Effects of host plant growth form on dropping behaviour in leaf beetles

Many leaf-eating insects drop from their host plants to escape predators. However, they must return to the leaves of the host plant after dropping, which represents a cost associated with this behaviour. In woody plants, the positioning of leaves is generally higher than that of herbaceous plants, which suggests that dropping from woody plants might be costlier for leaf-eating insects than dropping from herbaceous plants. Therefore, we predicted that dropping behaviour would be observed less frequently in insects that feed on woody plant leaves than in those that feed on herbaceous plant leaves. To test this prediction, we investigated dropping behaviour experimentally in larvae (23 species) and adults (112 species) of leaf beetles (Coleoptera: Chrysomelidae) on their host plants (86 species of 44 families) in field conditions. Larvae on woody plants exhibited dropping behaviour less frequently than those on herbaceous plants. However, this pattern was not detected in adults. Thus, host plant growth form might affect the evolution of dropping behaviour in leaf beetle larvae, but not in winged adults, perhaps owing to their higher mobility.


INTRODUCTION
Animals defend themselves against predators in various ways (Edmunds, 1974;Sugiura, 2020). The evolution of anti-predator defences can be driven by benefits gained from protection against predators and by the associated costs (Bowers, 1992;Camara, 1997;Zevereva et al., 2017). The benefits and costs of defences are closely associated with habitat. For example, phytophagous insects sequester host plant chemicals in their bodies to defend against predators (Nishida, 2002); the effectiveness of chemical defences can vary among host plant species because they have different enemies and access to different chemical components (Denno et al., 1990;Singer & Stireman, 2003). Although studies have investigated the effects of host plants on the effectiveness of anti-predator defences (Denno et al., 1990;Singer & Stireman, 2003), few studies have clarified the effects of host plants on the costs associated with anti-predator defences in phytophagous insects (Matsubara & Sugiura, 2018).
Host plants can affect the cost of dropping behaviour in phytophagous insects (Matsubara & Sugiura, 2018). When phytophagous insects drop to the ground, they must then return to an appropriate feeding position on the host plant. Underlying leaves can prevent phytophagous insects from dropping to the ground, thereby reducing the costs of dropping behaviour. For example, leaf beetles are much less likely to drop to the ground from large, oval leaves than from cleft leaves because the former can act as safety nets for the falling beetles (Matsubara & Sugiura, 2018). Given that the leaf positions on woody plants are generally higher than those on herbaceous plants, insects require more time and energy to move from the ground to their feeding position on woody plants. Consequently, the costs associated with dropping from woody plants are expected to be higher than those associated with dropping from herbaceous plants. Some arthropods that inhabit woody plants reportedly avoid dropping to the ground by gliding (Yanoviak et al., 2005(Yanoviak et al., , 2009(Yanoviak et al., , 2015Meresman et al., 2017) or using silk lifelines (Brackenbury, 1996;Sugiura & Yamazaki, 2006). Therefore, dropping behaviour is expected to occur less frequently in insects that feed on woody plant leaves than in those that feed on herbaceous plant leaves. However, this prediction has never been tested. Clarification of the relationships between host plants and dropping behaviour would contribute to our understanding of the evolution of defensive behaviours and host plant selection in phytophagous insects.
To elucidate the effects of host plant growth form (i.e. herbaceous vs. woody plants) on dropping behaviour in phytophagous insects, we investigated the anti-predator defences of leaf beetles (Coleoptera: Chrysomelidae) on their host plants in field conditions in Japan. The species diversity of leaf beetles is high; the family Chrysomelidae includes ~41 000 extant species (Jolivet et al., 2008), ~700 of which are recorded in Japan (Kimoto & Takizawa, 1994). Most leaf beetle species are dietary specialists that feed on particular families and genera of plants (Kimoto & Takizawa, 1994;Jolivet & Verma, 2002). Although closely related beetle species frequently use the same plant families, some congeneric species feed on different plant families (Kimoto & Takizawa, 1994). Leaf beetles defend themselves against predators in various ways, including dropping, flying, jumping, clinging, having spines and self-mimicking of feeding damage (Kimoto & Takizawa, 1994;Jolivet & Verma, 2002;Konstantinov et al., 2018;Shinohara & Takami, 2020;Sugiura, 2020). The secretion of chemical liquids from the body is well documented in some leaf beetle larvae as a chemical defence against predators such as ants (Sugawara et al., 1979;Pasteels et al., 1982;Pasteels et al., 1984;Kimoto & Takizawa, 1994). Thus, leaf beetles are an appropriate insect group for investigating the effects of host plant growth form on dropping behaviour and other anti-predator defences in phytophagous insects. Larvae and adults of many leaf beetle species are found on host plant leaves, making it possible to compare the effects of host plant growth form on dropping behaviour in larvae and adults.
In this study, we investigated experimentally the defensive behaviour of larvae (23 species) and adults (112 species) of leaf beetles on their host plants in field conditions in Japan. We examined whether larvae and adults showed dropping or other defensive behaviours in response to artificial stimuli on their host plants and explored the effects of host plant growth form (herbaceous or woody plants) on the evolution of dropping behaviour in leaf beetles.

