Molecular ecology and phylogenetics in formicine ants

The advent of molecular techniques has revolutionised most fields of biological sciences and has become indispensible for many areas of research. Evolutionary biology has especially benefited (Avise 2004). Comparative studies of any trait or character are possible only with sound knowledge about interrelationships of the organisms under investigation. DNA provides a completely independent approach to the traditionally morphology-based process to discern relationships of biological taxa and timelines of history. One advantage of molecular data over morphological ones is the amount of information that can be gathered, but also the relative independence of each data point to the others (Crozier 1983). Morphological techniques have had much and widespread success to establish natural groups but are bound to fail when characters converge because of similar ecological conditions and selection pressures. Further molecular data can detect differences in relatedness when little morphological variation is present (e.g. within populations and families). But molecular data are not a universally simple solution for all questions. As with morphological studies, only a thorough knowledge about the capabilities and potential pitfalls of the method can preclude wrong conclusions (Moreau 2009). Some problems like missing taxa or a lack of resolution are even the same for both approaches. But molecular data also suffer from other more specific problems. First, contaminations during the lab work are a source of error that may not always be traced back, but they can be mostly avoided by proper lab practices and documentation systems. Most problems, however, arise during the data analysis. For fragment analysis, for example stutter bands (resulting from mistakes of the polymerase) have to be separated from true data. But also null alleles (a missing band due to changes in the primer region) and a possible linkage of different markers have to be identified. Fortunately for all these problems statistical tests are available to exclude wrong inferences from incorrect usage of the raw data (Excoffier & al. 2005). For sequence-based methods other things have to be kept in mind. Here, pseudogenes (non-functional duplications of the gene of interest that underlie different selection pressures) can cause problems (e.g. Beckenbach 2009). Because of the inherent sequence similarity they often can be amplified with or instead of the correct fragment. Usually the distinction can be made when enough knowledge about the gene is available to detect stop codons, indels or frameshift mutations. Much care has to be placed on the choice of the markers, since the speed of change at the DNA level has to match the level of relatedness that should be inferred. Too little and too much change results in lack of resolution; in the first case through lack of information and in the latter through saturation of the marker and homoplasious characters (Felsenstein 2004). Therefore a test for a saturation effect needs to be carried out before proceeding to phylogenetic inference and the best model for the substitution process has to be tested (Posada 1998). Monotypic taxa can often have very long branches before they are connected to their extant relatives. If several such taxa are present in the data set, a problem called long-branch attraction can occur, when they share similar sequences just because they have undergone many changes and the number of characters is limited (Felsenstein 2004). This can happen within the group of interest or the outgroup taxa and can greatly influence the overall topology. In such cases taxon exclusion experiments can identify the extent of the problem and suggest solutions to reach the correct topology (Ward & al. 2010). With some care appropriate markers can be found to resolve any level of relatedness making DNA a uniformly applicable source of information. The advances in laboratory techniques have made this information easily accessible when standard equipment is at hand. Together with recent improvements in computer power, molecular data have gained the potential to successfully address fundamental questions in areas as diverse as systematics, biogeography, behaviour, ecology, population genetics and development (Avise 2004). While this list is by far not exhaustive, my thesis touches on most of these research fields. The bond uniting them is the evolutionary point of view from which they are addressed. The organisms of choice for my thesis are formicine ants. They share the eusocial life style of all ants. They live in colonies with reproductive division of labour among the females, cooperative brood care (by the workers) and overlapping adult generations. They are one of the most advanced eusocial groups and contain genera with highly specialized life histories (e.g. slave makers, weaver ants, honey pot ants, harvester ants, plant mutualists) (Hölldobler and Wilson 1990). The first three chapters are concerned with one such specialist. Oecophylla is the most advanced genus of weaver ants using larval silk for communal nest building (see Hölldobler and Wilson 1990 for a detailed discussion of weaving in ants). In Oecophylla the larvae are not fully-grown and almost completely passive during weaving. The possibility to control their immediate environment so closely has clearly contributed to the ecological dominance of these ants in the old world tropics. While their behaviour and chemical communication have been thoroughly studied by Hölldobler and Wilson (1977a, 1977b, 1983, 1990), we have hardly any knowledge about the breeding system, population genetics, dispersion and colony founding. The only reports are small studies (Peeters & Anderson 1989, Fraser & Crozier 1998) or side notes in old papers (Maxwell-Lefroy & Howlett 1909, Begg 1977). A thorough study of the genetic colony structure can afford us insights into breeding system, dispersal and colony founding of Oecophylla, but is also interesting to understand the evolution of social systems in general. Central topics like cooperation vs. conflict (e.g. sex allocation, reproductive competition), evolution of life history and selection on groups may be studied. In Oecophylla, like in most other ants, the workers specialise in certain behaviours, which leads to an increased efficiency of the whole group, i.e. the colony (Oster & Wilson 1978). This polyethism can arise through differences in morphology, age or genetics. Because of the influence of genetic composition of the colony, the breeding system and colony founding can have an impact on colony productiveness and reproductive success (Waibel & al. 2006). In chapter 1 my co-authors and I have reviewed the current knowledge about both species in the genus Oecophylla and in chapter 2 we rectify the omission of investigating the genetic colony structure and population genetics, infer breeding system characteristics and establish Oecophylla smaragdina as study system for ecology and evolution of social insects. In chapter 3 the impact of genetic differences on the nest building behaviour is addressed. There are about 3600 species of formicine ants described (Ward 2010), which makes them the second largest group of ants. They are very diverse in their behaviour and morphology. Hence they are an interesting study system for comparative approaches to the evolution of life history traits and morphological characters. In the recent years significant progress was made to test statistically independent contrasts, to reconstruct ancestral states on phylogenies or to measure character association, while taking phylogenetic uncertainty in to account (Pagel et al. 2004, Bollback 2006). These methods can be applied to any character or trait, including morphology, behaviour, ecology and development. In ants, one long-standing question is the evolutionary history of the metapleural gland. This gland provides a first line protection against microbial pathogens and is therefore thought to have boosted the radiation of the ants (Hölldobler and Wilson 1990). Given the strong selective advantages it confers, the loss of this gland in several very successful formicine genera has puzzled myrmecologists (Hölldobler and Engel-Siegel 1984). Also the possibility of evolutionary reversals of the loss of such a complex organ is an intriguing topic (Gould 1970). In chapter 4 we study the evolution of the metapleural gland by first constructing a phylogeny of the Formicinae and then mapping the presence and absence of the gland onto the tree. We also test potential reasons for the loss of the metapleural gland as suggested by Hölldobler and Engel-Siegel (1984). They hypothesised that because the selection pressure from microbes varies with the nesting habitat, species living in trees may loose the gland. In chapter 5 we address systematic issues of Camponotus, one of the largest and most diverse ant genera, to enable the reconstruction of the history of dispersion and diversification of this exceptionally species rich group. For a very long time taxonomists and systematists were unable to establish natural, monophyletic groups to manage this genus and to understand the evolutionary forces leading to the obvious convergence of morphological characters (Emery 1925). For many studies using DNA as a source of information, samples have to be stored and transported. DNA can be stabilised in buffers or other liquids to prevent degradation, but not all methods are equally capable preserving DNA (Nagy 2010). Chapter 6 details a comparison of methods for storage of DNA samples that are suitable for fieldwork and shipping of collection material