Resource availability and predator cues shape within- and transgenerational reaction norms

(a). Snail origin

Physa acuta (Physidae, Hygrophila, Gastropoda) is a freshwater snail with a worldwide geographical distribution [36]. It is a simultaneous hermaphrodite with preferential outcrossing, i.e. eggs are self-fertilized only when no mate is available to provide allosperm [37]. In this study, adult P. acuta snails from a laboratory outbred population have been used as a source (F₀ individuals). These snails originated from wild populations from a lentic backwater of the Rhône River near Lyon (south-eastern France) and have been reared in controlled conditions (25°C and a 12-hour light/dark cycle) for 21 generations. A total of 112 F₀ adult individuals were grouped together for three days in two 10 l aquariums filled with dechlorinated tap water for mating and laid eggs before being removed. Such a mass mixed breeding approach reflects the breeding in the natural environment but prevents the recording of family identity. The eggs hatched after approximately 7 days at 25°C and constituted the F₁ generation.

(b). Experimental design

F₁ generation. F₁ individuals developed during 10 days in dechlorinated tap water in the 10 l aquariums. Next, 112 F₁ snails were randomly assigned to each of the four treatments: control water (i.e. without predator cues) with ad libitum food (control non-exposed and high resource, (CH)), control water with restricted food (control non-exposed and low resource, (CL)), predator-cue exposure with ad libitum food (predator cues and high resource, (PH)) and predator-cue exposure with restricted food (predator cues and low resource, (PL)), resulting in a full-factorial design (n total = 112 snails × 2 predator-cues conditions × 2 resource conditions = 448 snails). Each individual was isolated in an 80 ml box, the water was renewed twice a week and snails were fed with chopped boiled lettuce. Exposure to predator cues was achieved by rearing snails in water containing predator cues. This predator-cue water was obtained by housing Faxonius limosus crayfish (one crayfish per 5 l) that were fed with snails for three days. After collecting the crayfish's water, we added ca. 1.5 g of live P. acuta snails crushed for 1 hour in the solution. Thus, the predator-cue water contained olfactory cues of crayfish presence, both crayfish kairomones and snail alarm cues [38]. For the high-resource treatment, snails were fed ad libitum at each water renewal (twice a week), low-resource treatment was achieved by feeding the snails only once a week during the first water renewal (no food was added at the second water renewal, leaving the snails without food for the next 3 days). This alternating availability and unavailability of resources is consistent with natural conditions (biofilm availability varies in space and time in freshwater ecosystems) and allows for precise control over resource availability while simulating low-resource availability throughout the snails’ lifetimes. At 25 days (15 days of treatment, called ‘juvenile stage’ thereafter), total mass and shell length were measured for each individual and then snails were put back in their boxes to continue the experiment. At 45 days old (35 days of treatment, called ‘adult stage’ thereafter), crawling-out of the water behaviour was recorded, which is a classic response to benthic predators such as crayfish [38,39]. Snails were again measured (total mass, shell length) and grouped together in a 10 l aquarium within each treatment to mate for 24 hours (i.e. mass mating allowing random copulations). Then, F₁ snails were isolated to lay eggs for 3 days, allowing us to assess the number of eggs laid and the number of juveniles produced, which subsequently became the F₂ generation. Finally, F₁ snails were sacrificed—put in boiling water to extract soft tissues from the shell—to subsequently measure shell crush resistance, a defence against crayfish predation [30,31,35].

F₂ generation. The F₂ experimental design was the same as for F₁, except that F₂ snails were all reared in control conditions (control non-exposed and high resource, i.e. in water without predator cues and with ad libitum food): at 10 days old, F₂ individuals within each parental treatment were pooled in aquarium, then 112 F₂ snails per parental treatment were randomly selected and isolated (n total = 112 F₂ snails × 2 parental predator-cues conditions × 2 parental resource conditions = 448 F₂ snails) and finally measured at the ‘juvenile stage’ (i.e. 25 days; total mass and shell length) and at the ‘adult stage’ (i.e. 45 days; crawling-out behaviour, total mass, shell length, shell crush resistance, number of eggs laid and number of juveniles produced).

The experiment was conducted in a temperature-controlled room, maintaining a constant temperature of 25°C and a 12-hour light/dark cycle.

