The broad goals of my research have been and will be understanding the consequences of life history evolution in natural populations, how adaptation in different environments gives rise to a few rather than many life history trajectories, and how adaptive differentiation is related to speciation. I am pursuing these questions in the Drosophila-cactus-yeast model system at many levels using a wide range of population genetic, physiological and ecological analyses. Thus, a fundamental understanding of the consequences of life history evolution for the origins of reproductive isolation is unfolding. On a much broader scale, I am collaborating with others to uncover the role of host plant shifts in large-scale evolutionary diversification using molecular techniques.

 1. Life history evolution and speciation. How reproductive isolation originates, which causes formation of new species, remains one of the most fundamental problems in biology. We are investigating the role of host plant use in the formation of reproductive isolation among populations of D. mojavensis, a resident of the Sonoran Desert. This species is a member of the well-studied D. repleta group, a large assemblage of about 100 taxa that originated in the New World and radiated into a diversity of habitats. Approximately half are associated with cactus, using different species of platyopuntias and columnar cacti throughout their ranges. Ongoing systematic analyses (see below) within and between species of the repleta group will allow us to include a phylogenetic perspective in all levels of work.

There are significant genetic differences in life history traits between Baja and mainland populations of D. mojavensis (Etges, 1990) and significant genetic variation persists for these traits expressed in a cactus-specific manner (Etges, 1993). Our working hypothesis is that D. mojavensis arose in Baja California and has recently (perhaps within the last 5-15,000 yr.) invaded mainland Sonora, Arizona, and southern California by switching host cacti (Etges and Heed, 1987). Our recent molecular phylogenetic (Fig. 1) and chromosomal inversion analyses (Etges et al., 1999) are congruent with this interpretation. A phylogenetic analysis of a 150 base pair stretch of Ace intron sequences from ten D. mojavensis cluster individuals produced patterns which are, in general, consistent with the biogeographic distribution of these individuals (Fig. 1; sequences not shown). The most derived groups are from Arizona, the most ancestral sequences are from the Mojave Desert, and isolates from Baja California (BCS and BCN) are intermediate. These sequences had 18 variable and 6 phylogenetically informative sites (Durando, DeSalle, and Etges, unpubl. data). We are in the process of completing this evaluation of phylogenetic relationships of multiple populations of the D. mojavensis species cluster, D. mojavensis, D. arizonae, and D. navojoa. Since these species use different host cacti, we plan to assess the patterns of host use in a phylogenetic context.

Ecology "matters." The significant influence of host substrates on expression of genetic and phenotypic expression of life histories suggested that other relevant phenotypes might also be so affected. Since there is low, but significant premating isolation between Baja and mainland populations of D. mojavensis reported in laboratory trials, I tested the hypothesis that cactus rearing substrates have no effect on reproductive isolation. This case of "incipient speciation" has been discussed in the literature for over 25 years, and even has found its way into some textbooks, but we have discovered that there is far more to be understood about the causes for behavioral isolation in this system (Etges, 1992; Brazner and Etges, 1993; Etges 1998; Etges and Ahrens 2001).

I rediscovered that premating isolation among populations of D. mojavensis is influenced by the type of food used to rear larvae (Etges, 1992; Brazner and Etges, 1993), i.e., larval experience affects adult behavior. Most significantly, we compared fermenting tissues of the host cacti of D. mojavensis with lab media used in all previous studies: in all cases, lab food induced premating isolation, and a cactus species used most widely in nature as a host decreased premating isolation to low and typically non-significant levels. Another major host causes significantly higher levels of premating isolation between populations. Thus, all previously published reports of premating isolation between geographical isolates of D. mojavensis are biased because the flies studied were reared on lab food (Brazner and Etges, 1993). We are currently trying to understand the causes for these differences in premating isolation in D. mojavensis by experimentally altering fatty acid precursors in larval diets. We are also focusing on cactus-induced shifts in the epicuticular hydrocarbons that have been postulated to serve as contact pheromones in this species (see below). These results underscore the value of the cactus-yeast Drosophila system; if we did not know the larval ecology of D. mojavensis, we could not fully understand the patterns of expression of premating isolation behaviors or know how to measure them. Surely, this research will lead to a clearer understanding of how species are formed.

 Assessing genetic variation for premating isolation. Of more importance to an understanding of how reproductive isolation arises in nature is the ability to rigorously evaluate alternate explanation based on hypothesis testing. A number of workers have proposed models for the evolution of reproductive isolation between Baja California and mainland populations of D. mojavensis, and I have approached this system experimentally (Etges 1998) . Once genetic differentiation in life characters between Baja and mainland populations had been documented revealing the presence of cactus "host races" (Etges, 1990), and significant genetic variability was revealed within populations (Etges, 1993), I proposed that sexual isolation has probably arisen as a correlated response to adaptation to different host cactus hosts in Baja California and mainland Mexico. This hypothesis was evaluated by performing artificial selection on egg-to-adult development time, a key life history character that has evolved in response to the use of different host cacti. A critical assumption here was that these populations harbor genetic variability for premating isolation despite the effects of larval substrates described earlier. Previous selection experiments by Roberta Koepfer suggested that such genetic variability should exist.

