Predator abundance
The number of predators on the aphid colonies varied spatiotemporally (Fig. 2). In particular, the number of predators in population A was significantly larger than that in population B in August but not in September (August, t20 = 3.93, P < 0.001; September, t29 = 1.44, P > 0.05). In population A, we found predators on the aphid colonies in August and September, but not in June and July. In August, the only predators found were A. ignipicta larvae (0.76 ± 0.19 individuals per aphid colony), whereas in September the predators comprised both A. ignipicta larvae (0.033 ± 0.033 individuals per aphid colony) and T. hamada larvae (0.033 ± 0.033 individuals per aphid colony). In population B, we found no predators in any of the months.
Temporal and between-population variation in the number of predators per aphid colony. The number of predators represents the sum of the numbers of A. ignipicta and T. hamada larvae. Error bars denote s.e. Asterisks indicate a significant difference between populations (***P < 0.001).
Soldier weapon size
The horns of soldiers from population A were longer than those of soldiers from population B in all months (June, F1,29 = 84.45, P < 0.001; July, F1,24 = 5.26, P = 0.030; August, F1,27 = 31.34, P < 0.001; September, F1,26 = 34.16, P < 0.001; Fig. 3a). Forelegs of soldiers from population A were longer than those of soldiers from population B in three of the months, with the exception of July (June, F1,29 = 55.32, P < 0.001; July, F1,24 = 0.56, P = 0.46; August, F1,27 = 88.29, P < 0.001; September, F1,26 = 20.72, P < 0.001; Fig. 3b).
Temporal and between-population variation in the weapon size of soldiers. The average (a) horn, (b) foreleg, and (c) body length of soldiers in populations A and B from June to September are shown. Error bars denote s.e. Asterisks indicate significant differences between populations (*P < 0.05; ***P < 0.001).
Furthermore, in population A, horn length varied temporally (F3,52 = 3.96, P = 0.012). Post hoc comparisons revealed that horn length was significantly longer in August than in June or September (June vs. August, F1,27 = 7.96, P = 0.008, August vs. September, F1,27 = 6.07, P = 0.020; Fig. 3a), but did it not vary between other months (June vs. September, F1,28 = 0.52, P > 0.05, July vs. August, F1,21 = 0.98, P > 0.05, July vs. September, F1,22 = 0.06, P > 0.05; Fig. 3a).
In addition, in population A, foreleg length varied temporally (F3,52 = 12.67, P < 0.001). Post hoc comparisons revealed that foreleg length was significantly longer in August than in the other 3 months (June vs. August, F1,27 = 37.09, P < 0.001, July vs. August, F1,21 = 15.32, P < 0.001, August vs. September, F1,27 = 25.86, P < 0.001; Fig. 3b), but it did not vary between other months (June vs. July, F1,25 = 1.00, P > 0.05, June vs. September, F1,28 = 1.80, P > 0.05, July vs. September, F1,22 = 0.06, P > 0.05; Fig. 3b).
In population B, horn length and foreleg length varied temporally (horn length, F3,56 = 26.29, P < 0.001; foreleg length, F3,56 = 31.49, P < 0.001). Post hoc comparisons revealed that horn length in June was significantly shorter than in the other months (June vs. July, F1,28 = 67.21, P < 0.001, June vs. August, F1,29 = 42.76, P < 0.001, June vs. September, F1,29 = 29.48, P < 0.001; Fig. 3a) and horn length was significantly longer in July than in the other months (July vs. August, F1,27 = 6.05, P = 0.020, July vs. September, F1,27 = 13.74, P < 0.001; Fig. 3a), whereas horn length in August was not significantly different from that in September (August vs. September, F1,28 = 1.66, P > 0.05; Fig. 3a).
Furthermore, post hoc comparisons revealed that foreleg length was significantly shorter in June than in the other months (June vs. July, F1,28 = 50.71, P < 0.001, June vs. August, F1,29 = 42.05, P < 0.001, June vs. September, F1,29 = 43.54, P < 0.001; Fig. 3b), and foreleg length was significantly longer in July than in the other months (July vs. August, F1,27 = 4.64, P = 0.039, July vs. September, F1,27 = 8.30, P = 0.007; Fig. 3b), but foreleg length in August was not significantly different from that in September (August vs. September, F1,28 = 0.63, P > 0.05; Fig. 3b).
The body length of soldiers from population A was greater than that of soldiers from population B in June and August (June, F1,29 = 22.02, P < 0.001; July, F1,23 = 1.57, P > 0.05; August, F1,27 = 7.15, P = 0.015; September, F1,28 = 1.40, P > 0.05; Fig. 3c).
