Predator Mimicry: Metalmark Moths Mimic Their Jumping Spider Predators

Cases of mimicry provide many of the nature's most convincing examples of natural selection. Here we report evidence for a case of predator mimicry in which metalmark moths in the genus Brenthia mimic jumping spiders, one of their predators. In controlled trials, Brenthia had higher survival rates than other similarly sized moths in the presence of jumping spiders and jumping spiders responded to Brenthia with territorial displays, indicating that Brenthia were sometimes mistaken for jumping spiders, and not recognized as prey. Our experimental results and a review of wing patterns of other insects indicate that jumping spider mimicry is more widespread than heretofore appreciated, and that jumping spiders are probably an important selective pressure shaping the evolution of diurnal insects that perch on vegetation.


INTRODUCTION
The phenomenon of mimicry, a high degree of resemblance due to selection, was first proposed in 1862 by Sir Walter Henry Bates upon his return from eleven years as a professional collector in Amazon. Writing about butterfly wing patterns, Bates noted, ''… on these expanded membranes Nature writes, as on a tablet, the story of the modifications of species…'' Bates proposed that longwings and other butterflies gain protection by mimicking distasteful species and that the resemblances among such unrelated insects lent support to Charles Darwin's newly proposed theory of natural selection [1]. Since Bates's initial contribution various cases of mimicry have been identified from across the tree of life. In this paper, we describe a curious form of Batesian mimicryagain involving the wing patterns of Lepidoptera -in which prey (metalmark moths) obtain protection by mimicking their predators (jumping spiders) (Fig. 1).
Many examples of Batesian and Müllerian mimicry and camouflage have been described [1][2][3][4]. Even cases of aggressive mimicry, where predators mimic prey, are known (e.g., females of Photuris fireflies lure males of different firefly species to their death by mimicking their courtship signals [5]). However, predator mimicry -cases in which prey have evolved to mimic their predators to thwart predation attempts -are both exceptional and rare.
Predator mimicry was suggested for owls, where owl ear tufts mimic mammalian predators for protection from such predators as lynx, fox, and marten [6]. Another potential case of predator mimicry is among South American cichlids: coloration and spotting of certain prey species makes them so similar to their predators that they are thought to be their mimics [7]. Eyespots on the wings of giant silk moths and other Lepidoptera undoubtedly mimic eyes of mammalian predators -but here the eyes may function not to mimic their would-be predators, but to resemble a much larger animal, one sizable enough to be a threat to lepidopteran would-be predators. Hence, these eyespots might be regarded as startle coloration [8]. However, in none of these cases are there experimental data demonstrating the efficacy of this mimicry. Our literature review suggests there are few wellsupported cases of predator mimicry: e.g., lycaenid butterflies that chemically mimic ants [9] and salticid spiders that mimic ants to avoid being preyed upon by them [10] (other ant mimics probably gain protection from all predators that tend to avoid ants [11]). Here we present evidence for another case of predator mimicry involving salticid spiders, but in this case salticids are predators and not prey.
Jumping spiders (Araneae: Salticidae) are visual predators of small arthropods. Among the cues that salticids use for distinguishing between prey, mates, rivals, and enemies are shape, symmetry, presence of legs and wings, size, and style of motion [12]. Their vision is highly acute -they can discriminate between objects even at 40 body lengths [13]. They maintain territories through ritualized displays, which are aimed at both con-and heterospecific individuals [14]. In many salticids, this display consists of a male raising and waving his forelegs at an intruding spider (see video S1 showing two males displaying). If salticids mistakenly recognize potential prey organisms as other jumping spiders, mimetic prey would enjoy higher chances of survival because, instead of an attack, a territorial display may ensue, making it more likely for the prey to escape predation. The results of our study provide evidence that this scenario occurs within Brenthia metalmark moths ( Fig. 2a and 2b).

