Direct and cross-generational effects of reproduction on fitness and behavioral variability in male-biased environments

Abstract

Population structure determines individuals’ interactions and trade-offs with evolutionary consequences. Male-biased populations increase intrasexual competition and intersexual harassment, reducing female resource acquisition, and thus, resources availability for the following generation. We analyzed direct and cross-generational effects of male harassment in two generations of damselflies (Odonata). We exposed adult females to treatments with different sex-ratio and density (balanced and male-biased) to modify the male harassment level. We analyzed female fecundity, fertility, and number of faecal deposits as an indirect measure of resources acquisition. We studied female flight performance after repeated exposures to males. We analyzed survivorship, development, exploration, thigmotaxis, and feeding latency of larvae produced by the experimental females. In both generations, we analyzed four metrics of behavior: mean value, interindividual differences in plasticity, intra-individual unpredictability, and repeatability. Mating duration increased in male-biased treatment, whereas female resources acquisition and fertility decreased. Females that mated longer showed higher fecundity when they were exposed to balanced treatment, but not if they were exposed to male-biased treatment. Females from the male-biased treatment showed interindividual differences in plasticity and no repeatability in flight performance. Offspring showed balanced sex-ratio and similar survivorship, development, and feeding latency independently of the parental treatment; however, females exposed to male-biased treatment produced offspring with higher differences in exploration plasticity and daughters less explorative and with higher unpredictable thigmotaxis. We propose prolonged copulation as courtship at balanced sex-ratio but a cost to females under male-biased sex-ratio. Cross-generational effects in behavioral variability may be a mechanism to cope with predicted future environments.

Environmental perception constitutes the first step in cognitive and adaptive processes leading to behavioral modifications (e.g., Stewart 1993). Population structure (e.g., sex-ratio and age distribution) is one of the main factors shaping behavior and fitness in organisms with scramble competition (Wolf and Weissing 2010; Engen et al. 2020). The individual perception of population structure may depend on the maturation time of each sex, particularly in species with protandry or protogyny. Balanced sex-ratios are expected if sexes mature synchronously and are equally costly to produce (Fisher 1930), but if one sex is more costly, or involves higher inclusive fitness (see Hamilton 1964), parents can modify their investment towards the sex with higher associated success (sex allocation, Charnov 1983). Consequently, offspring sex-ratio or sexual allocation could be biased in the next generation according to the parents’ experiences.

Operational sex-ratio is commonly biased towards males in many taxa (insects, Hurd et al. 1994; birds, Torres and Drummond 1999; mammals, Allainé et al. 2000; reptiles, Nelson et al. 2002; amphibians, Grayson et al. 2012; fish, Rundio et al. 2012). Male competition increases under male-biased sex-ratio. To obtain an optimal reproductive rate when the ratio of available females to males is low, males increase their matting attempts or force mating, reducing courtship and increasing harassment (e.g., Jirotkul 1999; Cordero-Rivera and Andrés 2002; but see Carrillo et al. 2012). Male harassment, defined as persisting male attempts towards females not interested in mating (Clutton-Brock and Parker 1995), is widespread in animals and produces direct and indirect costs. Directly, harassing males modify female behavior (Sundaresan et al. 2007), energy expenditure (reviewed in Arnqvist and Rowe 1995), time available for resources acquisition (Dadda et al. 2005), resource allocation (Córdoba-Aguilar 2009), immune response (Iglesias-Carrasco et al. 2019), and produce injuries over their female partners (Moldowan et al. 2020). Male harassment may be unproductive even for harassing males, as harassment direct benefits (e.g., increased number of matings) may not be outweighed by future costs. For instance, the sons of guppies with higher harassment behaviors possess shorter gonopodia and therefore, lower ability to harass (Gasparini et al. 2012). Indirectly, male harassment costs affect the next generation by different mechanisms as sexual allocation (Charnov 1983) or epigenetics (Holliday 1987), modifying the offspring fitness and consequently, population dynamics (reviewed in Svensson 2019; but see Gasparini et al. 2012). Therefore, the reproductive interactions may condition the experiences perceived by individuals of the same generation, but it may also condition the biological success and the population structure of future generations.

Given that male harassment is costly to females, females have evolved different mechanisms to counterbalance it (e.g., Arnqvist and Rowe 1995). In cases in which male harassment is not constant (e.g., when it is density-dependent), females first response by adapting their behavior (e.g., see Galicia-Mendoza et al. 2017). Behavioral modifications are frequently the first response that individuals perform to cope with environmental changes because behavior is highly plastic (e.g., Snell-Rood 2013; but see Sih et al. 2004). Females’ behavioral response is determined by its species-specific repertory, that is, the number of responses conditioned by the species skills and limits, but within this repertory, behavior can fluctuate in different dimensions. Formulae and techniques developed over the last decades may allow us to understand the behavioral response of females to male harassment beyond contrasting behavioral means, that is, by quantifying their variability at individual and population level. Differences in interindividual plasticity (differences between individuals in the way they change the level of behavioral performance over time; Dingemanse et al. 2010; Stamps and Biro 2016), intra-individual unpredictability (behavioral variance in relation to an expected pattern; Stamps et al. 2012), and repeatability (variability explained among the individuals considering the variability explained within individuals; Dingemanse et al. 2002; Nakagawa and Schielzeth 2010) are useful to highlight hidden responses that females could develop to confront male persistence. This behavioral variability could be inherited by genetics or epigenetics (Gräff and Mansuy 2008; Jensen 2014; Wang et al. 2017) and thus, conditioning the behavior of future generations.

In this study, we manipulated sex-ratio and population density to observe how the harassment exerted by males produces direct and cross-generational effects in fitness and, for the first time, behavioral variability. By modifying both sex-ratio and density towards an overrepresentation of males, we increased the potential effect of male harassment over females and thus, the likelihood to observe changes across generations. We selected Ischnura graellsii (Odonata: Zygoptera) as model species. This is a suitable organism for our purposes because 1) male harassment increases when sex-ratios are male-biased (Galicia-Mendoza et al. 2017); 2) sex-ratio and population densities influence copulation duration (Cordero-Rivera 1990a); 3) females need to mate only once to fertilize nearly all their eggs (Cordero-Rivera 1990a) and multiple matings are expected to provide females with limited fitness contributions (Takahashi and Watanabe 2010); 4) scramble competition is the only reproductive strategy (Cordero-Rivera 1989) with no courtship involved; and 5) females are able to cryptically select the sperm that will fertilize their eggs (Sánchez-Guillén et al. 2013). Female color polymorphism was interpreted as an evolutionary mechanism allowing I. graellsii females (as well as those from other Ischnurinae) to cope with male harassment (reviewed in Van Gossum et al. 2008). It was assumed that female color polymorphism was maintained in evolution due to the increase of male harassment towards the most common female phenotype in the population, which depends on the abundance, frequency, and density of males and females present in each season or years (reviewed in Galicia-Mendoza et al. 2017). In spite of the extensive study of male harassment in Ischnura, we are not aware of any previous studies trying to disentangle the underlying mechanisms of cross-generational effects that male-biased sex-ratios may have on damselflies.