Furthermore, in population A, body length varied temporally (F3,51 = 9.91, P < 0.001). Post hoc comparisons revealed that body length was significantly longer in August than in July or September (July vs. August, F1,23 = 4.64, P = 0.04, August vs. September, F1,27 = 7.32, P = 0.01; Fig. 3c) and it was longer in June than in July or September (June vs. July, F1,28 = 10.59, P = 0.002, June vs September, F1,29 = 10.66, P = 0.002; Fig. 3c). However, it did not vary between other months (June vs. August, F1,29 = 2.87, P > 0.05, July vs. September, F1,24 = 0.02, P > 0.05; Fig. 3c). Therefore, in population A, the soldiers’ weapon size, which was largest in August, was exaggerated in that month in comparison with body length because their body length was not greatest in August.
In population B, body length also varied temporally (F3,56 = 2.79, P = 0.05). Post hoc comparisons revealed that body length was significantly longer in June than in July or September (June vs. July, F1,24 = 21.21, P < 0.001, June vs September, F1,28 = 27.18, P < 0.001; Fig. 3c). However, it did not vary between other months (June vs. August, F1,27 = 3.74, P > 0.05, July vs. August, F1,27 = 0.72, P > 0.05, July vs. September, F1,27 = 0.02, P > 0.05, August vs. September, F1,28 = 0.61, P > 0.05; Fig. 3c). Therefore, in population B, the soldiers’ weapon size was exaggerated compared to their body length in August because their body length was not greatest in August.
Soldier aggressiveness
When encountering a predatory A. ignipicta larva, a soldier immediately grasped the predator and pierced it with the horns. In response, the predator attempted to tear the soldier off by biting or pulling at it with the mandibles. Once the soldier was repelled, it immediately became motionless and did not fight again. The duration of this aggressive interaction between a soldier and the predator differed between the populations (t11.91 = 2.58, P = 0.024; Fig. 4). The predator needed four times longer to tear a soldier from population A off its body than one from population B [population A (n = 11): 129.54 ± 36.35 s (mean ± s.e.); population B (n = 6): 30.94 ± 11.67 s (mean ± s.e.)]. No soldier killed the predator during the observation period in any of the treatments of this experiment.
Duration of aggressive interaction between a soldier and a specialist predator (A. ignipicta larva) in the two populations. The duration from when the soldier first grasped the predator until the predator tore the soldier off its body was measured in a Petri dish (population A: n = 6, population B: n = 11). The soldiers from population A (filled circle), which had a higher predator density in the wild and soldiers with larger weapons and higher aggressiveness, interacted with the predator for significantly longer than those from population B (open circle), which had a lower predator density in the wild (Welch’s t-test). Symbols and error bars indicate mean ± s.e. The asterisks indicate a significant difference between populations (*P < 0.05).
Soldier defensive prowess
In both treatments, the survival rate of first-instar reproductive individuals gradually decreased with time, but the pattern of decrease differed between the populations (repeated-measures ANOVA, F2.36,49.57 = 3.75, P = 0.024; Fig. 5). The survival rate of the first-instar reproductive individuals at 120 min was 44% in population A and 27% in population B. These results indicate that soldiers from population A improved the survival rate of first-instar reproductive individuals 1.5-fold compared with soldiers from population B. The survival rate of the soldiers at 120 min was 24% in population A and 17% in population B. The soldiers that did not survive were killed by the predator. No soldier killed the predator during the observation period in any of the treatments of this experiment.
The defensive prowess of soldiers against predatory A. ignipicta larvae. The survival rate of 44 first-instar reproductive individuals introduced into a Petri dish with six soldier aphids and a starved predator was monitored for 2 h. Soldiers from population A were used in 12 Petri dishes, and soldiers from population B were used in 11 Petri dishes. The first-instar reproductive individuals protected by soldiers from population A (filled circle), which had a higher predator density in the wild and soldiers with larger weapons and higher aggressiveness, had a significantly higher survival rate than those collected from population B (open circle), which had a lower predator density in the wild (repeated-measures ANOVA). Symbols and error bars indicate mean ± s.e. Asterisks indicate significant differences between populations (*P < 0.05).
The difference in the survival rate of first-instar reproductive individuals between the treatments was significant at all observation times (30 min: t21 = 2.70, P = 0.014; 60 min: t21 = 2.25, P = 0.036; 90 min: t21 = 2.57, P = 0.019; 120 min: t21 = 2.56, P = 0.020). Hence, the survival rate of first-instar reproductive individuals differed between the populations soon after the beginning of the experiment, and the difference became more pronounced over time.
Source: Ecology - nature.com