RESULTS/DISCUSSION
All species of Brenthia which we observed possess the same adult behavior: during the day both males and females perch on upper surfaces of vegetation and adopt a peculiar posture (Fig. 2b). Their hindwings are fanned outwards and brought forward, perpendicular to the forewings; the forewings are raised and held above the body at approximately a 45u angle [15,16]. In this position, the alternating white and black fascia on the hindwings are reminiscent of salticid legs (Fig. 1). Moreover, both sexes move with short, rapid, jerky motions, in much the same fashion as jumping spiders (see video S2 of live B. hexaselena and B. monolychna). Their exceptional wing posture, in conjunction with spiderlike wing markings and movement, makes Brenthia moths salticid look-alikes.
In 2003, we tested the hypothesis that Brenthia moths are salticid mimics by setting up trials in which one jumping spider (Phiale formosa) (Fig. 3) -the largest most common and conspicuous salticid at the study site -was paired with one moth in a small arena. Tested moths were either a presumed mimic (Brenthia) or a control species (other choreutids or comparably-sized moths, none of which exhibited wing patterns resembling salticids or engaged in the jerky movements characteristic of Brenthia). If Brenthia elicited a territorial display from spiders, we concluded that the moths were mistaken for salticids.
We carried out 78 trials (n mimics = 39, n controls = 39). Presumed mimics had much higher rates of survival than controls (p,0.0001, G = 18.96) ( Table 1), demonstrating an adaptive value to the   Brenthia morphology and/or behavior. However, the frequency of spider ''stalk'' and ''pounce'' behaviors indicated that the mimics were recognized as potential prey as often as controls (p = 0.1096, G = 2.56) ( Table 1). Not one P. formosa made a territorial display, suggesting all spiders used in the trials perceived Brenthia moths as prey. The position of wings in Brenthia adults may have afforded some protection -when pouncing on a moth, the spiders first hit the raised wings, and, as a consequence, missed the body of the moth. Their resemblance to jumping spiders may have been unimportant.
Since size has been shown to be a good predictor of the outcome for encounters between jumping spiders [17], one possible explanation for the lack of support for the hypothesis of salticid mimicry in the trials with P. formosa is the size difference between the prey and mimic -mature P. formosa are four times larger than the Brenthia moths used in the trials. To test how aggressive P. formosa is towards smaller spiders, we presented it with salticid spiders comparable in size to Brenthia moths in four trials. Although the smaller salticids engaged in a territorial display, in all four cases P. formosa individuals attacked and killed the smaller spiders without displaying back. These trials suggested that even if Brenthia were perfect salticid mimics, P. formosa would perceive them as prey.
Based on these results, we returned to the study site in 2005 to repeat our experiments using jumping spiders of similar size to Brenthia moths. As in our previous study, Brenthia had much higher survival rates when compared to controls: out of 77 Brenthia moths, only 5 were caught, whereas 43 of 69 controls were killed (n trials = 146, n mimic = 77, n control = 69, p,10 28 , G = 55.8) ( Table 2). More interestingly, in this set of trials many jumping spiders engaged in territorial displays after encountering a Brenthia; no spiders displayed to control moths (p,10 27 , G = 33.96) ( Table 2). In no trials in which spiders displayed toward Brenthia did Brenthia get caught. The difference in response to Brenthia versus controls was marked: spiders began stalking control moths soon after an individual was introduced into the arena and, usually after only a few pounces and without hesitation, caught and fed on the nonmimetic prey (see video S3 showing five trials with control moths). In trials with Brenthia moths the spiders first observed the moth, and then many (in 36% of the trials) raised and waved their forelegs towards the moth, i.e., engaged in territorial behavior (see video S4 of five trials with Brenthia moths). In several trials (n trials = 11), spiders even backed away from the mimics, demonstrating that many Brenthia are perceived as salticids -their ploy of donning wolf's clothing proves successful.
The adaptive value of jumping spider mimicry may extend beyond escaping predation by salticids. Mimicking jumping spiders may be helpful against birds and other vertebrate predators through evasive prey mimicry, wherein protection is gained through resemblance to prey that is too costly for predators to capture [18][19][20]. For example, if birds learn that it is not worthwhile to pursue salticids as prey, because these wary spiders often evade capture, then credible salticid mimics would be ignored as well. Lindroth [18] proposed this form of mimicry as the explanation for why some ground beetles in the genus Lebia (Coleoptera: Carabidae) resemble flea-beetles (Coleoptera: Chrysomelidae: Alticinae) and Balgooyen [19] suggested it as an explanation for why some short-horned grasshoppers mimic the alfalfa butterfly (Colias eurytheme). While no empirical studies have been conducted to support this mimicry type, Ruxton et al. [20] demonstrated mathematically that evasive prey mimicry would be selectively advantageous when there is alternative prey available and the evasiveness of the model is costly to the predator. We consider it likely that Brenthia and other insects that are salticid spider mimics are also favored through evasive prey mimicry, as well as predator mimicry.
Jumping spider mimicry appears to be taxonomically widespread among insects. Tephritid flies that mimic salticids have higher survival rates in encounters with jumping spiders when compared to other similar-sized, non-mimetic flies [21][22][23]. Salticid mimicry has been suggested for three fulgoroid homopterans: an issid [24], a fulgorid [25], and a derbid [26]. Both wing patterns and perching behaviors of other lepidopterans, including nymphuline pyralids, hilarographine tortricids, glyphipterigids, Fabiola (in the Oecophorinae), Ecballagonia bimetallica (in the Cosmopterigidae), and scattered gelechiid genera are suggestive of jumping spider mimicry. In addition to having lightly-coloured fascia along the leading edge of the forewing (which yield the impression of legs), these microlepidopteran taxa possess groups of eyespots, frequently including metallic scales, along the trailing edge of the forewing or an exposed, often elevated, portion of the hindwing that collectively resemble the clustered eyes of a salticid spider. As in Brenthia, the postures of the mimics are often dramatically showy relative to those of their non-mimetic sister taxa. The convergent acquisition of grouped eyespots and leg-like markings on wings of small moths and other insects offers further evidence that the vision of some invertebrates is significantly more sophisticated than generally appreciated. Presumably, the highly acute vision of jumping spiders is an integral part of the selective force shaping these mimicry systems. If even a portion of the various instances of salticid mimicry in the Diptera, Homoptera, and those proposed here for the Microlepidoptera prove to be valid, then jumping spiders can be regarded as important drivers in the evolution of diurnal insect phenotypes and behaviors.  Individuals from several lepidopteran families were used as controls. A moth qualified for control trials if it was similar in size to Brenthia and did not resemble jumping spiders. Control moths were obtained from mercury vapor light and diurnal net collections or by rearing. All moths were used in a single trial within 24 hours of collection or emergence.