In the parental generation (mothers), we analyzed fitness by measuring fertility (number of eggs containing embryos), fecundity (total number of eggs, with and without embryos), and fecal deposits as a proxy of resources intake. In the following generation (offspring), we measured larval development and survivorship. We measured behavioral variability in maternal (flying performance) and offspring (locomotion and thigmotaxis) generations taking into account four behavioral axes: 1) behavioral level (mean value), 2) interindividual plasticity (Stamps and Biro 2016), 3) intra-individual unpredictability (Stamps et al. 2012), and 4) repeatability (Nakagawa and Schielzeth 2010).

In a previous work using the same species in a similar setup, Galicia-Mendoza et al. (2017) observed an increase of male harassment in a male-biased treatment. In our case, the goal was to analyze if the female behavioral modification produced by male harassment could lead to long-term consequences. We first exposed females to males (Galicia-Mendoza et al. 2017), and then we observed their behavior when males were not present. We used mating duration, which is also dependent on male density and frequency (Cordero-Rivera 1990a; Galicia-Mendoza et al. 2017), as a proxy of male harassment. We expected females having lower direct fitness in male-biased scenarios due to the energy expenditure involved in avoiding male harassment and the lower opportunity to feed in presence of many males and thus, a decrease in their fertility and fecundity. We expected that females exposed to male-biased treatment would reduce their locomotion activity as a response towards risk (Schaffner and Anholt 1998) but also as a product of fat loss under stress (Stoks and Córdoba-Aguilar 2012). We predicted females showing low differences in interindividual behavioral plasticity if they similarly respond to male harassment as a specific-species way to cope with stress (for instance, individuals of the same species tend to develop similar anti-predator responses; e.g., Salvanes 2017; but see Hossie et al. 2017), but high differences if they adopt different strategies (for instance, by intensifying or modifying their personality traits after previous experiences; Frost et al. 2007). In any case, we expected differences between treatments in the way individuals differ in behavioral plasticity. We predicted females exposed to male-biased treatment showing higher behavioral unpredictability than those exposed to balanced treatments due to the alteration of their behavioral pattern (considering male harassment as a stressor; Köhler et al. 2011) or as a strategy to avoid males to learn their behavioral patterns (e.g., Fincke 2004). We predicted females exposed to male-biased treatment showing lower behavioral repeatability due to the high interindividual differences in plasticity and/or high unpredictability, that is, high within individual but low among individual behavioral variance. We expected that offspring belonging to females exposed to male-biased treatment would show lower survivorship and longer developmental times, as a consequence of poor investment by their mothers. We also predicted that this lower investment by females exposed to male-biased treatment would lead to offspring that would have lower locomotion, higher thigmotaxis, and higher time to perform the first food intake, but also higher differences in behavioral plasticity and higher behavioral unpredictability due to the maternal stress or different investment in eggs, and thus, lower repeatability. However, we expected offspring showing a similar behavioral variability to their mothers as an evolutionary response towards the adaptation to the predicted environments after assessing current population structure (Charnov 1983; Holliday 1987) or if these behavioral traits could be inherited by epigenetics (Gräff et al. 2008; Jensen 2014; Wang et al. 2017; but see Herman et al. 2014).

Materials and Methods

Adult collection and maintenance

Adult I. graellsii (30 females and 110 males) were collected using an entomological net at As Cachadas ponds (42°27ʹ13.2″N 8°50ʹ57.0″W, Pontevedra, Galiza, Spain) in July 2019. To avoid cofounder effects of female color phenotypes, we selected only aurantiaca females. This is one of the three female phenotypes that are found in I. graellsii, and one of the two showing different coloration than males. We chose this phenotype because, in a previous experiment, males spent more time mating with aurantiaca females than with any of the other female phenotypes (Galicia-Mendoza et al. 2017), and thus, we expected higher effects of mating duration. All collected females had the coloration typical of virgin females (orange), whereas all males were mature also according to the coloration pattern described in Cordero-Rivera (1987).

Adult treatments

Females were placed in pairs inside wooden insectaries (Sanmartín-Villar and Cordero-Rivera 2016) with Drosophila melanogaster added ad libitum, which avoided resource competition among the experimental individuals. Prior to the experiments, females spent 24 h alone in the insectaries for acclimation to the experimental setup and to achieve sexual maturation (brown thorax; Cordero-Rivera 1990b). The following day, males were added to each of the experimental insectaries containing two females. Two treatments were performed, based on Galicia-Mendoza et al. (2017): in the balanced treatment, we added two males to each insectary (1:1 male:female ratio; N = 16 females; N = 8 insectaries), whereas in the male-biased treatment, six males were added to each insectary (3:1 male:female ratio; N = 14 females, N = 7 insectaries). We modified sexual frequency and density at the same time to increase male competition and harassment probability. The control treatment included balanced sex-ratio and the same density among sexes, because a control treatment without males was inviable given the need to obtain offspring to analyze cross-generational effects in low harassed females. Balanced frequency and density mimic the natural population status at hatching and emergence in this species (Cordero-Rivera 1990b). Females were placed inside the insectaries between 10:00 and 18:00 h; and males between 12:00 and 18:00 h every day, during three consecutive days (see Figure 1). Every time a copula was observed, mating pairs were transferred to a separate plastic container after 10 min to avoid uncontrolled interactions with other males. Mating duration was recorded to the nearest minute. After a female had mated, individuals were constantly controlled by the observer to avoid double matings. After both females of each insectary had mated, the males in the insectary were substituted by males with non-functional aedeagus (i.e., blocked by adding a drop of glue or paint over it), which prevented genital assembling but not male harassment and tandem formation (see Galicia-Mendoza et al. 2017). In this way, differences between treatments were due to intersexual frequency-density instead to differences between the number of mating events. Note that with the replacement of functional males by non-functional males, we increased the differences between treatments in the number of new males exposed to females (two functional and two non-functional males in balanced treatment but 6 functional and 6 non-functional males in male-biased treatment). Males with functional and non-functional aedeagus were reused over different days and insectaries. Females only had one opportunity to mate per day, and not all females mated every day (see Table A1).

