False head complexity and evidence of predator attacks in male and female hairstreak butterflies (Lepidoptera: Theclinae: Eumaeini) from Mexico

Introduction

Predation is a major selective pressure responsible for numerous prey adaptations (Wickler, 1968; Ruxton, Sherratt & Speed, 2004). In butterflies, visually oriented predators have selected for several components of the colour pattern and morphology of the wings (Robbins, 1980; Howse, 2014; Rossato et al., 2018). The deflection of predator attacks to “expendable” parts of the body is a type of defence (Cooper Jr, 1998; Humphreys & Ruxton, 2018) that appears to be present in several butterfly species in which the posterior end of the hindwings resembles a butterfly head (at least to the human eye) when the butterfly perches with its wings closed (Fig. 1; Wickler, 1968; Robbins, 1980; Cordero, 2001; Ruxton, Sherratt & Speed, 2004). The ‘false head’ (FH hereafter) is composed of a number of wing pattern and wing shape elements (Robbins, 1980; Cordero, 2001) that vary between species and produce different degrees of similarity to a real head. The four FH components identified by Robbins (1981) (and slightly modified by us) are the following: (1) presence of colour patterns in the posterior end of the hindwings resembling eyes (= eyespots; Fig. 1: Pseudolycaena damo). (2) Presence of tails resembling antennae in the posterior end of the hindwings (Fig. 1). (3) Outline of the posterior end of the hindwings modified in a way that resembles the contour of a head (Fig. 1: both species, but especially clear in Pseudolycaena damo). (4) Presence, in the ventral side of both wings, of lines converging towards the posterior end of the hindwings (Fig. 1: Arawacus sito; Fig. 2: Micandra cyda and Laothus erybathis); these lines presumably direct the “predator’s eye” to the FH area. FHs are particularly frequent in species belonging to the subfamily Theclinae (Lycaenidae) (Robbins, 1980) and behaviours possibly enhancing deception by the FH have been identified (Robbins, 1980; Cordero, 2001).

Although there are observations of predators directing their attacks to the butterfly FH (Van Someren, 1922; Sourakov, 2013; López-Palafox & Cordero, 2017), published reports are scant. Further support for the deflection function of FHs comes from a recent study in which salticid spiders were exposed to virtual prey (resembling insects) on a computer screen (Bartos & Minias, 2016). The spiders preferred to attack the head of the virtual prey, but they were induced to increasingly attack the rear end as the experimenters added FH components (i.e., eyespots, tails resembling antennae, etc.) to the virtual prey. On the other hand, indirect evidence supporting the deflection hypothesis comes from two studies documenting symmetrical damage in the FH area, a likely result of failed predator attacks (Robbins, 1981; Tonner et al., 1993). Direct observations confirm symmetrical damage in the FH area of both hindwings resulting from predator attacks (Van Someren (1922) observations of lizard attacks).

Robbins (1981) reasoned that if the number of FH components in the wings of a butterfly species influences its degree of deceptiveness for predators, then that number should be directly proportional to the fraction of individuals exhibiting symmetrical damage in the FH area. This author analyzed rates of symmetrical damage in the FH area of Neotropical species of the tribe Eumaeini (Lycaenidae: Theclinae) that vary in the number of FH components. He used two samples, one of “more than 1,000 specimens of about 125 species” collected over a period of 5 weeks in an area of 0.5 km² in Colombia, and another of “almost 400 specimens of about 75 species” collected over a period of three months in Panama, and found support for his hypothesis. Here, we first provide further tests of Robbins’ hypothesis by analyzing two samples of Lycaenidae with characteristics different from those used by him. We used specimens belonging to two scientific collections, which were captured by different collectors in several localities covering many parts of Mexico, along a period of several decades. Secondly, since in many Lycaenidae species males employ a sit-and-wait mate location strategy, perching in consecutive days in the same places, at the same time of the day and returning to the same perches after inspecting flying objects (Johnson & Borgo, 1976; Alcock, 1983; Cordero & Soberon, 1990; Fischer & Fiedler, 2001; Takeuchi & Imafuku, 2005a; Takeuchi & Imafuku, 2005b; Salazar, 2011; Dinesh & Venkatesha, 2013), we also explored the hypothesis that males, due to their stationary, conspicuous and predictable behaviour, suffer higher rates of predator attacks than females.

Materials and Methods

Results

Effect of number of FH components on evidence of attacks in the FH area

As expected, in both the analyses at the species and at the genus level, we found highly significant positive effects of the number of FH components on the proportions of specimens with symmetrical damage in the FH area (Table 1, Figs. 3A and 3B). Analyses at both levels also detected an effect of the collection of origin of the data on the proportions of specimens with symmetrical damage (Table 1): these proportions were significantly larger in collection MZFC than in collection CNIN (Figs. 3A and 3B).

Table 1:

Results of models evaluating the effects of the number of false head (FH) components and the collection of origin (MZFC or CNIN) on the probability of symmetrical damage at the species (a) and genus (b) level.

In both cases, both the number of FH components and collection of origin had significant effects, such that the initial model is the minimum adequate model. Significant P-values are in bold type.

Fixed Effects	β± SEa	Za	PZa	Pχ2b
(a) Species level
Number of FH components	0.6291 ± 0.1701	3.757	1.72 × 10−4	3.103 × 10−4
Collection (MZFC-CNIN)	0.4000 ± 0.1952	2.049	0.0405	0.0325
Random effects: ${\sigma }_{\mathrm{species}}^2$= 0.6987 n = 197 total observations from 124 species
(b) Genus level
Number of FH components	0.7465 ± 0.2329	3.205	1.35 × 10−3	4.006 × 10−3
Collection (MZFC-CNIN)	0.5168 ± 0.1937	2.668	7.64 × 10−3	6.95 × 10−3
Random effects: ${\sigma }_{\mathrm{genus}}^2$= 0.4494 n = 81 total observations from 49 genera

DOI: 10.7717/peerj.7143/table-1

Notes:

aParameters from model.

bP-value resulting from χ² nested model comparisons following removal of each variable during backwards stepwise simplification.

