Discussion
The comparative analysis presented here documents both intra- and interspecific differences in shape complexity within the amblypygid pedipalp. A principal component analysis of Fourier harmonics shows frequent overlap between species, with intraspecific shape variation often exceeding shape differences between species, even in the relatively low numbers of individuals representing each taxon in this study. MANOVA does identify statistically significant pairwise shape differences between most species. However, unsupervised k-means struggles to define species clusters accurately. Of note, the twoDamon species, D. diadema and D. medius , could not be separated on the basis of MANOVA, k-means or within the PCA morphospace in either segment. Likewise, some species ofParaphrynus could not be differentiated on the basis of shape complexity in either segment. As such, our results do not support Hypothesis 1.
This result has potential implications for species recognition using pedipalp characters. MANOVA, a ‘supervised’ method which knows which specimens belong to each species a priori , could define differences between species. However, the unsupervised k-means methods, which has no prior knowledge of which species specimens belong to, could not. K-means is arguably a better analogue for the placement of a specimen within a species, as a taxonomist, like the k-means algorithm, would have no prior knowledge of specimen’s placement within a species. Notably, previous studies have identified species of Damon andParaphyrnus partially on the basis of the relative length of spines on the femur and tibia pedipalp segments (Weygoldt, 2000; Prendini, Weygoldt & Wheeler, 2005; Joya & de Armas, 2012). Although our results do not invalidate the use of any specific pedipalp tibia and femur characters to define species (as our analysis investigates overall pedipalp shape), we do urge caution when differentiating between closely related species using pedipalp traits.
In support of Hypothesis 2, a significant negative relationship between complexity and pedipalp length was identified in both segments. Visual inspection of Fourier outlines suggests that complexity predominantly reflects the relative length of major pedipalp spines (Figure 2). Interestingly, a similar pattern was revealed using GMM in D. variegatus , with longer pedipalps having shorter spines. Longer amblypygid pedipalps are therefore found to have relatively shorter ‘major’ spines once normalised to centroid size. Previous work has demonstrated the importance of display in the evolution of pedipalp morphology. Roughly, 80% of conflicts in P. longipes are decided via display in the favour of the individual with longer pedipalps (Chapin & Reed‐Guy, 2017). Pedipalp display has also been observed in territorial contests in a number of amblypygid species (Fowler-Finn & Hebets, 2006; Porto & Peixoto, 2013; Chapin & Hill-Lindsay, 2016) and during the first stages of courtship across the group (Weygoldt, 2000; Chapin & Hebets, 2016). This suggests that amblypygid pedipalps are likely under the influence of sexual selection via mate choice, and that resource holding potential may be correlated with pedipalp length.
It may therefore be possible that a trade-off between pedipalp length and complexity exists, with some species taking a less complex form, with relatively shorter spines, allowing for increased pedipalp length, developing under the pressures of sexual selection and contest, to be achieved at lower developmental cost. Hypothetically, a less complex form with shorter spines may be less energetically costly to produce per unit length, which may allow individuals to achieve longer pedipalps at lower energetic cost, though as yet we know nothing of genetics of development in amblypygids, and such inferences are therefore limited.
There is some evidence for structures becoming less complex with increasing size within arachnids, although little is known about interspecific trends of shape complexity in arachnids. For example, cheliceral dentition is known to be more pronounced in juveniles and relatively decreased in the large chelicerae of adults (Solifugae), though this may in part be a function of cheliceral wear (Bird, 2015). Some pedipalp spines also appear reduced in male thelyphonids with large pedipalps (Rajashekhar & Bali, 1982). However, it is often the case that structures under the influence of sexual selection are larger and more complex in arachnids. For example, male spider legs that are used in courtship display are longer than female conspecifics and often possess elaborations such as ridges of setae (Peckham & Peckham, 1889; Kronestedt, 1990; Girard & Endler, 2014). Furthermore, the fourth legs of male opiliones, used in male-male contest, are also longer and have larger elaborated coxal apophyses not seen in females (Willemart et al., 2009; da Silva Fernandes & Willemart, 2014). Thus, though a trade-off may be possible, evidence from other arachnids provides mixed support. However, it must be stressed that these examples are all intraspecific and thus may not be a good analogue for the interspecific trends we report in this work.