Figure 1.

Scheme for the adult females’ experiment. Experimental days in the upper left corner of each row. Schedule appears below each event. Field collection: females capture on their native habitat. Oviposition: females inside plastic containers with wet filter paper. Male exposures: addition of males into the insectaries. Behavioral observation: recording flight performance. Blank spaces after field collection: females into insectaries grouped by pairs.

Adult flight performance

One daily observation on female flight performance for each female and in each treatment was done for three consecutive days for females that were exposed to males the day before (see Figure 1), that is, female behavior was observed after 1 day of male interactions but not in the presence of males. We use this procedure as a proxy of long-term consequences of male interactions. These observations were carried out daily between 10:00 and 11:30 h. During this time, we performed a series of 10-min focal observations in which we recorded the duration in seconds of each female flight.

Adult fertility and feces production

Females were placed daily (N = 6 days; see Figure 1) inside individual plastic cups with a wet filter paper as a substrate for oviposition, to allow them to lay eggs. After the first day (field collection), females remained inside the oviposition chambers from 18:00 h until 10:00 h of the next day (N = 4 days). None of the females laid eggs in the first two ovipositions before being exposed to males, which further confirmed female virginity. After 3 days of exposure to males, females were placed inside the oviposition chambers from 14:00 h until their death (mean = 3.81 ± 0.19 days).

Eggs were maintained in a temperature and light controlled chamber (25 °C, 12:12 h photoperiod). After hatching, female fecundity was estimated by counting the total number of eggs laid by each female under a binocular microscope. We considered as fertile all those eggs that hatched or that contained a visible embryo, because the filter paper might affect the ability of larvae to hatch (Fincke 1984). Therefore, all empty, non-hatched eggs were considered to be non-fertile (see also Lorenzo-Carballa et al. 2016; Sanmartín-Villar and Cordero-Rivera 2016).

We counted the fecal deposits that females produced every day inside the oviposition chambers. We assumed that females eating fewer flies inside the insectaries would produce less feces inside the oviposition chambers (see the high correlation between fecal deposits and food intake in Berenbaum 1978). We did not consider the volume or weight of feces. Female body length was measured with a caliper after individuals’ death.

Larval maintenance

Just after hatching, larvae from the last oviposition of each female were individually placed in plastic seedbeds with cells of 4 × 4 × 4 cm (Sanmartín-Villar and Cordero-Rivera 2016). Larvae were kept in the laboratory at room temperature and fed daily with Artemia nauplii ad libitum. A total of 4,376 larvae from 23 females were individually isolated. One female of each treatment did not lay eggs, and two laid 7 and 10 eggs, again one of each treatment (see Table A1). We considered these females as anomalous compared with the others, so they were not considered for the statistical analyses. Due to space limitations, a set of larvae was reared in individual plastic cups (N = 1,338), but mortality was extreme. This produced the exclusion of larvae belonging to three mothers (one exposed to balanced treatment and two exposed to male-biased treatment). Therefore, the experimental offspring (N = 3,038 isolated larvae) belonged to 10 mothers exposed to male-biased treatment and 13 mothers exposed to balanced treatment.

Larval developmental time was measured as the number of days between egg hatching and final larval instar for the 342 larvae that reached the final larval instar. To calculate larval survivorship, we used the number of larvae that reached the last instar per treatment and sex. Larvae were sexed (by the presence of the future ovipositor in females) after testing them.

Larval exploration and thigmotaxis

Final instar larvae were daily placed in Petri dishes half-filled with water (Ø = 11 cm), and their locomotion was recorded using Canon and Sony handycams. Observations started at 15:00 h, lasted 10 min, and were performed during five consecutive days. Only individuals tested in five trials were considered for the analyses (N = 293 larvae observed, 49 excluded because they died before completing the tests). A maximum of twenty larvae per family were tested. We recorded the frequency of movement (percentage of time moving) as a proxy of exploration (locomotor activity under non-risky and novel environments; e.g., Réale et al. 2007). We assumed that the testing arenas continued to be new for larvae across the tests due to the short-term memory of the ischnurids (males do not preserve their mating preferences after 48 h; Van Gossum et al. 2001). We also measured the percentage of time that individuals stayed close to the border arena (thigmotaxis). This behavior was differently interpreted by previous authors as a proxy of shyness, stress, or as a product of experimental conditions (Hänzi and Straka 2018); or related with cognitive process linked to problem solving and learning (Doria et al. 2019). When swimming, the larvae tend to contact the walls of the Petri dish and continue their movement along the edge, in circles. We interpret larvae use the inner part of the experimental arena as a decision related to exploration (new environment) and boldness (larvae may feel protected by the walls of the dish or may seek exit through them). The edge of the arena was considered as 1.3 cm, which was assumed to be the maximum distance between the legs on both sides when the larvae were walking. We extracted the position of the individuals at a frame rate of 12.5. Movement was considered when individuals have moved more than 1 mm. We considered 0.06 s as the minimum time to consider individuals have stopped. Movement was tracked using Swistrack 4.0 (Correll et al. 2006). We use the terms “balanced treatment” and “male-biased treatment” to refer to the offspring of females exposed to the two treatments.

Larval feeding latency

After the first 10 min of the first test for each larva, we added 300 µL of water containing nauplii of Artemia to the Petri dish in distanced drops and we recorded the time needed to catch the first prey (feeding latency). Due to the quality of certain videos, we fail to identify the predatory behaviour in 50 larvae, and thus, they were excluded from the analysis (N = 243; balanced = 123; male-biased = 120).

Statistics

Statistical analyses were performed using R version 3.5.3 (R Core Team 2022) and Genstat 20th edition. Values are given as mean ± SE. The fecundity and fecal deposits of females that refused to mate once or twice over the 3 days of exposure to males (see Results) were compared to a similar, randomly selected number of females that mated in all three exposures using a t-test with 1,000 randomizations. Behavior was analyzed under four approaches: differences between means, interindividual differences in plasticity, intraindividual differences in unpredictability, and differences in repeatability (see above).

