Preference of antlion and wormlion larvae (Neuroptera: Myrmeleontidae; Diptera: Vermileonidae) for substrates according to substrate particle sizes

Sand-dwelling wormlion and antlion larvae are predators with a highly specialized hunting strategy, which either construct efficient pitfall traps or bury themselves in the sand ambushing prey on the surface. We studied the role substrate particle size plays in these specialized predators. Working with thirteen species of antlions and one species of wormlion, we quantified the substrate particle size in which the species were naturally found. Based on these particle sizes, four substrate types were established: fine substrates, fine to medium substrates, medium substrates, and coarse substrates. Larvae preferring the fine substrates were the wormlion Lampromyia and the antlion Myrmeleon hyalinus originating from desert habitats. Larvae preferring fine to medium and medium substrates belonged to antlion genera Cueta, Euroleon, Myrmeleon, Nophis and Synclisis and antlion larvae preferring coarse substrates were in the genera Distoleon and Neuroleon. In addition to analyzing naturally-occurring substrate, we hypothesized that these insect larvae will prefer the substrate type that they are found in. Specifically, we tested substrate preference in a choice experiment for four species of antlions. This was then compared to the substrate that the larvae were naturally found in. Larvae of the four antlion species were allowed to choose among eight substrates differing according to their particle size. Our findings show that the majority of M. hyalinus larvae built pits in the finest sand fraction, and the majority of N. microstenus larvae were found in a coarser fraction. The other two species, M. immaculatus and E. nostras, constructed majority of their pits in a medium sand fraction. These results support the hypothesis that neuropteran larvae prefer specific substrates for pitbuilding or ambushing prey, and that this – in combination with other abiotic and biotic factors – may drive selection of appropriate habitats.


INTRODUCTION
Organisms respond to biotic and abiotic environmental factors optimally to maximize their fitness. Natural selection will favour individuals who choose the most suitable habitat and therefore the greatest number of successful offspring can be raised. Habitat selection may be affected by the spatial variation in the distribution of resources, presence of other organisms, physiological state of the individual or abiotic factors, such as physical properties of the environment (Janetos, 1986;Rosenzweig, 1995). In the past, habitat selection theory mostly focused on biotic factors while abiotic factors (such as temperature) were often neglected. As shown for red flour beetles, temperature is clearly an important factor in habitat selection of ectotherms and should be considered whenever thermal differences exist between habitats (Halliday & Blouin-Demers, 2014).
In certain terrestrial animals, habitat substrate can play an important role. In desert scorpions, sand-dwelling antlions, and gerbils which are restricted to sandy substrates, selective pressures on choosing the appropriate substrate should be strong since sandy substrate may affect feeding strategy. It influences the rate and ease with which food can be found (Brownell & Farley, 1979; and plays a role in exposure to predators as well as the likelihood of successfully evading an encountered predator (Kotler et al., 2001).
Larvae of antlions (Neuroptera: Myrmeleontidae) are predators living in a variety of habitats, ranging from arboreal and rocky habitats, living in dust and detritus, and up to true psammophily (Mansell, 1996(Mansell, , 1999. Sanddwelling larvae, i.e. psammophily, culminate in the ability to modify the sandy environment into effective pitfall traps (Mansell, 1999).
While the majority of antlion species inhabit sandy substrate or loose soil, some build conical pitfall traps. These pit-building antlion larvae wait for their prey in the center of the pit. Using specialized traps is an advanced predatory strategy that evolved early in antlions (for review see Gepp & Hölzel, 1989;Scharf & Ovadia, 2006;Ruxton & Hansel, 2009;Gepp, 2010;Scharf et al., 2011). In contrast, the other sand-dwelling antlions are non-pitbuilders, which bury themselves in the sand waiting for prey on the surface.
The other sand-dwelling insects, wormlions (Diptera: Vermileonidae) resemble the pit-building antlions in their hunting strategy, offering a classical example of convergent evolution, as they belong to different insect orders. Their behavioural traits have rarely been studied (Devetak, 2008;Dor et al., 2014).

