Introduction

Autotomy is the self-amputation of a limb or other appendage, often as a reflexive action and along a predetermined breakage plane (Juanes and Smith 1995; Bateman and Fleming 2005). Autotomy has evolved independently many times, and has been observed in many taxa (Bateman and Fleming 2005). The loss of a part of the body incurs costs related to survival and reproduction. It is assumed autotomy may also generate long-term costs as a result of compromised locomotion and ability to escape from predators (such as reduced mobility, balance, sprint speed, or locomotory ability), reduced ability to forage or mate, lowered social status, loss of fat stores through loss of tissue, or death (Stoks 1999; Bateman and Fleming 2005). The amount of evidence related to the costs of autotomy continues to mount (Dixon 1989; Juanes and Smith 1995; Stoks 1999; Bateman and Fleming 2005, 2006, 2008, 2009; Clause and Capaldi 2006; Maginnis 2006; Fleming et al. 2007; Slos et al. 2009).

In insects, parasitoid attacks occur in various taxa (Godfray 1994; Hawkins et al. 1997) and are one of the greatest biotic sources of mortality for insect herbivores (Hawkins et al. 1997). Some host traits are under selection pressure by parasitoids (Gross 1993; Zuk et al. 1993, 1995; Wagner 1996). However, the effects of parasitoids on autotomy in hosts are unknown. Miura and Ohsaki (2007) found that autotomized adult Parapodisma subastris and P. tanbaensis (Orthoptera: Catantopidae) that had been attacked by the parasitoid fly Blaesoxipha japonensis (Diptera: Sarcophagidae) experienced higher mortality rates than were observed in intact adults in their natural habitat. B. japonensis larva egress through a hole made in the dorsal region of the intersegmental membrane between the head and thorax of a host grasshopper, causing host death. Thus, increased host mortality caused by B. japonensis could be one of the costs of autotomy for these two grasshopper species in their natural habitat. Autotomized adults may be less effective in attempting to remove the parasitic larvae deposited on their body surface than intact adults because the autotomized adults lacked hind leg(s) (Miura and Ohsaki 2007) and the host grasshopper on which B. japonensis deposits larvae may use its hind leg(s) to dislodge the deposited larvae (Danyk et al. 2000). Until now, we have not examined to see if this was one of the costs in each grasshopper species and whether there was an interspecific difference in the cost between P. tanbaensis and P. subastris adults.

Interspecific comparison has been extensively used in research in the fields of ecology and evolution (Endler 1986; Harvey and Pagel 1991). In particular, interspecific comparisons between closely related species allow the identification of factors and mechanisms underlying a biological phenomenon because of the basic assumption that closely related species share more phenotypic similarity and more similar responses to environmental factors than less closely related species (Endler 1986; Harvey and Pagel 1991). That is, the trait differences among closely related species are likely to be caused by fewer differences of factors or mechanisms among such species compared with the trait differences among more distantly related species. Moreover, because P. subastris and P. tanbaensis in our study are distributed sympatrically, they may share the same biotic and abiotic factors more frequently in their natural habitat than would allopatrically distributed and closely related species (Vogelweith et al. 2014). Thus, the interspecific comparison in our study between two closely related species, P. tanbaensis and P. subastris, should be advantageous during attempts to estimate the underlying factors or mechanisms generating their differences in autotomy.

In this study, we investigated the relationship between host mortality caused by B. japonensis and the condition of the hind legs in P. tanbaensis and P. subastris adults in their natural habitat. We simultaneously collected adults of both grasshopper species from the same locations and reared them in our laboratory. We allowed the collected grasshoppers to complete their lifespan and recorded the occurrence of death caused by B. japonensis. We examined whether host mortality caused by B. japonensis in the autotomized adults was higher than that in intact adults in each grasshopper species. Finally, we discuss the factors or mechanisms causing the difference in the host mortality between intact and autotomized adults and between the two species.

