0.68 0.72 0.72 0.65 0.49 0.10 0.72 A 6.98 5.82 7.89 8.59 5.89 4.85 1.99 10.72

GTII33 HO 0.03 0.00 0.05 0.00 0.00 0.00 0.00 0.23 HE 0.03 0.00 0.05 0.00 0.00 0.00 0.00 0.27 A 1.45 1.00 1.95 1.00 1.00 1.00 1.00 2.96

GAII41 HO 0.53 0.50 0.58 0.05 0.21 0.14 0.21 0.50 HE 0.64 0.56 0.43 0.05 0.19 0.28 0.27 0.47 A 4.88 3.89 2.95 1.90 2.00 2.00 2.95 4.16

GAI29A HO 0.73 0.80 0.79 0.70 0.58 0.62 0.79 0.71 HE 0.77 0.74 0.68 0.65 0.57 0.52 0.56 0.65 A 5.45 5.00 4.90 6.60 3.95 2.86 3.95 5.69

GAV18 HO 0.43* 0.35* 0.42 0.35 0.47 0.52 0.47 0.61 HE 0.58 0.47 0.50 0.49 0.52 0.49 0.48 0.66 A 4.15 2.99 2.00 2.90 2.95 5.68 2.00 6.53

GTI49 HO 0.13 0.10 0.05 0.00 0.00 0.00 0.00 0.22 HE 0.12 0.10 0.05 0.00 0.00 0.00 0.00 0.27 A 3.05 1.99 1.95 1.00 1.00 1.00 1.00 3.34

GAI26 HO 0.35* 0.20* 0.21 0.80 0.05* 0.19* 0.32 0.53 HE 0.69 0.53 0.24 0.89 0.10 0.34 0.68 0.66 A 11.63 7.60 4.84 15.18 2.90 6.39 4.90 11.44

GAIII28 HO 0.63* 0.68 0.58 0.70 0.63 0.48 0.00 0.60 HE 0.89 0.87 0.74 0.82 0.74 0.54 0.00 0.72 A 13.69 11.79 10.73 10.59 10.73 9.84 1.00 10.76

GTI13B HO 0.03 0.00 0.11 0.00 0.00 0.00 0.00 0.08 HE 0.03 0.00 0.10 0.00 0.00 0.00 0.00 0.12 A 1.45 1.00 2.00 1.00 1.00 1.00 1.00 1.85

Mean HO 0.36 0.40 0.34 0.35 0.25 0.25 0.19 across loci HE 0.47 0.44 0.35 0.38 0.29 0.27 0.21 A 5.47 4.41 4.02 5.17 3.43 3.56 2.08 *indicates significant deviations from Hardy-Weinberg Equilibrium after Bonferroni adjustment at an experiment-wise error rate of α=0.05.

76 Chapter 4: Adaptive divergence

Table 4.7. Pair-wise genetic divergence (FST-values) among 12 populations of P. mexicana. Microsatellite data (below diagonal) and cytochrome b sequence data (above diagonal). Statistically significant values are shown in bold (α’=0.0007). Non- sulfidic surface habitats: RA, Río Amatan; RO, Río Oxolotan; AB, Arroyo Bonita; AC, Arroyo Cristal; AT, Arroyo Tres. Sulfidic surface habitats: EA I, El Azufre I; EA II, El Azufre II. Sulfidic cave: V-XIII, cave chambers V-XIII of the Cueva del Azufre. Non-sulfidic cave: LA, Cueva Luna Azufre. RA RO AB AC AT EA I EA II V X XI XIII LA RA - 0.160 0.196 0.130 0.075 0.332 0.473 0.387 0.368 0.326 0.196 0.239 RO 0.016 - 0.231 0.163 0.107 0.370 0.528 0.427 0.415 0.369 0.231 0.277 AB 0.023 0.001 - 0.200 0.144 0.397 0.552 0.453 0.444 0.400 0.267 0.311 AC 0.104 0.092 0.096 - 0.078 0.339 0.486 0.395 0.378 0.333 0.200 0.244 AT 0.034 0.011 0.015 0.137 - 0.289 0.431 0.345 0.322 0.278 0.144 0.189 EA I 0.156 0.181 0.215 0.286 0.183 - 0.620 0.533 0.541 0.507 0.397 0.435 EA II 0.159 0.170 0.217 0.287 0.185 0.027 - 0.668 0.724 0.681 0.552 0.595 V 0.230 0.248 0.278 0.346 0.254 0.059 0.127 - 0.594 0.560 0.453 0.489 X 0.214 0.233 0.271 0.336 0.230 0.068 0.149 0.108 - 0.578 0.444 0.489 XI 0.276 0.289 0.321 0.389 0.293 0.072 0.152 0.022 0.096 - 0.400 0.444 XIII 0.290 0.300 0.334 0.404 0.305 0.092 0.178 0.052 0.105 0.022 - 0.311 LA 0.318 0.322 0.359 0.425 0.325 0.175 0.274 0.234 0.168 0.229 0.198 -

Chapter 4: Adaptive divergence 77

Table 4.8. Mean (± standard deviation, SD) number of first-generation migrants between different habitat types (based on microsatellite data) as determined by

GENECLASS2. ΔH2S indicates a change in the presence of hydrogen sulfide; Δlight indicates a change in the presence of light. ΔH2S Δlight Mean±SD from sulfidic to non-sulfidic from cave to surface 0.00±0.00 from sulfidic to non-sulfidic no change 0.21±0.58 from sulfidic to non-sulfidic from surface to cave 0.00±0.00 no change from cave to surface 0.92±1.44 no change no change 2.88±2.83 no change from surface to cave 0.00±0.00 from non-sulfidic to sulfidic from cave to surface 0.50±0.71 from non-sulfidic to sulfidic no change 0.07±0.27 from non-sulfidic to sulfidic from surface to cave 0.00±0.00

78 Chapter 5: Trophic ecology

Chapter 5

DIVERGENCE IN TROPHIC ECOLOGY CHARACTERISES

4 COLONISATION OF EXTREME HABITATS

Michael Tobler

Extreme habitats are characterised by the presence of physio-chemical stressors, but also differ in aspects of the biotic environment, such as resource availability or the presence of competitors. This study quantifies variation in trophic ecology of a small livebearing fish (Poecilia mexicana, Poeciliidae) across four different habitats that included non-sulphidic and sulphidic surface waters as well as a non-sulphidic and a sulphidic cave. Resource use in different habitat types was investigated using gut content analysis. Populations diverged in resource use from a diet dominated by algae and detritus in non-sulfidic surface habitats to a diet including invertebrate food items in the other habitats. Poecilia mexicana in cave habitats further exhibited a higher dietary niche width than conspecifics from surface habitats. Condition of P. mexicana was analysed using storage lipid extractions. Fish from sulphidic and cave habitats exhibited a very poor condition suggesting resource limitation and/ or high costs of coping with extreme conditions. Finally, divergence in resource use was correlated with variation in viscerocranial morphology. A common garden experiment indicated both a genetic and plastic basis to the morphological variation observed among field populations. It is suggested that the morphological diversification is an adaptation to the differential use of resources among populations.

4 Published as: M. Tobler (in press): Divergence in trophic ecology characterises colonisation of extreme habitats. Biological Journal of the Linnean Society.

Chapter 5: Trophic ecology 79

Introduction Extreme environments are characterised by the presence of physio-chemical stressors that require, of any organism tolerating them, costly adaptations absent in most other species (Townsend et al., 2003). Usually, microbes are associated with extreme environments, and they have evolved a diversity of physiological and biochemical mechanisms to cope with adverse conditions (Rothschild and Mancinelli, 2001). Likewise, metazoans – even vertebrates – have colonised habitats characterised by extremes in temperature, pressure, salinity, oxygen, pH as well as the presence of toxicants (Howarth, 1993; Van Dover, 2000; Laybourn-Parry and Pearce, 2007; Weber et al., 2007). Studies on metazoans inhabiting extreme environments have primarily focused on mechanisms that allow organisms surviving and coping with the particular physiochemical stressor(s) (e.g., Van Dover, 2000; Bergman et al., 2003; Ip et al., 2004). Such adaptations may occur in biochemical and physiological pathways as well as in morphology, behaviour, life history strategies, and symbioses with other organisms. Extreme habitats, however, differ from adjacent non-extreme habitats not only in the presence or absence of physio-chemical stressors; the environmental gradient in abiotic stressors is correlated with a suite of ecological differences. For example, extreme habitats are usually less productive and have a lower species diversity (McMullin et al., 2000; Tsurumi, 2003; Tobler et al., 2008c), which may lead to differences in resource availability and quality, changes in competitive interactions within and between species, as well as changes in the exposure to predators. Adaptations to extreme environments may thus not only include traits directly involved in coping with a particular physio-chemical stressor, but also traits that evolved in response to correlated differences in the biotic environment. In this study, I compared the trophic ecology of the small livebearing fish, Poecilia mexicana Steindachner (Poeciliidae), that inhabits four different habitat types in the southern Mexican state of Tabasco. Habitat types are characterised by the presence or absence of hydrogen sulphide and light (caves); besides non-sulphidic surface waters, this species has colonised sulphidic surface habitats as well as a non- sulphidic and a sulphidic cave, which can be considered as extreme (Gordon and Rosen, 1962; Tobler et al., 2006; Tobler et al., 2008b). Sulphidic habitats are characterised by high concentrations (up to 300 µM) of dissolved hydrogen sulphide (H2S, Tobler et al., 2006). This respiratory toxicant is lethal to most metazoans even 80 Chapter 5: Trophic ecology in micromolar concentrations (Bagarinao, 1992; Grieshaber and Völkel, 1998). Since oxygen available for respiration is generally limited in sulphidic habitats due to the correlated hypoxic conditions (Tobler et al., 2006) and at the same time oxygen is required for coping with the toxic effects of H2S (Bagarinao, 1992; Ip et al., 2004; Plath et al., 2007c), organisms in sulphidic environments are selected for an efficient oxygen uptake. Likewise, the cave environment is challenging (Poulson and White, 1969; Howarth, 1993; Langecker, 2000), and diurnally active species like P. mexicana that predominantly rely on visual senses in normal surface habitats should are for coping with complete darkness (Plath et al., 2004). Besides the described differences in abiotic conditions, the four habitat types differ in a number of biotic environmental factors. The energy basis is likely to differ both qualitatively and quantitatively among habitats. Compared to normal surface habitats, photosynthetic primary production is absent in the caves (Poulson and

Lavoie, 2000) and probably reduced in sulphidic surface habitats, because H2S is also toxic for most algae (Bagarinao, 1992). H2S in turn allows for chemoautotrophic primary production by sulphide oxidising bacteria in sulphidic habitats (Nelson and Jannasch, 1983; Sarbu et al., 1996), and mats of sulphide-oxidising bacteria are present in the sulphidic surface and cave habitats (Hose et al., 2000). Finally, allochthonous input from terrestrial habitats is probably present in all habitat types, but dominated by leaf litter and terrestrial insects in surface habitats and guano from bat colonies inside the caves. Furthermore, habitat types differ in the presence of potential competitors and predators of P. mexicana. Compared to normal surface habitats, sulphidic and cave habitats exhibit a reduction in fish species diversity with P. mexicana being the predominant species (Tobler et al., 2006; Tobler et al., 2008b). Hence, not only are competitors for similar food sources rare; piscine predators that may affect foraging behaviour are also absent. Here, several aspects of the trophic ecology of P. mexicana from different habitat types were investigated. (1) Was the colonisation of extreme habitats accompanied by a divergence in resource use? In non-sulphidic surface habitats, P. mexicana feeds predominantly on detritus and algae (Darnell, 1962; Miller, 2005b). Differential resource use among populations was predicted both due to differences in resource availability and fish community structure. (2) Do P. mexicana from extreme habitats exhibit a wider dietary niche than conspecifics from non-sulphidic surface habitats? Ecological factors such as competition and predation are primary

Chapter 5: Trophic ecology 81 determinants of a population’s niche width (Begon et al., 1996). Competitive release due to a reduced species diversity in extreme habitats (MacArthur et al., 1972; Schluter and McPhail, 1992) or a shift from inter- to intraspecific competition (Svanbäck and Bolnick, 2007) as well as resource limitation (MacArthur and Pianka, 1966; Schoener, 1971; Fenolio et al., 2006) could lead to trophic niche expansion. (3) Do P. mexicana from different habitat types differ in the body condition? Survival in stressful environments is considered costly (Townsend et al., 2003). Short-term survival of P. mexicana in sulphidic water critically depends on energy availability and the potential to perform aquatic surface respiration, whereby fish skim the surface of the water where diffusion maintains a better-oxygenated layer (Plath et al., 2007c). Likewise, energy availability may vary among habitat types. I examined the condition of P. mexicana from different habitat types using extraction of storage lipids as a proxy of energy availability. (4) Did P. mexicana in different habitat types diverge in their trophic morphology? Because viscerocranial morphology frequently reflects trophic and dietary characteristics in fishes (Winemiller et al., 1995; Horstkotte and Strecker, 2005), skull morphology and intestinal tract length in Poecilia mexicana from different habitat types were compared. I also investigated whether differences in jaw morphology have a heritable basis by comparing fish from a common garden experiment with fish collected in the field. The trophic morphology in poeciliids has previously been shown to be phenotypically plastic (Robinson and Wilson, 1995; Ruehl and DeWitt, 2007).

Materials & methods Study system and fish collections Poecilia mexicana is a common fish species inhabiting freshwater habitats on the Atlantic versant of Central America from northern Mexico to Costa Rica (Miller, 2005b). All study sites were located near the village of Tapijulapa in the southern Mexican state of Tabasco. Four habitat types differing in their abiotic conditions were sampled (Figure 1.1). Habitat types were characterised by the presence or absence of

H2S (sulphidic or non-sulphidic) and/ or light (surface or cave), thus proving a natural 2×2 factorial design: • The Cueva del Azufre is a sulphidic cave. The cave is structured into different chambers, the nomenclature of which follows Gordon and Rosen (1962). The front chambers obtain some dim light, whereas the rearmost cave chambers are 82 Chapter 5: Trophic ecology

completely dark. The cave is drained by a creek fed by a number of springs occurring throughout the cave, most of which contain high levels of dissolved

H2S (Tobler et al., 2006). The presence of H2S allows for bacterial chemoautotrophic primary production (Hose et al., 2000). Additional energy input into the cave comes from bats that reside in different cave chambers and deposit considerable amounts of bat guano. Poecilia mexicana occur throughout the cave, and for this study, they were collected in chambers V and X. • The Cueva Luna Azufre is, despite its name, a non-sulphidic cave (Tobler et al., 2008b). The creek in the Cueva Luna Azufre is also fed by springs, however,

these do not contain H2S. The energy basis of this cave is thus thought to rely entirely on allochthonous input, especially in form of bat guano. Poecilia mexicana were collected south of the Entrada Marabunda (Tobler et al., 2008b). • The El Azufre is a sulphidic surface habitat originating in the hills southwest of the two caves and is fed by multiple independent sulphidic springs. Both caves drain into the El Azufre, which eventually joins the Río Oxolotan. Hydrogen sulphide concentrations are comparable to those in the Cueva del Azufre. The

presence of H2S allows for chemoautotrophic primary production, but photoautotrophic production and allochthonous input from terrestrial habitats is possible. Poecilia mexicana were collected near the uppermost sulphur springs as well as at the resurgence of the Cueva del Azufre. • Two non-sulphidic surface habitats were sampled. These habitats include a large river, the Río Oxolotan (most proximate to the other habitat types) and Arroyo Tacubaya, a non-sulphidic creek similar in size and structure to the El Azufre. The energy basis in these habitats is photoautotrophic primary production and allochthonous input from terrestrial habitats. Poecilia mexicana populations from the four habitat types are morphologically and genetically distinct, and there is striking reduction of gene flow between habitat types (but not different sites within habitat types) despite their spatial proximity and the lack of physical barriers (Plath et al., 2007a; Tobler et al., in press). Fish were caught in January 2006 and May 2007. Because habitat structures differed between sampling sites, different methods were employed. In the caves, where the water is very shallow and low ceilings preclude seining, fishes were caught with dip nets (13 x 14 cm, 1 mm mesh-width). In the other habitats, fish were caught using a seine (4 m long, 4 mm

Chapter 5: Trophic ecology 83 mesh-width). Fish were euthanised using MS222 immediately after capture and fixed in a 10% formaldehyde solution.

Gut content and condition analyses To test for differences in resource use and dietary niche width among habitat types, a gut content analysis was performed on P. mexicana collected in 2007. All individuals were measured for standard length (SL) to the nearest 0.1 mm and weighed to the closest 0.01 g (wet weight). Fish were dissected and entire digestive tracts were prepared. The content of the first quarter of the intestine was investigated under a stereomicroscope. All visible diet items were sorted and spread on a mm-square grid. Percent area for each food item was recorded. The area each diet item occupied was assumed to be proportional to its volume (Hellawell and Abel, 1971; Gido and Franssen, 2007). Items too small to identify or sort under the stereomicroscope were subsequently placed on a glass slide for examination under a light microscope. The relative proportion of each identifiable item on the slide was estimated and then multiplied by the area previously occupied on the mm-square grid. Diet categories used in the gut content analysis were adapted from Winemiller (1990), but for the final analysis, the following groups were recognised: detritus (fine and coarse particulate organic matter), algae (filamentous algae, coccal algae, and diatoms), plant parts (parts of aquatic and terrestrial macrophytes, and seeds), aquatic arthropods (chironomid larvae, ostracods, isopods, annelids, and nematodes), terrestrial arthropods (collembola, adult chironomids, keroplatid larvae, and ants), gastropods, bat guano (lepidopteran scales and insect parts), and sand. Insect parts were classified as bat guano when they were present in small fragments (P. mexicana is not able to reduce insects into smaller pieces), while all other insects found were left intact. Relative proportions of diet items were arcsine-squareroot-transformed and subjected to a principle component analysis (PCA). Three principal components accounting for 60.7% of the variance were used as dependent variables in a multivariate analysis of covariance (MANCOVA) with habitat type and sex as independent variables, and SL as covariate. The assumptions of normal distribution and homogeneities of variances and covariances were met for this analysis. F-ratios were approximated using Wilks’ Lambda values. The interaction terms were not significant (F9,491≤1.79; P≥0.07), thus only main effects were analysed. 84 Chapter 5: Trophic ecology

The dietary niche width of each individual was calculated using the inverse of 2 -1 Simpson’s (1949) diversity measure (see Pianka, 1973, 1986): β=(∑ pi ) , where p is the proportional utilisation of each dietary item i. Niche width values were log10- transformed to normalise distributions and then subjected to an analysis of covariance (ANCOVA) using habitat type and sex as independent variables, and SL as a covariate. The assumptions of normal distribution and homogeneities of variances were met for this analysis. The interaction terms were not significant (F3,204≤2.55; P≥0.06), thus only main effects were analysed. To assess body condition, fish carcasses (without visceral organs) were dried at 55° C for 7 days after the gut content analysis. Soluble non-structural fats were then extracted during four 24-hour extractions in petroleum ether (Meffe and Snelson,

1993; Heulett et al., 1995). The relative fat content ([Mbefore extraction-Mafter extraction]/Mbefore extraction) served as a proxy for individual body condition of the fish. Relative fat content values were arcsine-squareroot-transformed to approximate normality and analysed using an ANCOVA with habitat type and sex as independent variables, and SL as a covariate. The interaction terms were not significant

(F3,227≤1.83; P≥0.14), thus only main effects were analysed.

