Substrate Selection for Pit Making and Oviposition in an Antlion, Myrmeleon Bore Tjeder, in Terms of Sand Particle Size

Entomological Science (2005) 8, 347–353

ORIGINAL ARTICLE

Substrate selection for pit making and oviposition in an antlion, Myrmeleon bore Tjeder, in terms of sand particle size

Toshiaki MATSURA1, Yoshitaka YAMAGA2 and Madoka ITOH3 1Department of Biology, Kyoto University of Education, Fushimi-ku, 2Asia Kohsoku Corporation, Osaka and 3Doshisha Junior High School, Sakyo-ku, Kyoto, Japan

Abstract The larvae of the pit-making antlion Myrmeleon bore Tjeder live in open sand in riverbeds with a substratum consisting of various particle sizes. We analyzed the spatial distribution of their pits in a sandy floodplain to determine their larval and adult responses to the heterogeneous substrate. The spatial distribution pattern of their pits had an aggregated distribution, and there was a significant positive correlation between pit density and the ratio of medium-size sand particles to total weight of sand. We examined the size of sand particles selected in the larval pit-building behavior and the oviposition behavior of the adult. Both larvae and adults selected medium-size sand particles. The larvae of M. bore are relatively sedentary predators and rarely move great distances. Thus, the present results suggest that habitat selection by adult females is a major factor causing the aggregative distribution of the pits. Key words: floodplain, habitat selection, Myrmeleontidae, Neuroptera, predator, spatial distribution.

INTRODUCTION (Lucas 1982; Allen & Croft 1985). Pitfall traps built in fine sand grains have a steep slope (Allen & Croft 1985) Most predacious insects capture prey directly using their and prey (ants) take a long time to escape from the pit mandibles (e.g. dragonfly larvae and lacewing larvae) or (Lucas 1982; Allen & Croft 1985; Botz et al. 2003). their specialized forelegs (e.g. praying mantids and giant Some studies have revealed that antlion larvae select fine waterbugs), but a few are predators that adopt hunting sand when they are presented with fine and coarse sand tools such as pitfall traps and webs. Larvae of several at the same time (Morisita 1952; Loiterton & Magrath genera (e.g. Myrmeleon, Hagenomyia and Euroleon) in 1996; Farji-Brener 2003). Although there have been Myrmeleontidae (antlions) build an inverted conical pit many behavioral and ecological studies of antlions on the sand surface. The larval antlions wait for prey at in the laboratory in relation to the characteristics of the base of a conical pitfall trap, opening their mandi- sand, how antlions in the field respond to the hetero- bles. Unlike non-pit-building antlion larvae, pit-building geneity of the substratum where they live has been antlions can move only backward on the sand when poorly investigated. they construct their pits or relocate them. The larvae of the antlion Myrmeleon bore Tjeder In general, capture efficiency is important for preda- build a pit actively in open sand such as in seaside dunes tors to enhance their fitness. The capture efficiency of and in floodplains. The substratum of a floodplain is antlion larvae is said to be affected markedly by the size composed of particles of various sizes, while the sand of the sand particles where their pits are constructed of seaside dunes is composed of particles of a relatively uniform size. When they live in floodplains, the larvae should make their pits in areas with adequate sand Correspondence: Toshiaki Matsura, Department of Biology, particle size to increase their foraging efficiency. The Kyoto University of Education, Fishimi-ku, Kyoto, 612-8522 previous studies have revealed that the larvae rarely Japan. Email: [email protected] relocate their pits, even under severe fasting in the lab- Received 27 April 2005; accepted 7 July 2005. oratory (Matsura 1987; Matsura & Murao 1994). If