Study Site and SpecieS
Field experiments were conducted at 102 sites in Japan (Matsubara & Sugiura, 2021; 25°51′-41°22′N, 127°42′-141°22′E, 1-1090 m a.s.l.). The study sites included various environments, such as forest, grassland and farm. Leaf beetle larvae are external leaf feeders, leaf/stem miners, seed borers, detritus feeders or root-feeders (Jolivet, 1988). In this study, we focused on external leaf feeders that potentially drop from host plant leaves. Early instar larvae were not included in this study because they were too small to be investigated in field conditions. Therefore, larvae with body length ≥ 5 mm were used for subsequent experiments.
Leaf beetles and their host plants were identified based on their morphological characteristics (Kimoto & Takizawa, 1994;Azegami et al., 2013a, b;Hayashi, 2014). The growth form (woody or herbaceous) of each plant species was determined based on the presence or absence of woody tissues in the stems (Clapham et al., 1987;Azegami et al., 2013a, b;Hayashi, 2014).

Field experimentS
To explore the effects of host plant growth form (woody or herbaceous) on the defensive behaviours of leaf beetles, we investigated the responses of beetles to simulated attacks in field conditions, following the procedure of Matsubara & Sugiura (2018). When larval or adult leaf beetles fed externally on plant leaves, we measured the feeding elevation (i.e. vertical distance from the ground to the feeding position). Individuals on leaves higher than 3.0 m were not included in our experiments because they could not be reached. We randomly selected beetles on leaves (height, ≤ 3.0 m) and poked the larval dorsal abdomen or adult elytra with forceps to simulate attacks by predacious insects (e.g. ants). Forceps have been used frequently to simulate predator attacks (e.g. Miyatake et al., 2008;Müller et al., 2016;Matsubara & Sugiura, 2018). To standardize the stimuli, the same researcher used the same forceps in all the field experiments. The same degree of stimulus was provided to all leaf beetles in this study. When beetles did not respond to a stimulus, they were poked repeatedly at 2 s intervals, up to five times. Beetles on the lowest leaves are more likely to drop to the ground than those on the overlying leaves, because underlying leaves can prevent beetles from reaching the ground (Matsubara & Sugiura, 2018). Therefore, to evaluate the effects of underlying leaves on dropping behaviour, we did not use beetles found on the lowest leaves in our experiments.
Initially, we recorded whether beetles exhibited dropping behaviour in response to artificial stimuli. Dropping behaviour was defined as departure from the feeding position attributable to gravity. When insects dropped from the host plant leaves, we also recorded the place to which they dropped (i.e. leaves or ground/ water) or whether they flew before landing ( Fig. 1). Dropping behaviour is frequently accompanied by thanatosis (Humphreys & Ruxton, 2018); for example, some adults of a leaf beetle species fold their legs and antennae to feign death as they drop from host plants (Matsubara & Sugiura, 2018). We included such death feigning as dropping behaviour in this study, because it was difficult to observe whether each leaf beetle feigned death during dropping. We also recorded other defensive behaviours, such as emission of visible liquids, flying, jumping and running.

data analySiS
The defensive behaviours of some beetle species were observed at several sites. However, site effects were not included in following models because our preliminary analysis detected no site effects on dropping behaviour.
Generalized linear mixed models (GLMMs) with binomial error distribution and logit link were used to determine the effects of host plant growth form on dropping and other behaviours of leaf beetles. As response variables, we used whether each individual exhibited dropping behaviour or not (one or zero), whether each individual dropped to the ground or not (one or zero), whether each larva exhibited chemical defence or not (one or zero) and whether each adult flew during dropping or not (one or zero). In all analyses, host plant growth form (woody or herbaceous) was used as a fixed factor. Leaf beetle tribe and species were fitted as nested random effects to account for phylogenetic constraints (Hiraiwa & Ushimaru, 2017). All analyses were performed using the software R v.3.5.2 (R Core Team, 2018). The GLMMs were run using the lme4 package 1.1.13 (Bates et al., 2017).

RESULTS
A total of 286 leaf beetle larvae (23 species from six tribes) on host plants (23 species from 16 families) were investigated in field conditions; 15 and eight beetle species fed on the leaves of woody and herbaceous plants, respectively (Appendix 1  Table 2). Host plant growth form did not significantly affect dropping behaviour in leaf beetle adults (Fig. 3A, B; Table 2). In addition, significantly more adults flew after dropping from woody (9.4%) than from herbaceous plants (2.9%; Fig. 3C; Table 2). Beetles that did not drop ran (22.1%), jumped (14.2%), flew (7.3%) or secreted visible chemical liquids from their bodies (0.6%) in response to stimuli. Other beetles (1.9%) used their legs to hold tenaciously to leaf surfaces.