Morphological differentiation To test whether variation in resource use among habitats was correlated with changes in trophic morphology, skull morphology and intestinal tract length in individuals from different habitat types were compared. For the examination of the skull, eyes and visceral organs of formalin fixed fish collected in 2006 were removed, and specimens were cleared and stained as per Turner (1984). A digital picture of the dorsal side of the head was then taken using a Spot Insight digital camera mounted on an Olympus stereomicroscope. Skull traits were measured to the closest 0.01 mm using an image analysing program (Spot Advanced 3.5, Diagnostic Instruments). Six measurements of the skull were taken following Robinson and Wilson (1995), who investigated trophically induced morphological changes in a closely related species, Poecilia reticulata: (1) the length of the dentary, (2) the length and (3) thickness of the premaxilla, (4) the length of the longest tooth on the premaxilla, (5) the width of the skull at the anterior margins of the orbitals, and (6) the length of the snout. Variation in skull morphology among populations of P. mexicana was examined using a MANCOVA, in which the individual skull measurements were

Chapter 5: Trophic ecology 85 used as dependent variables, SL as a covariate (to control for allometry), and sex as well as habitat type as independent variables. The assumptions of normal distribution and homogeneities of variances were met for this analysis. Furthermore, a discriminant function analysis (DFA) was performed to test whether individuals were correctly assigned to the population of origin based on their skull morphology. To remove the effects of allometry and sex, a MANCOVA with the skull measurements as dependent variables, SL as covariate, and sex as independent variable was performed. The residuals of the MANCOVA were then used as independent variables in the DFA (consequently the classification was only dependent on differences among habitats). Previous studies have shown that trophic morphology in poeciliids can be phenotypically plastic (Robinson and Wilson, 1995; Ruehl and DeWitt, 2007). It was therefore tested whether differences in skull morphology have a heritable component by comparing laboratory stocks with wild-caught specimens from the original collection site. If morphological differences were entirely caused by environmentally induced phenotypic plasticity, differences among fish from different habitat types should disappear in the laboratory stocks housed under identical conditions. Likewise, if morphological differences were entirely determined by genetics, no differences between laboratory raised and wild caught individuals would be expected. An intermediate results suggests the traits under investigation have heritable basis, but phenotypic plasticity also plays a role. For this analysis, only three of the four habitat types were included since no laboratory stocks from sulphidic surface habitats were available. Laboratory stocks were founded in January 2006 and maintained as randomly out-bred populations in 1500-liter tanks in a greenhouse at the Aquatic Research Facility of the University of Oklahoma. All stocks were exposed to identical environmental conditions. Algae, detritus and invertebrates (amphipods) were present in the stock tanks, and fish were supplemented with commercially available flake food. A random sample of fish from stocks of each habitat type was collected in May 2007. At this point the stocks were established in the laboratory for about three to four generations. Laboratory raised fish were subjected to the same procedure as outlined above for wild-caught specimens. Data were analysed with the same approach using a MANCOVA, in which the individual skull measurements were used as dependent variables, SL as a covariate (to control for allometry), and sex, habitat type as well as treatment (i.e., wild-caught or 86 Chapter 5: Trophic ecology laboratory-reared) as independent variables. The assumptions of normal distribution and homogeneities of variances were met for this analysis. The three-way interaction between sex, habitat, and treatment was not significant (F12,224<1.77; P>0.05) and was thus omitted from the analysis. Also, a discriminant function analysis (DFA) was performed to test whether laboratory-reared individuals cluster with wild-caught specimens from their original habitat type. Again, a MANCOVA with the skull measurements as dependent variables, SL as covariate, and sex as independent variable was performed to remove the effects of allometry and sex. The residuals were then used as independent variables in the DFA. Intestinal tract length was measured in all specimens that were subjected to a gut content analysis. Intestines were uncoiled and the length was measured from the esophagus to the anus to the closest 1 mm (Kramer and Bryant, 1995b). Variation in intestinal tract length was examined using an ANCOVA with intestinal tract length as dependent variable as well as sex and habitat as independent variables. An individual’s mass rather than its size has been suggested to be used as the base for comparisons of intestinal tract lengths (Kramer and Bryant, 1995a), thus a mass-based covariate (10*mass1/3) adapted from the Zihler index (Zihler, 1982) was included.

Results Gut content and condition analyses A total of 251 individuals were examined for their gut contents (SL±SD; non- sulphidic surface: 35.6±7.7 mm; sulphidic surface: 30.4±4.5 mm; non-sulphidic cave: 24.8±6.4 mm; sulphidic cave: 33.5±6.2 mm). 38 individuals had empty intestines (0 of 58 from non-sulphidic surface habitats; 9 of 74 from sulphidic surface habitats; 2 of 41 from the non-sulphidic cave; 27 of 78 from the sulphidic cave). The frequency 2 of empty intestines differed significantly among habitat types (χ 3,251=37.24, P<0.001). Individuals with empty intestines were excluded from the principle component analysis and the analysis of individual dietary niche width. Poecilia mexicana from all habitat types ingested large amounts of detritus, and in the surface habitats, detritus was the primary dietary item (Table 5.1A). But whereas individuals from non-sulphidic surface habitats consumed algae, those from sulphidic surface habitats consumed aquatic arthropods. Besides detritus, P. mexicana from cave habitats foraged primarily on bat guano. Based on gut contents, individuals from the two caves differed primarily in the amount of arthropods (abundant in the

Chapter 5: Trophic ecology 87 sulphidic cave) as well as the amount of algae, plant parts, and gastropods consumed (abundant in the non-sulphidic cave, Table 5.1A). The MANCOVA using PCA scores as dependent variables (see Table 5.1B for the component matrix of the PCA) suggested that the resource use of P. mexicana in different habitat types was significantly different, but that neither sex nor size had a significant influence (Table 5.2A). The individual trophic niche width also differed among habitat types (Table 5.2B, Figure 5.1A). Post-hoc tests indicated that individuals from surface habitats had a narrower spectrum of prey items in their intestines than those from cave habitats, and that fish from the non-sulphidic cave had a higher diversity than those from the sulphidic cave. Poecilia mexicana from non-sulphidic habitats further had significantly more storage lipids than fish from the other habitat types (Table 5.1C, Figure 5.1B). Also sex had a significant effect on the lipid content as males had less storage lipids than females.

Morphological differentiation Poecilia mexicana from non-sulphidic surface habitats were generally characterized by smaller heads (shorter head width and snout length), narrower mouths (shorter premaxilla and dentaries), and thinner premaxilla than fish from other habitat types (Tables 5.3 & 5.4A). Fish from sulphidic surface habitats were characterised by having longer teeth than individuals from the other habitat types. In the discriminant function analysis (DFA), 67.7% of the individuals were correctly assigned to the habitat type of origin based on skull morphology (compared to a random expectation of 25%). The DFA also suggests that the skull morphologies of P. mexicana inhabiting sulphidic and/ or cave habitats are more similar to one another than to the morphology of fish from the non-sulphidic surface habitat (separation along Function 1: Figure 5.2A; Table 5.3). Although skull morphology seems to be phenotypically plastic to some degree (Tables 5.4B & 5.5B), variation among populations appears to have a heritable basis. Laboratory-reared individuals clustered close to field-collected specimens of the respective habitat type (Figure 5.2B & Table 5.3). Gut length was measured in 58 individuals from non-sulphidic habitats (28 females), 74 from sulphidic surface habitats (46 females), 41 from the non-sulphidic cave (21 females), and 78 from the sulphidic cave (44 females). Both of the factors 88 Chapter 5: Trophic ecology included in the analysis (sex and population) as well as their interaction and the mass- based covariate had a highly significant effect on the gut lengths of P. mexicana (Table 5.2D, Figure 5.3). The significant interaction term between population and sex was driven by specimens collected in non-sulphidic surface habitats, where males had shorter intestines than females. Post-hoc tests revealed that average intestine lengths were significantly different between individuals from divergent habitat types (estimated marginal means of gut length in non-sulphidic surface > non-sulphidic cave > sulphidic surface > sulphidic cave; LSD: P<0.001 in all cases).

Discussion In southern Mexico, P. mexicana colonised a set of different habitat types, characterised by the presence or absence of light and H2S. This provides an excellent system for studying the ecological and evolutionary consequences of life under stressful conditions. Previous studies have shown that P. mexicana in the different habitat types diverged phenotypically and genetically, thereby adapting to life in sulphidic and/ or cave habitats (Plath et al., 2004; Plath et al., 2007a; Tobler et al., in press). This study confirmed that the colonisation of extreme habitats is accompanied by ecological differences that are indirectly related to the abiotic conditions of the different habitats. In particular, this study provides evidence that the colonisation of different habitat types in the Cueva del Azufre system was associated with a divergence in resource use and the trophic morphology of these fish.

Differences in resource use and niche expansion Poecilia mexicana in non-sulphidic surface habitats primarily fed on detritus and algae, which is consistent with previous studies investigating the food habits of closely related species and P. mexicana in other parts of its range (Darnell, 1962; Winemiller, 1993; Kramer and Bryant, 1995a; Bussing, 1998; Miller, 2005b). The most pronounced difference in diet was observed between the non-sulphidic surface habitats where diet was dominated by algae/detritus, and the divergent habitats where invertebrates were consumed by P. mexicana. Especially in the cave habitats, invertebrates made up the majority of the gut content. A substantial amount of this invertebrate diet (about 50%), however, stems from bat guano, not living invertebrates. This finding is also consistent with a previous study that investigated the gut content of P. mexicana from the sulphidic cave only (Langecker et al., 1996).

Chapter 5: Trophic ecology 89

Together these studies suggest that colonisation of extreme habitats in P. mexicana was accompanied by a differences in resource use. Similar differences towards the incorporation of invertebrate prey into the diet was found in a Cyprinodon species flock (Horstkotte and Strecker, 2005). Several limitations preclude conclusions on the strength of resource use differences across habitats. Firstly, the habitat effect in all analyses reflects multiple individuals from one or two sites per habitat type, thus replication of individual habitat types is low. Secondly, differences in resource use among habitats may be lower during different seasons. Seasonal differences in resource use have been documented in the closely related species Poecilia gillii Kner (in non-sulphidic surface habitats), but season affected only the frequency of different food items (detritus vs. algae) ingested, i.e., there was no shift towards an incorporation of invertebrates into the diet (Winemiller, 1993). Thirdly, the examination of gut contents may have underestimated the differences in resource use because small diet items are not readily quantifiable with the methodology used. Specifically, it was not possible to quantitatively differentiate between detritus and bacteria. Qualitatively, it was evident that specimens collected in sulphidic habitat had white filaments resembling the mats of sulphide oxidising bacteria (see Hose et al., 2000) in their intestines, which were absent in specimens from non-sulphidic habitats. Consequently, the diet of P. mexicana from sulphidic and non-sulphidic habitats may differ more than this dataset suggests. Future investigations will use stable isotope analyses to elucidate the trophic structure of communities in the different habitat types (Peterson and Fry, 1987; Post, 2002). It is likely that both differences in resources availability and the competitive regime contributed to differences in food resource use among habitats. Indeed, specimens collected in different habitat types did not only differ in what they fed on, but also the diversity of food items present in their intestines; and fish from cave habitats exhibited a significantly higher diversity (i.e., a higher individual trophic niche width). It is unlikely that this increase in the trophic niche width was caused by a higher diversity in resources available since there is ample evidence that caves (and sulphidic habitats for that matter) exhibit a reduced species diversity compared to surface habitats (Poulson and White, 1969; Gibert and Deharveng, 2002). Niche expansion as a response to competitive release (MacArthur et al., 1972; Schluter and McPhail, 1992), increased intraspecific competition (Svanbäck and Bolnick, 2007), or 90 Chapter 5: Trophic ecology as a response to resource scarcity (MacArthur and Pianka, 1966; Schoener, 1971; Fenolio et al., 2006) may drive the increase in the diversity of resources used, but the exact mechanisms are unclear so far. All scenarios are not necessarily consistent with P. mexicana from sulphidic surface habitats having a low individual dietary niche width. At least in terms of fishes, the communities in the sulphidic surface habitats are comparatively deprived as those in the cave habitats (Tobler et al., 2006; Tobler et al., 2008c), and P. mexicana from sulphidic surface habitats also are in an equally poor condition as conspecifics from the cave habitats (this study). High individual dietary niche widths in the two cave habitats may also be explained by non-selective foraging. Poecilia mexicana is a diurnal species relying on visual senses (Plath et al., 2004), and although fish from the sulphidic cave have evolved the ability for non- visual communication in the context of sexual selection (Plath et al., 2004; Plath et al., 2006), the derived cave-inhabiting P. mexicana may not be able to forage selectively on specific diet items. For example, the foraging efficiency of P. mexicana from the two caves in darkness is equally low as the efficiency of surface populations in darkness (Tobler, unpublished data).

Nutritional condition and energy limitation Previous studies examining body conditions in this system (by comparing length- weight regressions) found P. mexicana from non-sulphidic surface habitats to exhibit the highest body condition, while cave populations had the lowest, and specimens from sulphidic surface habitats were intermediate (Tobler et al., 2006; Plath et al., 2007c; Tobler et al., 2008b). Morphological differences among populations, however, affected these results since populations differ in body height (Tobler et al., in press), and P. mexicana from sulphidic surface habitats have equally low amounts of storage lipids as fish from the cave populations. Fish from cave and sulphidic habitats may have a low condition for different reasons. Poecilia mexicana from the non-sulphidic cave exhibited low amounts of storage fats likely because resources are scarce. Caves relying on energy input from surface habitats are known to be energy limited (Streever, 1996; Hüppop, 2000; Poulson and Lavoie, 2000). Bat guano is thought to be the trophic base of cave food webs whenever bats are present and provide an energy-rich food base (Culver, 1982; Willis and Brown, 1985), but recent work indicates that this is not necessarily the case (Graening and Brown, 2003).

Chapter 5: Trophic ecology 91

Sulphidic habitats in turn have been suggested to be resource-rich (Langecker et al., 1996). The paradox of fish with low body condition living in an apparently resource-rich environment may be explained in two (not mutually exclusive) ways. (1) Although resource-rich, sulphidic habitats may lack particular nutrients for fish or provide an imbalanced diet, which may negatively affect condition (Jeyasingh, 2007). (2) Coping with the toxic environment may be energetically costly. Although the physiological mechanisms of sulphide-tolerance are not well understood in P. mexicana, detoxifying H2S has been shown to be energetically costly under hypoxic conditions in the mudskipper, Boleophthalmus boddaerti Pallas (Ip et al., 2004). Short-term survival of P. mexicana in sulphidic water is directly dependent on energy-availability and possibility to perform aquatic surface respiration (ASR), where fish exploit the oxygen-rich air-water-interface (Plath et al., 2007c). ASR itself is physiologically costly and constrains an individual’s energy budget, leaving less time for foraging (Kramer, 1983; Weber and Kramer, 1983; Chapman and Chapman, 1993). Reduced foraging activity in oxygen deprived environments also reduces body condition in an African cyprinid (Barrow and Chapman, 2006). Poecilia mexicana in sulphidic habitats thus seem to be living in a resource-rich habitat but paying a cost for coping with the toxic conditions. The high resource availability in these habitats may, in fact, be one of the factors making life under such extreme environments possible at all. Future studies will need to examine the nutritional value of the food items ingested by P. mexicana as well as the costs of coping with the stressful conditions in the different habitat types.

Differentiation in trophic morphology The major difference in trophic morphology among populations living in different habitat types was found between P. mexicana from non-sulphidic surface habitats and those from extreme habitats, which parallels the major dietary differences with incorporation of invertebrates into the diet. Intestinal tract lengths in fishes are typically correlated with the amount of plant material ingested (Kramer and Bryant, 1995a), and P. mexicana from extreme habitats (which consume less plant material) had shorter intestinal tracts. They were also characterised by wider and thicker jaws, which may be advantageous in handling larger (Wainwright, 1996) and/ or more evasive prey items (Hulsey and Garcia de Leon, 2005; Higham et al., 2007). 92 Chapter 5: Trophic ecology

Previous studies on P. mexicana body morphology using geometric morphometric analyses did not find variation that was obviously related to trophic ecology (Tobler et al., 2008b; Tobler et al., in press). Differentiation in skull morphology among populations inhabiting different habitat types is also less pronounced than the differentiation in general body shape, which seems to be driven predominantly by abiotic environmental factors, i.e., the lack of light (eye size reduction) as well as the presence of H2S and hypoxia (increase in head and gill size, Tobler et al., in press). Thus, differences in jaw morphology as reported in this study are not simply explained by a correlated response to selection on other characteristics of the head (Chapman et al., 2000), but may actually be adaptive to differential use of resources among populations. The finding that skull morphology is at least partly determined by genetics supports the idea that evolutionary divergence in skull morphology by natural selection is possible.

Chapter 5: Trophic ecology 93

Figures and tables chapter 5

Figure 5.1. (A) Mean (± standard deviation) individual trophic niche width (β) of males and females from the different habitat types. (B) Relative fat content (± standard deviation) of males and females from the different habitat types. Pair-wise post-hoc tests (LSD, α < 0.05) revealed which populations differed as labeled by Greek letters.

94 Chapter 5: Trophic ecology

Figure 5.2. Discriminant function plots where functions 1 and 2 correspond to the discriminant functions from the analyses presented in Table 5.5A and 5.5B, respectively. Depicted are the mean (± standard deviation) discriminant function scores for each group. (A) Analysis of wild-caught individuals from all four habitat types. (B) Comparison between wild-caught and laboratory-reared individuals. Non- sulfidic surface habitats (), sulfidic surface habitats (), non-sulfidic cave (), and sulfidic cave (). Closed symbols represent wild-caught individuals, open symbols laboratory-reared individuals.