© 2005 The Entomological Society of Japan T. Matsura et al. such sedentary behavior is adopted in nature, habitat Life history of Myrmeleon bore selection at oviposition by the female becomes an Myrmeleon bore is a univoltine species (Matsura et al. important factor governing larval spatial distribution. 1991). Females lay eggs during mid-summer. First instar In the present study, we analyzed the spatial distribution larvae begin to emerge from late summer. The larval of M. bore pits in a sandy floodplain in relation to sand development rate mainly depends on feeding rate, and particle size, and we examined in the laboratory sand first to third instar larvae are found before hibernation. particle size preference by the larvae for pit construction Overwintered larvae pupate from June to July and the and also by the adult females for oviposition. period of adult emergence is from mid-June to early August. MATERIALS AND METHODS Field survey Research area We counted the number of pits in each quadrat once a The Kizu River flows through southern Kyoto Prefec- month (except three times in May) from April to August ture in Japan and forms many sandy floodplains in the in 1998. We did not dig out the antlion larvae from the middle course of the river. The research area was located sand, to avoid disturbing them. Therefore, their precise within a bush growing on one of the floodplains age structure was unknown, but the population was ° ° (38.8 N, 135.9 E). We had found many pits constructed assumed to consist mainly of second or third (last) instar by the larvae of M. bore in an open sandy field sur- larvae during the period from spring to early summer, rounded by grass before starting this research. This open based on the pit diameters and previous reports (e.g. sand was located at the top of a mound with a shape Matsura et al. 1991). All the antlion larvae were dug × that was nearly rectangular (4 16 m). We established out on 21 August 1998 and the number of each instar × 39 quadrats (each 1 1 m) within this sandy area larvae per quadrat was recorded. (Fig. 1). The quadrats were classified into the following three groups based on the pit density observed on 28 May 1998: None, a quadrat with no pits; Low, a quadrat with one to nine pits; and High, a quadrat with 10 or more pits. We selected five quadrats arbitrarily from these three groups (15 samples) and collected surface sand (approximately 150 mL) at the center of each quadrat. The sand samples were separated with an elec- tric sieve into five particle size classes (fine, <0.25 mm in diameter; medium, 0.25–0.5 mm; coarse, 0.5–1 mm; very coarse, 1–2 mm; and gravel, >2 mm); this defini- tion of sand particle size was adopted also in the laboratory experiments. We weighed the sand after drying it at approximately 150°C for 24 h, and calculated the ratio of sand weight of each particle size class to the total sand weight. The proportion of medium-size sand particles was arcsine transformed and analyzed by a one-way analysis of variance (ANOVA) to determine the significance of difference among the three groups defined by the pit density.

Sand particle size selected by larvae In May 1999, we examined the size of sand particles the larvae preferred as a pit-building site in a laboratory experiment under conditions of 25 ± 1°C, 50–60% rel- Figure 1 Monthly changes in the spatial distribution pattern ative humidity and 12 h light : 12 h dark (LD 12:12). A of the pits in the research area. “d”, number of pits per quadrat third instar antlion was put in a container containing (1 × 1 m). one of the above size classes of sand or non-sieved sand

348 Entomological Science (2005) 8, 347–353 © 2005 The Entomological Society of Japan Substrate selection in antlions for 1 week to adapt to the environment. The larva was oviposition experiment, we prepared 12, 13, 13 and 13 then placed in the center of a container (200 × cylindrical containers (90 mm in diameter and 50 mm 130 × 80 mm3) that contained two different kinds of high) and filled them with fine, medium-size, coarse and sand side by side to a depth of 40 mm: one was the sand very coarse sand, respectively, to a depth of 30 mm. One in which the larvae had been kept and the other was egg per container was placed on the sand surface and medium-size sand. covered with sand of the same particle size to approximately 5 mm. As the duration of egg in M. bore is Sand particle size selected by female approximately 2 weeks in midsummer (Matsura et al. adults in oviposition 2001), we examined 3 weeks after setting the experi- Females of M. bore flying into oviposition sites were ment whether the larvae had made pits or not. When caught at night and were brought to the laboratory. no pits were made, we observed the eggs under a ste- Because it is very difficult to find their eggs laid in sand, reoscopic microscope to determine whether they had we used the following method. We laid drawing paper hatched or not. This experiment was carried out under at the bottom of a container (200 × 130 × 80 mm3) and ambient room temperature. spread sand uniformly to a depth of 5 mm over it. The depth of eggs oviposited freely by M. bore is approximately 7.5 mm (Matsura et al. 2001). Accordingly, in RESULTS this experimental design, the inserted tip of the female abdomen would reach the paper at the bottom of the Spatial distribution of pits container and the oviposited eggs would adhere to it. The total number of pits in the research area varied The base of each container was divided into halves, and between 77 and 154 during the period from April to sand with a different particle size was put into each half. July (Table 1). Because the larvae of M. bore cease con- One female was released into each container and the structing their pits in wet sand, the number of observed containers were kept in a dark room. The next morning, pits varied partly depending upon when we carried out the sand was gently removed and the number of eggs the census since the last rain. The decrease in the num- adhering to the paper was counted. Three combinations ber of pits from June onward is probably due to pupa- of sand (medium and fine; medium and coarse; medium tion. The spatial distribution pattern of the pit was and very coarse) were prepared, and the total numbers analyzed by calculating Iδ-index according to Morisita of females used were 19, 22 and 4, respectively. Because (1959), which characterizes the spatial distribution as we could not obtain enough data for analysis in one uniform (Iδ < 1), random (Iδ = 1) or aggregated (Iδ > 1). season, we repeated the same experiment for 3 years in The values obtained were always more than 2, showing midsummer from 1999 to 2001, and used the pooled an aggregated distribution of the pits. data. The pit density tended to be lower in the northern and central quadrats (Fig. 1). In the field survey, few tracks Pit construction by the newly hatched larvae in of the larvae were found, and the correlation coefficient relation to sand particle size for pit density per quadrat on consecutive research dates To examine the pit construction rates of the newly was significantly high (Table 1), suggesting that larvae hatched larvae obtained from the eggs oviposited in the seldom move a great distance.

