Recurrent Collection of Drosophila melanogaster from Wild African Environments and Genomic Insights into Species History

Abstract

A long-standing enigma concerns the geographic and ecological origins of the intensively studied vinegar fly, Drosophila melanogaster. This globally distributed human commensal is thought to originate from sub-Saharan Africa, yet until recently, it had never been reported from undisturbed wilderness environments that could reflect its precommensal niche. Here, we document the collection of 288 D. melanogaster individuals from multiple African wilderness areas in Zambia, Zimbabwe, and Namibia. The presence of D. melanogaster in these remote woodland environments is consistent with an ancestral range in southern-central Africa, as opposed to equatorial regions. After sequencing the genomes of 17 wilderness-collected flies collected from Kafue National Park in Zambia, we found reduced genetic diversity relative to town populations, elevated chromosomal inversion frequencies, and strong differences at specific genes including known insecticide targets. Combining these genomes with existing data, we probed the history of this species' geographic expansion. Demographic estimates indicated that expansion from southern-central Africa began ∼10,000 years ago, with a Saharan crossing soon after, but expansion from the Middle East into Europe did not begin until roughly 1,400 years ago. This improved model of demographic history will provide an important resource for future evolutionary and genomic studies of this key model organism. Our findings add context to the history of D. melanogaster, while opening the door for future studies on the biological basis of adaptation to human environments.

Keywords: Drosophila; Africa; commensal evolution; demographic history; population genomics; wilderness collection.

Fig. 1.

Collection environment and genomic differentiation of the Kafue population of Drosophila melanogaster. (A) Locations of five D. melanogaster wilderness collections. (B–F) Satellite imagery of these sites. (G) Miombo woodland environment. (H) A sample of the diverse array of fruits present in this environment. (I) Climate graph for Livingstone, Zambia, depicting an extended dry season and a wide range of temperatures.

Fig. 2.

Elevated inversion frequencies in the Kafue population. (A) For all six chromosomal inversions common in this region, the Kafue park population yielded higher rearrangement frequencies than the Siavonga town population. (B) Kafue had a higher proportion of inverted chromosome arms (across X/2L/2R/3L/3R) than previous Drosophila melanogaster collections from Africa or elsewhere. Inversions were computationally inferred from individual genomes (dark blue) or pooled genomic samples (light blue), or scored cytologically (green). Frequencies were compiled for populations with sample sizes of at least 20, and for the six inversions in (A), plus In(1)Be, In(3L)P, and In(3R)Mo, which are not common in Zambia. Among pooled samples, only the populations yielding the highest and lowest inversion frequencies from each continent are depicted.

Fig. 3.

An estimated history of sampled Drosophila melanogaster populations. (A) For the focal population tree identified from distance relationships and model likelihood comparisons, median arm point estimates are depicted for each estimated divergence time and population size. (B) The locations of the included populations are indicated. Colors of population branches (A) and circles (B) indicate populations and lineages within the apparent ancestral range (red), and those in regions occupied by roughly 12 kya (yellow) and by ∼2 kya (blue).

Fig. 4.

Potential genomic targets of adaptive differences between wild and town populations. (A) A plot of PBE indicates genomic windows (averaging 4 kb in length) that have elevated genetic differentiation between the Kafue population and town populations, with labeled examples described in the text. The dotted line indicates genome-wide significance based on neutral demographic simulations. (B) The GO categories most enriched for PBE outliers are given along with raw P values. These GO categories include biological process (green) and molecular activity (blue) functions, after excluding redundant categories. Only the top-listed category has a P value low enough to attain genome-wide significance after correction for multiple tests (adjusted P = 0.003).