The ability to detach a body part in response to a predation attempt is known as autotomy, and it is perhaps the most intensively studied form of nonlethal injury in animals. Although autotomy enhances survival, it may impose reproductive costs to both males and females. We experimentally investigated how autotomy affects the reproductive success of males and females of a scorpion species. Individuals of Ananteris balzani autotomize the last abdominal segments (“tail”), losing the anus and leading to lifelong constipation since regeneration does not occur. Although male “tail” is used during courtship and sperm transfer, autotomy had no effect on male mating success. The combined effect of increased mortality and reduced fecundity resulted in autotomized females producing nearly 35% less offspring than intact females. In conclusion, the negative effects of “tail” autotomy are clearly sex-dependent, probably because the factors that influence reproductive success in males and females are markedly different.

Experiment I: Male reproductive success

To assess the possible implications of “tail” autotomy on the mating success of males, we first measured the cephalothorax length (a proxy of body size) of all individuals and paired males and females assortaively according to their size rank. As the individuals were collected in different years, the size-assortative pairing was done in blocks containing at least 24 individuals each. Then, we split the males into two experimental groups: intact (n = 40) and autotomized (n = 30). We induced “tail” autotomy in males of the autotomized group by repeatedly tapping their body with forceps for 1 min and then grasping and pushing their last metasomal segment with the same forceps (figure 1A). For intact males, we used the same protocol, but we grasped the first metasomal segment, where “tail” autotomy does not occur (figure 1B). All males were allowed to rest for at least 3 days to complete the healing of the fracture point.

The mating trials were always performed during the first six hours of the dark phase, under dim red illumination, and inside a glass box (base: 20 x 10 cm; height: 15 cm) with wet sand as substrate, and two pieces of flat wooden boards and two pieces of porous tiles to provide a suitable substrate for spermatophore deposition. We placed a male inside the box, and after 30 min of acclimatization, we placed the female. We followed and filmed each male-female sexual interaction with two cameras SONY (HDR–CX405), one placed above the mating pair and another placed laterally.

In some mating trials (27.5% of the total) males and females did not engage in courtship within 1 h after the first physical contact between them. This situation usually occurs because the female is pregnant or has already received the maximum number of inseminations (Peretti and Carrera 2005). Thus, whenever a female did not engage in courtship, we followed her during two months after the mating trial to identify signs of pregnancy (i.e., an increase in abdominal volume with embryos visible through the ventral intersegmental membranes of the mesosoma). Moreover, within 30 days after an unsuccessful mating trial, we tried to re-mate the supposedly unreceptive females with up to two other males (either intact or autotomized). We considered as unreceptive all females that were pregnant or did not accept any other male after the first mating trial. In these cases, we discarded the females from the sample and considered all mating trials accomplished with them as invalid. After removing these females (n = 19; 10 of them were pregnant), we had 28 mating trials with intact males and 23 mating trials with autotomized males.

We estimated male mating success using three measures. (1) Spermatophore deposition success, i.e., whether a male successfully deposited on the substrate his spermatophore, which is a sclerotized, flagelliform structure that is attached to a solid substrate at the end of the courtship (figure 1C). Considering only males that successfully deposited their spermatophore, we estimated (2) sperm transfer success, i.e., whether a male transferred all sperm contained in his spermatophore. Because the walls of the spermatophore are transparent in A. balzani, we used a stereomicroscope to score whether some sperm remained inside the spermatophore (figure 1C). Spermatophores containing any visible amount of remaining sperm were scored as incomplete transferences. Finally, from the videos, we estimated (3) courtship duration of all males that successfully deposited the spermatophore. Courtship duration was the period between pedipalpal grip (i.e., the beginning of the courtship; figures 1A,C) and the moment when both individuals retracted their pedipalps and separated their bodies after sperm transference (see the video in Supplementary Material S1).

Experiment II: Female reproductive success and maternal effects

To assess the possible implications of “tail” autotomy on the reproductive success of females, we used females from the previous experiment that successfully received a spermatophore during courtship (both from intact and autotomized males). We also included 9 additional females that were not used in the previous experiment, but that were paired with intact males following the same protocol described above. Because the females used in our experiments were collected in the field, we have no control on their previous pregnancy history. However, we reared females in the laboratory, from the last nymphal stage to their death, and detected that the tegument coloration of young adult females is lighter than that of older adult females (figure 1A,B). Using tegument coloration, we divided experimental females into two age categories: (a) young (n = 12), comprising 7 virgin females reared in the laboratory since they were nymphs, and 5 widecaught females that were between 1 to 3 months old and were unlikely to have produced a first litter, and (b) old (n = 37), comprising 29 females that produced a first litter in the laboratory before the experiment (when they were at least 8-10 months old), and 8 widecaught females that almost certainly produced a first litter in the field and were at least 7-8 months old when they were used in the experiment. Very old females, i.e., those showing the darkest tegument coloration and damages in the stinger, pedipalps, or pectens, were not included in the experiment. We equally split young and old adult females into two experimental groups: intact (n = 25) and autotomized (n = 24). This procedure ensured that a similar number of females in each experimental group would give birth to the first or second litter.