Table 1 Mean density per quadrat (1 × 1 m) and spatial distribution of the antlion pits on each census date Total no. Mean Maximum no. † ‡ Date pits density ± SD pits per quadrat Iδ-index r 30 April 77 2.0 ± 3.6 15 3.79 0.87* 7 May 122 3.1 ± 4.2 18 2.48 0.92* 21 May 116 3.0 ± 4.2 16 2.66 0.94* 28 May 154 3.9 ± 4.9 21 2.29 0.92* 9 June 96 2.5 ± 3.6 13 2.34 0.79* 9 July 104 2.7 ± 5.0 19 4.14 – *P < 0.0001 (F-test); †Morisita’s index of dispersion (see text for explanation); ‡correlation coefﬁcient between pit densities per quadrat on consecutive research dates.

Sand particle size and pit density There was a marked difference in the frequency distribution of sand particle size between the quadrats with a high pit density (High) and those with a low pit density (Low) or no pits (None) with their modes in the particle size classes of 0.25–0.5 mm, 0.5–1.0 mm and >2.0 mm, respectively (Fig. 2). The average proportion of medium particle size was significantly different among the three pit density levels (ANOVA, F = 100.22, P < 0.0001), with High quadrats having much higher proportion than Low and None quadrats (Table 2). The total numbers of pits occupied by first, second and third instar larvae that were dug out on 21 August were 480, 168 and 51, respectively. The average density of pits built by the first instar larvae on 21 August was also markedly higher in High quadrats than in None and Low quadrats. Namely, pits by the larvae of the next generation aggregated mainly in the quadrats where the large number of the previous generation had constructed pits. Sand particle size selected by the larvae Almost all of the larvae selected medium-size sand over fine or very coarse sand (Table 3). In the combination of medium-size sand and coarse sand, the results dif- fered with the size of sand particles larvae had been reared on before the experiment started. That is, the larvae reared in coarse or non-sieved (control) sand showed little bias in preference, while the larvae reared in medium-size sand tended to select medium-size sand. Sand particle size selected by adult females in oviposition In the combination of medium-size sand and very coarse Figure 2 Frequency distributions of the sand particle size in quadrats with (a) high pit density, (b) low pit density and sand, three out of four females oviposited in medium- (c) no pits. Each value is the average ratio of weight of sand size sand, whereas no females oviposited in very coarse of one particle size class to the total weight of sand per sand (Table 4). A female in each of experiment two sample.

Table 2 Proportion of medium-size sand particles at each density level classified by the pit density of the overwintered antlion larvae, and the pit density of the first instar larvae observed in late August No. quadrats Ratio of medium particle size (%)‡ No. first instar larvae per quadrat§ Pit density level† examined (mean ± SD) (mean ± SD) None 5 13.7 ± 3.6a 1.0 ± 1.3a Low 5 24.6 ± 8.9b 1.6 ± 1.9a High 5 60.9 ± 3.0c 67.0 ± 18.8b †Number of pits per quadrat (1 × 1 m) on 28 May 1998: Low, one to nine pits; High, more than 10 pits; ‡percent weight of medium-size sand particles to the total weight of sand; ‡,§different letters after SD in each column indicate significant difference at P = 0.05 (Tukey–Kramer’s test).