Chapter 5: Trophic ecology 95

Figure 5.3. Gut lengths in males and females of P. mexicana from different habitat types. Non-sulfidic surface habitats (), sulfidic surface habitats (), non-sulfidic cave (), and sulfidic cave (). Closed symbols represent males, open symbols females.

Table 5.1. (A) Proportions of dietary items averaged across individuals of each habitat type (including the sample size). (B) Component matrix of the principal component analysis on the proportion of food items in the guts. Axis loadings for the first three principle components are given (including the percent variation explained by each).

(A) Gut content (B) PC axis loadings surface surface cave cave 1 2 3 no H2S H2S no H2S H2S N / % variation explained 58 65 39 51 26.6 18.7 15.4 Detritus 0.71 0.70 0.29 0.29 -0.861 0.263 -0.020 Algae 0.12 0.00 0.09 0.00 0.036 0.746 -0.203 Plant parts 0.00 0.01 0.08 0.00 0.501 0.403 0.498 Aquatic arthropods 0.00 0.16 0.09 0.19 0.332 -0.578 0.450 Terrestrial arthropods 0.00 0.02 0.01 0.15 0.369 -0.315 -0.519 Gastropods 0.00 0.00 0.08 0.01 0.430 0.370 0.465 Bat guano 0.00 0.01 0.32 0.27 0.634 -0.081 -0.310 Sand 0.17 0.10 0.04 0.09 -0.549 -0.362 0.401

96 Chapter 5: Trophic ecology

Table 5.2. Multivariate analysis of covariance (MANCOVA) results examining gut contents of P. mexicana from different habitat types (A). Analysis of covariance (ANCOVA) results examining the dietary niche width (B), relative fat content (C), and gut length (D) of P. mexicana from different habitat types. Significant effects are bold. Effect F df P (A) Gut content SL 0.31 3, 499 0.817 Sex 2.19 3, 205 0.091 Habitat 40.68 9, 205 <0.001

(B) Niche width SL 2.78 1 0.098 Sex 0.74 1 0.390 Habitat 16.03 3 <0.001

(D) Gut length Mass 218.04 1 <0.001 Habitat 237.88 3 <0.001 Sex 24.49 1 <0.001 Habitat*Sex 47.98 3 <0.001

Table 5.3. Measurements of skull traits (mean ± standard deviation) in males and females of P. mexicana from different habitat types. Note that laboratory reared specimens from sulfidic surface habitats were not available during this study. Habitat Origin Sex N Standard Head Snout Dentary Premaxilla Premaxilla Tooth length width length length length thickness length surface Field ♀ 13 36.77±7.45 4.91±1.01 2.05±0.51 3.19±0.76 1.39±0.32 0.84±0.19 0.33±0.08 no H2S ♂ 17 25.00±2.81 2.98±0.35 1.36±0.24 1.95±0.26 0.89±0.16 0.56±0.07 0.22±0.04 Lab ♀ 5 38.40±3.78 5.12±0.55 2.71±0.44 3.63±0.47 1.57±0.24 0.88±0.13 0.30±0.08 ♂ 7 25.71±6.40 3.17±0.79 1.63±0.71 2.27±0.64 0.99±0.30 0.60±0.20 0.29±0.09 surface Field ♀ 22 32.05±5.38 4.77±0.86 2.32±0.45 3.28±0.68 1.44±0.29 0.87±0.16 0.33±0.07 H2S ♂ 14 28.50±3.77 3.91±0.60 2.03±0.42 2.76±0.48 1.22±0.22 0.74±0.13 0.28±0.04 cave Field ♀ 28 36.29±5.28 5.59±0.78 2.57±0.39 3.63±0.51 1.61±0.25 0.97±0.16 0.32±0.04 no H2S ♂ 4 24.00±1.41 3.24±0.19 1.71±0.26 2.06±0.06 0.88±0.04 0.60±0.08 0.23±0.02 Lab ♀ 8 26.50±3.21 4.08±0.52 1.75±0.37 2.30±0.33 1.06±0.15 0.68±0.11 0.28±0.06 ♂ 4 25.25±1.71 3.50±0.25 1.56±0.30 2.11±0.21 0.99±0.14 0.61±0.03 0.22±0.02 cave Field ♀ 19 36.00±6.86 5.75±1.02 2.41±0.56 3.80±0.75 1.66±0.33 0.99±0.17 0.30±0.08 H2S ♂ 16 28.94±4.30 4.00±0.72 2.00±0.54 2.73±0.58 1.18±0.24 0.79±0.18 0.26±0.07 Lab ♀ 6 30.67±4.84 4.87±0.88 2.10±0.31 2.66±0.44 1.16±0.16 0.74±0.13 0.32±0.07 ♂ 3 33.67±2.08 4.30±1.13 1.98±0.54 2.60±0.84 1.14±0.37 0.72±0.25 0.33±0.15

Chapter 5: Trophic ecology 97

Table 5.4. Multivariate analysis of covariance (MANCOVA) results examining the skull morphology of P. mexicana from different habitat types. For the tests of between subject effects, α-values were adjusted for multiple testing (α’=0.008). Significant effects are bold. (A) Analysis of wild-caught individuals from all four habitat types. (B) Comparison between wild-caught and laboratory-reared individuals. Due to the lack of laboratory stocks from sulfidic surface habitats, no specimens of this habitat types were included in this analysis. H=habitat, S=sex, and T=treatment. Multivariate tests Tests of between subject effects Dentary length Premaxilla length Premaxilla thickness (A) Comparison among wild-caught fish of different habitat types SL F6,119=182.08, P<0.001 F1,124=514.67, P<0.001 F1,124=596.59, P<0.001 F1,124=238.19, P<0.001 Habitat F18,337=9.76, P<0.001 F3,124=28.21, P<0.001 F3,124=26.19, P<0.001 F3,124=15.13, P<0.001 Sex F6,119=12.41, P<0.001 F1,124=12.72, P<0.001 F1,124=11.02, P<0.001 F1,124=2.53, P=0.115 H*S F18,337=1.92, P<0.001 F3,124=2.28,P=0.083 F3,124=4.34,P=0.006 F3,124=0.57,P=0.639

(B) Comparison between wild-caught and laboratory raised fish SL F6,114=157.03, P<0.001 F1,117=447.22, P<0.001 F1,117=487.23, P<0.001 F1,117=194.13, P<0.001 Habitat F2,228=7.90, P<0.001 F2,117=1.75, P=0.179 F2,117=3.50, P=0.033 F2,117=5.84, P=0.004 Sex F6,114=12.61, P<0.001 F1,117=12.36, P=0.001 F1,117=8.11, P=0.005 F1,117=2.14, P=0.146 Treatment F6.114=5.186, P<0.001 F1,117=13.82, P<0.001 F1,117=10.67, P=0.001 F1,117=7.70, P=0.006 H*S F12,228=2.24, P=0.011 F2,117=2.44, P=0.091 F2,117=4.82, P=0.010 F2,117=0.93, P=0.398 H*T F12,228=5.43, P<0.001 F2,117=20.31, P<0.001 F2,117=15.57, P<0.001 F2,117=4.84, P=0.010 S*T F6,114=1.01, P=0.424 F1,117=0.61, P=0.436 F1,117=0.63, P=0.430 F1,117=0.01, P=0.907

98 Chapter 5: Trophic ecology

Table 5.4. continued. Tests of between subject effects Tooth length Snout length Head width (A) Comparison among wild-caught fish of different habitat types SL F1,124=99.74, P<0.001 F1,124=192.22, P<0.001 F1,124=1025.73, P<0.001 Habitat F3,124=5.34, P=0.002 F3,124=20.03, P<0.001 F3,124=42.72, P<0.001 Sex F1,124=0.13, P=0.720 F1,124=0.05, P=0.829 F1,124=60.75, P<0.001 H*S F3,124=0.50,P=0.682 F3,124=0.51,P=0.676 F3,124=5.72,P=0.001

(B) Comparison between wild-caught and laboratory raised fish SL F1,117=75.10, P<0.001 F1,117=179.45, P<0.001 F1,117=855.22, P<0.001 Habitat F2,117=0.47, P=0.626 F2,117=3.64, P=0.029 F2,117=43.00, P<0.001 Sex F1,117=0.10, P=0.755 F1,117=0.89, P=0.348 F1,117=61.09, P<0.001 Treatment F1,117=9.19, P=0.003 F1,117=0.00, P=0.996 F1,117=2.14, P=0.146 H*S F2,117=0.82, P=0.443 F2,117=0.07, P=0.936 F2,117=9.24, P<0.001 H*T F2,117=0.92, P=0.403 F2,117=9.51, P<0.001 F2,117=1.73, P=0.723 S*T F1,117=0.078, P=0.378 F1,117=4.54, P=0.035 F1,117=0.13, P=0.723

Chapter 5: Trophic ecology 99

Table 5.5. Discriminant function analyses of the skull morphology in P. mexicana from different habitat types. (A) Analysis of wild-caught individuals from all four habitat types. (B) Comparison between wild-caught and laboratory-reared individuals. Significant effects are bold. (A) Wild-caught fish (B) Wild-caught and lab-reared fish Function 1 Function 2 Function 3 Function 1 Function 2 Canonical loadings Dentary length 0.754 0.193 0.592 0.430 -0.278 Premaxilla length 0.701 0.253 0.362 0.417 0.126 Premaxilla thickness 0.587 -0.021 0.252 0.480 0.024 Tooth length 0.045 0.567 0.051 -0.004 0.120 Snout length 0.636 0.339 -0.535 0.396 0.558 Head width 0.911 -0.288 0.046 0.930 -0.042

Canonical correlation 0.706 0.553 0.268 0.705 0.248 Eigenvalue 0.992 0.442 0.078 0.998 0.065 % variance 65.6 29.2 5.1 93.8 6.2 Chi-square 143.494 55.949 9.497 93.414 7.880 df 18 10 4 12 5 P <0.001 <0.001 0.050 <0.001 0.163

100 Chapter 6: Belostoma predation

Chapter 6

PREDATION OF A CAVE FISH (POECILIA MEXICANA,

POECILIIDAE) BY A GIANT WATER-BUG (BELOSTOMA, 5 BELOSTOMATIDAE) IN A MEXICAN SULFUR CAVE

Michael Tobler, Ingo Schlupp and Martin Plath

Abstract 1. Caves are often assumed to be predator-free environments for cavefishes. This has been proposed to be a potential benefit of colonizing these otherwise hostile environments. In order to test this hypothesis, the predator-prey interaction of a belostomatid (predator) and a cavefish (prey) occurring in the Cueva del Azufre (Tabasco, Mexico) was investigated with two separate experiments. 2. In one experiment, individual Belostoma were given a chance to prey on a cave fish, the cave form of the Atlantic molly (Poecilia mexicana), to estimate feeding rates and size-specific prey preferences of the predator. In the other experiment, population density of Belostoma was estimated using a mark-recapture analysis in one of the cave chambers. 3. Belostomatids were found to heavily prey on cave mollies and to exhibit a prey preference for large fish. The mark-recapture analysis revealed a high population density of the heteropterans in the cave. 4. The absence of predators is not a general habitat feature of cavefishes. Nonetheless predation regimes differ strikingly between epigean and hypogean habitats. The prey preference of Belostoma indicates that cave-dwelling P. mexicana experience size- specific predation pressure comparable to surface populations, which may have implications for life history evolution in this cavefish.

5 Published as: M. Tobler, I. Schlupp & M. Plath (2007): Predation of a cave fish (Poecilia mexicana, Poeciliidae) by a giant water-bug (Belostoma, Belostomatidae) in a Mexican sulfur cave. Ecological Entomology 32: 492-495.

Chapter 6: Belostoma predation 101

Introduction Due to the absence of photoautotrophic primary production, caves are usually considered energy-limited, and the food-web of subterranean ecosystems relies on nutrient-influx from epigean habitats (Poulson and Lavoie, 2000). As a consequence, caves often harbor comparatively simple food webs consisting of few specialized cave-dwelling species. Specializations of cave animals not only include traits that have evolved in response to permanent darkness itself (such as a reduction of pigmentation and the visual system and the elaboration of non-visual sensory organs) but also include adaptations enabling cave-dwellers to cope with the food scarcity commonly found in cave ecosystems, like an increased starvation resistance or reduced energy demands as a result of reduced metabolic rates (Hüppop, 2000; Langecker, 2000). The absence of light and the associated scarcity of food render caves into rather extreme environments (Howarth, 1993). Consequently, cave colonization is often viewed as an accidental process (Wilkens, 1979; Holsinger, 2000). However, cave colonization may also be an active and adaptive process providing specific benefits to cave colonizers that range from environmental stability and the presence of unoccupied niches (Romero and Green, 2005) to protection from predators (Romero and Green, 2005) and parasites (Tobler et al., 2007a). In the Cueva del Azufre in Tabasco, southern Mexico, the food web is based on chemoautotrophic bacterial primary production and the input of bat guano (Langecker et al., 1996). This cave is thought to be different from many other cave systems in that its ecosystem is energy-rich even compared to photoautotrophic epigean habitats (Langecker et al., 1996; but see Tobler et al., 2006). The most studied inhabitant of the Cueva del Azufre is a small livebearing fish, the cave molly (Gordon and Rosen, 1962; Parzefall, 2001), a cave-dwelling population of the Atlantic molly (Poecilia mexicana Steindachner). Other populations of the Atlantic molly frequently inhabit freshwater surface habitats in Central America (Miller, 2005b). Contrary to many other cavefishes, cave mollies occur at high densities. On average, twenty individuals per square-meter were recorded in the inner cave chamber X (Tobler et al., 2006), but densities are even higher towards the cave exit (chambers III-VI; Tobler et al., personal observation). High population densities are not explained by abundant food, because cave mollies show strong signs of malnutrition (Plath et al., 2005; Tobler et al., 2006). 102 Chapter 6: Belostoma predation

Compared to adjacent surface habitats, the Cueva del Azufre harbors a tremendously reduced fish community (Tobler et al., 2006). Besides the cave molly, only the synbranchid eel Ophisternon aenigmaticum Rosen & Greenwood was occasionally reported from the cave (Gordon and Rosen, 1962). The reduction of species diversity in this case seems not primarily be driven by the absence of light or food scarcity but by the presence of high concentrations of toxic hydrogen sulfide (Tobler et al., 2006). Consequently, interspecific competition with other fishes is reduced and predatory fish are lacking in the cave. This is also true for avian predators that are known to heavily prey on fishes (Trexler et al., 1994) and that are common in adjacent surface habitats (Tobler et al., 2007a). So far unstudied is the extent of filial cannibalism in the system; other poeciliid fishes are known to prey upon conspecific juveniles (Nesbit and Meffe, 1993). In the present study, it is tested if cave mollies live in a predator-free environment. Generally, few other metazoans share their habitat with the cave molly (Gordon and Rosen, 1962). These include larvae of the dipteran Tendipes fulvipilus Rempel (Tendipedidae) that are one of the primary food sources for the mollies (Tobler, personal observation) and the crab Avotrichodactylus bidens Bott (Trichodactylidae; referred to as a potamonid species by Gordon & Rosen, 1962). While the cave molly’s ecology and behavior are intensely studied (Parzefall, 1993, 2001; Plath et al., 2003a; Plath et al., 2004; Tobler et al., 2006), very limited information on the other inhabitants of the Cueva del Azufre is available so far. This paper focuses on a further inhabitant of the cave, a giant water-bug of the genus Belostoma, and its role as a molly predator. Belostoma are large aquatic hemipterans that prey on aquatic insects, snails, amphibians, and fish (Menke, 1979). Belostomatids are sit-and-wait predators that catch bypassing prey items with their raptorial forelegs that are strongly incrassate, with the femora often grooved to accept the tibiae. Upon capture, Belostoma inject toxins causing prey paralysis and digestive enzymes causing tissue necrosis (Swart and Felgenhauer, 2003). In the Cueva del Azufre, Belostoma has been found to prey on cave mollies and, with a limited sample size of eight individuals, to prefer large prey items over small ones (Plath et al., 2003a). Predation itself and size-specific predation in particular are known to affect life history traits in surface-dwelling fish. For example, early maturity and small body size at maturity in the prey species are positively selected under (size-specific)

Chapter 6: Belostoma predation 103 predation (Reznick and Endler, 1982; Johnson and Belk, 2001). Based on the finding that size distributions in cave molly populations appear not to differ from that of typical surface populations (Plath et al., 2003a) it was tested if a preference for large prey size could be detected in Belostoma. Furthermore, the population size of belostomatids in one cave chamber of the Cueva del Azufre was assessed. In summary, we tested whether cave mollies live in a predator free environment by examining if (1) Belostoma prey on cave mollies; (2) the water-bugs prefer large fish as prey and (3) the water-bugs occur at high density and thus are a potential selective factor in the evolution of cave mollies comparable to selection arising from avian and piscine predation in surface habitats.

Material and Methods All experiments were performed in the Cueva del Azufre, Tabasco, Mexico in August 2004. Nomenclature of cave chambers followed Gordon & Rosen (1962). Details about the abiotic habitat properties can be found in Tobler et al. (2006).

Population size of Belostoma To estimate population sizes, all visible Belostoma were collected by two people in cave chamber V during 30 minutes and were marked with a small dot of TippEx on the thorax. All marked individuals were then released at the collection site. Twenty four hours later, cave chamber V was re-sampled with the same effort. Population size was estimated based on the recapture rate of marked individuals following Bailey (Bailey, 1951):

N1 "(N 2 +1) Ntot = , N 3 +1 where Ntot is the estimated population size, N1 is the number of individuals initially caught and marked, N2 is the number of individuals caught after 24 h, and N3 is the number of marked individuals in N2.

Prey choice in Belostoma Fish and water-bugs for the prey choice experiments were collected in cave chamber V of the Cueva del Azufre. Prey choice experiments were performed in PET bottles (2 liters) that were perforated to allow exchange of water and air with the environment. Four large (standard length, mean ± SD: 34.8 ± 0.5 mm) and four small cave mollies 104 Chapter 6: Belostoma predation

(SL: 23.4 ± 0.6 mm) as well as a single individual of Belostoma (length from the tip of the head to the tip of the abdomen, mean ± SD: 22.5 ± 0.9 mm) were introduced into each bottle. Furthermore, a tablet of commercial fish food was added to each bottle. The bottles (N=20) were then partially submerged in a shallow area of cave chamber V and fixed with rocks. Partial submersion was used to allow water-bugs to breathe. The bottles remained in the cave for 48 h and were then checked for fish preyed upon by Belostoma. The numbers of large and small cave mollies consumed by Belostoma were compared with a Wilcoxon signed rank test using SPSS 11, SPSS Inc.