We ran linear models (LM) and linear mixed models (LMMs; function lmer in lme4 package; Bates et al. 2014) to compare the fitness traits and behavioral means. For adult females (mothers exposed to treatments), we ran a LM using fecundity as a response variable, body length as a covariable, and the experimental treatment as a fixed factor. We ran a LMM using the time spent flying by females during the 10 min of observation after 24 h of being exposed to males as a response variable, body length as a covariable, the treatments (balanced or male-biased) and repeated observations (three tests) as fixed factors, and the mother ID (repeated measures) and insectaries (shared by pairs of females) as random factors.

To analyze the larval behavior, we ran a LMM using behaviors performed during the first 10 min of test (exploration or thigmotaxis) as a response variable, larval development time as a covariable, the treatments experimented by their mothers (balanced or male-biased), larval sex, the interaction between treatment and larval sex, and the repeated observations (five tests) as fixed factors, and the larval ID (repeated measures) and mother ID as random factors.

Interindividual differences in plasticity within the experimental treatments was calculated by comparing models with and without the repeated tests as random slope and the individual ID as random intercept. Models were selected using AIC to unravel if the variability in the data was better explained considering or not considering repeated tests as random slope, that is, if individuals performed different behavioral patterns through the time or not (see also Sanmartín-Villar et al. 2021; Sanmartín-Villar and Jeanson 2021).

Unpredictability for flying performance (adults), exploration, or thigmotaxis (larvae) was calculated by the residual individual standard deviation (riSD), which “estimates the amount of variation around the expected values at every time step instead of estimating the amount of variation around the average of all observations” as do the standard deviation (Stamps et al. 2012), using the formula:

$${\displaystyle \begin{array}{l}{\displaystyle riSD=\sqrt{\frac{{(O_{ij}-E_{ij})}^2}{N_i-1}},}\end{array}}$$

where O_ij represents observation j for the ith individual, Eij the mean expected value produced by a regression model considering behavior as a variable and observations as a fixed factor, and Ni the number of observations for individual i. Unpredictability for adult flights was used as a response variable in a LMM with body length as a covariable and treatment as a fixed factor. Unpredictability for offspring exploration and thigmotaxis were used as response variables in LMMs using sex and treatment as fixed factors and mother ID as a random factor.

The repeatability for flying performance (adults) and exploration or thigmotaxis (larvae) was compared between treatments in adults and larvae using the intraclass correlation coefficient (ICC) which “quantifies the proportion of variance explained by a grouping (random) factor in multilevel/hierarchical data” (larval individual ID in our case; Nakagawa et al. 2017). To calculate it, we used rptGaussian function (package rptR; Stoffel et al. 2017). Values higher than zero mean lack of randomness and the existence of repeatability.

Larval development time and feeding latency were analyzed as survival curves using the Kaplan–Meier method (package survival; Therneau 2021). Cases in which larvae did not feed during the 10-min test received the maximum score of time (600 s) and were considered as censored data allowing the model to consider that the individual did not feed when the test was over.

Fertility was compared between treatments using a GLM with binomial errors using the total number of eggs laid by each female as binomials totals and body length, total mating duration, and total number of faecal deposits as covariables. The relative larval survivorship for the 293 larvae that reached the final instar was also analyzed using a GLM with binomial errors using the number of initial larvae isolated as binomials totals, also with overdispersion correction, as implemented in GenStat.

Variables that did not follow the assumptions for LMs (data following Normal distribution, homoscedasticity, and normality of residual errors) were transformed using transformTukey function from the rcompanion package (Mangiafico 2021). The best explanatory model was selected according to the AIC (see Table 1 and Table A2). LMM pairwise comparisons (post hoc analyses) were performed using emmeans function (Lenth 2021). For each model, we calculated the marginal (R²m; proportion of variance explained by the fixed factors) and the conditional coefficient of determination (R2c; the proportion of variance explained by both fixed and random factors; Nakagawa et al. 2017) using r.squaredGLMM function (package MuMIn; Barton 2009).

Table 1.

Linear mixed models selected according to AIC (see Table A2)

Analyses	Models	R2m	R2c
Mother flight time	Flight time ~ Body length + Treatment + (1 | Mother ID)	0.02	0.12
Mother unpredictability	Flight time riSD ~ Body length + Treatment + (1 | Mother ID)	0.19	0.19
Larval development	Larval development ~ Sex × Treatment + (1 | Mother ID)	0.01	0.48
Larval exploration	Movement frequency ~ Tests + Larval development + Treatment × Sex + (1 | Mother ID) + (1 + Tests | Larval ID)	0.03	0.44
Larval thigmotaxis	Border use ~ Larval development + Treatment × Sex + (1 | Mother ID) + (1 + Tests | Larval ID)	0.01	0.16
Larval exploration unpredictability	Movement frequency riSD ~ Larval development × Treatment × Sex	0.02	0.02
Larval thigmotaxis unpredictability	Border use riSD ~ Larval development + Treatment × Sex	0.02	0.02

Models are written following R language: response variable ~ fixed factors + (random slopes | random intercept). R²m (marginal coefficient of determination): proportion of variance explained by the fixed factors. R²c (conditional coefficient of determination): proportion of variance explained by the random factors.

Results

Adult flight performance

Females exposed to the two experimental treatments spent similar time flying (male-biased = 27.12 ± 2.90 s, balanced = 21.75 ± 2.25 s; F_1,27 = 0.215, P = 0.154). Females exposed to male-biased treatment showed interindividual differences in the time spent flying across the three tests (different plasticity), whereas no differences were found for the females exposed to the balanced treatment (Table 2; Figure 2A,B). No significant differences were found for unpredictability between treatments (F_1,26 = 0.00, P = 0.961) nor for the interaction between Treatment × Body length (F_1,27 = 0.07, P = 0.197). Females exposed to male-biased treatment showed zero behavioral repeatability in flight time (ICC = 0.00 ± 0.10, CI [0.00, 0.31]), whereas females exposed to balanced treatment showed a strong repeatability (ICC = 0.42 ± 0.16, CI [0.08,0.67]).

Table 2.