Table 3 Selection of sand particle size by antlion larvae Very coarse Sand used in rearing Fine sand Medium sand Coarse sand sand Control† Combination of sand‡ FMMCMCMVMC No. replications§ 13 (1) 19 (2) 17 (2) 5 (0) 17 (1) No. larvae selecting 1 11 11 6 7 8 5 0 8 8 Binomial test P = 0.032 P = 0.166 P = 0.500 P = 0.313 P = 0.598 †Field sand not sieved; ‡F, ﬁne sand particles; M, medium-size sand particles; C, coarse sand particles; V, very coarse sand particles; §ﬁgures in parentheses show the numbers of larvae constructing no pits.

Table 4 Sand particle size selected by females in oviposition Combination No. females Average no. Statistic (Wilcoxon Individual choice Statistic of sand oviposited eggs (±SD) signed rank test) of sand† (binomial test) Experiment 1 Fine 8 6.2 ± 6.9 P = 0.593 5 P = 0.500 Medium 11 9.5 ± 9.7 6 Experiment 2 Coarse 10 4.6 ± 4.3 P = 0.020 2 P = 0.033 Medium 11 12.5 ± 10.6 9 Experiment 3 Very coarse 0 0 P = 0.109 0 P = 0.875 Medium 3 5.7 ± 5.5 3 †Numerals are number of females who laid more eggs in each type of sand.

Table 5 Pit construction rates of newly hatched larvae in sand with various particle sizes Particle size No. eggs No. eggs No. pits Pit construction (%) of sand used hatched (A) constructed (B) (B/A × 100) Fine 12 11 8 72.7 Medium 13 11 10 90.9 Coarse 13 11 8 72.7 Very coarse 13 11 0 0

(coarse to medium) and experiment three (very coarse construction rate of first instar larvae was highest in to medium) laid no eggs. There was no significant dif- medium-size sand, while there was no significant differ- ference in the average numbers of eggs oviposited per ence in the pit construction rates among fine, medium- female between fine sand and medium-size sand, but 2.7 size and coarse sand (χ2 = 1.451, P = 0.48). times more eggs were oviposited in medium-size sand than in coarse sand. Nine of the 11 females oviposited DISCUSSION more eggs in medium-size sand than in coarse sand, but the number of females ovipositing in fine sand was not Prey density is in general one of the most important significantly different from that ovipositing in medium- factors governing the spatial distribution of arthropod size sand. predators. However, laboratory experiments have showed that M. bore larvae do not respond sensitively Pit construction by the newly hatched larvae in to temporal and spatial fluctuations of their prey avail- relation to sand particle size ability, and most larvae do not relocate their pits during Forty-four (86.3%) of the 51 eggs hatched and there a three-month fasting period (Matsura 1987; Matsura was no significant difference in the hatching rate with & Murao 1994). Pit-making antlion larvae must move the sand particle size (χ2 = 0.385, P = 0.94) (Table 5). backward beneath the sand surface when they relocate No larvae constructed pits in very coarse sand. The pit their pits. Movement in the sand requires high expendi-