Results Population size of Belostoma Of 35 Belostoma marked, four could be recaptured 24 h later. The total number of individuals caught after 24 h was 47. Thus, the estimated population size of Belostoma in chamber V of the Cueva del Azufre was 336 ± 130 (mean ± SE) individuals. Given a conservative estimate of the surface area of chamber V as 300 m2, this results in a density of 1.12 ± 0.43 Belostoma/m2.

Prey choice in Belostoma All but three Belostoma (two of which were males carrying eggs) consumed at least one molly. In total, 70 out of 160 mollies (44 %) were consumed by the belostomatids, equaling a per capitum capture rate of 1.75 mollies per water-bug per day. Mollies showed injuries at the tail, the body, as well as various parts of the head. An analysis of the captured fish revealed that the water-bugs consumed significantly more large mollies [median=2 individuals per trial (IQR=3.25)] than small mollies [median=1 individual per trial (IQR=1.25); N=20, Z=-2.355, P=0.019] during the period of 48 h.

Discussion Belostoma are common in the Cueva del Azufre. Although a density of about one individual/m2 may seem low, it has to be considered that these insects tend to accumulate on rocks along the water surface where they ambush prey with their front legs kept in the water and their abdomen in the air. Cave mollies can often be observed in the same microhabitat during foraging (midge larvae tend to accumulate

Chapter 6: Belostoma predation 105 along the rims of the water surface) and during aquatic surface respiration that allows survival in the sulfidic and hypoxic environment (Plath et al., 2007c). Thus, encounter rates between predators and prey may in fact be high. Protection from predation has been considered a potential benefit of cave colonization (Romero and Green, 2005); this, however, is not entirely the case for the Cueva del Azufre. The predation experiments indicated that may Belostoma prey heavily on cave mollies. The high capture rate may be partly caused by the spatial limitations of the bottles in which the experiments were performed. The small container might have limited the prey’s escape possibilities. The results, however, show that belostomatids are capable of preying on mollies multiple times over a period of 48 hours. Previous studies showed that Belostoma are able to feed multiply over prolonged periods of time without a depletion of salivary enzymes (Swart and Felgenhauer, 2003). Our experiment furthermore indicates that Belostoma have a preference for larger prey items. This prey preference has previously been hypothesized to maintain the striking size polymorphism present in male cave mollies (Plath et al., 2003a), even though large males are preferred in sexual selection (Plath et al., 2004). Size-specific predation has also been documented in surface habitats; e.g. herons preferentially prey on large mollies (Trexler et al., 1994) and predatory cichlids preferably select large guppies as prey (Johansson et al., 2004). Because belostomatids occur at high densities and are generalist predators, they have been considered keystone species structuring the communities in which they are found and affect morphological, behavioral, and life history modifications in their prey (Kehr and Schnack, 1991; Babbitt and Jordan, 1996; Chase, 1999). Certainly, Belostoma is the main, if not the only, predator of the cave molly. Other potential predators in the cave are either very rare (the synbranchid eel, Ophisternon aenigmaticum; Parzefall, personal communication) or did not prove to prey on cave mollies in comparable experiments (e.g., the crab, Avotrichodactylus bidens; Tobler et al., unpublished data). It is yet unclear to what extent Belostoma in the Cueva del Azufre rely on other prey species (e.g. midget larvae). In our experiment, the water bugs only had a chance to forage on cave mollies, but they may prefer other prey in a more natural situation. In that case, the predation pressure on cave mollies may be lower than implied by our experiment. Future work on the food web structure in the cave and prey preferences in Belostoma needs to clarify this. 106 Chapter 6: Belostoma predation

Definitely, the kind of predation on Poecilia mexicana differs strikingly between different habitat types. In adjacent surface habitats, Belostoma were only recorded sporadically while piscivorous fish and birds are common (Tobler et al., 2006; Tobler et al., 2007a). Beside darkness, hydrogen sulfide, and energy-limitation, the differential predatory regime is one potential driving force in the evolution of the cave molly. Inhabiting a cave – albeit rich in toxic hydrogen sulfide – appears to provide benefits in terms of protection from predators for Belostoma, which is the top predator in this cave ecosystem, but not necessarily for the cave fish, Poecilia mexicana.

Chapter 7: Parasite refuge hypothesis 107

Chapter 7

EXTREME HABITATS AS REFUGE FROM PARASITE 6 INFECTIONS? EVIDENCE FROM AN EXTREMOPHILE FISH

Michael Tobler, Ingo Schlupp, Francisco J. García de León, Matthias Glaubrecht and Martin Plath

Abstract Living in extreme habitats typically requires costly adaptations of any organism tolerating these conditions, but very little is known about potential benefits that trade off these costs. We suggest that extreme habitats may function as refuge from parasite infections, since parasites can become locally extinct either directly, through selection by an extreme environmental parameter on free-living parasite stages, or indirectly, through selection on other host species involved in its life cycle. We tested this hypothesis in a small freshwater fish, the Atlantic molly (Poecilia mexicana) that inhabits normal freshwaters as well as extreme habitats containing high concentrations of toxic hydrogen sulfide. Populations from such extreme habitats are significantly less parasitized by the trematode Uvulifer sp. than a population from a non-sulfidic habitat. We suggest that reduced parasite prevalence may be a benefit of living in sulfidic habitats.

6 Published as: M. Tobler, I. Schlupp, F. J. García de León, M. Glaubrecht & M. Plath (2007): Extreme habitats as refuge from parasite infections? Evidence from an extremophile fish. Acta Oecologica 31: 270-275.

108 Chapter 7: Parasite refuge hypothesis

Introduction Living under extreme environmental conditions is usually associated with costs, however very little is known about potential benefits. Townsend et al. (2003) point this out by defining an extreme environmental condition as one that requires, of any organism tolerating it, costly adaptations absent in most related species. These can include changes in morphology and physiological pathways that allow coping with a physiochemical stressor, as well as behavioral adaptations and shifts in life history strategies. But why do organisms colonize extreme habitats if there are immediate costs ranging from the investment towards specific adaptations to an increased risk of death? Given the associated costs, it could be argued that the organisms become trapped in an extreme habitat. However, many extreme habitats are not isolated but contiguous, and no permanent discontinuity prevents organisms from returning to their original habitat. Colonizing extreme habitats would be directly advantageous when individuals that have the ability to cope with the extreme environment confer an advantage compared to relatives living in non-extreme habitats. Selection favoring individuals with adaptations will lead to adaptive shifts within populations as colonizers of extreme habitats exploit new and unused niches (Romero and Green, 2005). In either way, if organisms persist in extreme habitats over evolutionary time scales, resource investment into the costly adaptations that allow for survival must be traded off. Thus, organisms living in extreme habitats have to invest more into specific adaptations, but they also may maintain or even increase their fitness compared to adjacent populations in non-extreme habitats. Since extreme habitats often harbor impoverished biocoenoses (Begon et al., 1996), the advantages of living in extreme habitats may include the reduction of competition and predation and the exploitation of new niches (Romero and Green, 2005), however, very few tests of these ideas have thus far been published. Another potential advantage of living in an extreme habitat that has received no attention so far is that such habitats may act as “refuge” from parasites and diseases. Parasites are ubiquitous, and infections often have significant consequences for the host. Parasites can directly affect viability and fertility of the host with consequences for the host’s reproductive fitness (Bush et al., 2001). By avoiding or at

Chapter 7: Parasite refuge hypothesis 109 least reducing the infection risk by parasites, hosts may increase their fitness and thereby trade off costly adaptations needed to survive in an extreme habitat. There are basically two proximate mechanisms that can lead to a decreased risk of a parasite infection in extreme habitats (Figure 7.1). Firstly, physiochemical stressors can have the same direct detrimental effect on free-living parasite stages as on every other organism. Thus, unless the parasite has the same ability to cope with the extreme environment as the host, it will be less successful and may even disappear from the habitat. Secondly, many parasites have indirect life cycles and rely on more than one host species as they go through different developmental stages. The lack of any necessary host species that does not survive in the extreme habitat interrupts the life cycle of the parasite. Thus, the absence of an obligate host species indirectly leads to the local extinction of the parasite. Based on this hypothesis, two empirically testable hypotheses can be made. (1) Given that a parasite species has at least one of the above mentioned characters (free-living stages or multiple host species), its prevalence should be reduced in more extreme habitats. (2) On the level of parasite communities, it is predicted that hosts in more extreme habitats harbor less parasite species. Furthermore, parasite communities should generally shift towards species with a direct life cycle, species lacking long lasting free-living stages and species living inside rather than on their hosts. A potential model system to test this hypothesis is a small livebearing fish, the Atlantic molly (Poecilia mexicana Steindachner, Poeciliidae), which is widely distributed on the Atlantic versant from northern Mexico to northern Costa Rica (Miller, 2005b). Besides normal stream and river habitats, this species also inhabits a limestone cave (the Cueva del Azufre) drained by a creek (Gordon and Rosen, 1962). The cave population of P. mexicana is also known as the cave molly (Parzefall, 2001). The creek running through the Cueva del Azufre is fed by several springs with high concentrations of hydrogen sulfide (Tobler et al., 2006). It eventually leaves the cave and forms a sulfurous surface creek, El Azufre. Hydrogen sulfide is highly toxic for all animals, because it binds to the iron of the heme to replace O2. It also binds at the cytochrome c oxidase, where it prohibits electron transport in aerobic respiration (Evans, 1967; National Research Council,

1979; Grieshaber and Völkel, 1998). In the cave, H2S ranges from 10 to 300 µM; such concentrations usually are considered toxic (Arp et al., 1992; Völkel and Grieshaber, 1992). Consequently, the Cueva del Azufre and El Azufre can be viewed as extreme 110 Chapter 7: Parasite refuge hypothesis

habitats (Tobler et al., 2006). How P. mexicana copes with H2S is so far not well understood. A costly behavioral adaptation, aquatic surface respiration (ASR), where fish exploit the air-water-interface, has been shown to be crucial for the survival in sulfidic water (Plath et al., 2007c). Other fish are able to detoxify sulfide to some extent, e.g. by oxidizing sulfide to thio-sulfate (Bagarinao and Vetter, 1990), but physiological adaptations to H2S remain to be studied in P. mexicana. It has been suggested that in the Cueva del Azufre chemoautotrophic primary production provides ad libitum amounts of food for the mollies (Langecker et al., 1996). Mollies from sulfurous habitats, however, are in a worse nutritional state than mollies from non-extreme habitats and have a lower body condition, indicating that energy may in fact be limited (Plath et al., 2005; Tobler et al., 2006). Compared to adjacent surface habitats, the Cueva del Azufre and El Azufre harbor an impoverished fish fauna with P. mexicana as the predominant species (Tobler et al., 2006). Hence, interspecific competition for resources and predation by piscivorous fishes is reduced in the sulfidic habitats. Other potential benefits of colonizing the Cueva del Azufre might also play a role in this system. Thus, we asked if living in an extreme habitat confers an advantage to P. mexicana with regard to a reduced risk of parasite infection by testing the first of the aforementioned predictions. Mollies are known to harbor a diverse parasite fauna (Tobler and Schlupp, 2005; Tobler et al., 2005 and unpublished data). One of the most prevalent species is the digenean trematode Uvulifer sp., the metacercariae of which provoke the production of a fibrous capsule of host tissue around the parasite, which is followed by the migration of melanocytes into the cyst’s wall, creating the characteristic appearance of a black spot (black spot disease, BSD) (Spellman and Johnson, 1987; Bush et al., 2001). This reaction of the host is assumed to be costly, since the penetration of the skin causes mechanical damage. Until the parasite becomes encapsulated, the host’s metabolic demand increases significantly so that energy reserves may decline. Uvulifer sp. has an indirect life cycle (Figure 7.1). After encapsulation in the fish host, the parasite remains dormant until the intermediate host is consumed by a piscivorous bird; the final host in which the parasite reproduces sexually. Water snails are the first intermediate hosts in which the parasite multiplies asexually, and free-swimming cercariae are produced. These cercariae infect fishes as second intermediate hosts by penetrating the skin and

Chapter 7: Parasite refuge hypothesis 111 transform into encysted metacercariae. This parasite thus has both characters to test the first prediction formulated above. In the present study, we compared parasitization of P. mexicana by Uvulifer sp. among different extreme and non-extreme habitats. Furthermore, we attempted to investigate if a potential reduction of parasitism in extreme habitats is caused by selection on free-living parasite stages or on other host species.

Materials and Methods Field sites All study sites are located near the village Tapijulapa in Tabasco, Mexico. The creeks studied eventually drain into the Río Oxolotan. The Río Oxolotan itself joins the Río Amatán and forms the Río Tacotalpa, a tributary of the Río Grijalva-system. We caught P. mexicana in cave chambers III, IV, V, X and XIII of the Cueva del Azufre (Gordon and Rosen, 1962). Additionally, fish were caught in El Azufre, a surface habitat containing toxic H2S. Currently, these are the only known sulfidic waters that harbor populations of P. mexicana. As a reference habitat, the closest comparable tributary to the Río Oxolotan on the opposite side of the river was chosen: Arroyo Cristal. This creek is comparable to the El Azufre in terms of size, structure and the adjacent surrounding. Details on the study sites can be found in Tobler et al.

(Tobler et al., 2006). H2S concentrations were highest within the cave (up to 300 µM); in El Azufre they ranged from 10 to 40 µM. H2S was absent in Arroyo Cristal (Tobler et al., 2006).

Data collection Fish were caught using small seines and dip nets. All trematode induced black spots on the body surface of the fish were counted. Furthermore, we estimated the occurrence of Uvulifer’s potential final and first intermediate hosts. Piscivorous bird species were qualitatively recorded in the morning and in the evening during 10 days in August 2004. Additionally, we identified the snail species inhabiting the different habitats, and we estimated their density by counting the numbers of snails in randomly selected patches (13 x 14 cm). A sub-sample of snails was checked for trematodes in the gonads by preparing the soft body parts of conserved specimens.

112 Chapter 7: Parasite refuge hypothesis

Data analysis Quantitative descriptors of parasitism were calculated according to Bush et al. (1997) and analyzed as suggested by Rózsa et al. (2000). The prevalence, as the proportion of individuals infected, was calculated and compared between populations using a Chi2 test. Furthermore, the mean abundance, as the mean number of parasites per host examined, and the mean intensity, as the mean number of parasites per infected host, were calculated. Mean abundance and mean intensity were analyzed using a GLM where “number of cysts” was the dependent variable and “population” and “sex” were independent variables. Snail densities were compared between habitats using a Mann-Whitney-U-test. Alpha levels were corrected according to the number of multiple comparisons using approximate Bonferroni adjustments [α' = 0.05/number of multiple comparisons (Grafen and Hails, 2002)]. Statistical analyses were performed using SPSS 11, SPSS Inc.

Results Parasite prevalence and intensity The prevalence of BSD differed significantly between populations, whereby Uvulifer sp. was most prevalent in Arroyo Cristal, less prevalent in El Azufre and absent in the cave (Table 7.1, Chi2=218.69, P<0.001; α'=0.0125). The significant difference between populations was not only driven by eminently low prevalence of Uvulifer sp. within the cave, because when the prevalence was only compared between El Azufre and Arroyo Cristal, the difference was still significant (Chi2=21.017, P<0.001; α'=0.0125). Furthermore, the mean abundance of Uvulifer sp. differed significantly between populations, showing the same pattern as the prevalence (Tables 7.1 and 7.2). The factor “sex” had no significant influence (Table 7.2). Since the parasite was absent in the Cueva del Azufre, the mean intensity of BSD was only compared between the populations from Arroyo Cristal and El Azufre; however, no significant differences between populations and sexes were detected (Tables 7.1 and 7.2). The correction of the alpha levels did not influence the results.

Further hosts in the life cycle of Uvulifer sp. Within the Cueva del Azufre, neither birds nor water snails could be observed. Around both surface habitats, however, several species of piscivorous birds (i.e.,

Chapter 7: Parasite refuge hypothesis 113 potential final hosts of Uvulifer sp.) were recorded. We observed Great egrets (Ardea alba Linnaeus), Green herons (Butorides virescens Linnaeus), Ringed kingfishers (Ceryle torquata Linnaeus), Snowy egrets (Egretta thula Molina), Least bittern (Ixobrychus exilis Gmelin) and Great kiskadees (Pitangus sulfuratus Linnaeus). One snail species (Pachychilus cf. indiorum Morelet, Pachychilidae) was recorded in the two surface habitats. The snail density was significantly higher in the Arroyo Cristal (median=55 snails/m2 (interquartile range, IQR=192), N=31) than in the El Azufre (median=0 snails/m2 (IQR=55), N=26; U=266.00, P=0.015). Trematodes could not be recorded in any of the examined P. cf. indiorum from El Azufre and Arroyo Cristal (N=29).

Discussion Parasitism in Poecilia mexicana through Uvulifer sp. was high in the non-extreme surface habitat and reduced in habitats containing hydrogen sulfide. Within the cave, fish infected with BSD were absent from our samples. H2S concentrations are especially high within the cave (Tobler et al., 2006) and potential final and intermediate hosts of Uvulifer sp. were absent. In contrast, potential final and intermediate hosts were present in both surface habitats, but snails were less abundant in the sulfur creek. The lower snail density in El Azufre could be due to toxic properties of the water or reduced food availability, as green algae cannot be found in sulfidic waters. We were not able to find any signs of trematode infections in the snails. This, however, may be due to the limited sample size. Poecilia mexicana from the sulfur creek were significantly less infected with BSD compared to the surface creek without sulfide compounds. The mechanism leading to a reduced prevalence of BSD infection in the extreme habitats cannot be verified on the basis of the present field data. Since Uvulifer sp. has an indirect life cycle, the reduction in prevalence in its second intermediate host (P. mexicana) can be caused by lower snail densities that may reduce the infection risk for fish, or the toxic

H2S that may have a direct detrimental effect on free-living parasite stages (Figure 7.1). Potentially, the difference in BSD prevalence between the surface habitats may not only be explained by a lower abundance of parasites in the sulfidic habitat, but may rather be due to higher parasite-induced mortality in this harsh environment (compare McKeown and Irwin, 1997). An infection with Uvulifer sp. is costly 114 Chapter 7: Parasite refuge hypothesis

(Spellman and Johnson, 1987; Bush et al., 2001), which might be especially relevant under extreme environmental conditions. BSD induced mortality might be higher in the sulfurous habitat, and increased parasite induced mortality would be an additional cost of living in an extreme habitat. If parasite induced mortality was higher in extreme habitats, then differences in the mean intensity of BSD would be expected, as the habitat with toxic H2S would lack heavily infected individuals. In our study, there was no significant difference in the mean intensity of BSD between the two surface habitats. Along with the low density of potential intermediate hosts, this suggests that lower BSD prevalence in the sulfidic habitat is not due to increased BSD induced mortality but rather due to a lower exposure to the parasite. Our study therefore suggests that reduced exposure to parasites may be a benefit of living in extreme habitats. Parasite communities of hosts living in extreme environments remain largely unexplored. For example, a low diversity of parasites has been documented from hosts of deep-sea hydrothermal vents compared to other deep-sea habitats (De Buron and Morand, 2004). However, it is as yet unclear if the reduced parasite diversity in this system is caused by the presence of physiochemical stressors (Van Dover, 2000), the reduced diversity of potential hosts (Biscoito et al., 2002), or simply the lack of research (De Buron and Morand, 2004).