Substrate particle sizes in natural habitats
Substrates of natural habitats of one wormlion and 13 antlion species from 17 sites were examined (Table 1). Substrate samples were collected in natural antlion habitats using a 12 ml volume spoon. Sand or loose soil was collected with the spoon so that sand solely (in the case of non-pit-building antlions) or sand together with an antlion forming a pit was taken. Substrate from a pit and its surroundings was taken with the spoon three times so that the total volume of one sample was approximately 30-40 ml. Special care was devoted to sampling some species (e.g. Myrmeleon formicarius and M. fasciatus) living in very small patches of loose soil. In such patches we avoided sampling substrate in the surroundings of the pit where often soil quality drastically differed from that in the pit. Samples were transported in plastic bags separately and dried for two weeks at room temperature (26 ± 3°C) and relative humidity 30%.
Antlion larvae together with the substrate samples were collected during spring and summer months (from May to August), with an exception of Myrmeleon yemenicus larvae which were collected at the end of November (for details of its habitat -see Devetak et al., 2010). The majority of the larvae were reared to adults. For identification we used fundamental literature (for the larvae: Lucas & Stange, 1981;Badano & Pantaleoni, 2014;for adults: Aspöck et al., 1980). Antlion natural substrate is a mixture of grains of different particle sizes. Substrate particle sizes were determined by sieving dry substrate samples with standard sieves used previously (see Devetak et al., 2005) and fractions composing a certain substrate were expressed in weight percent (w%). Mass of the fractions was determined by weighing dry substrate samples using a scale (Kern and Sohn GmbH, Balingen, Germany; accuracy of 0.001 g).

Preference for different sands of various particle sizes in choice experiments
In order to test the preference of larvae of different antlion species to different substrates, four antlion species were chosen: one naturally occurring in fine to medium sands (Myrmeleon hyalinus: Cyprus, n = 27), two occurring in medium sands (Euroleon nostras: Slovenia, n = 25; Myrmeleon immaculatus; USA, n = 26), and one typically found in coarse substrates (Neuroleon microstenus: Slovenia, n = 26). Prior to experimental manipulation Wormlion larvae lack long mandibles, but other parts of their elongated body are specialized for prey capture, including the mouthparts and pseudopodium (Wheeler, 1930;Ludwig et al., 2001;Devetak, 2008;Dor et al., 2014). The pseudopodium is probably involved in the mechanical detection and seizure of prey. An abdominal comb with a row of long spines on the posterior part of the larval body is used to anchor it in the fine substrate (Ludwig et al., 2001;Dor et al., 2014). Studying foraging behavior of wormlions, Dor et al. (2014) found that wormlions' body mass and length were not tightly correlated with pit size, in contrast to antlions. Wormlions rarely co-occur with antlion larvae and it seems that antlions avoid areas inhabited by wormlions and vice versa. The larvae may simply prefer different substrates independent of the presence of heterospecifics (Devetak, 2008).

Myrmeleon immaculatus
De Geer, 1773 -PB G4 G5, G6 Medium the larvae were held in plastic cups filled with sand from their natural habitat, separated to avoid cannibalism. Room temperature was 26 ± 3°C. Only third instar larvae were used in the experiments and larval stages were determined by measuring body length and head capsule width . Choice experiments started with a habituation period during which antlions were fed with ants. One Lasius fuliginosus ant was fed to each antlion per day for five days after which five days of food deprivation occurred. This handling was conducted to standardize the hunger level of the larvae and ensure that the physiological state of all larvae was approximately the same at the beginning of the experiment. After these ten days, a single antlion was placed in the center of a container with eight different sand fractions.
To obtain fractions with different particle sizes, comminuted sand (Gradis d.o.o., Maribor, Slovenia; grain sizes 0-4 mm) was sieved. The sand was dried for two weeks at room temperature (26 ± 3°C); the moisture content of the substrate was less than 1.5% by weight.
We anticipated that antlions should be capable of distinguishing among substrate types. We tested this hypothesis by giving antlions the chance to choose one of eight particle-size fractions. Each antlion was treated only once. Round plastic containers (diameter 20 cm; height 11 cm) were partitioned using a plexiglass partitioner divided into eight parts of equal size. Each compartment was carefully filled with 100 ml of sand of a certain particle-size fraction. The fractions G1-G8 were used in the experiment. Substrates were partitioned as shown in Fig. 1. The plexiglass partitioner was then carefully removed and an antlion was dropped into the center. After 24 h we recorded the fraction in which an antlion was present simply by watching the antlion's reaction to an ant released on the sand surface, or in the case of pit-builders, where a pit was built.

