Study species and collection
In this experiment we used the damselfly I. elegans as focal species. As top predator, we used the European perch (Perca fluviatilis) to impose non-consumptive predator effects on the damselfly. I. elegans a common insect species in Europe, occurring from northern Spain to central Sweden (Dijkstra et al. 2020). Central Europe populations are uni- and bivoltine (one or two generations per year, respectively), depending on the thermal conditions (Corbet et al. 2006). Larvae hatch 2-3 weeks after egg laying. Eggs and larval stages commonly share habitats with predatory fish (Zwick 2001). Fish cues can affect egg and larval life histories and physiology in the study species (Stoks et al. 2015; Fontana-Bria et al., 2017; Sniegula et al., 2020).
Adult I. elegans females were collected at a pond in Zabierzów Bocheński, Poland (50°03'16.3"N, 20°19'45.7"E). This fish pond contains P. fluviatilis. In total, 40 and 36 females were caught in copula on 22 June 2019 (i.e., early group) and on 7 July 2019 (i.e., late group). Females were individually placed in plastic cups with perforated lids and wet filter paper for egg laying, and transported by car in a Styrofoam box to the Institute of Nature Conservation PAS (INC PAS), Krakow, Poland. Adult females were kept in a room at a temperature of 24°C and natural daylight (photoperiod). Females laid eggs within three days after they had been field-collected. In total 22 clutches were used for the early group treatment, and 26 clutches for the late group treatment. After egg laying, females were released in their natural population.
Ten P. fluviatilis (age: 1+) were caught in Dobczyce lake (49°52′27″N, 20°2′55″E) on 19 June 2019. Five fish were used in the experiment, another five were used as a backup. Fish collection and housing were done with a permission from the Local Ethical Committee (ref. 261/2019).
Egg clutches from early collected females were pooled, and the same was done with eggs from late collected females. The two hatching phenology groups, early (E) and late (L), had 16 days difference in hatching dates, corresponding to the time interval between adult damselfly field collection dates. Such difference in hatching dates occurs in the natural populations because of the long I. elegans mating season and mixed voltinism in the sampling region (Corbet et al. 2006; Mikolajczuk 2014). We also created mixed phenology groups, where early hatched individuals shared the same container with late hatched individuals. Note that for the statistical analyses early and late hatched individuals in mixed phenology groups were considered as two different groups, E+L and L+E, where the group E+L referred to the early larvae in the presence of late larvae, and the group L+E to the late larvae in the presence of early larvae. This resulted in four phenology groups: non-mixed E and L, and mixed E+L and L+E. In the non-mixed phenology groups sets of 16 larvae of the same phenology group (E or L) were placed in separate containers, and in mixed phenology groups 8 larvae from E and 8 larvae from L phenology groups were placed in the same container, creating E+L and L+E phenology groups. This way all containers contained 16 larvae. Each phenology group was studied under the four combinations of two top predator treatments (fish predator cues present and absent) and two temperature treatments (22°C and 26°C, hereafter, low and high temperature). This created the following full factorial crossed design: 4 phenology groups × 2 predator cue treatments × 2 temperature treatments × 12 replicated containers × 16 larvae = 2304 individuals at the start (Fig. 1). Throughout the experiment we used a constant photoperiod of L:D 16:8h, which corresponds to the summer photoperiod, i.e., peak of the larval growth season, at the collection site. We used two climate incubators (Pol-Eko ST 700) for damselfly rearing.
Hatching took place on 6 July 2020 (E group) and 22 July 2020 (L group) at the high temperature, and on 11 July 2020 (E group) and 27 July 2020 (L group) at the low temperature treatment. At hatching, we randomly chose 16 larvae from E and L groups and transferred them to separate containers (16 × 12 cm, height 8 cm) filled with 600 mL of dechlorinated tap water and two nylon net strips, providing hiding space for larvae and climbing structure during emergence. In E+L and L+E groups, we randomly choose 8 larvae from the E group and 16 days later added 8 larvae from the L group. Larvae were fed twice a day (morning and afternoon feeding) with Artemia salina nauplii. During the feeding, E and L groups received 10 portions/container (mean = 201.9 nauplii/portion, SD = 17.2). In mixed groups, early hatched larvae received five portions until late hatched larvae were introduced to the same containers. From this time, mixed phenology groups received 10 portions/container.
