Lake Level Fluctuations Synchronize Genetic Divergences of Cichlid Fishes in African Lakes “Introduction”

Introduction The Great East African Lakes have been established as major model systems for the study of adaptive radiation. Each lake harbors flocks of cichlid fishes including hundreds of endemic species. It is now clear that the flocks of cichlid fishes in Lakes Victoria, Malawi, and Tanganyika arose independently via intralacustrine speciation (Kocher et al. 1993; Meyer 1993; Nagl et al. 2000). Despite several stunning similarities, the three species flocks differ from each other in age, species number, complexity, and overall degree of morphological diversity (Fryer and Iles 1972; Greenwood 1980, 1984; Poll 1986; Eccles and Trewavas 1989). All three lakes have complex geological histories, characterized by dynamic basin morphology, as well as by lake level fluctuations, caused by variations in amount of rainfall, temperature, evaporation, and, for some lakes, tectonic activity (Scholz and Rosendahl 1988; Gasse et al. 1989; Tiercelin and Mondeguer 1992; Cohen, Soreghan, and Scholz 1993; Delvaux 1995; Johnson et al. 1996; Lezzar et al. 1996; Cohen et al. 1997). Lake Tanganyika is the oldest of the three major East African lakes. Its central basin was formed 9–12 MYA. The structure of the lake basin suggests that a meandric river gave rise to at least three shallow, swampy proto-lakes (9–12 MYA), which progressively deepened to fuse finally into a single deep lake (5–6 MYA; Tiercelin and Mondeguer 1991; Cohen, Soreghan, and Scholz 1993). The Lake Malawi rift basin started to develop about 8.6 MYA, but deepwater conditions were acquired only 4.5 MYA. With an estimated age of about 400,000 years, Lake Victoria is substantially younger than the other two lakes, is much shallower, and has a different geological origin (Johnson et al. 1996). The age of a species flock may not correspond to the geological age of a lake, since climatic or geological events may have caused a temporary dry-up, such that preexisting species flocks may have gone extinct.  moreover, it was also shown by recent molecular studies that the dynamics of diversification events in African cichlid fishes are likely to be connected to fluctuations in the lake level (Sturmbauer and Meyer 1992; Johnson et al. 1996; Sturmbauer et al. 1997; Ru¨ber et al. 1998; Nagl et al. 2000). The evolutionary consequences of such severe environmental changes are well known for European terrestrial faunas (Hewitt 1996) but are much less understood for tropical ecosystems. Lakes Malawi and Tanganyika were severely affected by the change to a drier climate in the late Pliocene/early Pleistocene, resulting in a drop of the lake level by 650–700 m about 1.1 MYA in Lake Tanganyika (Lezzar et al. 1996; Cohen et al. 1997) and an almost complete dry-up of Lake Malawi from 1.6 MYA until 1.0–0.57 MYA (Delvaux 1995). After that minimum, both lakes rose until about 400,000 years ago. New regressions started about 390,000 years ago in Lake Tanganyika (Cohen et al. 1997) and 420,000 years ago in Lake Malawi (Delvaux 1995), followed by a period of fluctuating lake level in both lakes. In Lake Tanganyika, the minima were dated at 390,000–360,000 years ago, 290,000–260,000 years ago, and 190,000–170,000 years ago; those of Lake Malawi were not precisely dated by Delvaux (1995). Lake Tanganyika rose to its present level 170,000–40,000 years ago, and the rise of Lake Malawi to its present level was estimated at about 250,000–120,000 years ago. Concerning their most recent history, several studies agree that the lake levels of all three lakes were substantially lower during the late Pleistocene ice ages, when the climate in much of north and equatorial Africa became progressively more arid. Two recent studies carried out in the north part of Lake Tanganyika demonstrated three periods of low lake level for Lake Tanganyika in its recent past, the first 40,000–35,000 years ago, the second 23,000 years ago, and the third 18,000 years ago (Lezzar et al. 1996; Cohen et al. 1997). The lowstands at 23,000 and 18,000 years ago were also found in a sediment study in the very south part of the lake by Gasse et al. (1989). A minimum water level of about 400 m below the present level was reached 18,000 years ago, and the lake was lower until 13,000 years ago abruptly rising in two steps 13,000 and 10,600 years ago (Gasse et al. 1989). An earlier study tentatively dated the latest major lowstand for Lakes Malawi (250–500 m) and Tanganyika (600 m) at about 25,000 years ago (Scholz and Rosendahl 1988; C. A. Scholz, personal communication). This age estimate has been obtained by extrapolating