Methods

Study species

The adults of the two grasshopper species, P. tanbaensis and P. subastris, are brachypterous (Miyatake and Kanô 1992). Both species are univoltine, overwinter as eggs, and feed on a wide range of dicotyledonous plants (Miyatake and Kanô 1992; Miura and Ohsaki 2004). P. tanbaensis occurs only in the Kinki district, whereas P. subastris occurs in the Kinki and Tohkai districts, Honshu Island, Japan (Miyatake and Kanô 1992). In Kyoto in the Kinki district, both adult grasshopper species occur from July to October (Miura and Ohsaki 2007) and are distributed sympatrically (Miyatake and Kanô 1992; Miura 2003; Miura and Ohsaki 2007).

In orthopteran insects, a lack of hind leg(s) has been reported to occur in natural habitats (Dixon 1989; Bateman and Fleming 2005; Hochkirch et al. 2007; Miura and Ohsaki 2007), and juveniles may be able to cover the cost of loss of a leg by regenerating it in successive instars. However, limb re-generation appears to be rare or incomplete in orthopteran insects (Bateman and Fleming 2008). P. tanbaensis and P. subastris adults are often observed to lack a hind leg (Miura and Ohsaki 2007), and leg regeneration has not been observed (Miura, personal observation). Because the hind legs are used for jumping in orthopteran insects (Bennet-Clark 1990; Bateman and Fleming 2005) and because Parapodisma adults are brachypterous, the hind legs of Parapodisma grasshopper adults might be more important functionally than those of macropterous grasshopper adults for moving away from natural enemies that they encounter.

The parasitoid fly B. japonensis occurs in Japan, China, and Far East Russia (Pape 1994). Many Blaesoxipha species parasitize various grasshopper species (Pape 1994; Miura 2003; Miura and Ohgushi 2010). B. japonensis is a gregarious parasitoid that parasitizes fifth-stadium Parapodisma nymphs and adults of both P. tanbaensis and P. subastris in Kyoto (Miura 2003).

B. japonensis larvae egress through a hole made in the dorsal region of the intersegmental membrane between the head and thorax of the dead grasshopper host (Miura, personal observation). The internal organs of the host body tissue become liquefied immediately before the emergence of B. japonensis (Miura, personal observation). The processes of host use by other Blaesoxipha flies (Rees 1973) suggest that the larvae of B. japonensis can be placed on the host body surface and then feed on the hemolymph and/or fat body after they burrow into the host body (Miura and Ohsaki 2014).

Sampling and rearing methods

Once a month, we captured Parapodisma adults in 2001 and 2002 from July to October in each of three locations: Kibune, Kumogahata, and Mt. Hiei in Kyoto. After looking for Parapodisma adults and carefully capturing them using a net, we identified the species and measured the prothorax length in our laboratory with a caliper as the index of body size. The presence or absence of the hind leg(s) of adults of both Parapodisma species was recorded. The grasshoppers were captured almost individually. In the net, there was no hind leg of grasshopper after being captured. So, it seemed that they had no trouble in losing their hind legs during the sweeping attempts we made to capture them. Captured Parapodisma adults were assumed to have been parasitized after adult eclosion, when B. japonensis larvae egressed. It took less than 8 days for B. japonensis larvae to egress from the captured hosts (Miura 2003; Miura and Ohsaki 2014); therefore, the B. japonensis adults appear to emerge quite quickly after infestation of the host. Miura and Ohsaki (2007) explained the details of the sampling and rearing procedures used in our study.

Statistical analysis

We set the value for “injury” in intact hosts as zero, the value in hosts lacking one hind leg as one, and the value in hosts lacking both hind legs as two, as a ranked variable. Host body size was a continuous variable. The other factors except “month” were categorical variables. Month was a continuous variable because we examined to see if the host mortality rate caused by B. japonensis might have changed with late season capture. We examined the sampled data of July as those of 15 days, the sampled data of August as those of 45 days, and the sampled data of September as those of 75 days for these statistical analyses. Because few hosts were captured in October, the data from October were combined with those from September for statistical analysis. The categorical choice of either a dead host caused by the emergence of B. japonensis or a surviving host of each individual was treated as the dependent variable. We used JMP version 9.03 for all statistical analyses (SAS Institute, Cary, NC, USA, 2011).