(c). Trait measurements

Snail mass at juvenile and adult ages: the total fresh mass of each snail (shell mass + body mass) was determined using a precision scale to the nearest 0.001 g. Prior to measurement, snails were gently dried with tissue to remove excess moisture. Shell length at juvenile and adult stages: the maximum length of the snail shell was measured using ImageJ software to the nearest 0.001 mm from a photograph of the shell with its aperture upward, taken by a camera mounted on a binocular loop. Crawling-out behaviour was scored once at the adult stage (45 days old) as a binary variable (the snail was out of the water or touching the surface versus in the water [35]), approximately 3 hours after the water renewal. Shell crush resistance at adult age: the resistance of empty shells to crushing was assessed by a device designed to apply pressure on the shell (with the aperture pointing downward), and the force was recorded to the nearest 0.1 g. The force recorded at the peak represents the force needed to crush the shell (see [35] for more details). The number of eggs laid by each snail was counted before hatching. The number of juveniles was counted 10 days after laying (i.e. approx. 2–3 days after hatching).

(d). Statistical analyses

Statistical analyses tested the effects of predator exposure, resource availability and their interaction on the expression of traits in parents (F₁) and in offspring (F₂). Data from the F₁ and F₂ generations were analysed separately. Snail size: shell length and snail mass (cube root transformed) were highly correlated in both F₁ and F₂ (supplementary material S1). Thus, to reduce the number of variables, a principal component analysis (PCA) was performed. The first axis of this PCA, which well summarized the variability information contained in the two variables (PC1 > 95%; supplementary material S1), was interpreted as an estimation of snail size. The coordinate on the first PCA axis was extracted for all individuals (juvenile and adult stages separately) and analysed in two linear models (LM; one for the juvenile stage and one for the adult stage) using an analysis of variance (ANOVA) with predator exposure (factor variable with two levels), resource availability (factor variable with two levels) and their interaction. To give a biological estimate of the effects, the raw length and weight (and not the first PCA axis coordinate) were used to calculate the percentage of differences between means of environmental treatment. The figures of raw length and weight and the analyses are available in supplementary material S2. Crawling-out behaviour was a binary variable (0 or 1) and was then analysed with a generalized linear model (GLM) using a binomial distribution (logit link function) and with the same predictors as for models of snail size. The significance of effects was tested using likelihood ratio tests. Shell crush resistance was analysed in an LM using an ANOVA with the same predictors as the other models. Reproductive outputs: the numbers of eggs and juveniles were analysed with a GLM using a negative binomial distribution to account for overdispersion and with the same predictors at the other models. The significance of effects was tested using likelihood ratio tests. Post hoc pairwise comparisons (using a Tukey test) were performed on all models to identify differences between the combinations of environmental treatments. The versions of R and RStudio, as well as the R packages used and their respective versions, are detailed in the ReadMe file provided at the data and code repository.

(a). Size of F₁ snails at juvenile and adult stages

F₁ snail size at juvenile and adult stages was significantly influenced by the interaction between predator-cue exposure and resource availability (table 1 and figure 2). In the high-resource environment, juvenile and adult snails exposed to predator cues were smaller than non-exposed snails (juvenile PH-CH: 12% shorter and 29% lighter, t[432] = –7.85, p < 0.001; adult PH-CH: 14% shorter and 36% lighter, t[432] = –14.47, p < 0.001). In the low-resource environment, juvenile snails exposed to predator cues had a similar size than non-exposed snails (PL-CL: t[432] = –1.91, p = 0.221), but adult snails exposed to predator cues were smaller than non-exposed snails (PL-CL: 7% shorter and 19% lighter, t[432] = –3.58, p = 0.002). Overall, juvenile and adult snails reared in low-resource environments were significantly smaller than those in high-resource environments (figure 2; juvenile L-H: 40% shorter and 77% lighter; adult L-H: 40% shorter and 79% lighter).

(b). Anti-predator behaviour and shell resistance of F₁ adult snails

The number of adult F₁ snails exhibiting the crawling-out behaviour was influenced by the interaction between predator-cue exposure and resource availability (table 1 and figure 3A). Predator cues led to an increase in crawling-out behaviour in both resource environments (PH-CH: odds.ratio = 11.07, z = 7.38, p < 0.001; PL-CL: odds.ratio = 51.39, z = 9.38, p < 0.001). This increase is slightly higher for low-resource snails. However, the low-resource availability did not significantly influence the behaviour both in non-exposed individuals (CL-CH: odds.ratio = 0.43, z = –2.21, p = 0.120) and in predator-cue exposed ones (PL-PH: odds.ratio = 1.99, z = 1.87, p = 0.242) relative to the high-resource environment.

