Taste adaptations associated with host specialization in the specialist Drosophila sechellia

Abstract

Chemosensory-driven host plant specialization is a major force mediating insect ecological adaptation and speciation. Drosophila sechellia, a species endemic to the Seychelles islands, feeds and oviposits on Morinda citrifolia almost exclusively. This fruit is harmless to D. sechellia but toxic to other Drosophilidae, including the closely related generalists D. simulans and D. melanogaster, because of its high content of fatty acids. While several olfactory adaptations mediating D. sechellia's preference for its host have been uncovered, the role of taste has been much less examined. We found that D. sechellia has reduced taste and feeding aversion to bitter compounds and host fatty acids that are aversive to D. melanogaster and D. simulans. The loss of aversion to canavanine, coumarin and fatty acids arose in the D. sechellia lineage, as its sister species D. simulans showed responses akin to those of D. melanogaster. Drosophila sechellia has increased taste and feeding responses towards M. citrifolia. These results are in line with D. sechellia's loss of genes that encode bitter gustatory receptors (GRs) in D. melanogaster. We found that two GR genes which are lost in D. sechellia, GR39a.a and GR28b.a, influence the reduction of aversive responses to some bitter compounds. Also, D. sechellia has increased appetite for a prominent host fatty acid compound that is toxic to its relatives. Our results support the hypothesis that changes in the taste system, specifically a reduction of sensitivity to bitter compounds that deter generalist ancestors, contribute to the specialization of D. sechellia for its host.

Keywords: Drosophila sechellia; Behavior; Bitter receptor; Chemosensation; Ecological adaptation; Gustatory receptors; Host plant specialization; Noni.

Fig. 1.

Proboscis extension responses (PER) are taste organ and species specific. (A) Image of the fly preparation. Food-deprived female flies were mounted on a glass slide (the image shows a male), and the taste organs were stimulated with food solution (here dyed blue for visualization) applied to all leg tarsi or the proboscis; flies were not allowed to drink. The arrow indicates proboscis extension upon stimulation. Each fly was tested with one condition only. (B) In Drosophila melanogaster, PER to 1 mol l⁻¹ sugar was independent of the taste organ stimulated (P>0.001), but higher upon proboscis stimulation at lower concentrations; Fig. S1A). (C) In Drosophila sechellia, PER was higher upon tarsi stimulation (Fisher exact tests, *P<0.05, **P<0.01). (D) Stimulation with 1 mol l⁻¹ sucrose + 0.5 mmol l⁻¹ denatonium (a bitter compound) reduced PER in both species (***P<0.001); data obtained from different animals to those in B and C. Data in B–D show the percentage of flies that extended their proboscis at least once; numbers in parentheses indicate the number of flies tested.

Fig. 2.

Drosophila sechellia has comparatively increased taste and feeding responses to noni. (A) PER across species. Flies were prepared as in Fig. 1A and their tarsi were stimulated first with 1 mol l⁻¹ sugar solution (sucrose for the generalists and glucose for D. sechellia given their differential responsiveness; Fig. 1C and Fig. S1B), and then with noni (3 times/stimulus solution). Flies were not allowed to drink the sugar or noni, but could drink water between presentation of these food solutions. The proportion of flies showing PER to sugar stimulation was higher than that showing PER to noni in D. melanogaster and D. simulans (McNemar tests, ****P<0.001 in both cases), while D. sechellia showed similar PER to the two stimuli (n.s., P>0.05). (B) Reduced PER to noni in the generalist D. melanogaster does not require olfaction. Parallel cohorts of intact and anosmic flies (olfactory organs ablated 2 days before tests) were assayed as in A; the two groups showed similarly reduced PER to noni in comparison with 1 mol l⁻¹ sucrose (***P<0.005 in both cases). (C–E) Temporal consumption assay of 24 h food-deprived restrained flies. Flies were prepared as in Fig. 1A; their tarsi were stimulated up to 10 consecutive times with 750 mmol l⁻¹ glucose or noni juice and allowed to drink. We stimulated tarsi because it evoked stronger responses than proboscis stimulation (Fig. 1). The timing of feeding responses was digitally recorded and analyzed off-line. Here and in all upcoming experiments, we used glucose instead of sucrose because D. sechellia has a greater responsiveness to this sugar (which occurs in noni). (C) Example of the temporal sequence of ingestion of individual D. melanogaster flies offered noni or glucose. The dotted vertical line at zero indicates the beginning of the tests. Horizontal bars indicate feeding events (3 and 7, respectively); the summed duration of all feeding events was 5 and 10 s in these examples. (D) In D. melanogaster, the percentage of flies that consumed noni was lower than the percentage that consumed glucose (Fisher exact test, ***P<0.005; n=19, 21), but not in D. sechellia (n.s., P>0.05; n=23–22). (E,F) The number of feeding events (E) and the feeding duration (F) were higher in D. melanogaster when glucose was offered (Mann–Whitney U-tests, *P<0.05, **P<0.01), but were similar upon stimulation with either stimulus in D. sechellia (P>0.05, n.s.). Symbols are individual data, boxes indicate the 25% and 75% quartiles, the horizontal line inside boxes indicates the median, and the whiskers indicate the 10% and 90% quartiles.

Fig. 3.