DISCUSSION
Previous studies have investigated the effects of host plants on anti-predator defences in phytophagous insects (Denno et al., 1990). However, few studies have focused on host plant architecture as a factor that affects dropping behaviour in leaf-feeding insects (Matsubara & Sugiura, 2018). In the present study, we conducted field experiments to test the prediction that dropping behaviour would be observed less frequently among insects that feed on woody plant leaves than among those that feed on herbaceous plant leaves. Leaf beetle larvae on woody plants exhibited dropping behaviour less frequently than those on herbaceous plants ( Fig. 2A; Table 1), supporting our prediction. However, this pattern was not detected in adults ( Fig. 3A; Table 2).

deFenSive StrategieS in leaF beetleS
Leaf beetle larvae exhibited chemical defence more frequently than dropping behaviour in our experiments ( Fig. 2C; Table 1). Thus, larvae frequently avoided leaving their feeding sites. Most larvae that dropped from host plants ultimately landed on the ground (Fig. 2B), necessitating a return to the host plant leaves by walking. Given that leaf positions on woody plants are generally higher than those on herbaceous plants, dropping from woody plants might be costlier compared with dropping from herbaceous plants. Therefore, larvae that feed on woody plants were more likely to avoid leaving their feeding sites than those that feed on herbaceous plants ( Fig. 2A). However, larvae that feed on woody plants did not use chemical defences more frequently than those that feed on herbaceous plants (Fig. 2C). Given that chemicals emitted by leaf beetle larvae can effectively repel predators such as ants and spiders (Sugawara et al., 1979;Pasteels et al., 1982Pasteels et al., , 1988Kimoto & Takizawa, 1994), the larvae of many leaf beetle species prefer to adopt chemical defences. However, we might have overestimated the importance of chemical defences in larval leaf beetles because our study was limited to the larvae of 23 species (six tribes of Chrysomelidae).
Leaf beetle adults exhibited dropping behaviour more frequently than larvae (Figs 2A, 3A). However, the drop rates of adult beetles did not differ significantly between woody and herbaceous plants ( Fig. 3A; Table 2). Dropped adults frequently returned to host plant leaves by flying before landing on the ground, suggesting that the costs associated with dropping behaviour are lower for adults that can fly. The results of our previous study showed that larvae of the leaf beetle Phaedon brassicae Baly required more time than did adults to return to feeding sites (i.e. leaves) on host plants (Matsubara & Sugiura, 2018). Given that the costs associated with dropping to the ground are lower for adults than for larvae, adults might not avoid leaving their host plants in response to predator attacks. Alternative defensive behaviours of adults included running, jumping and flying. Although very few adults emitted visible chemical liquids from their bodies, adults of some leaf beetle species reportedly secrete small amounts of defensive chemicals (Pasteels et al., 1988). Thus, our study might have underestimated the importance of chemical defences in adult leaf beetles.

eFFectS oF plant growth Form on dropping behaviour
We observed effects of host plant growth form (i.e. woody or herbaceous plants) on dropping behaviour  in leaf beetle larvae ( Fig. 2A) but not in the adults (Fig. 3A). Given that few leaf beetle larvae can move among host plants, host plants can strongly influence their survival. The costs associated with dropping from woody plants might be higher than those associated with dropping from herbaceous plants, and differential costs between woody and herbaceous plants should be higher in leaf beetle larvae than in flying adults. Dropping behaviour in leaf beetle larvae might have evolved via at least one of four potential processes: (1) species that feed on herbaceous plants acquiring dropping behaviour; (2) species that feed on woody plants losing dropping behaviour; (3) species that exhibit dropping behaviour shifting from woody to herbaceous host plants; or (4) species that do not exhibit dropping behaviour shifting from herbaceous to woody host plants. The drop rate can vary among larvae of the same leaf beetle species (Appendix 1; Matsubara & Sugiura, 2018), which suggests that this behaviour is frequently acquired or lost among species (i.e. hypothetical process 1 or 2). We did not observe whether oviposition site preferences (e.g. woody or herbaceous plants) varied among adults of the same leaf beetle species in the present study. However, oviposition preferences for host plant species reportedly vary among adult females of the same leaf beetle species Vencl et al., 2011, suggesting that shifts in host plant occur frequently among species (i.e. hypothetical process 3 or 4). Molecular phylogenetic analyses (e.g. ancestral reconstruction) of defensive behaviours and host plants would help to elucidate the selective processes promoting the evolution of dropping behaviour in leaf beetles.

concluSion
The results of this study indicate that host plant growth form affected the evolution of dropping behaviour in leaf beetle larvae but not in winged adults. However, the evolution of dropping behaviour in phytophagous insects can be influenced by other factors, such as host plant range. When generalist species drop from the host plant to the ground, they can find other plants to eat more easily than can specialist species (Bernays & Graham, 1988). Therefore, the costs associated with dropping from host plants might be higher for specialists than for generalists. In addition, primary defences, such as body colour, might affect the evolution of dropping behaviour in phytophagous insects. Insect species with cryptic body colour are less easily detected by predators compared with those having aposematic body colour when they drop to the ground. Consequently, dropping from host plants might have evolved more frequently in cryptic species than in aposematic species. Further studies are needed to test these effects on the evolution of dropping behaviour in phytophagous insects.