When flies were reared under four different temperatures, wings appeared to be larger at lower temperatures (Fig. 1, S2). We calculated the centroid size and found that it became smaller as the rearing temperature was increased (Fig. 2a, b). That tendency was observed both in males and in females with one exception of a nonsignificant difference between the centroid size at 18 °C and 21 °C in males (Fig. 2a).

Wings from male and female flies reared at 18 ℃, 21 ℃, 25 ℃, and 28 ℃. a: Wings from flies at 18 ℃. The left black arrowheads indicate “Proximal” spots and the right black arrowheads indicate “Middle” spots. b: Wings from flies at 21 ℃. c: Wings from flies at 25 ℃. d: Wings from flies at 28 ℃. To make pictures shown here, the brightness of pictures in Fig. S2 was increased with ImageJ. Scale bars indicate 400 μm

Centroid size of wings and the spot size from flies reared at 18 ℃, 21 ℃, 25 ℃, and 28 ℃. a: Centroid size of wings from male flies. b: Centroid size of wings from female flies. c: Size of “Proximal” spots on wings from male flies. d: Size of “Proximal” spots on wings from female flies. e: Size of “Middle” spots on wings from male flies. f: Size of “Middle” spots on wings from female flies. In all categories, there were significant differences between temperatures (p < 0.001, one-way ANOVA, degree of freedom = 3, F = 383.9 in a, 331.1 in b, 6.546 in c, 20.83 in d, 31.25 in e, 18.6 in f). Different letters indicate significant differences (p < 0.05, Tukey’s HSD test). Black bars indicate mean values

We measured the spot size and found that there was no clear tendency correlated with temperature although significant differences between different temperatures were detected in all categories by one-way ANOVA (Fig. 2c-f). The size of “Proximal” spots was almost stable in male flies, as no significant difference was detected between the spot size at 18 °C, 25 °C, and 28 °C by Tukey’s HSD test (Fig. 2c). The size of “Proximal” spots in female flies was not as stable as that in male flies (Fig. 2d). It is possible to interpret that the size of “Middle” spots in males became smaller when the temperature got higher (Fig. 2e), but it is not possible to interpret the data on the size of “Middle” spots in females in the same way (Fig. 2f).

As centroid size of wings and the area of the polygon mentioned above are highly correlated (Fig. S3), we divided the spot size by the area of the polygon to adjust the spot size by the wing size. After the adjustment, we found that the ratio of the spot size to the wing size was higher at higher temperatures (Fig. 3a-d). Both in males and females, a significant difference between the ratio of “Proximal” size to wing size at 25 °C and the ratio at 28 °C was observed by Wilcoxon rank sum test with Bonferroni correction (Fig. 3a, b). For “Middle” spots, a significant difference between the ratio at 18 °C and that at 21 °C was detected by Wilcoxon rank sum test with Bonferroni correction (Fig. 3c, d).

The spot size adjusted with wing size and the ratio of “Proximal” spot size to “Middle” spot size. Flies were reared at 18 ℃, 21 ℃, 25 ℃, and 28 ℃. a: Size of “Proximal” spots adjusted with wing size from male flies. b: Size of “Proximal” spots adjusted with wing size from female flies. c: Size of “Middle” spots adjusted with wing size from male flies. d: Size of “Middle” spots adjusted with wing size from female flies. e: The ratio in male flies. f: The ratio in female flies. In all categories, there were significant differences between temperatures (p < 10–13, Kruskal–Wallis rank sum test, degree of freedom = 3, χ2 = 120.35 in a, 127.11 in b, 65.207 in c, 77.059 in d, 101.22 in e, 102.96 in f). Different letters indicate significant differences (p < 0.05, Wilcoxon rank sum test with Bonferroni correction). Black bars indicate mean values

As a conspicuous result, we noticed that “Proximal” spot was smaller than “Middle” spot at 18 °C and 21 °C (Fig. 1a, b). When we analyzed the ratio of “Proximal” size to “Middle” size, we found that the ratio became higher when the rearing temperature became higher (Fig. 3e, f). Other than the comparison between ratios at 18 °C and those at 21 °C, significant differences were detected by Wilcoxon rank sum test with Bonferroni correction both in males and females (Fig. 3e, f).

When log10 (spot size) is regressed against log10 (polygon size), we found that the correlation between log10 (spot size) and log10 (polygon size) was the strongest in samples for male “Proximal” spots (Fig. 4, Table S1). As results for testing whether there is an interaction between log10 (polygon size) and temperature (testing whether regression lines have different slopes or not), no significance was detected in any categories by two-way ANOVA (Table S2). By ANCOVA, it was shown that the effect of temperature was significant in all categories (Table 1). The results of the post hoc analyses are written in Table 2. For “Proximal” spots in males and females, other than the comparison between 18 °C and 21 °C, significant differences were detected. For “Middle” spots in males, significant differences were detected in the comparison between 18 °C and 21 °C, and that between 21 °C and 25 °C. In female “Middle” spots, the significant difference was detected only in the comparison between 18 °C and 21 °C.