Interindividual plasticity analyses comparing models with and without tests as random slopes, that is, with different or homogeneous between individuals’ plasticity

Analyses	Treatment	Models	Interindividual variability	AIC	R2m	R2c
Mother flight time	Balanced	Flight time ~ Tests + (Tests | Mother ID) Flight time ~ Tests + (1 | Mother ID)	Homogeneous	86.30 82.58	0.00 0.00	0.45 0.44
	Male-biased	Flight time ~ Tests + (Tests | Mother ID) Flight time ~ Tests + (1 | Mother ID)	Homogeneous	77.97 82.23	0.02 0.01	0.48 0.01
Larval exploration	Balanced	Movement ~ Larval development + Tests × Sex + (1 | Mother ID) + (1 + Tests | Larval ID) Movement ~ Larval development + Tests × Sex + (1 | Mother ID) + (1 | Larval ID)	Homogeneous	−785.58 −786.30	0.02 0.02	0.44 0.41
	Male-biased	Movement ~ Larval development + Tests × Sex + (1 | Mother ID) + (1+ Tests | Larval ID) Movement ~ Larval development + Tests × Sex + (1 | Mother ID) + (1 | Larval ID)	Different	−711.16 −701.36	0.04 0.06	0.45 0.42
Larval thigmotaxis	Balanced	Border use ~ Larval development × Sex + (1 | Mother ID) + (Tests | Larval ID) Border use ~ Larval development × Sex + (1 | Mother ID) + (1 | Larval ID)	Homogeneous	331.86 327.86	0.02 0.02	0.19 0.19
	Male-biased	Border use ~ Larval development × Sex + (1 | Mother ID) + (Tests | Larval ID) Border use ~ Larval development × Sex + (1 | Mother ID) + (1 | Larval ID)	Homogeneous	365.59 361.84	0.01 0.01	0.13 0.13

The most explanatory model was selected according to the parsimony principle and when AIC differences between models were >2 (in bold) following the classic convention (but see Arnold 2010). In bold, selected models (lower AIC). Models are written following R language: response variable ~ fixed factors + (random slopes | random intercept) R²m (marginal coefficient of determination): proportion of variance explained by the fixed factors. R²c (conditional coefficient of determination): proportion of variance explained by the random factors.

Figure 2.

Plasticity interindividual differences in adult flight performance (A and B) and larval exploration (C and D). Lines represent modeled individuals’ regressions for the time flying (adults) or movement frequency (larvae) across tests. Lines slopes differed for mothers and offspring from male-biased treatment (A and C; tests selected as random slopes) while did not differ for mothers and offspring from balanced treatment (B and D; tests were not selected as random slopes).

Mother reproduction

Nine of the 30 experimental females refused to mate on 1 (N_male-biased = 5; N_balanced = 3) or 2 days (N_balanced = 1) over the 3 days of male exposures. Fecundity (total number of eggs laid) and the number of feces produced by these females were similar to the remaining 21 females that mated three times (P > 0.05; in 100% of cases of the analysis performed with 1,000 randomizations).

Females of male-biased treatment mated on average 2.64 h longer in overall (male-biased: 9.55 ± 1.00 h; balanced: 6.91 ± 0.70 h, F_1;28 = 4.86, P = 0.036; Figure 3A) and tended to produce less fecal deposits than females exposed to the balanced treatment (male-biased: 48.86 ± 3.65 feces; balanced: 56.81 ± 1.98, F_1;28 = 3.93, P = 0.057; Figure 3B). Females exposed to both treatments showed similar fecundity (male-biased: 340.83 ± 26.37 eggs, N = 14; balanced: 379.93 ± 34.60, N = 12; F_1,24 = 0.76, P = 0.390). However, females exposed to male-biased treatment showed lower fertility than those exposed to balanced treatment (male-biased: 91.58% ± 0.05; balanced: 97.36% ± 0.01; deviance= 259.22, df = 24, P = 0.037). The total mating time and total number of faecal deposits had no significant effect on fertility (deviance = 1.63, df = 23, P = 0.869; deviance = 4.49, P = 0.784, respectively). Overall, fecundity (r = 0.09, P = 0.629) and fertility (deviance = 50.43, P = 0.358) were not affected by female body size.

Figure 3.

Differences between experimental treatments (male-biased and balanced) for the time spent mating (A) and the total number of faeces (B). The horizontal line in each box represents the median, and the lower and upper hinges indicate the first and third quartiles. Lower and higher whiskers extend to the most extreme values within 1.5 interquartile ranges from the first and third quartiles, respectively.

Time spent mating reduced the production of faeces of females exposed to both treatments (r = −0.46, P = 0.015; Figure 4A), whereas it highly increased the fecundity of those females exposed to balanced treatment (balanced: r = 0.56, P = 0.038; male-biased: r = 0.17, P = 0.593; Figure 4B). No significant correlations were found between the number of faecal deposits and fecundity (balanced: r = −0.05, P = 0.860; male-biased: r = 0.12, P = 0.713; Figure 4C) or fertility (balanced: r = −0.05, P = 0.843; male-biased: r = 0.16, P = 0.596) or between faeces production and female body length (balanced: r = 0.23, P = 0.386; male-biased: r = −0.11, P = 0.709).

Figure 4.

Correlations between the time spent mating and the total number of feces produced by the experimental females of both treatments (male-biased and balanced) (A), between the time spent mating and fecundity (B), and between the total number of feces produced and fecundity (C).

Larval development

Females exposed to male-biased and balanced treatment produced an overall similar offspring sex-ratio (male:female; balanced: 1.20, N = 174; male-biased: 1.31, N = 164; χ²₁= 0.08, P = 0.779). The percentage of offspring that reached the last instar was low and similar between treatments (mean ± SE: balanced= 11.15% ± 1.05; male-biased: 16.77% ± 3.35; Wilcoxon test: W = 89, P = 0.148; Levene test: F = 2.99, df = 21, P = 0.098).

Larval exploration and thigmotaxis

A total of 293 larvae were tested five times (62 daughters and 82 sons of females of male-biased treatment and 67 daughters and 82 sons of females of balanced treatment). Exploration was similar between sexes in the balanced treatment (males: 32.78 ± 0.81%; females: 34.96 ± 7.71%; estimate difference: 0.03 ± 0.02, df = 277.5, P = 0.230), whereas female larvae from the male-biased treatment explored less than males (males: 35.49 ± 0.73%; females: 30.35 ± 0.96%; estimate difference: 0.05 ± 0.02, df = 280.9, P = 0.040, Treatment × Sex: F_1;280.41 = 10.69, P = 0.001, Figure 5).

Figure 5.

Larval behavior between sexes and mother treatments. (A) Larval exploration (transformed movement frequency). (B) Larval thigmotaxis unpredictability. The horizontal line in each box represents the median, and the lower and upper hinges indicate the first and third quartiles. Lower and higher whiskers extend to the most extreme values within 1.5 interquartile ranges from the first and third quartiles, respectively.