We assessed the relationships between the host mortality caused by B. japonensis and other factors using a generalized linear model (GLM) with a binomial error distribution and a logit link function. In this analysis we selected the GLM with overdispersion tests and intervals and with Firth bias-adjusted estimates in the model specification (SAS Institute, Cary, NC, USA, 2011). We assessed the data in each host species separately because we did not examine the interaction effect between the injury and the other factors because degrees of freedom were lacking. First, we examined a model that had six main effects (year, month, location, body size, host sex, and injury) and 15 first-order interaction effects between the main effects in P. tanbaensis. Second, we examined a model that had six main effects (year, month, location, body size, host sex, and injury) and 10 first-order interaction effects between the main effects (except injury) in P. subastris because degrees of freedom were lacking.

Results

In 2001, 76 P. tanbaensis and 40 P. subastris were collected in Kibune, 16 and 21, in Kumogahata, respectively, and 40 and 34 on Mt. Hiei, respectively. In 2002, 85 P. tanbaensis and 34 P. subastris were collected in Kibune, 19 and 4 in Kumogahata, respectively, and 15 and 9 on Mt. Hiei, respectively. These totaled to 251 P. tanbaensis and 142 P. subastris individuals.

In P. tanbaensis, injury type was not significant, while there was a significant interaction effect between year and body size, and between host sex and body size (Table 1). In P. subastris, injury type was significantly related to host mortality (χ 22  = 7.407, P = 0.025) and there were no significant interaction effects (P > 0.60) for any factors. That is, if one considers the factors of year (Y), month (M), host sex (H), and body size (B), as well as for the interactions of Y × M, Y × H, Y × B, M × H, and M × B, then df = 1, χ 2 = 0.000, and P = 1.000. Similarly, for location (L), L × B, M × L, and Y × L, df = 2, χ 2 = 0.000, and P = 1.000. The exceptions were injury type (I), L × H, and H × B, where df = 2, 2 and 1, χ 2 = 7.407, 0.426 and 0.204, and P = 0.025, 0.808 and 0.651, respectively.

Table 1 Relationship between various factors and mortality caused by Blaesoxipha japonensis in Parapodisma tanbaensis adults in a generalized linear model
Full size table

In P. tanbaensis, 7.3 % of the 232 intact adults and 10.5 % of the 19 adults lacking one hind leg were killed by B. japonensis, respectively (Table 2). In P. subastris, 1.6 % of the 128 intact adults, and 7.7 % of the 13 adults lacking one hind leg, and 100 % (one of one adult) lacking both hind legs were killed by B. japonensis (Table 2). Thus, autotomized adults incurred higher mortality caused by B. japonensis than intact adults of P. subastris, whereas those of P. tanbaensis did not incur higher mortality.

Table 2 Number of adults lacking a hind leg(s) in Parapodisma tanbaensis and P. subastris adults captured at each site, and the mortality caused by Blaesoxipha japonensis in the two grasshoppers
Full size table

Discussion

Autotomy is a widespread anti-predator behavior with future costs including reduced mating success and survival that also provides an organism with obvious escape benefits (Fleming et al. 2007; Slos et al. 2009). The present study suggests that P. subastris adults lacking hind leg(s) were more frequently killed by B. japonensis than intact ones in their natural habitat, and that host mortality caused by B. japonensis could be one of the future costs of autotomy in P. subastris adults. Autotomy is considered to have an effective escape function against attack by predators (Bateman and Fleming 2005), and recent studies suggest that defenses that are effective against predators are not as effective against parasitoids (Gentry and Dyer 2002; Barbosa and Caldas 2007; Smilanich et al. 2009). Accordingly, defense against parasitoids might be less effective in autotomized adults when they are compared with intact adults. That is, an effective predator avoidance response involving autotomy might have resulted in enhanced susceptibility to the parasitoid, although there have been no reports of differences in behavior or physiology between intact P. subastris adults and ones lacking hind leg(s) during the process of host use by B. japonensis.