Chapter 7: Parasite refuge hypothesis 115

Figures and tables chapter 7

Figure 7.1. Life cycle of Uvulifer sp., a trematode parasite with an indirect life cycle. Hosts, e.g. the fish as second intermediate host, can escape infections by entering extreme habitats with environmental conditions that cannot be tolerated by either free living parasite life stages or other hosts on which the parasite relies.

116 Chapter 7: Parasite refuge hypothesis

Table 7.1. Prevalence, mean (±SD) abundance, and mean intensity of BSD infection within the three populations of Poecilia mexicana studied.

H2S Prevalence Mean abundance Mean intensity Range N [number of cysts] [number of cysts] [number of cysts] Arroyo Cristal absent 0.45 0.98±1.72 2.17±2.00 0-14 159 El Azufre low 0.20 0.55±1.60 2.80±2.65 0-11 128 Cueva del Azufre high 0.00 0.00±0.00 0.00 0 456

Table 7.2. Generalized linear models (GLM) on the mean abundance and mean intensity of BSD infections in the Poecilia mexicana populations studied. The interaction effect of “population” and “sex” was not significant in either case (F=0.012, P=0.91 and F=1.42, P=0.243, respectively) and thus only main effects were analyzed. Factor df Mean square F P α' Mean abundance Population 2 59.51 55.425 <0.001 0.0125 Sex 1 0.09 0.086 0.77 0.0125

Mean intensity Population 1 9.53 2.00 0.16 0.0125 Sex 1 3.28 0.69 0.41 0.0125

Chapter 8: Conclusions and perspectives 117

Chapter 8

PERSPECTIVES

P. mexicana in the Cueva del Azufre system provide an ideal system to investigate ecological and evolutionary consequences of the colonization of novel habitats. This thesis highlights that comparative studies in this system should not only include a comparison between epigean and cave habitats, but the presence or absence of H2S is equally important as the presence or absence of light. P. mexicana in the different habitat types are differentiated genetically as well as phenotypically. Divergent traits cannot only be attributed to the differences in abiotic conditions among habitats, but also to correlated biotic environmental factors. Future investigations will have to elucidate how the different selective factors shape different traits, how they interact with each other, and how adaptation to one selective factor may constrain adaptation to another. To date, one of the biggest unanswered questions is why gene-flow among populations is low despite the spatial proximity and lack of physical barriers among habitats. I propose that divergent selection caused by the abiotic environmental conditions gives rise to reproductive isolation. Several mechanisms, which may act in synchrony, seem possible in this system. (1) Selection could act directly on immigrants from divergent populations causing prezygotic isolations (Nosil et al., 2005). For example, P. mexicana from non-sulfidic habitats have been shown to have a high susceptibility to the toxic effects of H2S (Tobler et al., 2008c). Likewise, fish from surface habitats lack the ability for intraspecific communication in darkness, which may lower their reproductive success in the cave environment (Plath et al., 2004; Plath et al., 2006). Selection against immigrants, however, should be unidirectional, unless adaptations to sulfidic and cave habitats come at a cost in non- sulfidic and surface habitats, respectively. For example, the reduced eye size could make cave P. mexicana more susceptible to predation in surface habitats. Individuals from cave populations further have been shown to be less efficient foragers in light compared to conspecifics from surface habitats (Tobler et al. unpublished data). (2) P. mexicana from different habitat types may be less attracted to conspecifics from divergent habitat types, which again may cause prezygotic isolation (Schluter, 2000; 118 Chapter 8: Conclusions and perspectives

Rundle and Nosil, 2005). For example, in poeciliid fishes of the genus Gambusia, assortative mating for divergent body shapes caused by different predatory regimes causes reproductive isolation among allopatric populations (Langerhans et al., 2007b). Whether mating preferences for divergent phenotypes are present in the P. mexicana system remains to be studied. Beside intersexual selection, intrasexual selection by male-male competition may be operating, since all populations from cave and/ or sulfidic habitats have reduced male sexual activity (Plath et al., 2003a; Plath et al., 2007b; Plath, 2008), and Cueva del Azufre fish are less aggressive (Parzefall, 1974). Reproductive success in poeciliids is determined by the number of successful copulations (Magurran and Seghers, 1994). Males from divergent habitats may be outcompeted by fish from non-sulfidic surface fish due to higher aggression and higher rates of sexual behavior. (3) Divergent selection could act on hybrids of P. mexicana from different habitats (Hatfield and Schluter, 1999; Schluter, 2000), but to date no empirical evidence for this mechanism is available. Future studies will need to pay careful attention to the mechanisms of reproductive isolation causing the observed small-scale population differentiation in the Cueva del Azufre system in order to test whether parapatric ecological speciation is indeed occurring. The current study, however, provides strong evidence for population differentiation and local adaptation along abiotic environmental gradients. Most model systems studied to date found biotic ecological factors, i.e. interactions between species, such as resource use, predation, or parasitism to be driving adaptive divergence among lineages. However, divergent selection by abiotic environmental conditions, as reported in this thesis, is probably more common than previously thought.

References 119

REFERENCES

Abdel-Latif H, Hassan E, von Campenhausen C, 1990. Sensory performance of blind Mexican cave fish after destruction of the canal neuromasts. Naturwissenschaften 77:237-239. Abel DC, Koenig CC, Davis WP, 1987. Emersion in the mangrove forest fish Rivulus marmoratus: a unique response to hydrogen sulfide. Environmental Biology of Fishes 18:67-72. Adams GL, Johnson JE, 2001. Metabolic rate and natural history of Ozark cavefish, Amblyopsis rosae, in Logan Cave, Arkansas. Environmental Biology of Fishes 62:97-105. Adelman IR, Smith Jr. LL, 1970. Effect of hydrogen sulfide on northern pike eggs and sac fry. Transactions of the American Fisheries Society 99:501-509. Affonso E, Rantin F, 2005. Respiratory responses of the air-breathing fish Hoplosternum littorale to hypoxia and hydrogen sulfide. Comparative Biochemistry and Physiology C - Toxicology and Pharmacology 141:275-280. Alexander HJ, Breden F, 2004. Sexual isolation and extreme morphological divergence in the Cumana guppy: a possible case of incipient speciation. Journal of Evolutionary Biology 17:1238-1254. Allegrucci G, Todisco V, Sbordoni V, 2005. Molecular phylogeography of Dolichopoda cave crickets (Orthoptera, Rhaphidophoridae): A scenario suggested by mitochondrial DNA. Molecular Phylogenetics and Evolution 37:153-164. Antonovics J, 2006. Evolution in closely adjacent plant populations X: Long-term persistence of prereproductive isolation at a mine boundary. Heredity 97:33- 37. Arp AJ, Hansen BM, Julian D, 1992. Burrow environment and coelomic fluid characteristics of the chiurian worm Urechis caupo from populations at three sites in northern California. Marine Biology 113:613-623. Astall C, Anderson S, Taylor A, Atkinson R, 1997. Comparative studies of the branchial morphology, gill area and gill ultrastructure of some thalassinidean mud-shrimps (Crustacea: Decapoda: Thalassinidea). Journal of Zoology 241:665-688. 120 References

Babbitt K, Jordan F, 1996. Predation on Bufo terrestris tadpoles: effects of cover and predator identity. Copeia 1996:485-488. Bagarinao T, 1992. Sulfide as an environmental factor and toxicant: tolerance and adaptations in aquatic organisms. Aquatic Toxicology 24:21-62. Bagarinao T, Lantin-Olaguer I, 1999. The sulfide tolerance of milkfish and tilapia in relation to fish kills in farms and natural waters in the Philippines. Hydrobiologia 382:137-150. Bagarinao T, Vetter R, 1990. Oxidative detoxification of sulfide by mitochrondria of the California killifish Fundulus parvipinnis and the speckled sanddap Citharichthys stignaeus. Journal of Comaparative Physiology B 160:519-527. Bagarinao T, Vetter RD, 1989. Sulfide tolerance and detoxification in shallow water marine fishes. Marine Biology 103:291-302. Bailey N, 1951. On estimating the size of mobile populations from recapture data. Biometrica 38:293-306. Baker CF, Montgomery J, 1999. The sensory basis of rheotaxis in the blind Mexican cave fish, Astyanax fasciatus. Journal of Comparative Physiology 184:519- 527. Barr TC, 1968. Cave ecology and the evolution of troglobites. Evolutionary Biology 2:35-102. Barr TC, Holsinger JR, 1985. Speciation in cave faunas. Annual Review of Ecology and Systematics 16:313-337. Barrow LM, Chapman LJ, 2006. Foraging costs of hypoxia acclimation in the swamp- dwelling African cyprinid, Barbus neumayeri. Copeia:552-557. Begon ME, Harper JL, Townsend CR, 1996. Ecology, third edition. Oxford: Blackwell Science. Bergman AN, Laurent P, Otiang'a-Owiti G, Bergman HL, Walsh PJ, Wilson P, Wood CM, 2003. Physiological adaptations of the gut in the Lake Magadi tilapia, Alcolapia grahami, an alkaline- and saline-adapted teleost fish. Comparative Biochemistry and Physiology - Part A: Molecular & Integrative Physiology 136:701-715. Berlocher S, Feder J, 2002. Sympatric speciation in phytophagous insects: moving beyond controversy? Annual Reviews of Entomology 47:773-815.

References 121

Biscoito M, Segonzac M, Almeida A, Desbruyeres D, Geistdoerfer P, Turnispeed M, Van Dover CL, 2002. Fishes from the hydrothermal vents and cold seeps: an update. Cahiers de Biologie Marine 43:359-362. Blais J, Rico C, Van Oosterhout C, Cable J, Turner GF, Bernatchez L, 2007. MHC adaptive divergence between closely related and sympatric African cichlids. PLoS ONE 2:e734. Bohonak A, 1999. Dispersal, gene flow, and population structure. Quarterly Review of Biology 74:21-45. Bond AN, 1960. An analysis of the response of salamander gills to changes in the oxygen concentration of the medium. Developmental Biology 2:1-20. Bookstein F, 1991. Morphometric tools for landmark data. Cambridge: Cambridge University Press. Boughman J, Rundle HD, Schluter D, 2005. Parallel evolution of sexual isolation in sticklebacks. Evolution 59:361-373. Brauner CJ, Ballantyne CL, Randall DJ, Val AL, 1995. Air-breathing in the armored catfish (Hoplosternum littorale) as an adaptation to hypoxic, acidic, and hydrogen sulfide-rich waters. Canadian Journal of Zoology-Revue Canadienne De Zoologie 73:739-744. Brodie ED, 1992. Correlational selection for color pattern and antipredator behavior in the Garter snake Thamnophis ordinoides. Evolution 46:1284-1298. Brönmark C, Miner J, 1992. Predator-induced phenotypical change in body morphology in crucian carp. Science 258:1348-1350. Burggren WW, Mwalukoma A, 1983. Respiration during chronic hypoxia and hyperoxia in larval and adult bullfrogs (Rana catesbeiana): I. Morphological responses of lungs, skin, and gills. Journal of Experimental Biology 105:191- 203. Burt de Perera T, 2004. Spatial parameters encoded in spatial map of the blind Mexican cave fish, Astyanax fasciatus. Animal Behaviour 68:291-295. Bush AO, Fernández JC, Esch GW, Seed JR, 2001. Parasitism: The diversity and ecology of animal parasites. Cambridge: Cambridge University Press. Bush AO, Lafferty KH, Lotz JM, Shostak AW, 1997. Parasitology meets ecology on its own terms: Margolis et al. revisted. Journal of Parasitology 83:575-583. 122 References

Bussing WA, 1998. Peces de las aguas continentales de Costa Rica - Freshwater fishes of Costa Rica, 2nd ed. San José: Editorial de la Universidad de Costa Rica. Carrico R, Blumberg W, Peisach J, 1978. The reversible binding of oxygen to sulfhemoglobin. Journal of Biological Chemistry 253:7212-7215. Caskey L, Riedel R, Costa-Pierce B, Butler J, Hurlbert SH, 2007. Population dynamics, distribution, and growth rate of tilapia (Oreochromis mossambicus) in the Saltin Sea, California, with notes on bairdiella (Bairdiella icistia) and orangemouth corvina (Cynoscion xanthulus). Hydrobiologia 576:185-203. Chapman LJ, Chapman CA, 1993. Desiccation, flooding, and the behavior of Poecilia gillii (Pisces: Poeciliidae). Ichthyological Exploration of Freshwaters 4:279- 287. Chapman LJ, Chapman CA, Brazeau DA, McLaughlin B, Jordan M, 1999. Papyrus swamps, hypoxia, and faunal diversification: variation among populations of Barbus neumayeri. Journal of Fish Biology 54:310-327. Chapman LJ, Galis F, Shinn J, 2000. Phenotypic plasticity and the possible role of genetic assimilation: Hypoxia-induced trade-offs in the morphological traits of an African cichlid. Ecology Letters 3:387-393. Chapman LJ, Hulen KG, 2001. Implications of hypoxia for the brain size and gill morphometry of mormyrid fishes. Journal of Zoology 254:461-472. Chase J, 1999. Food web effects of prey size refugia: variable interactions and alternative stable equilibria. American Naturalist 154:559-570.

Chen K, Morris J, 1972. Kinetics of oxidation of aqueous sulfide by O2. Enviromental Science and Technology 6:529-537. Christiansen K, 1965. Behavior and form in the evolution of cave collembola. Evolution 19:529-537. Clement M, Posada D, Crandall K, 2000. TCS: a computer program to estimate gene genealogies. Molecular Ecology 9:1657-1659. Cline J, 1969. Spectrophotometric determination of hydrogen sulfide in natural waters. Limnology and Oceanography 14:454-458. Cline J, Richards F, 1969. Oxygenation of hydrogen sulfide in seawater at constant salinity, temerature, and pH. Enviromental Science and Technology 3:838- 843.

References 123

Colby PJ, Smith Jr. LL, 1967. Survival of walleye eggs and fry on paper fiber sludge deposits in the Rainy River, Minnesota. Transactions of the American Fisheries Society 96:278-296. Crispo E, Bentzen P, Reznick DN, Kinnison MT, Hendry AP, 2006. The relative influence of natural selection and geography on gene flow in guppies. Molecular Ecology 15:49-62. Crispo E, Chapman LJ, 2008. Population genetic structure across dissovled oxygen regimes in an African cichlid fish. Molecular Ecology 17:2134-2148. Culver DC, 1982. Cave life - evolution and ecology. Cambridge: Harvard Uiversity Press. Culver DC, Fong DW, 1986. Why all cave animals look alike. Stygologia 2:208-216. Culver DC, Kane TC, Fong DW, 1995. Adaptation and natural selection in caves. Cambridge: Harvard University Press. Curtis C, Bartholomew T, Rose F, Dodgson K, 1972. Detoxication of sodium 35S- sulphide in the rat. Biochemical Pharmacology 21:2313-2321. Danielopol DL, Rouch R, 2005. Invasion, active versus passive. In: Encyclopedia of caves (Culver DC, White W, eds). Amstserdam: Elsevier Academic Press. Dare MR, Hubert WA, Meyer JS, 2001. Influence of stream flow on hydrogen sulfide concentrations and distributions of two trout species in a Rocky Mountains tailwater. North American Journal of Fisheries Management 21:971-975. Darnell RM, 1962. Fishes of the Rio Tamesi and related coastal lagoons in east- central Mexico. Publications of the Institute of Marine Sciences, University of Texas 8:299-365. Darwin C, 1859. On the origin of species based on natural selection , or the preservation of favoured races in the struggle of life, 6th ed. London: John Murray. Day T, 2000. Competition and the effect of spatial resource heterogeneity on evolutionary diversification. American Naturalist 155:790-803. De Buron I, Morand S, 2004. Deep-sea hydrothermal vent parasites: why do we not find more? Parasitology 128:1-6. DeWitt TJ, 1998. Costs and limits of phenotypic plasticity: Tests with predator- induced morphology and life history in a freshwater snail. Journal of Evolutionary Biology 11:465-480. 124 References

DeWitt TJ, Langerhans RB, 2003. Multiple prey traits, multiple predators: keys to understanding complex community dynamics. Journal of Sea Research 49:143-155. DeWitt TJ, Scheiner S, 2004. Phenotypic plasticity: Functional and conceptual approaches. Oxford: Oxford University Press. Dhuyvetter H, Hendrickx F, Gaublomme E, Desender K, 2007. Differentiation between two salt marsh beetle ecotypes: evidence for ongoing speciation. Evolution 61:184-193. Dias P, 1996. Sources and sinks in population biology. Trends in Ecology & Evolution 11:326-330. Dias P, Blondel J, 1996. Local specialization and maladaptation in the Mediterranean blue tit (Parus caeruleus). Oecologia 107:79-86. Dieckmann U, Doebeli M, Metz JAJ, Tautz D, 2004. Adaptive speciation. Cambridge: Cambridge University Press. Donohue K, Hammond Pyle E, Messiqua D, Heschel M, Schmitt J, 2001. Adaptive divergence in plasticity in natural populations of Impatiens capensis and its consequences for performance in novel habitats. Evolution 55:692-702. Dries LA, 2003. Peering through the looking glass at a sexual parasite: Are Amazon mollies Red Queens? Evolution 57:1387-1396. Eklöv P, Jonsson P, 2007. Pike predators induce morphological changes in young perch and roach. Journal of Fish Biology 70:155-164. Evanno G, Regnaut S, Goudet J, 2005. Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Molecular Ecology 14:2611-2620. Evans C, 1967. The toxicity of hydrogen sulphide and other sulphides. Quarterly Journal of Experimental Physiology 52:231-248. Excoffier L, Smouse P, Quattro J, 1992. Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics 131:479-491. Fenolio D, Graening G, Collier B, Stout J, 2006. Coprophagy in a cave-adapted salamander: the importance of bat guano examined through nutritional and stable isotope analyses. Proceedings of the Royal Society B-Biological Sciences 273:439-443.