Entomological Science (2005) 8, 347–353 351 © 2005 The Entomological Society of Japan T. Matsura et al. ture of energy. Larvae cannot assess prey availability of important. Our current knowledge about movement of a new foraging site unless they build a pit, but pit antlion larvae in the field is, however, limited. construction carries a high energy cost (Lucas 1985). The cost of pit relocation would often exceed its benefit. As a result, most larvae might adopt a sit-and-wait ACKNOWLEDGMENTS strategy, rarely relocating their pits. The spatial distribution of antlion larvae has been This study was supported by a grant from the Founda- reported to be strongly influenced by abiotic environ- tion for Riverfront Improvement and Restoration mental factors such as physical disturbance and degree (Tokyo). of dryness of the soil (Gotelli 1993; Day & Zalucki 2000; Gatti & Farji-Brener 2002). Of these abiotic factors, sand particle size appears to be one of the major REFERENCES factors affecting the spatial distribution of the pits con- Allen GR, Croft DB (1985) Soil particle size and the pit mor- structed by M. bore larvae. Farji-Brener (2003) found phology of the Australian ant-lions Myrmeleon diminutus that larvae of Myrmeleon crudelis Walker in coarse and M. pictifrons (Neuroptera: Myrmeleontidae). Austra- soil (50% of soil particles between 2 and 6 mm and lian Journal of Zoology 33, 863–874. 50% >6 mm) required much longer to start digging a Botz JT, Loudon C, Barger JB, Olapsen JS, Steeples DW (2003) pit and to finish constructing it than those in fine soil Effects of slope and particle size on ant locomotion: impli- (75% of soil particles <2 mm). It takes longer for ants cations for choice of substrate by antlions. Journal of the to escape from pits built on fine sand (Lucas 1982; Allen Kansas Entomological Society 76, 426–435. & Croft 1985; Ito 2001). Larvae of some antlion species Day MD, Zalucki MP (2000) Effect of density on spatial build smaller pits (Kitching 1984; Ito 2001) and often distribution, pit formation and pit diameter of Myrmeleon relocate their pits made on too fine sand (Ito 2001). acer Walker, (Neuroptera: Myrmeleontidae): patterns and process. Austral Ecology 25, 58–64. These findings suggest that too coarse or too fine sand Farji-Brener AG (2003) Microhabitat selection by antlion lar- is not suitable for their pit building behavior. The pref- vae, Myrmeleon crudelis: effect of soil particle size on pit- erence of M. bore larvae for medium-size sand for mak- trap design and prey capture. Journal of Insect Behavior ing pits, shown in the laboratory experiments, would 16, 783–796. reasonably explain their higher pit density in the field Gatti MG, Farji-Brener AG (2002) Low density of ant lion quadrats with a higher ratio of medium-size sand. larva (Myrmeleon crudelis) in ant-acacia clearings: high The substratum of floodplains in the Kizu River con- predation risk or inadequate substrate? Biotropica 34, sisted of particles ranging from pebbles to clay, and a 458–462. sandy area was found only at the top or the downstream Gotelli N (1993) Ant lion zones: causes of high-density pred- side of each mound in the floodplain. Because these ator aggregations. Ecology 74, 226–237. sandy areas are randomly distributed, the distribution Ito M (2001) Structure of pits and prey capture efficiency in antlion larvae (Masters Degree Thesis). Kyoto University pattern of larval populations of M. bore are probably of Education. (In Japanese.) established mainly by the oviposition of adult females Kitching RL (1984) Some biological and physical determinants in summer rather than by migration of larvae. There of pit size in larvae of Myrmeleon pictifrons Gerstaecker was a small difference in preference for sand particle (Neuroptera: Myrmeleontidae). Journal of the Australian size between the adults and the larvae of M. bore: the Entomological Society 23, 179–184. larvae preferred medium-size and coarse sand as a pit- Loiterton SJ, Magrath RD (1996) Substrate type affects partial building site, whereas the adults preferred fine and prey consumption by larvae of the antlion Myrmeleon medium sand as an oviposition site. Their habitats actu- acer (Neuroptera: Myrmeleontidae). Australian Journal ally consisted of variously sized sand particles, and the of Zoology 44, 589–597. proportion of fine (<0.25 mm) sand was very low. It Lucas JR (1982) The biophysics of pit construction by antlion appears that the proportion of medium-size sand, which larvae (Myrmeleon, Neuroptera). Journal of Animal Ecol- ogy. 30, 651–664. was preferred commonly at both stages, is an important Lucas JR (1985) Metabolic rates and pit-construction costs of factor governing spatial distribution of the pits. We sup- two antlion species. Journal of Animal Ecology 54, 295– pose that the selection of oviposition sites (or substrata) 309. by adult females could be the primary factor governing Matsura T (1987) An experimental study on the foraging spatial distribution of pits in a field, and that larval behavior of a pit-building antlion larva, Myrmeleon bore. movement (if they move at all), would not be very Researches on Population Ecology 29, 17–26.

Matsura T, Murao T (1994) Comparative study on the behav- lions living in seaside dunes: tolerance to high temperature ioral response to starvation in the three species of antlion (Neuroptera: Myrmeleontidae). Entomological Science 4, larvae (Neuroptera: Myrmeleontidae). Journal of Insect 17–23. Behavior 7, 873–884. Morisita M (1952) Habitat selection and evaluation of envi- Matsura T, Satomi T, Fujiharu K (1991) Control of the ronment: an experimental study on the density of ant-lion life cycle in a univoltine antlion, Myrmeleon bore larvae. Physiology and Ecology 5, 1–16. (In Japanese.) (Neuroptera). Japanese Journal of Entomology 59, 275– Morisita M (1959) Measuring of the dispersion of individuals 287. and analysis of the distributional pattern. Memoirs of the Matsura T, Ogasawara Y, Higashi M, Arahori Y (2001) Eco- Faculty of Science, Kyushu University, Series (E) 2, 215– logical characteristics of oviposition and eggs in the ant- 235.