The results of the present interspecific comparison between the two grasshopper species suggests that some mechanism(s) and factor(s) affecting the process of host use by B. japonensis from the search for a host until emergence from the host differ between the two host species. The condition of the hind legs in P. subastris adults was related to the adult mortality, whereas that in P. tanbaensis adults was not (Table 1). No significant factor or interaction effect was observed in P. subastris, whereas significant interaction effects were found between year and body size, and between host sex and body size, in P. tanbaensis (Table 1). Thus, some factor(s) related to host mortality caused by B. japonensis, such as physiological or behavioral trait(s), or ecological factor(s), might differ between these two host grasshopper species. Below, we will discuss two such possible factors: immune responses in the host grasshopper, and vegetation composition and structure in the habitat.

While immunity is often thought of in the sense of genetic or physiological immunity, insects often exhibit behavioral immunity (Roode and Lefèvre 2012). First, P. subastris adults lacking hind leg(s) may have weaker behavioral immune responses against B. japonensis than adults of P. tanbaensis or than intact adults in either species. Immune response often differs between closely related species and is dependent on various factors (Zuk et al. 2004) such as age (Chun et al. 1995; Rolff 2001; Doums et al. 2002; Hillyer et al. 2005), spatial location (Berggren 2009; Vogelweith et al. 2013a), and sex (Rolff 2001; Schmid-Hempel 2005). In our study, the adult age, spatial location, and annual variations in adults of both grasshopper species appeared to not be related to host mortality caused by B. japonensis, except for the significant interaction between year and body size in P. tanbaensis (Table 1). No indication of a sexual difference in the immune responses was observed in P. subastris adults. However, some difference may exist in P. tanbaensis between the sexes in the immunological responses, depending on host sex and body size (Table 1); for example, as Miura and Ohsaki (2014) reported, B. japonensis caused less mortality in larger female adults, whereas the mortality was independent of male body size. Stronger behavioral (Lee et al. 2008) and physiological (Vogelweith et al. 2013b) immune responses in larger lepidopteran larvae and physiological immune responses in larger house crickets (Ryder and Siva-Jothy 2001) have been reported, and further research will be needed to clarify whether physiological and behavioral immune response differences cause the body-size-dependent sexual differences in the host mortality caused by B. japonensis.

Autotomized damselfly (Lestes viridis) larvae show possibly reduced immune responses compared with intact ones, because of reduced phenoloxidase activity (Slos et al. 2009). Phenoloxidase is a key component of immune responses in insects, and catalyses melanin formation to encapsulate pathogens (Slos et al. 2009). Autotomy in insects could also lead to lower foraging by the autotomized insects (Stoks 1998, 1999), low nutritional status (Moret and Schmid-Hempel 2000; Siva-Jothy and Thompson 2002), as well as energy limitation and the reallocation of resource reserves, resulting in reduced immune responses. Thus, phenoloxidase activity in autotomized grasshoppers could also be reduced, although perhaps transiently (Slos et al. 2009). Therefore, even if B. japonensis larvae infested a host grasshopper before the loss of a hind leg by autotomy, reduced host immune responses could lead to a higher survival rate and/or greater growth of B. japonensis larvae in a subsequently autotomized host than in an intact one. Therefore, autotomized adults could incur higher mortality caused by B. japonensis than intact ones.