No differences between treatments were found for larval thigmotaxis (balanced = 56.19 ± 1.11%; male-biased: 53.85 ± 1.14%; P = 0.219). Males were slightly less likely to follow the walls than females (males: 53.49 ± 1.04% inner arena use; females: 57.01 ± 1.22%; F_1;281.191 = 4.60, P = 0.033).

Larval feeding latency

Larval feeding latency did not differ between mother treatments (balanced = 269.37 ± 18.66 s; male-biased: 267.16 ± 18.95 s; χ²₃ = 1.3, P = 0.700).

Larval behavioral plasticity

Offspring from females of male-biased treatment showed interindividual differences in their exploration pattern across the five tests (different plasticity), whereas no differences were found for the offspring of the balanced treatment (Table 2; Figure 2C,D). No interindividual differences in plasticity were found for thigmotaxis in larvae from either treatment (see Table 2).

Larval behavioral unpredictability

Larvae showed low unpredictability for exploration (riSD = 0.09 ± 0.05 estimated variance) and thigmotaxis (0.18 ± 0.09). No differences were found between treatments for exploration (Table 1; Table A2). Thigmotaxis unpredictability slightly differed in the interaction between Treatment × Sex (F_1;288 = 4.00, P = 0.046, Figure 5). Daughters from females exposed to male-biased treatment showed higher unpredictability in their thigmotaxis than those from females exposed to balanced treatment. However, these differences were not significant after applying Tukey correction in post hoc analyses (0.04 ± 0.02 estimated variance, t = 2.18, no correction: P = 0.030; Tukey correction: P = 0.132).

Larval behavioral repeatability

Larvae showed high values of repeatability in their exploratory behavior across the five tests performed (0.39 ± 0.03, CI [0.33, 0.44], P <0.001) and lower values for thigmotaxis (0.15 ± 0.02, CI [0.10, 0.20], P < 0.001), but no differences were found between treatments for these variables (overlapping confidence intervals; activity _male-biased= 0.38 ± 0.04, CI [0.29, 0.46]; activity _balanced= 0.40 ± 0.04, CI [0.32, 0.47]; thigmotaxis _male-biased= 0.12 ± 0.04 CI [0.05, 0.19]; thigmotaxis _balanced= 0.18 ± 0.04, CI [0.11, 0.25]).

Discussion

We found an increase of mating time and a reduction in fertility (but not in fecundity) in male-biased scenarios as well as modifications of both female and offspring behavioral variability in Ischnura graellsii. The increase of interindividual differences in plasticity was only observed in females exposed to male-biased treatments and in their offspring. Females exposed to male-biased treatment produced females that showed less exploration and greater unpredictability. Females from the male-biased treatment trended to defecate less and had lower fertility, suggesting that they were resource-limited. Some of the cross-generational changes in behavioral variability could be interpreted as a product of an adaptive response to counteract male harassment.

Female flight performance (mean values) was not affected 1 day after being exposed to males in the two treatments here performed. Similar results were found in previous studies where female flight performance was measured when males were present (Galicia-Mendoza et al. 2017). Females from both treatments showed similar flight unpredictability. However, females exposed to balanced treatments showed moderate-to-high repeatability in flight performance (ICC = 0.42 ± 0.16; see Martin et al. 2019), whereas females exposed to male-biased treatments showed no repeatability (ICC = 0.00 ± 0.10). This suggests that the variability explained within females was higher than the variability explained among females in those exposed to male-biased treatments (Nakagawa and Schielzeth 2010), which is supported by the interindividual differences in plasticity only found in these females. As behavioral repeatability decreases after exposure to stressors (for instance, in the red flour beetle; Wexler and Scharf 2017) and male harassment function as a stressor for females (Sapolsky 2004), we conclude that our results point to male harassment as a stressor capable of increasing female behavioral variability to values above population levels. The ecological consequences of our findings could differ according to the scale at which they are analyzed. Although within-individual variability increased in females exposed to male-biased treatment, suggesting the plasticity of their behavioral response towards environmental conditions, the among-individual variability was reduced, which suggest a standardization of population behavioral response. Future studies might focus on the reduction of behavioral diversity according to the population structure to disentangle if the overrepresentation of males could lead to a loss of intraspecific ethodiversity (see Cordero-Rivera 2017).

Females exposed to male-biased treatments remained in mating for longer, tended to produce less faecal deposits, and showed a reduction of their fertility. The increase of mating duration in male-biased scenarios corroborates results from previous studies performed with the same species and similar methods (Cordero-Rivera 1990a; Galicia-Mendoza et al. 2017). Increased mating duration is a mechanism by which males, who control mating duration in this order of insects (Andrés and Cordero-Rivera 2000), retain females under highly competitive conditions. Prolonged mating in I. graellsii does not increase sperm transference or removal, but increase female guarding (Cordero-Rivera 1990a). Female guarding also increases under male-biased sex-ratio in other taxa and reduces the probability of males removing sperm in posterior matings with the same female (reviewed in Jirotkul 1999). Frequency dependency of female guarding involves males modifying the time spent mating after assessing population structure, that is, increasing guarding when more competitors are perceived. In our study, mating duration differed between treatments even if pairs were isolated after 10 min of the start of copulation (note that copulations may last for hours), suggesting that males adapt their reproductive behaviour according to the context perceived before or at the beginning of the copula. In the field, it is common that males in copula escape flying and hide to decrease the harassment from other males (Sanmartín-Villar I, personal communication), and thus, males might experience similar situations as we observed in the laboratory.

In contrast to our predictions and to what happened to females exposed to male-biased treatments, mating duration was positively correlated with egg production in the balanced treatment. Females could increase their egg oviposition after assessing the male ability to compensate for costs (“handicap” hypothesis) or his heritable traits (“sexy sons” hypothesis) (Eberhard 1996). In fact, increasing the time invested in mating might also produce costs for males by losing time to feed and opportunities to mate with other females. Female guarding might constitute an investment in reproduction for males, but also could be interpreted as courtship during mating. According to this, males mating for longer time had higher fertilization success in the damselfly Ceriagrion tenellum (Andrés and Cordero-Rivera 2000). However, the reduction of fertility and production of faecal deposits suggest that female guarding might become a cost for females exposed to male-biased sex-ratios, in which the optimal guarding time might be overpassed (Elias et al. 2014), and thus, no correlation between mating duration and fecundity was observed for the females exposed to this treatment. Other effects caused by the male-biased treatment could also be responsible for the decrease in egg production. For instance, females suffer a higher number of tandem attempts by males under male-biased sex-ratios (Galicia-Mendoza et al. 2017). In such situation, females must reject males more frequently, probably losing energy and feeding opportunities and thus, obtaining less resources to invest in reproduction.