sedimentation rates and may correspond to the minimum at 18,000 years ago found in the more recent studies (Gasse et al. 1989; Lezzar et al. 1996; Cohen et al. 1997). As suggested for Lake Tanganyika, Lake Malawi was also at least 400 m lower 18,000–10,700 years ago (Brooks and Robertshaw 1990; Owen et al. 1990; Finney and Johnson 1991), and seismic reflection profile and piston core analyses from Lake Victoria suggest that Lake Victoria was (almost) dry from 17,300 years ago until 12,400 years ago (Johnson et al. 1996; see also and Stager, Reinthal, and Livingstone 1986; Talbot and Livingstone 1989). In summary, the lake levels of all three Great East African Lakes seem to have been influenced in a similar way by the same global climatic changes. They were generally low 18,000–12,000 years ago, quickly rising to present levels with few and less severe fluctuations, with a maximum of 150 m in the Holocene and in historic times (Owen et al. 1990). In order to compare the evolutionary consequences of the most recent minima of the lake levels on the cichlid faunas of Lakes Tanganyika, Malawi, and Victoria, we investigated the geographic distribution of closely related genotypes. Therefore, we selected species and populations of littoral cichlid fishes in each of the three lakes that have been shown to be weak dispersers and thus likely to be greatly affected by lake level changes (Meyer et al. 1990; Sturmbauer and Meyer 1992; Moran and Kornfield 1993; Verheyen et al. 1996; Sturmbauer et al. 1997; van Oppen et al. 1997; Albertson et al. 1999; Arnegard et al. 1999; Markert et al. 1999). The samples included in our study were chosen according to the basin structure of the lakes. Our approach was based on the observation that lake level fluctuations temporarily form or break down barriers among habitats and thus either promote or prevent gene flow among adjacent populations and/or incipient species (Sturmbauer 1998). The degree of habitat change enforced by water level fluctuations may range from small-scale effects to major vicariant events that affect species communities in most habitats. Any drop in the lake level will establish secondary contact and admixis among previously isolated populations in shallow regions of a lake, leading to an increase in genetic diversity in admixed populations. A rise in the lake level may promote population subdivision due to the colonization of new habitats. Newly formed ecological barriers interrupt gene flow, such that genetic differences can accumulate independently and lineage sorting can proceed. Populations of cichlid fishes specialized to particular habitats such as rocks in the littoral zone are likely to become isolated to a higher degree than less stenotopic and thus more mobile species. Only populations that have become isolated in the very recent past should share identical or closely related haplotypes, even if they are now separated by long distances or are situated on opposite shores. When the time of divergence between two populations is short, not many new mutations arise and lineage sorting is likely to be incomplete. In this case, any individual sampled from one population is expected to be as similar to individuals from different populations that were formed by the same founder event as it is to some individuals sampled from its own population. Any set of genetically heterogeneous populations is expected to contain several clusters of equally closely related genotypes, since lineage sorting is likely to be incomplete, as long as the number of generations after the split is equal to or less than twice the population size (Tajima 1983). It is important to note that our approach is not affected by ancient DNA sequence polymorphisms due to recent speciation events, because only gene trees referring to the geographic distribution of genotypes are used to derive relative time estimates for allopatric divergence. While most rock-dwelling cichlid species of Lake Tanganyika are easily distinguishable by means of mtDNA sequences, and incipient speciation mostly proceeds in absence of ecomorphological innovation, species of Lakes Victoria and Malawi are much younger and still tend to share mitochondrial haplotypes, even if they sometimes differ dramatically in terms of their trophic morphologies and are placed in different genera (Greenwood 1980, 1984; Eccles and Trewavas 1989; Meyer et al. 1990; Moran and Kornfield 1993; Kornfield and Parker 1997; Parker and Kornfield 1997; Nagl et al. 2000). Thus, mtDNA-based phylogenies still represent gene trees and not species trees for Lake Victoria and Lake Malawi cichlids.

Updated: January 27, 2015 — 5:57 pm © 2014 Frontier Theme