TITLE: Eyespot peek-a-boo: false eyes improve the survival of caterpillars in leaf rolls

RUNNING TITLE: Eyespots protect caterpillars in rolled leaves

AUTHOR: *Elizabeth G. Postema^a

a. Department of Entomology and Nematology, Animal Behavior Graduate Group, University of California, Davis.

KEY WORDS: color, eyespots, predator-prey interactions, leaf roll, deimatic display, environmental deimatism, lepidoptera

WORD COUNT: 150 in abstract, 4378 in main text

# OF FIGURES, TABLES, AND REFERENCES: 3 figures, 64 references.

STATEMENT OF AUTHORSHIP: EGP conceived the study, collected the data, performed the analyses, and wrote all parts of the manuscript.

DATA AVAILABILITY STATEMENT: Upon article acceptance, I will upload all raw data files and R scripts used in this study to Dryad.

*Corresponding Author:

Phone: 734-972-9595, Email: egpostema@ucdavis.edu

Full Address: Elizabeth G. Postema, Department of Entomology and Nematology, Briggs Hall 380K, 1 Shields Avenue, UC Davis, Davis, CA 95616.

ABSTRACT

Deimatic displays typically involve body parts that can conceal or reveal visual signals, potentially reducing detectability at a distance while startling predators up close. Some species may achieve this “conceal-then-reveal” effect using modified aspects of their environment (environmental deimatism hypothesis). The larvae of spicebush swallowtail butterflies (Papilio troilus ) possess large eyespots, and rest in leaf rolls during the day. I tested the hypothesis that leaf rolls reduce eyespot conspicuousness while maintaining eyespot effectiveness by comparing avian predation on 659 artificial larvae: eyespotted and non-eyespotted, presented in leaf rolls or on open leaves. Leaf rolls reduced predation regardless of color pattern. Eyespots also reduced predation, but only for artificial larvae in leaf rolls. On open leaves, eyespots neither increased nor decreased predation. These results suggest that eyespots and leaf rolls can combine to create a deimatic display – and that this strategy likely evolved to enhance existing antipredator effects of leaf rolls.

INTRODUCTION

Conspicuous color patterns have well-known advantages in social, sexual, and antipredator contexts (Prudic et al. 2007; Stuart-Fox & Moussalli 2008; Aronsson & Gamberale-Stille 2009; Caro & Allen 2017). However, strong signals may also attract the attention of unwanted receivers, e.g., predators (Endler 1983; Justin Marshall 2000; Halfwerket al. 2014; de Lira et al. 2018). Even in aposematic species, high detectability can result in high mortality through the attraction of naïve, specialist, or otherwise undeterred predators (Ruxton et al. 2009; Mappes et al. 2014; Fabricant & Herberstein 2015; Umbers et al. 2015). The inbuilt tradeoffs of detectability are cited as potential drivers for the evolution of intermediate aposematic signals (Ruxton et al. 2009), distance-dependent crypsis (Barnett et al. 2017, 2018), and even seasonal trends in proportions of aposematic versus cryptic species (Mappes et al. 2014). Beyond color traits alone, deimatic displays may allow organisms to resolve detectability tradeoffs behaviorally (Umbers et al. 2015, 2017; Umbers & Mappes 2015; Badiane et al. 2018).

Deimatic displays are broadly defined by Umbers and Mappes (2016) as “momentary, transient, conspicuous” signals that induce “a startle response” or overload “the senses of an attacking predator, such that the predator pauses, slows or stops the attack.” Deimatism generally occurs late in the predation sequence – i.e., when the initial defense(s) have failed – and may or may not involve honest advertisements of toxicity and (Umbers et al. 2015, 2019). These displays often blur lines between traditional categories of antipredator signalling, combining distinct types of visual defenses (aposematism, crypsis, masquerade, and/or mimicry) and sometimes additional modalities (e.g., acoustic, chemical) in a single complex display (Dookie et al. 2017, Badiane et al. 2018, Vidal-García et al. 2020, Whiting et al. 2022, Drinkwater et al. 2022). Deimatic displays thus offer sensory ecologists a way to study (1) how behavior intersects with morphology to shape the perception of visual signals, (2) what conditions select for the evolution of multicomponent and/or multimodal signals, and (3) how evolutionary tradeoffs can produce and maintain signal diversity (Rowe 1999, Cuthill et al. 2017, Stevens and Ruxton 2018, Postema et al. 2022). Despite a surge of theoretical interest in the past 10 years, the ecology and evolution of deimatic displays remains poorly understood relative to other forms of visual defenses (e.g., aposematism, crypsis; Umbers et al. 2015, 2017, Skelhorn et al. 2016b, Umbers and Mappes 2016). In particular, experimental evidence for the fitness consequences of deimatic displays is lacking – and, even more so, how these displays function under natural conditions (but see Umbers et al. 2019).