Second, B. japonensis may more frequently encounter P. subastris adults lacking hind leg(s) than intact adults or adults of P. tanbaensis. The vegetation composition and structure can affect parasitoid prevalence in grasshoppers (Laws and Joern 2012). Searching for better foodplants, microclimate, and mates can differ between sexes and between species (Krasnov et al. 2005; Walker and Allen 2014). Sometimes, the parasitism rates in closely related oligophagous insect herbivores depend on their host plants (Ohsaki and Sato 1990, 1994; Sisterson and Gould 1999; Oppenheim and Gould 2002; Pélissié et al. 2009). Habitat segregation or difference in plant use among closely related polyphagous species can occur although their food habits are polyphagous, partly because of reproductive interference among them (Gröning et al. 2007). Additionally, species-specific nutritional needs among closely related polyphagous insect herbivores (Behmer and Joern 2008) can cause them to prefer different microhabitats, although a predation risk can alter species-specific nutritional needs (Hawlena and Schmitz 2010). The vegetation composition (total grass biomass) and structure (canopy height and spatial variability in canopy structure) can affect the behaviors of grasshoppers, because grasshopper species density differs according to the characteristics of vegetation (Joern 2004, 2005). In particular, vegetation structure can affect the success of grasshoppers attempting to escape from predators and parasitoids, and may also affect their ability to regulate body temperature and to maintain water balance (Joern 2004). Thermoregulation is a critical activity that affects food processing, nutrient acquisition rates, and the potential to fend off pathogens (Joern 2004).

In the present study, the seasonal, spatial, and annual variation of the vegetation in the habitat may not have been related to the host mortality caused by B. japonensis in adults of either grasshopper species, except for the significant interaction between year and body size in P. tanbaensis (Table 1). Rather, sex and body size are importantly related to the behavior of intact and autotomized grasshoppers (Table 1). Autotomized insects must respond to two conflicting demands, the needs to forage and to escape from predators (Stoks 1999). After autotomy, larvae of the damselfly Lestes sponsa changed their behaviors and used a complex microhabitat to seek refuge from predators (Stoks 1999). Thus, Parapodisma grasshopper adults in our study might have differentially selected microhabitats for escape space and for thermoregulation depending on the body size of each species, as well as on hind leg condition, and sex. These factors could lead to autotomized P. subastris adults being more frequently attacked and killed by B. japonensis than intact ones.

In addition, it has recently been proposed that host plants affect the level of immunity of insect herbivores (Ojala et al. 2005; Karimzadeh and Wright 2008; Vogelweith et al. 2011), as well as their non-immunological defenses (Parker et al. 2011), such as plant-mediated resistance of insect herbivores against pathogens and parasitoids. Insect herbivores often defend themselves against their enemies not only by feeding on high-quality host plants (Ojala et al. 2005; Karimzadeh and Wright 2008; Vogelweith et al. 2011) but also by storing secondary metabolites obtained from their host plants in their bodies (Sword 2001; Singer and Stireman 2003; Smilanich et al. 2011). Thus, the difference in the food plants between intact and autotomized adults or between the two grasshopper species studied here could have influenced the immune responses against B. japonensis. We should perform combined studies of the host insect’s immune response and the structure or composition of plant species in the habitat to examine whether plant-mediated immune traits in adults of Parapodisma species could affect the mortality in them caused by B. japonensis.

Little is known about the effect of parasitoids on autotomized insects. However, our methods using Parapodisma and Blaesoxipha species are useful for making a detailed interspecific and intraspecific comparisons among Parapodisma species to aid in the investigation of the effects of parasitoids on autotomized hosts and for identifying the environmental factors and ecological mechanisms responsible for the patterns observed in the present study. Fourteen Parapodisma species occur in Japan (Miyatake and Kanô 1992) with interspecific (Miyatake and Kanô 1992) and intraspecific (Tatsuta et al. 2001) variation in their respective morphological traits. Some parasitoids, including Blaesoxipha species, have been reported to use grasshoppers as hosts in Japan (Shima 1999; Miura 2003; Miura and Ohgushi 2010). Thus, the experimental tests and field data in our study system can be helpful for clarifying the function of autotomy in these species’ natural habitat, and for stimulating further research on the significance of ecological interactions in host defense against parasitoids.