The results unambiguously showed correlated decreases in sexual isolation and female mating discrimination in the lines selected for fast and slow development time and no behavioral changes in the control lines after 13 generations of artificial selection (Fig. 2; details in Etges, 1998). Here, Yule’s V is the behavioral isolation statistic and OP and AG refer to the fast (F) and slow (S) replicate lines reared on different cacti. If we assume that past natural selection has shaped the genetic differences in development time observed between mainland and Baja California populations (development time differences closely match rates of cactus tissue breakdown; see Etges, 1989 Evolutionary Ecology), then this selection experiment suggests that shifting gene frequencies influencing development time has altered levels of adult behavioral isolation. Thus, adaptation to alternate host plants can cause reproductive isolation, and in this case, incipient speciation.

 A mechanism for premating isolation. A determinant of mating success in many Drosophila and other insect species is a class of contact pheromones composed of cuticular hydrocarbons. We have characterized the epicuticular hydrocarbons involved in mate recognition in D. mojavensis and its closest relatives, D. arizonae and D. navojoa (Etges and Jackson, 2001). By experimentally altering adult hydrocarbon composition, we demonstrated that one or more hydrocarbon components have pheromonal activity (Etges and Ahrens, 2001) and like premating isolation, hydrocarbon amounts are also influenced by larval rearing substrates (Stennett and Etges, 1997). Future studies will reveal the exact "blend" of hydrocarbons that influence courtship success.

Genetic analysis of population differentiation of these hydrocarbons will reveal the nature of genetic differences between geographical regions where local mate recognition systems have evolved. The Etges-Noor-Ritchie collaboration has produced a microsatellite DNA genetic map (Staten et al., 2004 BMC Genetics 5:12) that we are using to identify quantitative trait loci (QTLs) influencing sexual isolation, epicuticular hydrocarbon variation, mating songs, and egg to adult development time differences between these populations. D. mojavensis is especially suitable for such an evolutionary study because so much is known about the effects of environmental cues on its sexual behavior. Hence, we can combine the tools and utility of a model system with extensive ecological data to make stronger inferences about how sexual isolation evolves in nature.

Genetic analysis of characters most likely responsible for premating isolation will reveal the kinds of genetic changes associated with the early stages of reproductive isolation within species. In future studies, we aim to identify the genes influencing the chemical and behavioral mechanisms that are responsible for changes in mating systems in D. mojavensis. These results will provide a fundamental understanding for how new species are formed.

2. Molecular systematics and host use of the Drosophila repleta group. Over ten years ago, Rob DeSalle, Bill Heed, and I initiated a long-term project to resolve the systematic relationships of the repleta group of Drosophila, the largest species group in the genus (ca 100 species) besides the Hawaiian Drosophila. In the first phase of the project, our higher level phylogenetic analyses are so far completed (Fig. 3; from Durando et. al. 2000), and we are moving on to finish a number of lower level analyses that require more sequence data. Overall, we have scrutinized the patterns of divergence in the group as a whole in relation to the patterns established by previous patterns of chromosomal inversion sharing. Lower level studies include within-species coalescent approaches, particularly in the well-resolved D. mojavensis species cluster. Of particular importance is the identification of host cactus use for many of these flies in central and southern Mexico, as well as South America, and the role of host plant use in patterns of species formation. An NSF funded international collaboration with Miguel Armella at the Universidad Autonoma Metropolitana - Iztapalapa in Mexico City has allowed identification of the host cacti for most of these species. We have mapped the host cacti onto the phylogenetic trees of the flies to compare any ecological similarities among close relatives or whether particular clades share ancestral or derived types of cacti. Results indicate that host shifts from Opuntia using species to columnar cactus breeders have occurred independently at least 9 times throughout the group. We also plan to evaluate the role of biogeography of the flies in relation to host use because many close relatives of Mexican species reside in South America.

 Future Directions. The advantages of working with Drosophila, and in particular the D. repleta group, are the ability to synthesize field and laboratory analyses and to have access to a tremendous variety of tools, from cuticular hydrocarbons to chromosomes to current analyses of the D. mojavensis genome. Our systematic work has revealed a number of areas within the phylogeny of this group that require further data to increase resolution. The lower level analyses of population differentiation within the D. mojavensis species cluster will establish the evolutionary history of these species and will augment our ongoing understanding of species divergence. Certainly, D. mojavensis has become one of the premier species in the analysis of life history evolution and incipient speciation.

With the identification of more genes associated with hydrocarbon biosynthesis in Drosophila, it will be possible to begin analyzing nucleotide sequence variation associated with the geographic differences between populations of D. mojavensis. Mapping these genes using in situ hybridization techniques is currently being planned with Dr. Alfredo Ruiz at the Universitat Autonoma de Barcelona. My collaboration with Mohamed Noor and Mike Ritchie will provide essential genetic information concerning the genetic bases for cuticular hydrocarbon variation, mating song differences, and sexual isolation. This effort will extend our abilities to understand the number and influence of genes for these traits involved in reproductive isolation. Our QTL studies are a preliminary attempt to identify the types of gene differences that arise early during the formation of species. These studies will lead us to an understanding of the genetic bases of sexual isolation within and between species with known ecologies. This system holds great promise for understanding the initial stages of speciation in terms of its ecological and genetic causes. This is perhaps the greatest unsolved problem of evolutionary biology.