To induce “tail” autotomy in females we followed the same protocol described in the experiment with males. Contrary to males, however, we autotomized females only after mating because we were interested in the effect of autotomy on offspring development and not on the mating success of females, which are never rejected by the males once the courtship process has begun (Polis and Sissom 1990; see also Peretti and Carrera 2005 and references therein). Moreover, considering that the females used here are the same used in the previous experiment, the lack of the “tail” in females during the courtship process could interfere with their receptivity and consequently with males’ behavior. Thus, we would be unable to disentangle the effects of male autotomy and other variables that could influence the response variable ‘courtship duration’.

After mating, all females were maintained individually in plastic containers with sand as substrate, wet cotton, and a dry leave to provide shelter. We inspected the females at 3 day-intervals during the entire period of experiment to record if they were alive and if they gave birth to the nymphs. We estimated female reproductive success using three measures: (1) female survival until offspring dispersion, i.e., the probability of a female remaining alive until her offspring is independent from maternal care, which occurs 4-5 months after mating and two weeks after offspring birth; (2) offspring number, which is the mean number of nymphs produced by the females (figure 1D); and (3) female fitness, which is the mean number of nymphs produced by the females (including those that died before giving birth) and that survived until dispersion. We stress that if a female dies before giving birth, the offspring number produced is zero. If a female dies after giving birth, but before nymphs disperse from her dorsum, the offspring does not survive, probably because the nymphs, which are overly sensitive to dehydration, need to absorb water directly through the mother’s cuticle (Vannini et al. 1985). The nymphs produced by these females (n = 3) were used to estimate offspring number, but to estimate fitness we considered that they left no offspring because the nymphs did not survive until dispersion.

Finally, we estimated maternal effects using two measures: (1) offspring mass, which is the mean weight of all second instar nymphs of each litter; and (2) offspring size, which is the cephalothorax width of all second instar nymphs of each litter. The size of the nymphs was measured from digital photos using the software ImageJ (Schneider et al. 2012).

Data analysis

We tested for the implications of “tail” autotomy on the reproductive success of males using a different model to each proxy. For all models, the experimental group (intact or autotomized) was the predictor variable. For the response variables spermatophore deposition success (no = 0, yes = 1) and sperm transfer success (incomplete = 0, complete = 1), we adjusted generalized linear models (GLMs) with binomial error distribution and logit link function. For the response variable courtship duration, we adjusted a linear model with Gaussian error distribution, and included an error structure to account for heteroscedasticity using the gls() function of the package nlme (Pinheiro et al. 2019). Given that the variation in mating duration was markedly different between intact and autotomized males, we performed a Levene test using the package car (Fox and Weisberg 2019) to compare the variance of the two experimental groups.

To test for the implications of “tail” autotomy on the reproductive success of females, we used a different model for each proxy. In all models the predictor variable was the experimental group. For the response variable female survival until offspring dispersion, we first performed a survival analysis using the Cox-Proportional Hazards method. Because data collection ended when a female died or when the offspring dispersed, survival time was right-censored (censoring status: 1 = alive, 2 = dead). The result of this survival analysis is a ratio that describes the relative probability of dying of one group compared to another group. Thus, when the confidence interval includes 1 we can conclude that survival is not affected by the predictor variables (i.e., experimental groups or age classes). Second, we adjusted a GLM with binomial error distribution and logit link function to estimate the survival probability until offspring dispersion (no = 0, yes = 1). In both analyses, we included female age (young or old) as a categorical covariate. Finally, for the response variables offspring number and female fitness, we adjusted GLMs with Poisson error distribution. In the analysis of offspring number, we also included female age and female body size as covariates. The dataset used in the analysis of offspring number included 13 intact and 12 autotomized females that survived until giving birth, and the dataset used in the analysis of female fitness included all females of both experimental groups.

To test for the implications of “tail” autotomy on offspring phenotype, we used a different model to each proxy. For the response variable offspring mass, we adjusted a GLM with Gaussian error distribution. For the response variable offspring size, we adjusted a generalized linear mixed model (GLMM) with Gaussian error distribution and female identity as a random factor because each female has several nymphs in her litter. Given that offspring size may also be influenced by offspring number, we included the total number of nymphs in each litter as a continuous covariable. In the models of offspring mass and offspring size we also included female age and female body size as covariates. We did not include the experimental group of the father as a covarible because exploratory analyses showed no effect of this variable on offspring phenotype (electronic supplementary material S2).

All GLMs were performed using the package stats (R Core Team 2019). The GLMM for offspring size was performed using the package lme4 (Bates et al. 2015). Because the results of the models adjusted with binomial and Poisson error distributions are in logit and log scale, respectively, we used the package emmeans (Lenth 2019) to transform them into probabilities and number of individuals. Using the package emmeans we also calculated effect size (Cohen’s d) for all analyses. The fit of the survival curve and the Cox-Proportional Hazards model were performed in the package survival (Therneau 2020).

The complete methods are in the original paper of The American Naturalist, and the code has several comments that will you help to understand the data set.

If you have any doubt about it, please write to me at solimarygh@alumni.usp.br

Good luck with your research,

Soly