The caterpillars of swallowtail butterflies (family Papilionidae) exhibit a wide variety of defensive color strategies (Gaitonde et al. 2018) and complementary defensive behaviors (Hossie & Sherratt 2012, 2013, 2014; Hossie et al. 2013, 2015). Many species possess eyespots in the final instars, presumably to deter predators by mimicking the eyes of more threatening animals (Wagner 2005; Hossie & Sherratt 2013; De Bona et al. 2015; Skelhorn et al.2016a). Spicebush swallowtail (Papilio troilus ) eyespots are large compared to other North American species in the same genus, such as P. eurymedes, glaucus, or rutulus (Wagner 2005).Like other eyespotted Papilio larvae, spicebush swallowtails inflate their thoraxes in response to agitation (Hossie & Sherratt 2013). This reflex increases the size of the eyespots, and may heighten the larva’s resemblance to a snake (Hossie & Sherratt 2014). However, even at rest, P. troilus eyespots may still be conspicuous to predators unless the larvae are otherwise concealed (Fig. 1a). Unlike most Papilio species, P. troilus larvae construct shelters from the leaves of their host plant. They rest in these leaf rolls during the day (Fig. 1d), emerging at night to feed (Wagner 2005). Larvae in leaf rolls point their heads up towards the leaf petiole, generally making their eyespots partially visible at the small opening at the top of the roll (Wagner 2005; see supplemental table 1).

I hypothesize that the combination of eyespots and leaf rolls inP. troilus larvae creates the same antipredator effect as a deimatic display. I refer to this potential defensive strategy as “environmental deimatism”: the use of environmental materials to create a sudden, startling visual signal that stops or slows predation. Similar to other deimatic displays, P. troilus eyespots are concealed at rest, but may become suddenly apparent once a predator opens or looks into the leaf roll (Fig. 1a,d). This effect would fulfill the definition of a “momentary, transient signal” that startles a predator enough to slow or halt its attack (Umbers & Mappes 2016). Additionally, leaf rolls may offset potential detectability costs of large, conspicuous eyespots (Hossie et al. 2013, Postema 2022). While the startling “conceal-then-reveal” effect of eyespots within leaf rolls has been presumed for both P. troilus larvae and other lepidopteran species (Wagner 2005; Janzen et al. 2010), this assumption remains untested. In other arthropod systems, leaf rolls have been shown to play a role in predator defense (Murakami 1999; Tvardikova & Novotny 2012) among other functions (Kobayashi et al. 2015; Romero et al. 2022), but their influence on the perception of prey color patterns is not well-known. I expect that environmental deimatism may be a relatively widespread (but not well-characterized) defensive strategy among shelter-building species – e.g., spiders that spring out suddenly from leaf rolls (Postema personal observations ), or other leaf-sheltering caterpillars with conspicuous markings (Janzen et al. 2010).

To test the environmental deimatism hypothesis, I conducted an artificial prey experiment using a combination of eyespotted and non-eyespotted clay caterpillars presented on either open or rolled host plant leaves (Fig. 1b-e). I predicted an overall protective effect of leaf rolls, as well as interactions between prey presentation (roll vs. open leaf) and color (eyespots vs. no eyespots). Specifically, I expected eyespots to decrease predation on leaf-rolled prey. For prey on open leaves, I predicted that eyespots would either increase or have no effect on predation – depending on how well eyespots deter predators without additional defensive components (Hossie and Sherratt 2013, Postema 2022). Overall, my primary aims for this experiment were (1) to conduct a test of the environmental deimatism hypothesis in the field, and (2) to quantify the potential costs of conspicuous color signals that lack additional behavioral components.

MATERIALS AND METHODS