As discussed above, in the male-biased treatment, females remained longer in mating, diminishing their time to feed (negative correlation between mating duration and number of faecal deposits in our study) or suffered higher male harassment (Galicia-Mendoza et al. 2017), but they produced a similar number of eggs as females exposed to balanced treatments. Females might modify their investment in quality instead in quantity, providing different resources to their eggs instead of reducing their number. Thus, the lower fertility observed in females exposed to male-biased treatment could be an example of such different investment. Differences in female investment could explain the cross-generational differences that we observed. However, more studies are needed to disentangle if the differences in fertility could also depend on the quality of sperm (Pike et al. 2010; Walczak-Jedrzejowska et al. 2013) and if this quality depends on the competitors’ frequency (see Carrillo et al. 2012; Firman et al. 2020): males may invest less in their sperm in a male-biased population due to the high probability that other males will remove sperm from the females with whom they have copulated.

Females exposed to certain environmental conditions—social context included—may modify the offspring sex-ratio by hormonal mediation (Trivers and Willard 1973; reviewed in Navara 2013). Therefore, and knowing the different resource intake (indirectly measured as the number of faecal deposits) and investment (fertility) of females exposed to our experimental treatments, we expected different sex-ratio and larval development in the offspring of the two treatments. However, females exposed to both treatments produced offspring in a balanced sex-ratio. Balanced sex-ratio occurs in field conditions for most odonates (Cordero-Rivera and Stoks 2008) and in captivity (I. graellsii, Cordero-Rivera 1990b). In addition, we did not observe different development between larvae from mothers exposed to both treatments. Both results suggest that larval sex and development are canalized by genetics and evolutionary pressures rather than environmental conditions (at least for the male overrepresentation perceived in a single generation). It must be noted that we sexed the offspring in their final instar when the external genitalia are visible, so any differences in sex-ratio that could be present at hatching time would go unnoticed. A parallel explanation for the similarity of larval development even when the mothers possessed different resources could be that the mothers invested similar resources only in fertile eggs.

Offspring belonging to females exposed to the two experimental treatments showed similar mean values for thigmotaxis and feeding latency, while differed in the other traits measured. Mothers exposed to male-biased treatments produced individuals more differentiated in exploration plasticity, their daughters showed lower exploration than their sons and lower thigmotaxis unpredictability than females from balanced treatments. Considering the costs of production and maintenance of plasticity (DeWitt et al. 1998), interindividual differences in offspring behavioral plasticity could be caused by resources limitation in the previous generation, that is, mother differently investing resources in the offspring, and thus, some larvae did not have enough resources to develop as their siblings, but also by resource allocation, that is, mothers allocate resources differently to increase offspring plasticity differences as a mechanism to increase behavioral variability to cope with the environmental condition perceived. However, the behavioral variability found in offspring could be a consequence of an evolutionary mechanism increasing the likelihood of their adaptation (e.g., bet hedging; e.g., Hopper 1999; Simons 2011). Under a male-biased scenario, differences in behavioral plasticity and high values of behavioral unpredictability may increase the ability of females and males to avoid harasser males (see Richardson et al. 2018). Future studies might focus on unravelling the underlying mechanisms that drive cross-generational behavioral variability in this species.

Differences in offspring according to maternal treatment varied between the sexes. Males showed lower thigmotaxis than females. This could be related to a higher risk taking (considering that larvae perceived more dangerous the inner part of the experimental arena as happens in other taxa; Valle 1970; Walsh and Cummins 1976; Sneddon 2003; Carlson and Langkilde 2013; Detrain et al. 2019), but also to other traits as higher variability in the angle of their movement or learning abilities (Sneddon 2003; Doria et al. 2019). Daughters of mothers exposed to male-biased treatments showed lower mean values for exploration than their brothers, whereas these daughters showed higher unpredictability for thigmotaxis than daughters belonging to mothers exposed to balanced treatment. Daughters showing low exploration (or locomotion) and increased thigmotaxis unpredictability might be more adapted to risky environments as could be the male-biased populations. Lower exploration might diminish the encounters with males, whereas higher thigmotaxis unpredictability could allow females to increase the variability of their behavioral patterns towards risky conditions (assuming its relationship with cognitive abilities; Doria et al. 2019) and thus develop different patterns against male encounters.

Larvae showed high repeatability in exploration but low repeatability in thigmotaxis as well as low unpredictability for both traits. This indicates that I. graellsii larvae 1) differed in their exploratory behavior at the individual level, 2) most of them tended to move close to the wall, and 3) the pattern of both traits was constant (predictable) across the time. This suggests that each I. graellsii larva possesses its own pattern of exploration and thus different patterns within the same population. The preference for using the arena border could depend on individuals searching for an orientation reference (thigmotaxis), the security that guaranties being close to the wall (considering thigmotaxis as a synonym of shyness; Valle 1970; Walsh and Cummins 1976; Sneddon 2003; Carlson and Langkilde 2013; Detrain et al. 2019) or because they were looking to escape from the arena. The high repeatability and low unpredictability observed in I. graellsii larvae suggest this species as a good model to study consistent behavioral patterns and personality traits (Chenard and Duckworth 2021).

The offspring of females exposed to male-biased treatment modified their behaviour in the same way as their mothers did, suggesting epigenetic inheritance as a possible cause of behavioral changes (Gräff et al. 2008; Jensen 2014; Wang et al. 2017; but see Herman et al. 2014). It must be noted that in this experiment offspring behavior was only analyzed during larval stage. Even if behavioral patterns are consistently maintained through metamorphosis in heterometabolous insects (Amat et al. 2018), more studies are needed to understand the evolutionary function of the observed behaviors and the underlying mechanisms driving cross-generational behavioral modulations.

The fact that sex-ratio and development time of the offspring were not affected by the environment experienced by the mothers, but their behavior was, points to behavior as a more plastic phenotypic trait in contrast to lifespan traits (however, the opposite was found in other experiments working with I. graellsii, Debecker et al. 2016; or ants, Sanmartín-Villar and Jeanson 2021). Our results suggest that male-biased ratios modify reproductive strategies with potential consequences in population dynamics as well as highlight the relevance of considering personality traits when studying reproduction costs across generations. We were able to detect the effect of biased sex-ratios by using different approaches of behavioral analyses, but other behaviors and conditions should be considered to better understand the effect of sex-biased populations in the field. In addition, more studies should be performed to unravel separately the effects of sex-ratio and density and to disentangle the role of paternal effects in cross-generation behavioral variability (Braun 2014).