MATERIALS AND METHODS
In trials that were done in 2003, we used 78 individuals of the jumping spider Phiale formosa (Banks). In 2005 trials we used 80 individuals representing 12 species of jumping spiders that were similar in size to Brenthia (body length 5-8 mm). Vouchers of all spider individuals used in 2005 trials are deposited in the UCMS collection. Spiders were held in 15-dram plastic vials and fed small arthropods initially, then kept without food for 48 h before trials. In 2003 each spider was used only once. In 2005, 62 of 80 spiders were used in two trials, once with a mimic and once with a control; 12 were used with mimics only; 4 were used with controls only; and 2 were used in 3 trials (either with 2 mimics and 1 control or vice versa). When used in more than one trial, spiders were randomly assigned first to mimic or control trials and they were held for at least two days between successive trials. Both male and female moths and spiders were used in the trials.
Trials in 2003 were carried out in a plastic arena (136867 cm) with a transparent top; in 2005 we used a slightly smaller PlexiglasH arena (106564 cm). In every trial, a spider was first placed into the arena and then later the moth was introduced. Trials were terminated upon capture of the moth or after 5 minutes. Four behaviors were recorded: 1. ''stalk'' -spider stealthily approached the moth; 2. ''pounce'' -spider jumped towards the moth and attempted to catch it; 3. ''catch'' -spider captured the moth; 4. ''display'' -spider raised and waved its forelegs. All trials in 2005 were filmed with a mini DV camera and reviewed afterwards. Video footage of five trials with mimics and five trials with controls is available as supporting information, as well as footage of live B. hexaselena and monolychna and spider territorial displays. All video footage was made at La Selva.
In 2003 we carried out 78 trials, 39 with mimics and 39 with controls, and in 2005, 146 trials, 77 with mimics and 69 with controls. The data were organized into a 262 contingency table with the marginal totals for the number of trials fixed (Tables 1  and 2). We analyzed the data with the G-test of independence, applying Williams's correction [27]. For the calculation of the G value for data shown in Table 2, we replaced the 0 value for the number of displays to control moths with the number 1 to facilitate calculation.

SUPPORTING INFORMATION
Video S1 Territorial behavior of two male salticids.