Gene reuse facilitates rapid radiation and independent adaptation to diverse habitats in the Asian honeybee

Abstract

Animals with recent shared ancestry frequently adapt in parallel to new but similar habitats, a process often underlined by repeated selection of the same genes. Yet, in contrast, few examples have demonstrated the significance of gene reuse in colonization of multiple disparate habitats. By analyzing 343 genomes of the widespread Asian honeybee, Apis cerana, we showed that multiple peripheral subspecies radiated from a central ancestral population and adapted independently to diverse habitats. We found strong evidence of gene reuse in the Leucokinin receptor (Lkr), which was repeatedly selected in almost all peripheral subspecies. Differential expression and RNA interference knockdown revealed the role of Lkr in influencing foraging labor division, suggesting that Lkr facilitates collective tendency for pollen/nectar collection as an adaptation to floral changes. Our results suggest that honeybees may accommodate diverse floral shifts during rapid radiation through fine-tuning individual foraging tendency, a seemingly complex process accomplished by gene reuse.

Fig. 1. Population structure, phylogeny, and demographic history of mainland A. cerana.

(A) Sampling sites. The pie charts indicate ancestry fractions inferred by ADMIXTURE when K = 9. Arrows refer to the plausible expansion routes for each peripheral group independently derived from the Central group. Habitat type and average altitude for each group were marked for each genetically distinct group. LN, Liaoning. (B) Population networks based on minimum-spanning tree. (C) Ancestry fractions inferred by ADMIXTURE when K = 7 to 9. (D) Neighbor-joining (NJ) tree constructed from identity-by-state (IBS) distances. The scale bar represents the raw genetic distance. (E) Maximum likelihood population tree inferred by TreeMix using Apis mellifera as the outgroup. (F) Estimated divergence times between each peripheral group and the Central group after 100 bootstraps. The widths indicate probability densities. (G) Dynamic of N_e inferred by SMC++. The dashed line represents temperature fluctuation reflected by δ¹⁸O contents. Genetically distinct population groups: Malay (ML), Hainan (HN), Taiwan (TW), Bomi (BM), Aba (AB), Qinghai (QH), Northeast (NE), and Central groups.

Fig. 2. Habitat distribution of A. cerana peripheral groups.

(A) The locations of three highland groups distributed in canyons around the edge of the Q-T Plateau. Different altitudes are indicated by different colors. The arrows indicate the opening of the canyons. (B) A schematic diagram of the population radiation model.

Fig. 3. Bioclimates and flowering phenology.

(A) PCA of 19 bioclimatic variables for all sampling sites at present and LGM. H&J, Heilongjiang and Jilin. (B) The long-term phenological observation results of flowering plants in China. The flowering season was reported as year-round on Hainan island (15).

Fig. 4. Genes under selective sweep in peripheral groups.

(A) Distribution of genes under selective sweep in peripheral population groups. Horizontal bars to the right represent genes under selection in each group. Vertical bars indicate the numbers of selected genes exclusively associated with one or multiple groups. (B) Number of genes under selection in a single peripheral group (1) or across multiple groups (2, 3, 4, and 5). Asterisks indicate cases where the observed (black) number is significantly larger than expected (red) number (***P < 0.001; exact multinomial test for goodness of fit). (C) Distribution of CLR values on the scaffold including Leucokinin receptor (Lkr). The horizontal dashed lines indicate the top 1% thresholds of CLR values on the whole genome. (D) Positions of beneficial variations and hard/soft sweep classification of the Lkr gene inferred by LASSI analyses. The peak point of the T value and its corresponding m value is marked by vertical dashed lines, indicating where beneficial variation has most likely occurred. Selective sweeps are classified as hard when m = 1 (marked by horizontal dashed lines) or soft when m > 1.

Fig. 5. Differential Lkr expressions and PER thresholds to sucrose with RNAi.

(A) Relative expression levels of Lkr in various tissues of nurse (NB) and forager bees (FB) (n = 12 for each tissue) quantified by quantitative polymerase chain reaction (qPCR). An, antennae. (B) Relative Lkr expressions in the brain of pollen (n = 31) and nectar (n = 31) foragers. The Actin gene was used as control. (C) Thresholds for gustatory sucrose response in foragers (sucrose solution placed on antennae). The control (n = 10), RNAi-dsGFP (n = 15), and RNAi-dsLkr (n = 15) groups were presented here. (D) Thresholds for sucrose in foragers possibly triggered by cues of surface humidity (sucrose solution not touching antennae). The control (n = 15), RNAi-dsGFP (n = 12), and RNAi-dsLkr (n = 15) groups were presented here. GFP, green fluorescent protein gene. *P < 0.05, **P < 0.01, ***P < 0.001 (two-tailed Mann-Whitney test). NS, not significant.

Fig. 6. A proposed model showing how Lkr under selection translates to a colony-level adaptive phenotype.