Shell crush resistance was independently influenced by predator-cue exposure and resource availability (non-significant interaction, table 1 and figure 3B). Shell crush resistance was significantly 10% higher in snails exposed to predator cues relative to non-exposed snails regardless of the resource environment and was 7.9% lower in low resource compared to high resource regardless of predator exposure (non-significant interaction, table 1 and figure 3B).

(c). Reproductive outputs of F₁ adult snails

Egg and juvenile numbers produced by the F₁ adult snails were only influenced by resource availability (table 1 and figure 4). Snails raised in low resource laid significantly fewer eggs and produced fewer juveniles than those raised in high resource (–84 and –83%, respectively).

(d). Size of F₂ snails at juvenile and adult stages

F₂ snail size at the juvenile and adult stages was significantly influenced by the interaction between parental predator-cue exposure and parental resource availability (interaction is marginally significant for adults, see table 2 and figure 5). The parental exposure to predator cues significantly increased the juvenile and adult size of F₂ snails from both resource parental environment (juvenile PH-CH: 9% longer and 3% heavier, t[444] = 2.73, p = 0.033; juvenile PL-CL: 35% longer and 92% heavier, t[444] = 8.66, p < 0.001; adult PH-CH: 3% longer and 12% heavier, t[444] = 3.17, p = 0.010; adult PL-CL: 7% longer and 26% heavier, t[444] = 5.60, p < 0.001); this increase was slightly higher for F₂ snails from low-resource parental environment (table 2 and figure 5). Overall, F₂ juvenile and adult snails from the low-resource parental environment were significantly smaller than F₂ ones from the high-resource parental environment (figure 5; juvenile CL-CH: 41% shorter and 80% lighter, t[444] = –24.85, p < 0.001; juvenile PL-PH: 32% shorter and 62% lighter, t[444] = –18.91, p < 0.001; adult CL-CH: 8% shorter and 27% heavier, t[444] = –7.51, p < 0.001; adult PL-PH: 4% shorter and 18% lighter, t[444] = –5.11, p < 0.001).

(e). Anti-predator behaviour and shell resistance of F₂ adult snails

The number of F₂ adult snails exhibiting the crawling-out behaviour was marginally influenced by the interaction between parental resource and parental exposure to predator cues (table 2 and figure 6A). Parental exposure to predator cues increased the crawling-out behaviour of F₂ adult snails in the high-resource parental environment (PH-CH: odds.ratio = 3.51, z = 4.30, p = 0.001) but did not significantly affect their behaviour in the low-resource parental environment (PL-CL: odds.ratio = 1.62, z = 1.76, p = 0.293). The parental resource environment did not significantly affect the offspring behaviour from non-exposed parents (CL-CH: odds.ratio = 1.64, z = 1.68, p = 0.333) and exposed parents (PL-PH: odds.ratio = 0.75, z = –1.03, p = 0.730).

Shell crush resistance of F₂ snails was independently influenced by the parental resource and the parental exposure to predator cues (non-significant interaction, table 2 and figure 6B). The shell resistance increased (+7%) for F₂ snails from parents exposed to predator cues relative to those from non-exposed parents. Shell was 18% less resistant for F₂ snails from the low-resource parents.

(f). Reproductive outputs of F₂ adult snails

Egg number laid and juvenile number produced by F₂ snails were influenced by the parental resource (interaction with parental predator-cue exposure marginally significant for juvenile number; table 2 and figure 7). F₂ snails from the low-resource parental environment produced fewer eggs (−17%) but more juveniles (+31%) than those from the high-resource parental environment (figure 7). In addition, F₂ snails from exposed parents produced more juveniles (+12%) than those from non-exposed parents (table 2 and figure 7B). In the high-resource parental environment, there was no difference between F₂ snails from exposed and non-exposed parents (egg number PH-CH: ratio = 0.99, z = –0.11, p = 0.999; juvenile number PH-CH: ratio = 1.01, z = –0.19, p = 0.997). In the low-resource parental environment, F₂ snails from exposed parents laid a similar number of eggs but produced more juveniles than those from non-exposed parents (egg number PL-CL: ratio = 1.17, z = 2.17, p = 0.132; juvenile number PL-CL: ratio = 1.23, z = 2.73, p = 0.032).