Discussion

The shapes and colours that make butterfly wings aesthetically appealing to humans are, in large part, products of the evolutionary pressures imposed by predators (Howse, 2014; Rossato et al., 2018). The “false head” of Lycaenidae is considered an adaptation for deflecting predator attacks to less vulnerable areas of the animal. Robbins (1980) found that the complexity of FHs is positively associated with the proportion of specimens with symmetrical damage in the FH area, a likely evidence of failed predator attacks (Van Someren, 1922). Here, using larger and more heterogeneous samples, we confirm that the probability of finding symmetrical damage in the FH area is larger in species and genera with more FH components (Tables 1 and 2, Fig. 3), providing further evidence for the role of predator deflection in the evolution of butterfly false heads and for the idea that FHs with more components are better at deceiving predators.

In contrast, our hypothesis that the sit-and-wait mate location strategy employed by males of many Lycaenidae species (Johnson & Borgo, 1976; Alcock, 1983; Cordero & Soberon, 1990; Fischer & Fiedler, 2001; Takeuchi & Imafuku, 2005a; Takeuchi & Imafuku, 2005b; Salazar, 2011; Dinesh & Venkatesha, 2013) results in males being more stationary and behaviourally conspicuous than females, and thus more prone to be attacked by visually oriented predators, was not supported by the data. Males and females did not differ in the proportion of specimens with symmetrical damage (Table 2). It is possible that our generalization is wrong and the number of species in which males exhibit sit-and-wait mate location behaviour is much smaller than we think. If this is the case, the prediction that males will suffer higher rates of attack should be tested only in species exhibiting sit-and-wait male mate location strategy. Furthermore, female behaviour, especially egg laying and nectar feeding, could expose females to a predation risk similar to that experienced by males. In fact, the lack of obvious sexual differences in FH traits observed in our samples also point against our hypothesis.

If we accept the conclusion that a FH with all four components is better at deflecting predator attacks, then we need to understand why species with different numbers of FH components frequently coexist in the same location (e.g., Robbins, 1981; López-Palafox, Luis-Martínez & Cordero, 2015). Robbins (1981) considered that genetic restrictions or ecological factors could be responsible for explaining variation in FH complexity. Robbins (1981) considers unlikely that genetic restrictions provide a general explanation, because evidence indicates that FH components evolve rapidly. Our data are consistent with this point of view: 11 of the 26 genera (42%) with more than one species in our database have species belonging to two (8 genera) or three (3 genera) different categories of number of FH components (Table 3). Thus, ecological factors could provide better explanations for the diversity observed in FH patterns. Species with different numbers of FH components could represent alternative adaptive solutions to selective pressures from visually oriented predators when we consider additional traits. For example, a species with a “complete” FH could be rather conspicuous when perched (Fig. 1), whereas a species like Callophrys xami with only eyespots and tails (López-Palafox & Cordero, 2017), but with the ventral side of the wings of green colour, could be hard to spot when posed on or near leaves, as well as being able to deflect predator attacks to the FH once it is detected. The morphological components of the FH considered in this paper are complemented with behaviours that possibly enhance the deceptive effect (Robbins, 1980; Cordero, 2001; see also Hoskins & Bálint, 2016), such as the back-and-forth movement of the hindwings along the sagittal plane frequently performed while resting by Theclinae species (López-Palafox & Cordero, 2017) and a few species of other groups (Riodinidae, Robbins, 1985; Nymphalidae, Sourakov, 2015). The interaction of these behavioural traits with the morphological components could also help explain the interspecific variation in the complexity of the FH.

One of the referees of this paper (Dr. A. Krupitsky) noted the interesting fact that in some species the degree of development or even the number of some of the FH components varies between individuals (the reviewer mentions Cyanophrys spp., Micandra cyda and Oenomaus ortygnus among these species). We did not consider this possibility in our study and, thus, we did not look for intraspecific variation in the FH components. In our opinion, quantitative documentation of the extent of this variation and its distribution between the different taxa, as well as its possible relationships with the average number of FH components and sex (among other variables), is an important subject for future research. Speculate about the possible causes of such intraspecific variation seems premature, although relaxed selection from reduced predation pressure comes to mind as one possible hypothesis. Additionally, species with substantial intraspecific variation could be used as “natural treatments” in the functional study of FH components, in experiments similar to that of López-Palafox & Cordero (2017) on “false antennae”.

Conclusions

The false head (FH) hindwing pattern is considered an adaptation to deflect predator attacks away from the real head to less vulnerable parts of the body. We tested one prediction of this hypothesis using two large samples of museum specimens. Since FHs vary between species in the number of morphological and colour components (“complexity”), we predicted a positive correlation between the number of FH components and the proportion of specimens with evidence of predator attacks directed to the FH area. We found the predicted association whether we used species or genus as a sample unit. We also tested the related hypothesis that behavioural differences between the sexes makes males more prone than females to receive predator attacks. Contrary to our prediction, we found that the proportion of specimens with evidence of predator attacks directed to the FH area was not different between males and females. Overall, our results support the deflection hypothesis and the idea that FHs with more elements are better at deceiving predators, and suggest that behavioural differences between the sexes produce no differences in the risk of being attacked.

Supplemental Information