Another possibility is that that spine length is constrained by selective regimes or developmental factors, meaning spines have simply remained at a similar absolute size across species. In this case, species with longer pedipalps would have relatively shorter spines. For example, spine length could be limited by developmental constraints relative to pedipalp length. Additionally, it is possible that absolutely longer spines may provide limited additional functional benefits to prey capture, and are not important in display-based contest, meaning there is little advantage to spine length increasing with isometry relative to pedipalp length. Pedipalp spines are thought to primarily function in prey capture, with a number of species forming pedipalp ‘catching baskets’, which are hypothesised to help capture and secure prey items. Little is known about amblypygid diets, but they are thought to be largely composed of primary consumer arthropods, especially of the orders Orthoptera and Blattodea (Chapin & Hebets, 2016). If the primary function of spines is indeed in prey capture, spine size may scale more closely with prey size than pedipalp size. Therefore, should prey size remain similar or equal across the group, spine size may also remain similar across species leading to relatively smaller spines in species that have attained longer pedipalps due to pressures of sexual selection and contest. However, this hypothesis can only be properly investigated by addressing the current paucity in data related to diet in amblypygids.
An analogue to the palpal form in Amblypygi is found in scorpion pedipalp chelae, which increase in size at rates much lower than isometry with relation to body size (Van der Meijden, Herrel & Summers, 2010). This may be because species with vastly different body sizes feed on similarly sized prey (Polis & McCormick, 1986), suggesting there is little added benefit to larger species producing equivalently sized chelae. If a similar overlap in prey size exists across amblypygid species, spines (which are hypothetically under the most direct selection for prey capture) may also scale at rates lower than isometry relative to pedipalp length. More work is needed into amblypygid life history and social dynamics to determine the relative importance of sexual selection, selection via territorial contest, and natural selection via prey capture.
As part of this study, sexual shape dimorphism was quantified across amblypygids for the first time. Significant sexual dimorphism in shape complexity is identified in the amblypygid pedipalp tibia, with female tibiae found to be characterised by greater complexity than their male counterparts. As previously discussed, complexity appears to be influenced by the length of the major spines, which are relatively larger in the female tibia. This may be due to the increased length of the male pedipalp, which develops under the pressure of sexual selection. A similar pattern has already been identified using GMM analysis of the pedipalps of D. variegatus , with the observed sexual shape dimorphism being thought to arise due to female optimising for prey capture while males optimise for pedipalp display (McLean, Garwood & Brassey, 2020). The occurrence of this form of dimorphism in numerous species across the group may suggest that these driving forces behind sexual dimorphism in size and complexity are pervasive across the group. Interestingly, contest and courtship both involve display and are very similar in behaviour across amblypygid species, suggesting this could be a common selection pressure (Weygoldt, 2000). However, more work into amblypygid life history is needed to fully understand the drivers of pedipalp morphology.
Conclusion
In conclusion, our results highlight the importance of considering intra- and interspecific variation in terms of shape, by highlighting a number of previously undocumented shape relationships across amblypygid pedipalps. Here we find that within-species shape complexity variation in pedipalps can occasionally exceed differences between species, and thus caution needs to be taken when defining species on the basis of pedipalp characters. We also find that gross complexity decreases with increasing pedipalp length, potentially uncovering a trade-off between investment in pedipalp total length (for use in sexual selection and territorial contest) and pedipalp spine length (primarily for use in prey capture). Sexual dimorphism in gross complexity is also present in the tibia and follows similar patterns seen intraspecifically in other studies, once again highlighting the pressures of display based contest and courtship, and trophic niche partitioning. Future studies that look to address the paucity of data on amblypygid life history, ecology and social dynamics will be valuable in further understanding the functional morphology in this unique system.