(a). The combination of information- and state-based plasticity shapes the within-generational plasticity reaction norms

Developmental exposure to predator cues is well known to induce information-based WGP (28), including shifts in life-history traits (e.g. growth, fecundity) as well as specific behavioural and morphological anti-predator traits, across many organisms ranging, for example, from birds [50], lizards [51] and amphibians [52] to molluscs [42], insects [53] and crustaceans [54]. Changes in life-history traits may arise either as direct responses to predation cues (e.g. they can be defensive) or as by-products of the induction of behavioural and morphological defences through energetic trade-offs, thereby intertwining state- and information-based plasticity [27]. For instance, Tollrian [55] showed, in Daphnia pulex, that the life history shifts in response to predator cues were independent of morphological changes (‘neckteeth’). Conversely, Teplitsky et al. [56] demonstrated that the induction of morphological defences in tadpoles of Rana dalmatina was associated with reduced growth and larval development rates, reflecting potential energetic costs of producing anti-predator traits.

In the same way, our results revealed the induction of anti-predator traits—namely increased crawling-out behaviour and shell crush resistance—in response to predator cues associated with reduced growth, consistent with our previous studies (e.g. [32,38,40]). The reduced individual size is a non-consumptive effect commonly reported under predation risk in freshwater snails [57,58] (but see [3,30,59]), and reflects energetic trade-offs behind information-based WGP of defences [24]. By experimentally manipulating resource availability and predator cue exposure, our study also provided evidence that the WGP reaction norms were shaped by a combination of state- and information-based plasticity:

First, the reduction in individual size in response to predator cues observed in high-resource environments was absent (juveniles) or less pronounced (adults) in low-resource environments, revealing a masking non-additive pattern between the state- and information-based plasticity. The smaller baseline size of snails exposed to limited resources may impose physical constraints that limit further reduction owing to a reallocation of energy away from growth. For instance, other studies on freshwater snails and amphibians found such a masking non-additive pattern by revealing that the negative effects of predation risk on growth were strongest when prey were at low density (i.e. low competition for resources) but weakened or disappeared completely when prey were at high density (i.e. high competition for resources [10,60–62]).

Second, despite resource limitations, individuals from the low-resource environment still expressed the anti-predator traits (crawling-out behaviour, shell crush resistance) at levels comparable to those from the high-resource environment. Interestingly, a revealing non-additive pattern was observed for crawling-out behaviour: the slope of the reaction norm increased for individuals from the low-resource environment, a pattern already observed in another Physa species [63]. This may suggest a compensation for heightened vulnerability to crayfish predation owing to a smaller size in low-resource environments with an increase in crawling-out behaviour. Such trait compensation in prey facing predators has been observed in several studies on freshwater snails (e.g. [64–66]) and also, for example, on birds [67] and fishes [68,69]. By contrast, we found no interaction between information- and state-based plasticity in shell crush resistance (i.e. an additive pattern). The shell crush resistance increased with predator-cue exposure with the same magnitude in individuals from low- and high-resource environments.

Third, exposure to predator cues did not influence the reproductive output, but the low-resource environment strongly reduced it. This suggests that individuals exposed to predator cues still find the resources they need to invest in reproduction, even in a low-resource environment. Brönmark et al. [70] consistently found that defence traits were expressed independently of environmental quality (individual density) in the freshwater snail Radix balthica and resulted in reduced growth rates but also decreased fecundity, particularly with limited resources (see also [41]).

Overall, our results demonstrate that exposure to predator cues influences the somatic state, but above all, even under resource scarcity, the induction of anti-predator traits remains a priority, drawing energy away from growth (but not from reproduction), which suggests strong selection from predators. Such investment in defences might improve individual fitness by improving survival under predation risk and thus by extending lifespan, thereby allowing reduced reproductive output, particularly in low-resource environments, to be compensated across multiple reproductive events [71]. This echoes the general prediction that selection maximizes parental fitness rather than offspring fitness, especially in species that lack parental care and that reproduce many times [23]. Additional research evaluating lifetime reproductive success of P. acuta under predation risk and resource scarcity must be conducted before definite conclusions about the adaptive value of phenotypic responses observed.