Bitter compounds evoke different levels of feeding aversion across species. (A) Schematic representation of the group feeding assay. Flies were food deprived for 24 h (n=12–15 per vial) and then transferred to vials containing a disk of filter paper impregnated with 160 µl of 750 mmol l⁻¹ glucose dyed blue (control vials), or 750 mmol l⁻¹ glucose plus a bitter or fatty acid compound (test vials) dyed blue. After 30 min, vials were frozen and then flies in each vial were scored blind to treatment according to the amount of blue dye in their abdomen using a five-point scale (0–2) (as in Reisenman and Scott et al., 2019); a single feeding score was calculated for each vial (biological replicate). (B–D) Feeding scores of female D. melanogaster (B), D. simulans (C) and D. sechellia (D) offered control (white boxes) or test (gray boxes) food solutions. Test vials had 750 mmol l⁻¹ glucose plus one of the following: 0.5 mmol l⁻¹ denatonium (Den.), 10 or 25 mmol l⁻¹ caffeine (Caff.), 1 mmol l⁻¹ lobeline (Lob.), 10 mmol l⁻¹ l-canavanine (Can.), 10 mmol l⁻¹ coumarin (Cou.), 10 mmol l⁻¹ theophylline (The.), 100 mmol l⁻¹ octanoic acid (OA, 1.3% v/v) or 100 mmol l⁻¹ hexanoic acid (HA, 1.6% v/v). Box plot description as in Fig. 2. Asterisks indicate differences from the control for each species (Kruskal–Wallis ANOVA and post hoc Dunn's tests; ****P<0.001, ***P<0.005, **P<0.01, *P<0.05; n=12–28 per species and food solution). In D. melanogaster and D. simulans (B,C), but not in D. sechellia (D), the two fatty acids (OA and HA) significantly reduced feeding. In D. melanogaster, all bitter compounds reduced feeding (B), while caffeine, lobeline and coumarin reduced feeding in D. simulans (C). Drosophila sechellia consumed similar food amounts in the absence and presence of canavanine or coumarin (D; P>0.05).

Fig. 4.

Drosophila sechellia has a comparatively reduced feeding aversion to most bitter compounds and noni fatty acids. Data represent the glucose-normalized (on a day-to-day basis) feeding scores of female D. melanogaster, D. simulans and D. sechellia flies (calculated from data in Fig. 3), to allow interspecific comparisons. Box plot description as in Fig. 2; n=12–25 per species and test compound; concentrations as in Fig. 3. The horizontal dotted line at 1 indicates that flies consume similar amounts of control (glucose only) and test solutions (glucose+bitter/fatty acid compound), i.e. no feeding aversion or enhancement to the test solutions. Drosophila melanogaster and D. simulans showed feeding aversion to solutions containing canavanine, coumarin, OA or HA (gray bars, P<0.005, one-sample signed rank tests against median=1), while D. sechellia had lost the aversion to these compounds (white bars, n.s., P>0.05 in all cases). Drosophila sechellia flies retained aversion (feeding scores significantly <1) to caffeine, lobeline, theophylline and denatonium (gray bars). In all cases except for lobeline and denatonium, the responses of D. sechellia flies were divergent from those of D. simulans and D. melanogaster (boxes outlined red, P<0.05 in all cases; different letters indicate inter-specific differences, Kruskal–Wallis ANOVA and post hoc Dunn's tests).

Fig. 5.

GR28b and GR39a D. melanogaster null mutants have reduced aversion to bitter compounds that correlates with the D. sechellia’s behavioral phenotype. Glucose-normalized feeding scores (calculated as in Fig. 4) of GR28b (A) and GR39a (B) null mutants and their respective genetic background controls to solutions containing 750 mmol l⁻¹ glucose plus a bitter or fatty acid compound (concentrations as in Fig. 3; 10 mmol l^-1 for caffeine). Box plot description as in Fig. 2; the horizontal dotted lines at 1 indicate no feeding aversion or enhancement. Addition of any bitter/fatty acid compound reduced feeding in all (one-sample signed rank tests, P<0.005) but one case (GR39a mutants offered canavanine; P>0.05); n=12–21 for each genotype and food solution. Both mutants consumed larger amounts of solutions containing canavanine or coumarin (and theophylline in GR28b mutants) than their respective genetic controls (Mann–Whitney U-tests, *P<0.05, ***P<0.005, ****P<0.001); aversive responses to caffeine and theophylline were slightly reduced in GR39a mutants (B, P=0.066 and 0.051, respectively). Responses to fatty acids were not different between mutants and their respective controls for the most part (Mann–Whitney U-test, P>0.05; GR39a mutants consumed less HA than the control, *P<0.05).

Fig. 6.

Drosophila sechellia has increased appetite for noni fatty acids in comparison with D. melanogaster. (A) Feeding scores of D. melanogaster (left) and D. sechellia (right) offered 750 mmol l⁻¹ glucose, water or water+OA (100 mmol l⁻¹, 1.3% v/v). Flies were prepared as before, and all groups were tested with parallel cohorts. Both species fed the most on glucose (Kruskal–Wallis ANOVA followed by Dunn's tests, n=13–21 per species and food solution; different letters indicate significant differences, P<0.05). (B) Water-normalized feeding scores of flies offered glucose (left) or water+OA (right). The horizontal dotted line at 1 indicates no aversion or preference in comparison to water; gray shading indicates P-values (one-sample signed rank tests against median=1). Relative to water, both species have a strong appetite for glucose (i.e. median>1; left), but D. melanogaster fed less on OA (i.e. median<1), while D. sechellia fed more (i.e. median>1, right). Normalized responses to glucose and OA differed between species (Mann–Whitney U-tests, ****P<0.001).