Log–log plot of polygon area (μm2) and spot size (μm2). The spot size is from flies reared at 18 ℃, 21 ℃, 25 ℃, and 28 ℃. a: “Proximal” spots on wings from male flies. b: “Proximal” spots on wings from female flies. c: “Middle” spots on wings from male flies. d: “Middle” spots on wings from female flies

By rearing D. guttifera under different temperatures, it was shown that the wing size and the spot size of D. guttifera exhibit phenotypic plasticity. To investigate which stage is sensitive to temperature, we changed the rearing temperature during the pupal period. The wing size was the largest when flies were reared under “Condition 1” (reared at 18 °C until P4 (i)) (Fig. 5). By Tukey’s HSD test, significant differences were detected between the wing size of the flies reared under “Condition 1” and the other two conditions (Fig. 6a, b). No significant difference was detected between “Condition 2” and “Condition 3” (Fig. 6a, b). The same tendency was observed both in males and females (Fig. 6a, b).

Wings from male and female flies whose rearing temperatures were changed during the pupal period. a: Wings from male and female flies reared under “Condition 1”. Until P4 (i), flies were reared at 18 ℃. From P4 (i), they were reared at 25 ℃. b: Wings from male and female flies reared under “Condition 2”. Until P4 (i), flies were reared at 25 ℃. From P4(i) to P14-15, they were reared at 18 ℃. From P14-P15, they were reared at 25 ℃. The left black arrowheads indicate “Proximal” spots and the right black arrowheads indicate “Middle” spots. c: Wings from male and female flies reared under “Condition 3”. Until P14-15, flies were reared at 25 ℃. From P14-15, they were reared at 18 ℃. For all pictures, the brightness of the background was increased with ImageJ. Scale bars indicate 400 μm

The results form analyses of flies reared under “Condition 1”, “Condition 2”, and “Condition 3”. Centroid size of wings from flies, the spot size, the spot size adjusted with wing size, and the ratio of “Proximal” spot size to “Middle” spot size are indicated. a: Centroid size of wings from male flies. b: Centroid size of wings from female flies. c: Size of “Proximal” spots on wings from male flies. d: Size of “Proximal” spots on wings from female flies. e: Size of “Middle” spots on wings from male flies. f: Size of “Middle” spots on wings from female flies. g: Size of “Proximal” spots adjusted with wing size from male flies. h: Size of “Proximal” spots adjusted with wing size from female flies. i: Size of “Middle” spots adjusted with wing size from male flies. j: Size of “Middle” spots adjusted with wing size from female flies. k: The ratio in male flies. l: The ratio in female flies. From a to f, there were significant differences between temperatures (p < 10–5, one-way ANOVA, degree of freedom = 2, F = 24.64 in a, 15.43 in b, 39.75 in c, 78.34 in d, 20.07 in e, 17.77 in f). Different letters indicate significant differences (p < 0.05, Tukey’s HSD test). From g to l, there were significant differences between temperatures (p < 10–5, Kruskal–Wallis rank sum test, degree of freedom = 2, χ2 = 45.315 in g, 62.827 in h, 15.805 in i, 16.804 in j, 6.414 in k, 42.698 in l). Different letters indicate significant differences (p < 0.05, Wilcoxon rank sum test with Bonferroni correction). Black bars indicate mean values

When we measured spot size, we found that the mean spot size of wings from flies reared under “Condition 2” (reared at 18 °C from P4 (i) to P14-15) became the smallest (Fig. 6c-f). Both in males and females, all comparisons of “Proximal” spot size showed significant differences by Tukey’s HSD test (Fig. 6c, d). For “Middle” spot size, a significant difference between the spot size of the flies reared under “Condition 1” and “Condition 3” was detected in males by Tukey’s HSD test (Fig. 6e), but it was not detected in females (Fig. 6f).

When the spot size is adjusted with wing size, the mean spot size of the wings from flies reared under “Condition 2” became the smallest in all categories (Fig. 6g-j). No significant difference between the spot size of flies reared under “Condition 1” and “Condition 3” was detected in any category (Fig. 6g-j).

Under “Condition 2”, “Proximal” spot was smaller than “Middle” spot (Fig. 5b). Analyzing the ratio of “Proximal” size to “Middle” size, we found that the ratio becomes smaller under “Condition 2” both in males and females (Fig. 6k, l). A significant difference between the spot size of the flies reared under “Condition 1” and “Condition 3” was not observed (Fig. 6k, l).

When log10 (spot size) is regressed against log10 (polygon size), we found that the correlation between log10 (spot size) and log10 (polygon size) was the strongest in male “Proximal” spots (Fig. 7, Table S1). As results for testing whether there is an interaction between log10 (polygon size) and conditions (testing whether regression lines have different slopes or the not), no significance was detected in any categories by two-way ANOVA (Table S2). By ANCOVA, it was shown that the effect of conditions was significant in all categories (Table 1). In the results of the post hoc analyses, significant differences between “Condition 1” and “Condition 2”, and between “Condition 2” and “Condition 3” were detected in all categories. No significant difference between “Condition 1” and “Condition 3” was detected in any category (Table 2).

Log–log plot of spot size (μm2) and polygon area (μm2). The spot size is from flies whose rearing temperatures were changed during the pupal period. a: “Proximal” spots on wings from male flies. b: “Proximal” spots on wings from female flies. c: “Middle” spots on wings from male flies. d: “Middle” spots on wings from female flies