(b). The combination of information- and state-based plasticity shapes the transgenerational plasticity reaction norms

Our results provided evidence that, overall, the parental environments influenced the offspring traits, indicating TGP as previously observed in this system (e.g. [32,35,40]). Parental exposure to predator cues increased the size, crawling-out behaviour and shell crush resistance of F₂ individuals. This suggests adaptive transgenerational information-based responses, whereby parental information ‘anticipates’ the risk of predation that offspring will experience and induces the defences of offspring accordingly [17,23]. Indeed, increased size might enable individuals to reach a ‘refuge size’, reducing vulnerability to crayfish predation (as crayfish preferentially consume smaller snails [3,30,64]), while enhanced crawling-out behaviour and shell crush resistance—resulting from a complex combination of shifts in shell size, shape and thickness—improve the likelihood of surviving predator attacks [31,35,72]. Additional work is nevertheless necessary to test the adaptive value of transgenerational responses of P. acuta to crayfish.

Above all, our results demonstrated that the combination of parental resource availability and exposure to predator cues shaped the TGP reaction norms via an additive pattern for shell crush resistance and non-additive masking or revealing patterns for size, crawling-out behaviour and marginally for reproductive outputs. The overall weaker shell crush resistance of F₂ from low-resource parents may be a physical consequence of their smaller size compared to F₂ from high-resource parents. By constraining the absolute level of size, state-based TGP influenced in turn the absolute level of shell crush resistance in the absence of actual predator cues, which might influence the predator-induced WGP responses (not tested in this study but see [32]). The increase in size in response to parental predator-cue exposure was slightly higher for F₂ snails from low-resource parents (revealing pattern), suggesting that these individuals may take advantage of their favourable high-resource current environment to partly compensate for the state-based TGP constraining their size. By contrast, the crawling-out behaviour of F₂ individuals exhibited a masking pattern: parental exposure to predator cues induced crawling-out only in F₂ from high-resource parents, while no such response was observed in F₂ from low-resource parents. The lack of significant behavioural response to parental predator exposure in F₂ from low-resource parents might illustrate the trade-off between resource acquisition and anti-predator traits. Indeed, a poor state-driven effect may cause an elevated baseline crawling-out behaviour in control individuals and a lower crawling-out behaviour in parentally exposed individuals. The elevated baseline crawling-out behaviour might reflect a behavioural compensation for morphological vulnerability (i.e. small size) even in the absence of parental and actual predator exposure [64]. The poor state of F₂ from low-resource parents might also constrain the time spent crawling out at the expense of safety to prioritize foraging in the water. For example, Relyea [60] showed that increased competition (i.e. low-resource condition) caused higher activity of Rana sylvatica tadpoles, even under predation risk. These results illustrate well that the intertwining between information- and state-based effects can drive TGP reaction norms through the resource acquisition-safety trade-off: current environmental conditions, such as the risk of predation, can be perceived by parents in order to predict the future environment of their offspring and adjust their phenotype accordingly, but these anticipatory responses may be constrained because such environmental conditions also influence the body state of the parents and, consequently, that of their offspring [16,18,20,48].

Finally, the results may suggest a revealing pattern for juvenile number: parental exposure to predator cues increased the number of juveniles only in F₂ snails from low-resource parents. We further showed that this predator-induced TGP on juvenile number was mediated by offspring size, since the effect of parental exposure to predator cues disappeared once mass was accounted for (supplementary material S3). The positive relationship between size and reproductive outputs is well documented in P. acuta (e.g. [71,73]) and in ectotherms in general [74]. Consequently, the increased juvenile number produced by F₂ individuals from low-resource and predator-exposed parents may not reflect an energy reallocation from growth to reproduction, but rather a by-product of changes in size and defences driven by the interaction between parental state and parental predator information. Interestingly, F₂ snails from low-resource and predator-exposed parents exhibited the highest juvenile number overall. While they produced a similar number of eggs as snails from high-resource and predator-exposed parents, they produced a greater number of surviving juveniles. This suggests a higher hatching success and early juvenile survival (for 3 days post-hatching) in the offspring (F₃) of F₂ from low-resource and predator-exposed parents. Thus, F₂ individuals may improve their overall reproductive success (by increasing F₃ offspring fitness) in the absence of current predator cues under combined parental state- and information-based effects by modulating the trade-off between offspring number and offspring quality. This result is rare evidence that the fitness benefits of TGP are condition-dependent (here a combination of ancestral exposure to predator cues and low-resource availability) and can be revealed only three generations later (see evidence for grand-parental effects in P. acuta [38] or in other taxa e.g. [75–77]), extending the widely documented condition-dependent parental effects on reproductive strategy in many organisms (see reviews [20,22,23]). Further investigation of maternal provisioning in eggs (e.g. egg size, energetic reserves in eggs) would be valuable to unravel the proximal mechanisms behind the transgenerational modulation of the trade-off between offspring number and offspring quality.