References 125

Feulner P, Kirschbaum F, Tiedemann R, 2005. Eighteen microsatellite loci for endemic African weakly electric fish (Campylomormyrus, Mormyridae) and their cross species applicability among related taxa. Molecular Ecology Notes 5:446-448. Fuller R, McGhee K, Schrader M, 2007. Speciation in killifish and the role of salt tolerance. Journal of Evolutionary Biology 20:1962-1975. Funk DJ, 1998. Isolating a role for natural selection in speciation: host adaptation and sexual isolation in Neochlamisius bebbianae leaf beetles. Evolution 52:1744- 1759. Gannon AT, Demarco VG, Morris T, Wheatly MG, Kao YH, 1999. Oxygen uptake, critical oxygen tension, and available oxygen for three species of cave crayfishes. Journal of Crustacean Biology 19:235-243. Geiger SP, Torres JJ, Crabtree RE, 2000. Air breathing and gill ventilation frequencies in juvenile tarpon, Megalops atlanticus: responses to changes in dissolved oxygen, temperature, hydrogen sulfide, and pH. Environmental Biology of Fishes 59:181-190. Gibert J, Deharveng L, 2002. Subterranean ecosystems: a truncated functional biodiversity. Bioscience 52:473-481. Gido K, Franssen N, 2007. Invasion of stream fishes into low trophic positions. Ecology of Freshwater Fish 16:457-464. Gordon MS, Rosen DE, 1962. A cavernicolous form of the Poeciliid fish Poecilia sphenops from Tabasco, México. Copeia 1962:360-368. Goudet J, 2002. FSTAT, Version 2.9.3.2. Lausanne: Institute for Ecology, University of Lausanne. Gough L, Shaver G, Carroll J, Royer D, Laundre J, 2000. Vascular plant species richness in Alaskan arctic tundra: the importance of soil pH. Journal of Ecology 88:54-66. Graening G, Brown A, 2003. Ecosystem dynamics and pollution effects in an Ozark cave stream. Journal of the American Water Resources Association 39:1497- 1507. Grafen A, Hails R, 2002. Modern statistics for the life sciences. Oxford: Oxford University Press. Graham J, 2005. Aquatic and aerial respiration. In: The physiology of fishes, 3rd ed (Evans D, Claiborne J, eds). London: CRC Press. 126 References

Greenfield DW, Greenfield T, Wildrick D, 1982. The taxonomy and distribution of the species of Gambusia (Pisces: Poeciliidae) in Belize, Central America. Copeia 1982:128-147. Greenfield DW, Wildrick D, 1984. Taxonomic distinction of the Antilles Gambusia puncticulata complex (Pisces: Poeciliidae) from the G. yucatana complex of Mexico and Central America. Copeia 1984:921-933. Grieshaber MK, Völkel S, 1998. Animal adaptations for tolerance and exploitation of poisonous sulfide. Annual Review of Physiology 60:33-53. Hatfield T, Schluter D, 1999. Ecological speciation in sticklebacks: environment dependent hybrid fitness. Evolution 53:866-873. Hays C, 2007. Adaptive phenotypic differentiation across the interstitial gradient in the alga Silvetia compressa. Ecology 88:149-157. Hellawell J, Abel R, 1971. A rapid volumetric method for the analysis of the food of fishes. Journal of Fish Biology 3:29-37. Hendry AP, Kelly ML, Kinnison MT, Reznick DN, 2006. Parallel evolution of the sexes? Effects of predation and habitat features on the size and shape of wild guppies. Journal of Evolutionary Biology 19:741-754. Heulett ST, Weeks SC, Meffe GK, 1995. Lipid dynamics and growth relative to resource level in juvenile eastern mosquitofish (Gambusia holbrooki, Poeciliidae). Copeia:97-104. Higham TE, Hulsey CD, Rican O, Carroll AM, 2007. Feeding with speed: prey capture evolution in cichlids. Journal of Evolutionary Biology 20:70-78. Holsinger JR, 2000. Ecological derivation, colonization, and speciation. In: Ecosystems of the world 30 Subterranean ecosystems (Wilkens H, Culver DC, Humphreys WF, eds). Amsterdam: Elsevier Science. Holt R, M B, Gomulkiewicz R, 2004. Temporal variation can facilitate niche evolution in harsh sink environments. American Naturalist 164:187-200. Holtmeier CL, 2001. Heterochrony, maternal effects, and phenotypic variation among sympatric pupfishes. Evolution 55:330-338. Horstkotte J, Strecker U, 2005. Trophic differentiation in the phylogenetically young Cyprinodon species flock (Cyprinodontidae, Teleostei) from Laguna Chichancanab (Mexico). Biological Journal of the Linnean Society 85:125- 134.

References 127

Hose L, Palmer AN, Palmer M, Northup DE, Boston PJ, DuChene HR, 2000. Microbiology and geochemistry in a hydrogen-sulphide-rich karst environment. Chemical Geology 169:399-423. Howarth FG, 1987. The evolution of non-relictual tropical troglobites. International Journal of Speleology 16:1-16. Howarth FG, 1993. High-stress subterranean habitats and evolutionary change in cave-inhabiting arthropods. American Naturalist 142:S65-S77. Howarth FG, Stone F, 1990. Elevated carbon dioxide levels in Bayliss Cave, Australia: Implications for the evolution of obligate cave species. Pacific Science 44:207-218. Hubbs C, 1959. Population analysis of a hybrid swarm between Gambusia affinis and G. heterochir. Evolution 13:236-246. Hulsey CD, Garcia de Leon F, 2005. Cichlid jaw mechanics: linking morphology to feeding specialization. Functional Ecology 19:487-494. Humphreys WF, 1993. Cave fauna in semi-arid tropical Western Australia: a diverse relict wet-forest litter fauna. Mémoires de Biospéologie 20:105-110. Hüppop K, 2000. How do cave animals cope with the food scarcity in caves? In: Ecosystems of the world 30: Subterranean ecosystems (Wilkens H, Culver DC, Humphries WF, eds). Amsterdam: Elsevier Science; 159-188. Hüppop K, Wilkens H, 1991. Bigger eggs in subterranean Astyanax fasciatus (Characidae, Pisces): their significance and genetics. Zeitschrift für Zoologische Systematik und Evolutionsforschung 29:280-288. Ip YK, Kuah SSL, Chew SF, 2004. Strategies adopted by the mudskipper Boleophthalmus boddaerti to survive sulfide exposure in normoxia or hypoxia. Physiological and Biochemical Zoology 77:824-837. Jeffery WR, 2001. Cavefish as a model system in evolutionary developmental biology. Developmental Biology 231:1-12. Jeffery WR, 2005. Adaptive evolution of eye degeneration in the Mexican blind cavefish. Journal of Heredity 96:185-196. Jeffery WR, Strickler AG, Yamamoto Y, 2003. To see or not to see: Evolution of eye degeneration in Mexican blind cavefish. Integrative and Comparative Biology 43:531-541. Jeyasingh PD, 2007. Plasticity in metabolic allometry: the role of dietary stoichiometry. Ecology Letters 10:282-289. 128 References

Jiggins C, Naisbit R, Coe R, Mallet J, 2001. Reproductive isolation caused by colour pattern mimicry. Nature 411:302-305. Jimenez-Ambriz G, Petit C, Bourrie I, Dubois S, Olivieri I, Ronce O, 2006. Life history variation in the heavy metal tolerant plant Thlaspi caerulescens growing in a network of contaminated and noncontaminated sites in southern France: role of gene flow, selection and phenotypic plasticity. New Phytologist 173:199-215. Johansson J, Turesson H, Persson A, 2004. Active selection for large guppies, Poecilia reticulata, by the pike cichlid, Crenicichla saxatilis. Oikos 105:595- 605. Johnson JB, Belk MC, 2001. Predation environment predicts divergent life-history phenotypes among populations of the lifebearing fish Brachyrhaphis rhabdophora. Oecologia 126:142-149. Johnson P, Hoppensteadt F, Smith J, Bush G, 1996. Conditions for sympatric speciation: A diploid model incorporating habitat fidelity and non-habitat assortative mating. Evolutionary Ecology 10:187-205. Johnson S, Hulsey CD, Garcia de Leon FJ, 2007. Spatial mosaic evolution of snail defensive traits. BMC Evolutionary Biology 7:50. Kawecki TJ, Ebert D, 2004. Conceptual issues in local adaptation. Ecology Letters 7:1225-1241. Kehr A, Schnack J, 1991. Predator prey relationship between Giant water bugs (Belostoma oxyurum) and larval anurans (Bufo arenarum). Alytes 9:61-69. Keller L, Grant P, Grant B, Petren K, 2001. Heritability of morphological traits in Darwin's finches: misidentified paternity and maternal effects. Heredity 87:325-336. Kittell M, Harvey M, Contreras-Balderas S, Ptacek MB, 2005. Wild-caught hybrids between sailfin and shortfin mollies (Poeciliidae, Poecilia): morphological and molecular verification. Hidrobiologica 15:131-137. Körner KE, Schlupp I, Plath M, Loew ER, 2006. Spectral sensitivity of mollies: comparing surface- and cave-dwelling Atlantic mollies, Poecilia mexicana. Journal of Fish Biology 69:54-65. Kramer DL, 1983. The evolutionary ecology of respiratory modes in fishes: an analysis based on the costs of breathing. Environmental Biology of Fishes 9:145-158.

References 129

Kramer DL, 1987. Dissolved oxygen and fish behavior. Environmental Biology of Fishes 18:81-92. Kramer DL, Bryant MJ, 1995a. Intestine length in the fishes of a tropical stream: 2. Relationships to diet - the long and short of a convulted issue. Environmental Biology of Fishes 42:129-141. Kramer DL, Bryant MJ, 1995b. Intestine length in the fishes of a tropical stream: 1. Ontogenic allometry. Environmental Biology of Fishes 42:115-127. Kramer DL, Mehegan J, 1981. Aquatic surface respiration, an adaptive response to hypoxia in the guppy, Poecilia reticulata (Pisces, Poeciliidae). Environmental Biology of Fishes 6:299-313. Langecker TG, 1989. Studies on the light reaction of epigean and cave populations of Astyanax fasciatus (Characidae, Pisces). Mémoires de Biospéologie 16:169- 176. Langecker TG, 2000. The effect of continuous darkness on cave ecology and cavernicolous evolution. In: Ecosystems of the world 30: Subterranean Ecosystems (Wilkens H, Culver DC, Humphreys WF, eds). Amsterdam: Elsevier Science. Langecker TG, Longley G, 1993. Morphological adaptation of the Texas blind catfishes Trogloglanis pattersoni and Satan eurystomus (Siluriformes: Ictaluridae) to their underground environment. Copeia 1993:976-986. Langecker TG, Wilkens H, Parzefall J, 1996. Studies on the trophic structure of an energy-rich Mexican cave (Cueva de las Sardinas) containing sulfurous water. Mémoires de Biospéologie 23:121-125. Langerhans RB, Chapman LJ, DeWitt TJ, 2007a. Complex phenotype-environment associations revealed in an East African cyprinid. Journal of Evolutionary Biology 20:1171-1181. Langerhans RB, Gifford M, Joseph E, 2007b. Ecogical speciation in Gambusia fishes. Evolution 61:2056-2074. Langerhans RB, Layman CA, DeWitt TJ, 2005. Male genital size reflects a tradeoff between attracting mates and avoiding predators in two live-bearing fish species. Proceedings of the National Academy of Sciences of the United States of America 102:7618-7623. Langerhans RB, Layman CA, Shokrollahi AM, DeWitt TJ, 2004. Predator-driven phenotypic diversification in Gambusia affinis. Evolution 58:2305-2318. 130 References

Laybourn-Parry J, Pearce D, 2007. The biodiversity and ecology of Antarctic lakes: models for evolution. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 362:2273-2289. Lenormand T, 2002. Gene flow and the limits of natural selection. Trends in Ecology & Evolution 17:183-189. Levene H, 1953. Genetic equilibrium when more than one ecological niche is available. American Naturalist 87:331-333. Levins R, 1968. Evolution in changing environments. Princeton: Princeton University Press. Lovatt Evans C, 1967. The toxicity of hydrogen sulphide and other sulphides. Quarterly Journal of Experimental Physiology 52:231-248. Luther G, Ma S, Trouwborst R, Glazer B, Blickley M, Sarborough R, Mensinger M,

2004. The roles of anoxia, H2S, and storm events in fish kills of dead-end canals of Delaware inland bays. Estuaries 27:551-560. Macalady JL, Lyon EH, Koffman B, Albertson LK, Meyer K, Galdenzi S, Mariani S, 2006. Dominant microbial populations in limestone-corroding stream biofilms, Frasassi cave system, Italy. Applied and Environmental Microbiology 72:5596-5609. MacArthur R, Diamond J, Karr J, 1972. Density compensation in island faunas. Ecology 53:330-342. MacArthur R, Pianka E, 1966. On optimal use of patchy environment. American Naturalist 100:603-609. MacArthur RH, Wilson EO, 1967. The theory of island biogeography. Princeton: Princeton University Press. Macnair MR, Christie P, 1983. Reproductive isolation as a pleiotropic effect of copper tolerance in Mimulus guttatus. Heredity 50:295-302. Magurran AE, Seghers BH, 1994. Sexual conflict as a consequence of ecology: evidence from Guppy, Poecilia reticulata, populations in Trinidad. Proceedings of the Royal Society of London Series B-Biological Sciences 255:31-36. Manier M, Seyler C, Arnold S, 2007. Adaptive divergence within and between ecotypes of the terrestrial garter snake, Thamnophis elegans, assessed with FST-QST comparisons. Journal of Evolutionary Biology 20:1705-1719.

References 131

Martel C, Réjasse A, Rousset F, Bethenod M, Bourguet D, 2003. Host-plant- associated genetic differentiation in Northern French populations of the European corn borer. Heredity 90:141-149. Matthews WJ, 1998. Patterns in freshwater fish ecology. Boston/ Dordrecht/ London: Kluwer Academic Publisher. Mayland HJ, 1984. Mittelamerika: Cichliden und Lebendgebärende. Hannover: Landbuch Verlag. McDonald P, Fonseca C, Overton J, Westoby M, 2003. Leaf size divergence along rainfall and soil-nutrient gradients: is the method of size reduction common along clades? Functional Ecology 17:50-57. McKeown CA, Irwin SWB, 1997. Accumulation of Diplostomum spp. (Digenea: Diplostomatidae) metacercariae in the eyes of 0+ and 1+ roach (Rutilus rutilus). International Journal of Parasitology 27:377-380. McKinnon J, Mori S, Blackman B, David L, Kingsley D, Jamieson L, Chou J, Schluter D, 2004. Evidence for ecology's role in speciation. Nature 429:294- 298. McMullin E, Bergquist D, Fisher CR, 2000. Metazoans in extreme environments: adaptations of hydrothermal vent and hydrocarbon fauna. Gravitational and Space Biology Journal 13:13-23. McPeek MA, 1995. Morphological evolution mediated by behavior in the damselflies of two communities. Evolution 49:749-769. McPeek MA, Wellborn GA, 1998. Genetic variation and reproductive isolation among phenotypically divergent amphipod populations. Limnology and Oceanography 43:1162-1169. Meffe GK, Snelson FF, 1993. Lipid dynamics during reproduction in two livebearing fishes, Gambusia holbrooki and Poecilia latipinna. Canadian Journal of Fisheries and Aquatic Sciences 50:2185-2191. Menke A, 1979. Family Belostomatidae - Giant water bugs. In: The semiaquatic and aquatic hemiptera of California (Heteroptera: Hemiptera) (Menke A, ed). Berkeley: University of California Press. Miller J, 2005a. Cave adaptation in the spider genus Anthrobia (Araneae, Linyphiidae, Erigoninae). Zoologica Scripta 34:565-592. Miller RR, 1976. Geographical distribution of Central American freshwater fishes. In: Investigations of the ichthyofauna of Nicaraguan Lakes - A monumantal work 132 References

on Nicaraguan Fishes (Thorson TB, ed). Lincoln, Nebraska: The school of life sciences, University of Nebraska Lincoln; 125-155. Miller RR, 2005b. Freshwater fishes of Mexico. Chicago: University of Chicago Press. Moller PR, Schwarzhans W, Iliffe TM, Nielsen JG, 2006. Revision of the Bahamian cave-fishes of the genus Lucifuga (Ophidiiformes, Bythitidae), with description of a new species from islands on the Little Bahama Bank. Zootaxa 1223:23-46. Montgomery JC, Coombs S, Baker CF, 2001. The mechanosensory lateral line system of the hypogean form of Astyanax fasciatus. Environmental Biology of Fishes 62:87-96. Moore J, Gow J, Taylor EB, Hendry AP, 2007. Quantifying the constraining influence of gene flow on adaptive divergence in the lake-stream threespine stickleback system. Evolution 61:2015-2026. Mühlenberg M, 1993. Freilandökologie, 3. Auflage ed. Heidelberg: UTB, Quelle und Meyer. National Research Council, 1979. Hydrogen sulfide. Baltimore: University Park Press. Nelson D, Jannasch H, 1983. Chemoautotrophic growth of marine Beggiatoa in sulfide-gradient cultures. Archives of Microbiology 136:262-269. Nesbit D, Meffe G, 1993. Cannibalism frequencies in wild populations of the Eastern Mosquitofish (Gambusia holbrooki: Poeciliidae) in South Carolina. Copeia 1993:867-870. Nicholls P, 1975. The effect of sulphide on cytochrome aa3. Isosteric and allosteric shifts of the reduced alpha-peak. Biochemistry Biophysics Acta 396:24-35. Nosil P, Crespi B, 2006. Experimental evidence that predation promotes divergence in adaptive radiation. Proceedings of the National Academy of Sciences USA 103:9090-9095. Nosil P, Crespi B, Sandoval C, 2002. Host-plant adaptation drives the parallel evolution of reproductive isolation. Nature 417:440-443. Nosil P, Vines TH, Funk DJ, 2005. Perspective: Reproductive isolation caused by natural selection against immigrants from divergent habitats. Evolution 59:705-719.