Every other day, 150 mL of water in every container was refilled with water containing predator cues or no predator cues. Earlier studies have shown that chemical cues of aquatic predators have an average half-life degradation time of ca. 36.5 h (Buskirk et al. 2014). Previous experiments on non-consumptive predator effects in damselfly larvae supported this (Sniegula et al. 2019b; Sniegula et al. 2020).
To distinguish early from late hatched individuals in E+L and L+E groups, we cut the tibia of either one left or one right middle leg. Individuals from E and L groups were marked the same way. The larvae were marked when 30 days old. This marking persists until emergence and does not impact the measured traits (Hagler and Jackson 2001; Sniegula et al. 2019a).
Freshly emerged individuals were individually transferred to a dry plastic cup and kept for 24 hours until the cuticle hardened. Next, damselflies were weighted and frozen at -80°C for physiology analyses. The experiment ended when the last damselfly larvae emerged.
The survival was noted daily between hatching and emergence. Individuals that emerged with fully developed body and wing parts were considered to have emerged successfully. Larval development time was measured as the number of days between hatching and emergence. One day after emergence, damselfly wet mass was measured to the nearest 0.1 mg with the use of an electronic balance (Radwag AS.62). The growth rate was calculated as adult wet mass divided by the number of days between hatching and emergence.
For physiological analyses, damselfly bodies without legs and wings were grinded with phosphate buffer solution (15 µL for each milligram of wet mass) and centrifuged at 10,000 g for 5 minutes at 4°C. All physiology analyses were done on homogenates.
The classical procedure for measuring total body fat in insects (Marsh and Weinstein 1966) was optimized for damselfly bodies. A volume of 8 µL homogenate was mixed with 56 µL 100% sulfuric acid, and heated for 20 minutes at 150°C. After cooling down, 64 µL Milli-Q-Water was added. Of this mixture, 30 µL was put in a well of a 384-well microliter plate, and absorbance was measured at 340 nm. The measurements were made on an Infinite M2000 (TECAN) plate reader. To convert absorbances into fat contents, the standard curve of glyceryl tripalmitate was used. The average of three technical replicates per sample was used in the statistical analyses.
Protein content (µg of protein/mg of body mass) was determined using the Bradford (1976) method. Of the homogenate, 1µL was mixed with 160 µL of Milli-Q-Water and 40 µL of Bio-Rad Protein Dye. After five minutes of incubation at 25°C, the absorbance was measured at 595 nm and converted into protein contents using standard curves of bovine serum albumin. The measurements were repeated three times per sample, and the average values used for statistical analyses.
A modified version of the assay described in Stoks et al. (2006) was used for determining PO activity. Of the homogenate, 10 µL was mixed with 10 µL of phosphoric buffered saline and 5 µL of chymotrypsin. The mixture was put in wells of a 384-well microtiter plate. Afterwards, the samples were incubated for 5 minutes at room temperature. After incubation, the substrate L-DOPA (1.966 mg dihydroxyphenyl-L-alanine per 1 mL of PBS-buffer) was added and mixed with the samples. Immediately afterwards, the linear increase in absorbance was measured at 490 nm every 20 seconds for 30 minutes at 30°C. The PO activity was quantified as the slope of the reaction curve, and the average of two technical replicates was used for statistical analyses.
All analyses were run using R 4.0.4. (R Development Core Team 2019). Generalized mixed models with a binomial distribution were used to separately analyse the survival and emergence success (glmer function in the lme4 package, Bates et al., 2015). The other life history traits (development time, wet mass and growth rate) and the physiological traits (PO activity, fat and protein contents) were analysed using linear mixed models (lmer function in the lme4 package, Bates et al., 2015). In all models, phenology group, top predator cue treatment, temperature treatment and sex were entered as fixed effects. Initially, models with all possible interactions were run. Interaction terms with p > 0.1 were removed from the final models. In all models, container nested within phenology groups were used as random variable. SMPE would be indicated by the pattern where the trait value in the E+L group would be statistically higher (survival, mass, growth rate, fat content, protein content and PO activity) or lower (development time) than in the other phenology groups. If a factor with more than two levels or any interaction term was found statistically significant, post hoc Tukey HSD tests (function lsmeans) were run to test pairwise between-level differences. Because of low number of surviving larvae in the L+E group, this group was excluded from all analyses, except for survival until emergence and emergence success.