Intraspecific weapon polymorphisms that arise via conditional thresholds may be affected by juvenile experiences such as predator encounters, yet this idea has rarely been tested. The New Zealand harvestman Forsteropsalis pureora has three male morphs: majors (alphas and betas) are large-bodied with large chelicerae used in male-male contests, while minors (gammas) are small-bodied with small chelicerae and scramble to find mates. Individuals use leg autotomy to escape predators and there is no regeneration of the missing leg. Here, we tested whether juvenile experience affects adult morph using leg autotomy scars as a proxy of predator encounters. Juvenile males that lost at least one leg (with either locomotory or sensory function) had a 45 times higher probability of becoming a minor morph at adulthood than intact juvenile males. Leg loss during development may affect foraging, locomotion, and/or physiology, potentially linking a juvenile’s predator encounters to their final adult morph and future reproductive tactic.

Adult males (N = 86) of F. pureora were collected between Waitomo and Te Anga, New Zealand, across four sites: Ruakuri Bushwalk, Waitomo (38°15'53.7"S 175°04'46.4"E); private land, Te Anga (38°15'41.4"S 175°00'53.6"E); Tawarau Forest, Te Anga (38°17'24.8"S 174°56'50.2"E); and Marokopa Falls Track, Te Anga (38°15'33.6"S 174°50'54.6"E). Individuals were hand collected between 21:00 and 02:00 hours in January and February 2018. 

Sexual maturity was confirmed by the presence of a functional genital operculum, which is fused during the juvenile stage but can be opened upon the adult moult (Gnaspini 2007). Following Powell et al. (2020), we assigned each male to either major morph (alphas and betas, N = 63) or minor morph (gamma, N = 23). Morph frequency in our sample, with about 75% of majors and 25% of minors, closely followed that of the larger sample in Powell et al. (2020). Males were randomly collected, but it is possible that fewer minors were located because they were moving cryptically on the forest floor rather than resting on top of vegetation, as occurs with majors. We did not consider alphas and betas separately because, according to our hypothesis, both morphs probably had access to a greater amount of food resources during development. Moreover, Painting et al. (2015) hypothesized that trimorphism in New Zealand neopilionids arises via both a genetic polymorphism (which promotes the existence of two major morphs) and polyphenism (which promotes the existence of majors and minors via environmental conditions during development). Therefore, because leg loss during development should act exclusively on the polyphenic mechanism, our data considers specifically the dichotomy between majors and minors.

To determine the “age” of autotomy, we inspected the remaining trochanter and coxa of the specimens that had experienced autotomy. Because harvestmen do not regenerate legs, scars remain at the location of legs lost across the individual’s life, even if the leg was autotomised early on in development. Thus, it is possible to tell if legs were autotomised prior to or after the last moult to adulthood (R. Macías-Ordóñez, personal communication). Fresh scars representing autotomy during adulthood were characterised by a fully intact trochanter. If autotomy occurred before adulthood, the trochanter was reduced or even absent, with a sclerotised patch of cuticle over the location where the trochanter joint used to be. Further, this patch of cuticle sometimes had associated setae and in some specimens the coxa joint of the missing leg was highly reduced and compressed by the coxa of the intact legs, making it clear that legs were autotomized at least one moult before adulthood (E. C. Powell, personal observation). For the 50 individuals that experienced any autotomy, we classified scars as recent, if they occurred during adulthood, or old, if they occurred during development, i.e., as a juvenile at least one moult before adulthood.

To test whether leg loss (yes or no) during development correlates with adult male morph, we generated a generalised linear model (GLM) with a binomial response variable (logit link function) describing male morph identity: small-bodied minor morph (gamma) and large-bodied major morphs (alpha and beta). We did not include the number of missing legs in the model because only two males in our sample had more than one juvenile autotomy (see ‘Results’). We also tested whether the type of leg autotomized during development affects adult male morph (response variable) using a GLM with binomial distribution (logit link function). We used leg type as a predictor because, as mentioned above, legs serve different functions in harvestmen. The predictor variable ‘leg type’ was divided into two levels: (a) locomotory legs, which includes legs III and IV, and (b) sensory legs, which includes legs II (Willemart et al. 2009). Legs I are used both as a locomotory and a sensory appendage (Willemart et al. 2009). Thus, to test the sensitivity of our results to the classification of legs I, we ran two separate GLMs, first classifying them as a locomotory and then as a sensory leg. As mentioned above, only two males had more than one juvenile autotomized leg in our sample, and thus it was not possible to include male identity as a random factor in the analyses. To avoid recording the multiple autotomies of these two males as independent points, we removed them from the analyses. All statistical analyses were performed in the software R (R Core Team 2020).

The .csv file can be opened with Microsoft Excel, Notepad, OpenOffice Calc, among other programs.