References 133

Nybakken J, 2001. Marine biology, an ecological approach. San Francisco: Benjamin Cummings. Opsahl S, Chanton J, 2006. Isotopic evidence for methane-based chemosynthesis in the Upper Floridan aquifer food web. Oecologia 150:89-96. Oseid DM, Smith Jr. LL, 1974. Chronic toxicity of hydrogen sulfide to Gammarus pseudolimneatus. Transactions of the American Fisheries Society 103:819- 822. Palaima A, 2007. The fitness cost of generalization: present limitations and future possible solutions. Biological Journal of the Linnean Society 90:583-590. Park C, Nagel R, Blumberg W, Peisach J, Maliozzo R, 1986. Sulfhemoglobin: properties of partially sulfurated tetramers. Journal of Biological Chemistry 261:8805-8810. Parzefall J, 1969. Zur vergleichenden Ethologie verschiedener Mollienesia-Arten einschliesslich einer Höhlenform von M. sphenops. Behaviour 33:1-37. Parzefall J, 1970. Morphologische Untersuchungen an einer Höhlenform von Mollienesia sphenops (Pisces, Poeciliidae). Zeitschrift für Morphologie der Tiere 68:323-342. Parzefall J, 1973. Attraction and sexual cycle of poeciliids. In: Genetics and mutagenesis of fish (Schroeder J, ed). Berlin: Springer Verlag; 177-183. Parzefall J, 1974. Rückbildung aggressiver Verhaltensweisen bei einer Höhlenform von Poecilia sphenops (Pisces, Poeciliidae). Zeitschrift für Tierpsychologie 35:66-84. Parzefall J, 1993. Behavioural ecology of cave-dwelling fishes. In: Behaviour of teleost fishes, 2nd Edition (Pitcher TJ, ed). London: Chapman and Hall; 573- 608. Parzefall J, 2001. A review of morphological and behavioural changes in the cave molly, Poecilia mexicana, from Tabasco, Mexico. Environmental Biology of Fishes 62:263-275. Parzefall J, Kraus C, Tobler M, Plath M, 2007. Photophilic behaviour in surface- and cave-dwelling Atlantic mollies Poecilia mexicana (Poeciliidae). Journal of Fish Biology 71:1225-1231. Peakall R, Smouse P, 2001. GenAIEx v5; genetic analysis in Excel. Population genetic software for teaching and research. Canberra: Australian National University. 134 References

Peek AS, Feldmann R, Lutz R, Vrijenhoek RC, 1998. Conspeciation of chemoautotrophic bacteria and deep sea clams. Proceedings of the National Academy of Sciences USA 95:9962-9966. Peters N, Peters G, 1968. Zur genetischen Interpretation morphologischer Gesetzmässigkeiten der degenerative Evolution. Zeitschrift für Morphologie der Tiere 62:211-244. Peters N, Peters G, Parzefall J, Wilkens H, 1973. Über degenerative und konstruktive Merkmale bei einer phylogenetisch jungen Höhlenform von Poecilia sphenops (Pisces, Poeciliidae). Internationale Revue der Gesamten Hydrobiologie 58:417-436. Petersen L, 1977. The effect of inhibitors on the oxygen kinetics of cytochome c oxidase. Biochimica Et Biophysica Acta 460:299-307. Peterson B, Fry B, 1987. Stable isotopes in ecosystem studies. Annual Review of Ecology and Systematics 18:293-320. Pfennig DW, Rice AM, 2007. An experimental test of character displacement's role in promoting postmating isolation between conspecific populations in contrasting competitive environments. Evolution 61:2433-2443. Pianka E, 1973. The structure of lizard communities. Annual Review of Ecology and Systematics 4:53-74. Pianka E, 1986. Ecology and natural history of desert lizards. Princeton: Princeton University Press. Pigliucci M, 1996. How organisms respond to environmental changes: from phenotypes to molecules (and vice versa). Trends in Ecology & Evolution 11:168-173. Piry S, Alapetite A, Cornuet J, Peatkau D, Baudouin L, Estoup A, 2004. GENECLASS2: a software for genetic assignment and first-generation migrant detection. Journal of Heredity 95:536-539. Pisarowicz J, 2005. Return to Tabasco. Association for Mexican Cave Studies Newsletter 28:27-57. Plath M, 2008. Male mating behavior and costs of sexual harassment for females in cavernicolous and extremophile populations of Atlantic mollies (Poecilia mexicana). Behaviour 145:73-98. Plath M, Hauswaldt S, Moll K, Tobler M, Garcia de Leon F, Schlupp I, Tiedemann R, 2007a. Local adaptation and pronounced genetic differentiation in an

References 135

extremophile fish, Poecilia mexicana, inhabiting a Mexican cave with toxic hydrogen sulfide. Molecular Ecology 16:967-976. Plath M, Heubel KU, Garcia de Leon FJ, Schlupp I, 2005. Cave molly females (Poecilia mexicana, Poeciliidae, Teleostei) like well-fed males. Behavioral Ecology and Sociobiology 58:144-151. Plath M, Makowicz A, Schlupp I, Tobler M, 2007b. Sexual harassment in livebearing fishes (Poeciliidae): comparing courting and noncourting species. Behavioral Ecology 18:680-688. Plath M, Parzefall J, Körner K, Schlupp I, 2004. Sexual selection in darkness? Female mating preferences in surface- and cave-dwelling Atlantic mollies, Poecilia mexicana (Poeciliidae, Teleostei). Behavioral Ecology and Sociobiology 55:596-601. Plath M, Parzefall J, Schlupp I, 2003a. The role of sexual harassment in cave- and surface-dwelling populations of the Atlantic molly, Poecilia mexicana (Poeciliidae, Teleostei). Behavioral Ecology and Sociobiology 54:303-309. Plath M, Schlupp I, 2008. Parallel evolution leads to reduced shoaling behavior in two cave dwelling populations of Atlantic mollies (Poecilia mexicana, Poeciliidae, Teleostei). Environmental Biology of Fishes in press. Plath M, Seggel U, Burmeister H, Heubel KU, Schlupp I, 2006. Choosy males from the underground: male mating preferences in surface- and cave-dwelling Atlantic mollies (Poecilia mexicana). Naturwissenschaften 93:103-109. Plath M, Tobler M, Riesch R, Garcia de Leon FJ, Giere O, Schlupp I, 2007c. Survival in an extreme habitat: the role of behaviour and energy limitation. Naturwissenschaften 94:991-996. Plath M, Wiedemann K, Parzefall J, Schlupp I, 2003b. Sex recognition in surface and cave dwelling male atlantic mollies Poecilia mexicana (Poeciliidae, Teleostei). Behaviour 140:765-781. Porter M, Crandall K, 2003. Lost along the way: the significance of evolution in reverse. Trends in Ecology & Evolution 18:541-547. Post D, 2002. Using stable isotopes to estimate trophic position: models, methods, and assumptions. Ecology 83:703-718. Poulson TL, 1963. Cave adaptation in amblyopsid fishes. American Midland Naturalist 70:257-290. 136 References

Poulson TL, 2001. Morphological and physiological correlates of evolutionary reduction of metabolic rate among amblyopsid cave fishes. Environmental Biology of Fishes 62:239-249. Poulson TL, Lavoie KH, 2000. The trophic basis of subterranean ecosystems. In: Ecosystems of the world 30: Subterranean ecosystems (Wilkens H, Culver DC, Humphries WF, eds). Amsterdam: Elsevier Science; 231-249. Poulson TL, White WB, 1969. The cave environment. Science 165:971-981. Price A, 2002. Simultaneous ‘hotspots’ and ‘coldspots’ of marine biodiversity and implications for global conservation. Marine Ecology Series 241:23-27. Pritchard JK, Stephens M, Donnelly P, 2000. Inference of population structure using multilocus genotype data. Genetics 155:945-959. Protas ME, Conrad M, Gross J, Tabin C, Borowsky R, 2007. Regressive evolution in the Mexican Cave tetra, Astyanax mexicanus. Current Biology 17:452-454. Protas ME, Hersey C, Kochanek D, Zhou Y, Wilkens H, Jeffery WR, Zon LI, Borowsky R, Tabin CJ, 2006. Genetic analysis of cavefish reveals molecular convergence in the evolution of albinism. Nature Genetics 38:107-111. Proudlove G, 2006. An account of the subterranean (hypogean) fishes described up to 2003 with a bibliography 1541-2004. Moulis: International Society for Subterranean Biology. Ptacek MB, 2002. Patterns of inheritance of mating signals in interspecific hybrids between sailfin and shortfin mollies (Poeciliidae: Poecilia: Mollienesia). Genetica 116:329-342. Quesada H, Posada D, Caballero A, Moran P, Rolan-Alvarez E, 2007. Phylogenetic evidence for multiple sympatric ecological diversification in a marine snail. Evolution 61:1600-1612. Railsback S, Lamberson R, Harvey B, Duffy W, 1999. Movement rules for individual-based models of stream fish. Ecological Modelling 123:73-89. Rajakaruna N, Baldwin B, Chan R, Desrochers A, Bohm B, Whitton J, 2003. Edaphic races and phylogenetic taxa in the Lashenia californica complex (Asteraceae: Heliantheae): an hypothesis of parallel evolution. Molecular Ecology 12:1675- 1679. Rannala B, Mountain J, 1997. Detecting immigration by using multilocus genotypes. Proceedings of the National Academy of Sciences USA 94:9197-9221.

References 137

Räsänen K, Hendry AP, 2008. Disentangling interactions between adaptive divergence and gene flow when ecology drives diversification. Ecology Letters 11:624-636. Reiffenstein R, Hulbert W, Roth S, 1992. Toxicology of hydrogen sulfide. Annual Reviews of Pharmacology and Toxicology 1992:109-134. Reis R, Trajano E, Hingst-Zaher E, 2006. Shape variation in surface and cave populations of the armoured catfishes Ancistrus (Siluriformes: Loricariidae) from the Sao Domingos karst area, upper Tocantins River, Brazil. Journal of Fish Biology 68:414-429. Relyea KA, 2001. Morphological and behavioral plasticity of larval anurans in presonse to different predators. Ecology 82:523-540. Reznick D, Endler JA, 1982. The impact of predation on life history evolution in trinidadian guppies (Poecilia reticulata). Evolution 36:160-177. Rice W, Salt G, 1990. The evolution of reproductive isolation as a correlated character under sympatric conditions: experimental evidence. Evolution 44:1140-1152. Riesch R, Tobler M, Schlupp I, Plath M, in press. Offspring number in a livebearing fish (Poecilia mexicana, Poeciliidae): reduced fecundity and reduced plasticity in a population of cave mollies. Environmental Biology of Fishes. Roast S, Jones M, 2003. Morphometrics and gill function in Gammarus duebeni (Peracarida: Amphipoda) inhabiting a sewage treatment works. Marine Ecology Progress Series 260. Robinson BW, Wilson DS, 1995. Experimentally-induced morphological diversity in Trinidadian guppies (Poecilia reticulata). Copeia 1995:294-305. Rodriguez R, Sullivan L, Cocroft R, 2004. Vibrational communication and reproductive isolation in the Enchenopa binotata species complex of treehoppers (Hemiptera: Membracidae). Evolution 58:571-578. Rohlf F, 2004. tpsDig. Available from http://lifebiosunysbedu/morph/. Rohlf F, 2005. tpsRegr. Available from http://lifebiosunysbedu/morph/. Rohlf F, 2006. tpsUtil. Available from http://lifebiosunysbedu/morph/. Rohlf F, 2007. tpsRelw. Available from http://lifebiosunysbedu/morph/. Romero A, Green SM, 2005. The end of regressive evolution: examining and interpreting the evidence from cave fishes. Journal of Fish Biology 67:3-32. 138 References

Romero A, Singh A, McKie A, Manna M, Baker R, Paulson KM, Creswell JE, 2002. Replacement of the troglomorphic population of Rhamdia quelen (Pisces : Pimelodidae) by an epigean population of the same species in the Cumaca Cave, Trinidad, West Indies. Copeia:938-942. Rosen D, Bailey R, 1963. The poeciliid fishes (Cyprinodontiformes), their structure, zoogeography and systematics. Bulletin of the American Museum of Natural History 126:1-176. Rosenthal GG, de la Rosa X, Kazianis S, Stephens MJ, Morizot DC, Ryan MJ, Garcia de Leon FJ, 2003. Dissolution of sexual signal complexes in a hybrid zone between the swordtails Xiphophorus birchmanni and Xiphophorus malinche (Poeciliidae). Copeia 2003:299-307. Rothschild L, Mancinelli R, 2001. Life in extreme environments. Nature 409:1092- 1101. Rózsa L, Reiczigel J, Majoros G, 2000. Quantifying parasites in samples of hosts. Journal of Parasitology 86:228-232. Ruehl C, DeWitt TJ, 2007. Trophic plasticity and fine-grained resource variation in populations of western mosquitofish, Gambusia affinis. Evolutionary Ecology Research 7:801-819. Rundle HD, Nagel L, Boughman J, Schluter D, 2000. Natural selection and parallel speciation in sympatric sticklebacks. Science 287:306-308. Rundle HD, Nosil P, 2005. Ecological speciation. Ecology Letters 8:336-352. Ryan P, Bloomer P, Moloney C, Grant T, Delport W, 2007. Ecological speciation in south Atlantic island finches. Science 315:1420-1422. Sarbu SM, Kane TC, Kinkle BK, 1996. A chemoautotrophically based cave ecosystem. Science 272:1953-1955. Sarrazin J, Juniper S, 1999. Biological characteristics of a hydrothermal edifice mosaic community. Marine Ecology Series 185:1-19. Schartl M, 1995. Platyfish and swordtails: a genetic system for the analysis of molecular mechanisms in tumor formation. Trends in Genetics 11:185-189. Schilthuizen M, Cabanban A, Haase M, 2005. Possible speciation with gene flow on tropical cave snails. Journal of Zoological Systematics and Evolutionary Research 43:133-138. Schluter D, 2000. The ecology of adaptive radiation. Oxford: Oxford University Press.

References 139

Schluter D, 2001. Ecology and the origin of species. Trends in Ecology & Evolution 16:372-380. Schluter D, McPhail J, 1992. Ecological character displacement and speciation in sticklebacks. American Naturalist 140:85-108. Schmidt T, Bielawski J, Gold J, 1998. Molecular phylogenetics and evolution of the cytochrome b gene in the cyprinid genus Lythrurus (Actinopterygii: Cypriniformes). Copeia 1998:14-22. Schneider S, Roessli D, Excoffier L, 2000. Arlequin: a software for population genetics data analysis, Version 2.0. Geneva: Genetics and Biometry Laboratory, Department of Anthropology, University of Geneva. Schoener T, 1971. Theory of feeding strategies. Annual Review of Ecology and Systematics 2:369-404. Servedio MR, Noor MAF, 2003. The role of reinforcement in speciation: Theory and data. Annual Review of Ecology, Evolution and Systematics 34:339-364. Silvertown J, Servaes C, Biss P, Macleod D, 2005. Reinforcement of reproductive isolation between adjacent populations in the Park Grass Experiment. Heredity 95:198-205. Simpson E, 1949. Measurements of diversity. Nature 163:688. Sinervo B, Bleay C, Adamopoulou C, 2001. Social causes of correlational selection and the resolution of a heritable throat clolor polymorphism in a lizard. Evolution 55:2040-2052. Sket B, 1985. Why all cave animals do not look alike: a discussion on adaptive value of reduction processes. NSS Bulletin 47:78-85. Smith Jr. LL, Oseid DM, Kimball GL, El-Kandelgy SM, 1976. Toxicity of hydrogen sulfide to various life history stages of the bluegill (Lepomis macrochirus). Transactions of the American Fisheries Society 105:442-449. Smith L, Kruszynah H, Smith RP, 1977. The effect of methemoglobin on the inhibition of cytochrome c oxidase by cyanide, sulfide or azide. Biochemical Pharmacology 26:2247-2250. Spellman SJ, Johnson AD, 1987. In vitro encystment of the black spot trematode Uvulifer ambloplitis (Trematoda, Diplostomatidae). International Journal of Parasitology 17:897-902. 140 References

Spitze K, 1992. Predator-mediated plasticity of prey life history and morphology: Chaoborus americanus predation on Daphnia pulex. American Naturalist 139:229-247. Stawikowski R, Werner U, 1998. Die Bundbarsche Amerikas, Band I. Stuttgart: Eugen Ulmer Verlag. Steiniger T, Haldimann P, Leiss K, Müller-Schärer H, 2002. Does natural selection promote population divergence? A comparative analysis of population structure using amplified fragment length polymorphism markers and quatitative traits. Molecular Ecology 11:2583-2590. Storfer A, Sih A, 1998. Gene flow and ineffective antipredator behavior in a stream- breeding salamander. Evolution 52:558-565. Strecker U, 2003. Genetic divergence between cave and surface populations of Astyanax in Mexico (Characidae, Teleostei). Molecular Ecology 12:699-710. Streelman JT, Danley PD, 2003. The stages of vertebrate evolutionary radiation. Trends in Ecology and Evolution 18:126-131. Streever WJ, 1996. Energy economy hypothesis and the troglobitic crayfish Procambarus erythrops in Sim's sink cave, Florida. American Midland Naturalist 135:357-366. Sultan S, Spencer H, 2002. Metapopulation structure favors plasticity over local adaptation. American Naturalist 160:271-283. Svanbäck R, Bolnick D, 2007. Intraspecific competition drives increased resource diversity within a natural population. Proceedings of the Royal Society B - Biological Sciences 274:839-844. Swart C, Felgenhauer B, 2003. Structure and function of the mouthparts and salivary gland complex of the Giant Waterbug, Belostoma lutarium (Stal) (Hemiptera: Belostomatidae). Annals of the Entomological Society of America 96:870- 882. Swenson N, Enquiest B, 2007. Ecological and evolutionary determinats of a key plant functional trait: wood density and its community-wide variation across latitude and elevation. American Journal of Botany 94:451-459. Templeton A, Crandall K, Sing C, 1992. A cladistic analysis of phenotypic associations with haplotypes inferred from restriction endonuclease mapping and DNA sequence data. III. Cladogram estimation. Genetics 132:619-633.