Appendix

Table A1.

Matings occurred during the 3 days of treatment and number of larvae from the last oviposition individualized after hatching in days 14 to 17 of the experiment

Mother ID ID	Treatment	Male ID			N larvae (day of eclosion)
		Day 2	Day 3	Day 4	Day 14	Day 15	Day 16	Day 17	Total
1	Balanced	19	37	49	—	93	44	—	137
2	Balanced	20	36	25	46	34	14	—	94
3	Male-biased	50	44	20	199	72	—	—	271
4	Male-biased	21	19	36	254	73	—	—	327
5	Balanced	28	27	35	176	61	—	—	237
6	Balanced	26	28	31	164	144	—	—	308
7	Male-biased	27	25	33	—	—	211	—	211**
8	Male-biased	—	26	32	34	76	91	—	201
9	Balanced	32	33	47	1	115	56	—	172**
10	Balanced	—	49	19	—	26	272	—	298
11	Male-biased	—	31	34	—	—	—	—	10*
12	Male-biased	49	35	27	211	60	—	—	271
13	Balanced	—	41	45	256	117	—	—	373
14	Balanced	37	42	48	—	—	—	—	0
15	Male-biased	41	47	22	—	2	18	24	44**
16	Male-biased	39	45	40	—	24	150	—	174
17	Balanced	44	—	—	11	—	—	—	11*
18	Balanced	—	40	42	240	84	—	—	324
19	Male-biased	42	48	43	—	—	—	—	0
20	Male-biased	26	25	49	34	16	—	—	50
21	Balanced	47	26	65	45	57	—	—	102
22	Balanced	19	42	32	6	18	68	15	107
23	Male-biased	74	40	42	36	66	—	—	102
24	Male-biased	—	65	47	24	76	—	—	100
25	Balanced	—	19	40	100	—	—	—	100
26	Balanced	25	69	34	100	—	—	—	100
27	Male-biased	49	36	70	100	—	—	—	100
28	Male-biased	32	78	19	25	55	29	—	109
29	Balanced	40	32	66	100	—	—	—	100
30	Balanced	78	49	26	100	—	—	—	100

Around 100 larvae were only individualized as maximum from females 51 to 62. Treatment: balanced or male-biased sex-ratio.

*No larvae were used from females 11 and 17, who only laid 10 and 11 eggs.

**Larvae were raised using different methodology and they were not considered in the experiment.

Table A2.

Linear mixed models compared according to AIC

Analyses	Models	AIC	R2m	R2c
Mother flight time	Flight time ~ Body length + Tests × Treatment + (1 | Mother ID) + (1 | Insectaries)	173.61	0.05	0.14
	Flight time ~ Body length + Tests × Treatment + (1 | Mother ID)	171.96	0.04	0.13
	Flight time ~ Body length + Tests + Treatment + (1 | Mother ID)	168.23	0.04	0.14
	Flight time ~ Body length + Treatment + (1 | Mother ID)	164.42	0.02	0.12
Mother unpredictability	Flight time riSD ~ Body length + Treatment + (1 | Mother ID)	−2.60	0.19	0.19
Larval development	Larval development ~ Sex × Treatment + (1 | Mother ID)	3162.03	0.01	0.48
	Larval development ~ Sex + Treatment + (1 | Mother ID)	3166.53	0.01	0.48
	Larval development ~ Treatment + (1 | Mother ID)	3171.68	0.01	0.48
Larval exploration	Movement frequency ~ Larval development × Treatment + Tests × Treatment × Sex + (1 | Mother ID) + (1 + Tests | Larval ID)	−1483.68	0.05	0.44
	Movement frequency ~ Tests + Larval development × Treatment × Sex + (1 | Mother ID) + (1 + Tests | Larval ID)	−1508.75	0.05	0.44
	Movement frequency ~ Tests + Larval development + Treatment × Sex + (1 | Mother ID) + (1 + Tests | Larval ID)	−1541.81	0.03	0.44
Larval thigmotaxis	Border use ~ Larval development × Treatment + Tests × Treatment × Sex + (1 | Mother ID) + (1 + Tests | Larval ID)	724.42	0.02	0.16
	Border use ~ Larval development × Treatment + Tests × Treatment × Sex + (1 | Mother ID) + (1 | Larval ID)	720.52	0.02	0.16
	Border use ~Tests+ Larval development × Treatment × Sex + (1 | Mother ID) + (1 | Larval ID)	695.78	0.02	0.16
	Border use ~ Larval development × Treatment × Sex + (1 | Mother ID) + (1 + Tests | Larval ID)	685.10	0.02	0.16
	Border use ~ Larval development + Treatment × Sex + (1 | Mother ID) + (1 + Tests | Larval ID)	646.15	0.01	0.16
Larval exploration unpredictability	Movement frequency riSD ~ Larval development × Treatment × Sex + (1 | Mother ID)	−862.80	0.03	0.03
	Movement frequency riSD ~ Larval development × Treatment × Sex	−961.21	0.03	0.03
	Movement frequency riSD ~ Larval development × Treatment × Sex	−963.79	0.02	0.02
Larval thigmotaxis unpredictability	Border use riSD ~ Larval development × Treatment × Sex + (1 | Mother ID)	−420.55	0.03	0.07
	Border use riSD ~ Larval development × Treatment × Sex	−504.58	0.03	0.03
	Border use riSD ~ Larval development + Treatment × Sex	−508.40	0.02	0.02

Factor/interaction deleted in the following model appears in bold. The most explanatory model was selected according to the parsimony principle and when AIC differences between models were >2 (in bold) following the classic convention (but see Arnold 2010). Models are written following R language: response variable ~ fixed factors + (random slopes | random intercept). R²m (marginal coefficient of determination): proportion of variance explained by the fixed factors. R²c (conditional coefficient of determination): proportion of variance explained by the random factors.

In memory of Dalia Ivette Galicia-Mendoza (1973–2020), pioneer and mentor of this work.

Conflict of Interest Statement

We declare no conflict of interests.

Authors’ Contributions

I.S.V. and A.C.R. conceived the ideas and designed methodology; I.S.V. and X.Y. collected the data; I.S.V. and A.C.R. analyzed the data; I.S.V. and A.C.R. led the writing of the manuscript. All authors contributed critically to the drafts and gave final approval for publication.

Data Availability

Data analyzed in this study are included in this published article.