References 141

Theede H, 1973. Comparative studies on the influence of oxygen deficiency and hydrogen sulphide on marine bottom invertebrates. Netherlands Journal of Sea Research 7:245-252. Tiedemann R, Moll K, Paulus KB, Schlupp I, 2005. New microsatellite loci confirm hybrid origin, parthenogenetic inheritance, and mitotic gene conversion in the gynogenetic Amazon molly (Poecilia formosa). Molecular Ecology Notes 5:586-589. Timmerman CM, Chapman LJ, 2004. Hypoxia and interdemic variation in Poecilia latipinna. Journal of Fish Biology 65:635-650. Tobler M, in press. Divergence in trophic ecology characterises colonisation of extreme habitats. Biological Journal of the Linnean Society. Tobler M, DeWitt TJ, Schlupp I, Garcia de Leon FJ, Herrmann R, Feulner P, Tiedemann R, Plath M, in press. Toxic hydrogen sulfide and dark caves: Phenotypic and genetic divergence across two abiotic environmental gradients in Poecilia mexicana. Evolution. Tobler M, Franssen C, Plath M, 2008a. Male-biased predation on a cave fish by a giant water bug. Naturwissenschaften 95:775-779. Tobler M, Riesch R, Garcia de Leon FJ, Schlupp I, Plath M, 2008b. A new and morphologically distinct cavernicolous population of Poecilia mexicana (Poeciliidae, Teleostei). Environmental Biology of Fishes 82:101-108. Tobler M, Riesch R, Garcia de Leon FJ, Schlupp I, Plath M, 2008c. Two endemic and endangered fishes, Poecilia sulphuraria (Alvarez, 1948) )and Gambusia eurystoma Miller, 1975 (Poeciliidae, Teleostei), as only survivors in a small sulfidic habitat. Journal of Fish Biology 72:523-533. Tobler M, Schlupp I, 2005. Parasites in sexual and asexual mollies (Poecilia, Poeciliidae, Teleostei): a case for the Red Queen? Biology Letters 1:166-168. Tobler M, Schlupp I, Garcia de Leon FJ, Glaubrecht M, Plath M, 2007a. Extreme habitats as refuge from parasite infections? Evidence from an extremophile fish. Acta Oecologica 31:270-275. Tobler M, Schlupp I, Heubel K, Riesch R, Garcia de Leon FJ, Giere O, Plath M, 2006. Life on the edge: Hydrogen sulfide and the fish communities of a Mexican cave and surrounding waters. Extremophiles 10:577-585. 142 References

Tobler M, Schlupp I, Plath M, 2007b. Predation of a cave fish (Poecilia mexicana, Poeciliidae) by a giant water-bug (Belostoma, Belostomatidae) in a Mexican sulfur cave. Ecological Entomology 32:492-495. Tobler M, Wahli T, Schlupp I, 2005. Comparison of parasite communities in native and introduced populations of sexual and asexual mollies of the genus Poecilia. Journal of Fish Biology 67:1072-1082. Torrans E, Clemens H, 1982. Physiological and biochemical effects of acute exposure of fish to hydrogen sulfide. Comparative Biochemistry and Physiology 71C:183-190. Townsend CR, Begon ME, Harper JL, 2003. Essentials of ecology, 2nd edition. Oxford: Blackwell Publishing. Trajano E, 1995. Evolution of tropical troglobites: Applicability of the model of Quaternary climatic fluctuations. Mémoires de Biospéologie 22:203-209. Trexler J, Tempe R, Travis J, 1994. Size-selective predation of Sailfin mollies by two species of heron. Oikos 69:250-259. Tsurumi M, 2003. Diversity at hydrothermal vents. Global Ecology and Biogeography 12:181-190. Turgeon J, Estoup A, Bernatchez L, 1999. Species flock in the Norh American Great Lakes: Molecular ecology of Lake Nipigon Ciscoes (Teleostei: Coregonidae: Coregonus). Evolution 53:1857-1871. Turner BJ, 1984. Evolutionary genetics of fishes. New York: Plenum Press. Tyerman J, Bertrand M, Spencer C, Doebeli M, 2008. Experimental demonstration of ecological character displacement. Bmc Evolutionary Biology 8:34. Van Dover CL, 2000. The ecology of deep-sea hydrothermal vents. Princeton: Princeton University Press. Van Tienderen P, 1991. Evolution of generalists and specialists in spatially heterogeneous environments. Evolution 45:1217-1331. Verovnik R, Sket B, Trontelj P, 2004. Phylogeography of subterranean and surface populations of water lice Asellus aquaticus (Crustacea: Isopoda). Molecular Ecology 13:1519-1532. Völkel S, Grieshaber MK, 1992. Mechanisms of sulfide tolerance in the peanut worm Sipunculus mudus (Sipunculida) and in the lungworm Arenicola marina (Polychaeta). Journal of Comparative Physiology B 162:469-477.

References 143

Wainwright PC, 1996. Ecological explanation through functional morphology: The feeding biology of sunfishes. Ecology 77:1336-1343. Walters L, Walters V, 1965. Laboratory observations on a cavernicolous poeciliid from Tabasco, Mexico. Copeia 1965:214-233. Wang H, McArthur E, Sanderson S, Graham J, Freeman D, 1997. Narrow hybrid zone between two subspecies of big sagebush (Artemisia tridentata: Asteraceae). IV. Reciprocal transplant experiments. Evolution 51:95-102. Weber A, Horst R, Barbier G, Oesterhelt C, 2007. Metabolism and metabolomics of eukaryotes living under extreme conditions. International Review of Cytology 256:1-34. Weber JM, Kramer DL, 1983. Effects of hypoxia and surface access on growth, mortality, and behavior of juvenile Guppies, Poecilia reticulata. Canadian Journal of Fisheries and Aquatic Sciences 40:1583-1588. Weir B, 1996. Genetic data analysis II. Sunderland: Sinauer. Weir B, Cockerham C, 1984. Estimating F-statistics for the analysis of population structure. Evolution 38:1358-1370. West-Eberhard M, 1989. Phenotypic plasticity and the origins of diversity. Annual Review of Ecology and Systematics 20:249-278. Wilkens H, 1973. Über das phylogenetische Alter von Höhlentieren: Untersuchungen über die cavernicole Süsswasserfauna Yucatans. Zeitschrift für zoologische Systematik und Evolutionsforschung 11:49-60. Wilkens H, 1979. Reduktionsgrad und phylogenetisches Alter: Ein Beitrag zur Besiedlungsgeschichte der Limnofauna Yukatans. Zeitschrift fur zoologische Systematik und Evolutionsforschung 17:262-272. Wilkens H, 1988. Evolution and genetics of epigean and cave Astyanax fasciatus (Characidae, Pisces): support for the neutral mutation theory. Evolutionary Biology 23:271-367. Wilkens H, 2007. Regressive evolution: ontogeny and genetics of cavefish eye rudimentation. Biological Journal of the Linnean Society 92:287-296. Wilkens H, Strecker U, 2003. Convergent evolution of the cavefish Astyanax (Characidae, Teleostei): genetic evidence from reduced eye-size and pigmentation. Biological Journal of the Linnean Society 80:545-554. Willis L, Brown A, 1985. Distribution and habitat requirements of the Ozark cavefish, Amblyopsis rosae. American Midland Naturalist 114:311-317. 144 References

Wilson DS, Yoshimura J, 1994. On the coexistence of specialists and generalists. American Naturalist 144:692-707. Winemiller KO, 1990. Spatial and temporal variation in tropical fish trophic networks. Ecological Monographs 60:331-367. Winemiller KO, 1992. Life-history strategies and the effectiveness of sexual selection. OIKOS 63:318-327. Winemiller KO, 1993. Seasonality of reproduction by livebearing fishes in tropical rainforest streams. Oecologia 95:266-276. Winemiller KO, Kelso-Winemiller LC, Brenkert AL, 1995. Ecomorphological diversification and convergence in fluvial cichlid fishes. Environmental Biology of Fishes 44:235-261. Xiao H, Chen SY, Liu ZM, Zhang RD, Li WX, Zan RG, Zhang YP, 2005. Molecular phylogeny of Sinocyclocheilus (Cypriniformes : Cyprinidae) inferred from mitochondrial DNA sequences. Molecular Phylogenetics and Evolution 36:67- 77. Zeiske E, 1968. Prädispositionen bei Mollienesia sphenops (Pisces, Poeciliidae) für einen Übergang zum Leben in subterranen Gewässer. Zeitschrift für vergleichende Physiologie 58:190-222. Zelditch M, Swiderski D, Sheets H, Fink W, 2004. Geometric morphometrics for biologists. Amsterdam: Elsevier Academic Press. Zihler F, 1982. Gross morphology and configuration of digestive tracts of cichlidae (Teleostei, Perciformes): phylogenetic and functional significance. Netherlands Journal of Zoology 32:544-571.

Acknowledgements 145

ACKNOWLEDGEMENTS First of all, I thank Ingo Schlupp for giving me the opportunity for this thesis, his scientific and financial support, as well as the freedom to develop myself and my projects in any direction I desired. I also thank Uli Reyer and William Matthews for hosting me at the Universities of Zürich and Oklahoma, respectively. Without their continuous support in many aspects of my work, this thesis would not have been possible. My sincerest thanks further go to Martin Plath for the years of excellent collaboration, his financial support and – most importantly – his friendship even in rougher times. With the unconditional help of Francisco García de León, I was able to pursue my research in the most beautiful country on earth. A suite of collaborators, faculty members, fellow researchers, and graduate and undergraduate students kept my mind going, challenged me, introduced me to new methods and ways of thinking, provided equipment and other resources, spent their free time helping in the field as well as in the laboratory, and read through stacks of manuscript drafts. Specifically, I want to thank Rickey Cothran, Thom DeWitt, Matt Dugas, Ola Fincke, Courtney Franssen, Nate Franssen, Dave Gillette, Hartmut Greven, Puni Jeyasingh, Sarah Kern, Rosemary Knapp, Lee Krumholz, Hernan Lopez-Fernandez, Amber Makowicz, Edie Marsh-Matthews, Bill Matthews, Rudy Riesch, Laura Rosales la Garde, Mike Ryan, Annette Sauter, Manfred Schartl, Gioia Schwarzenbach, Don Shepard, Jon Shik, Peter Unmack, Christoph Vorburger, Larry Weider, Gary Wellborn, and Rich Zamor. I thank Anni Mäder in Zürich as well as Carol Baylor, George Davis, George Martin, Wendal Porter, and Robbie Stinchcomb in Oklahoma for all kind of administrational and technical support. I also thank the people of Tapijulapa and Teapa for their hospitality as well as the Municipal in Tacotalpa for logistical support. I am deeply indebted to my extended family in Switzerland – especially my dad – and the US as well as my friends. You gave me moral support whenever needed, made me laugh and even relax at times. Ally, Lea, and Courtney provided me with the fuel that kept me going every day. 146 Acknowledgements

The Mexican government issued the permits to conduct this research (291002-613-1577, DGOPA/5864/260704/-2408, and 16986/191/205/-8101). The project was approved by the Animal Care and Use Committee of the University of Oklahoma (AUS R07-004). Financial support came from the American Livebearer Association, the Basler Foundation for Biological Research, the German Ichthyological Association, the Janggen Pöhn Foundation, the Roche Research Foundation, the Universities of Oklahoma and Zürich, and the Wolfermann Nägeli Foundation.

Curriculum vitae 147

CURRICULUM VITAE

Personal Name Michael Tobler Birth 26. November 1980 in Zürich, Switzerland Citizenship Thal, SG, Switzerland

Education 2004-2008 Dissertation at the Institute of Zoology, University of Zürich: “The evolutionary ecology of Poecilia mexicana in the Cueva del Azufre system: Effects of abiotic and biotic environmental conditions”. Supervised by Prof. Dr. I. Schlupp (University of Oklahoma), Prof. Dr. H.-U. Reyer (University of Zürich), and Prof. Dr. L. Keller (University of Zürich). Affiliated research scholar at the Department of Zoology, University of Oklahoma (Norman, USA). 2003-2004 Diploma thesis at the ETH Zürich: “Sex and parasites in a gynogenetic complex of fishes". Supervision: Prof. Dr. P. Schmid-Hempel (ETH Zürich) and PD Dr. Ingo Schlupp (University of Zürich). 2000-2004 Studies of biology at the ETH Zürich with special emphasis on biosystematics, behavioral ecology, evolutionary biology, population biology, and aquatic ecology. 1995-2000 Kantonsschule Romanshorn, Matura, Typus C.

148 Curriculum vitae

Publications • Tobler, M., T. J. DeWitt, I. Schlupp, F. J. Garcia de Léon, R. Herrmann, P. G. D. Feulner, R. Tiedemann & M. Plath (in press): Toxic hydrogen sulfide and dark caves: Phenotypic and genetic divergence across two environmental gradients in Poecilia mexicana. Evolution. • Riesch, R., M. Tobler, M. Plath & I. Schlupp (in press): Offspring number in a livebearing fish (Poecilia mexicana, Poeciliidae): reduced fecundity and reduced plasticity in a population of cave mollies. Environmental Biology of Fishes. • Tobler, M. (in press): Divergence in trophic ecology characterises colonisation of extreme habitats. Biological Journal of the Linnean Society. • Tobler, M., I. Schlupp & M. Plath (in press): Does divergence in female mate choice affect male size distributions in two cave fish populations? Biology Letters. • Franssen, C. M., M. Tobler, R. Riesch, F. J. Garcia de Léon, R. Tiedemann, I. Schlupp & M. Plath (in press): Sperm production in an extremophile fish, the cave molly (Poecilia mexicana, Poeciliidae, Teleostei). Aquatic Ecology. • Tobler, M. & I. Schlupp (2008): Expanding the horizon: The Red Queen and potential alternatives. Canadian Journal of Zoology 86 (8): 765-773. • Tobler, M., C. M. Franssen & M. Plath (2008): Male-biased predation on a cave fish by a giant water bug. Naturwissenschaften 95 (8): 775-779. • Tobler, M., I. Schlupp, R. Riesch, F. J. García de León & M. Plath (2008): A new and morphologically distinct cavernicolous population of Poecilia mexicana (Poeciliidae, Teleostei). Environmental Biology of Fishes 82: 101-108. • Tobler, M., R. Riesch, F. J. García de León, I. Schlupp & M. Plath (2008): Two endemic and endangered fishes, Poecilia sulphuraria and Gambusia eurystoma (Poeciliidae, Teleostei), as only survivors in a small sulfidic habitat. Journal of Fish Biology 72: 1-11. • Tobler, M. & I. Schlupp (2008): Influence of black spot disease on shoaling behaviour in female western mosquitofish, Gambusia affinis (Poeciliidae, Teleostei). Environmental Biology of Fishes 81 (1): 29-34. • Plath, M., M. Tobler, R. Riesch, F. J. Garcia de Léon, O. Giere & I. Schlupp (2007): Survival in an extreme habitat: the role of behavior and energy limitation. Naturwissenschaften 94: 991-996.

Curriculum vitae 149

• Tobler, M., I. Schlupp & M. Plath (2007): Predation of cave fish (Poecilia mexicana, Poeciliidae) by a giant water-bug (Belostoma, Belostomatidae) in a Mexican sulfur cave. Ecological Entomology 32 (5): 492-495. • Parzefall, J., C. Kraus, M. Tobler & M. Plath (2007): Photophilic behaviour in a cave fish, Poecilia mexicana (Poeciliidae). Journal of Fish Biology 71 (4): 1225- 1231. • Plath, M. & M. Tobler (2007): Sex recognition in surface- and cave-dwelling Atlantic molly females (Poecilia mexicana, Poeciliidae, Teleostei): influence of visual and non-visual cues. Acta Ethologica 10 (2): 81-88. • Tobler, M. (2007): Reversed sexual dimorphism and female courtship in the Topaz cichlid, Archocentrus myrnae (Cichlidae, Teleostei), from Costa Rica. Southwestern Naturalist 52: 371-377. • Schlupp, I., R. Riesch & M. Tobler (2007): Quick guide – Amazon mollies. Current Biology 17: R536-R537. • Plath, M., A. M. Makowicz, I. Schlupp & M. Tobler (2007): Sexual harassment in livebearing fishes (Poeciliidae): comparing courting and non-courting species. Behavioral Ecology 18: 680-688. • Tobler, M., I. Schlupp, F. J. Garcia de Leon, M. Glaubrecht & M. Plath (2007): Extreme habitats as refuge from parasite infections? Evidence from an extremophile fish. Acta Oecologica 31: 270-275. • Plath, M., S. Hauswaldt, K. Moll, M. Tobler, F. J. Garcia de Léon, I. Schlupp & R. Tiedemann (2007): Local adaptation and pronounced genetic differentiation in an extremophile fish, Poecilia mexicana, inhabiting a Mexican cave with toxic hydrogen sulfide. Molecular Ecology 16: 967-976. • Tobler, M., H. Burmeister, I. Schlupp & M. Plath (2006): Regressive evolution of visually mediated preferences in the Cave molly (Poecilia mexicana, Poeciliidae, Teleostei). Subterranean Biology 4: 59-65. • Plath, M. & M. Tobler (2006): Coercive mating and genitalia size in two populations of a livebearing toothcarp (Poecilia mexicana): do Cave molly males have shorter gonopodia? Zeitschrift für Fischkunde 8 (1-2): 103-107. • Tobler, M. (2006): The eggspots of cichlids: Evolution through sensory exploitation? Zeitschrift für Fischkunde 8 (1-2): 39-46. 150 Curriculum vitae

• Schlupp, I., J. Poschadel, M. Tobler & M. Plath (2006): Male size polymorphism and testis weight in two species of mollies (Poecilia latipinna, P. mexicana, Poeciliidae, Teleostei). Zeitschrift für Fischkunde 8 (1-2): 9-16. • Tobler, M., I. Schlupp, K. U. Heubel, R. Riesch, F. J. Garcia de Leon, O. Giere & M. Plath (2006): Life on the edge: Hydrogen sulfide and the fish communities of a Mexican cave and surrounding waters. Extremophiles 10 (6): 577-585. • Riesch, R., I. Schlupp, M. Tobler & M. Plath (2006): Reduction of the association preference for conspecifics in cave-dwelling Atlantic mollies, Poecilia mexicana. Behavioral Ecology and Sociobiology 60: 794-802. • Tobler, M., M. Plath, H. Burmeister & I. Schlupp (2006): Black spots and female association preferences in a sexual/ asexual mating complex (Poecilia, Poeciliidae, Teleostei). Behavioral Ecology and Sociobiology 60: 159-165. • Tobler, M., K. Wiedemann & M. Plath (2005): Homosexual behaviour in a cavernicolous fish, Poecilia mexicana (Poeciliidae, Teleostei). Zeitschrift für Fischkunde 7 (2): 95-99. • Tobler, M., T. Wahli & I. Schlupp (2005): Comparison of parasite communities in native and introduced populations of sexual and asexual mollies of the genus Poecilia (Poecliidae, Teleostei). Journal of Fish Biology 67 (4): 1072-1082. • Tobler, M. & I. Schlupp (2005): Parasites in sexual and asexual mollies (Poecilia, Poeciliidae, Teleostei): A case for the Red Queen? Biology Letters 1 (2): 166-168. • Tobler, M. (2005): Feigning death in the Central American cichlid Parachromis friedrichsthalii. Journal of Fish Biology 66 (3): 877-881. • Plath, M., M. Tobler & I. Schlupp (2004): Cave fish looking for mates: A visual mating preference in surface- and cave-dwelling Atlantic mollies (Poecilia mexicana, Poeciliidae). Zeitschrift für Fischkunde 7 (1): 61-69.