Statistical analysis
The programs GraphPad Prism and SPSS 17.0 for Windows were used for statistical analyses and graphical presentation. A chi-square test was used to determine differences in proportion of antlions present in each substrate type after 24 h.

Distribution of antlion species in different types of substrates in natural habitats
We determined substrate particle size composition of substrates in natural habitats for thirteen antlion and one wormlion species from eleven countries and four continents (Table 1; Fig. 2). Loose soil or sand from various habitat types was investigated -e.g. sand beaches, sand dunes, sand deserts, stony semideserts, river banks, grasslands, and under rock overhangs. According to the presence of certain substrate particle sizes the following four substrate groupings were recognized (Table 1): (i) Fine substrates (fractions G1-3 ≥ 30 w% of the substrate).
The finest substrate is a mixture of powder and fine sand from Parc Grüel (Spain) with dominant fractions G2 and G1. The second finest sand is a Saharan sand from Douz (Tunisia) which is composed of G2 as a dominant fraction and G3 as important fraction (mean ± SD: 73.5 ± 4.5 w%; 15.1 ± 3.9 w%).
These results show that antlions occupy substrates of varying substrate size composition regardless of their prey capture strategy. Four of the investigated species (Nophis teillardi, Neuroleon microstenus, Distoleon tetragrammicus, and Synclisis baetica) were non-pit-builders, and the rest of the antlions were pit-builders.
The wormlion Lampromyia iberica was found in the finest sand with two dominant fractions G1 and G2 (mean ± SD: 23.4 ± 6.4 w%; 27.7 ± 3 w%; Barcelona, Parc Grüel, Spain). The second finest sand is a Saharan sand from Douz (Tunisia) which is composed of G2 as a dominant fraction (73.5 ± 4.5 w%). In the second finest substrate pit-building antlion Myrmeleon hyalinus was very common.
One non-pit-builder Nophis teillardi and the pit-building antlions M. hyalinus, M. fasciatus, M. inconspicuous, M. bore, Euroleon nostras and Cueta lineosa were all found naturally occurring in fine to medium substrates. In medium substrates with important finer fractions pit-builders (M. hyalinus, M. yemenicus, C. lineosa) and a non-pitbuilder (Synclisis baetica) were found. In medium substrates with important medium fractions two pit-builders were collected (M. hyalinus, M. immaculatus). Pit-building species were also found in medium substrates with important coarser fractions, namely M. yemenicus, M. formicarius and E. nostras. According to the substrate particle sizes in which larvae were naturally found, the investigated larvae can be classified into four categories (Table 2). A few larvae were recorded in only one type of substrate -in fine substrate (wormlion Lampromyia) or in coarse substrate (Neuroleon and Distoleon). The antlion M. hyalinus tolerates a wide range of finer substrates, from fine sands to medium substrates with finer (or medium) fractions. Euroleon nostras is an antlion species tolerating a wide range of medium substrates with important fine or coarser fractions.

Choice experiment examining substrate particle-size preference
Four antlion species were tested for substrate particlesize preference by giving antlion larvae the choice to choose one of eight substrate fractions.
Each larva dropped into the centre of the fractions moved in an apparently haphazard direction on, or just underneath, the substrate surface. This locomotion, called doodling (Topoff, 1977) in pit-builders, may assist in finding a suitable location for submerging or constructing the pit. During pit construction, antlions moved in large circles or loops, and some larvae, in the eight substrate-choice tests, encountered more than one particle-size fraction before constructing their pits in the most convenient one (Fig. 3). The larvae never built the pits between two neighbouring fractions but always chose one fraction.
Our results show that antlion species have a clear preference for a certain substrate type: M. hyalinus occurs naturally in fine to medium substrates and predominantly chose Fig. 3. Choice of different substrate fractions according to their particle sizes in four antlion species.  (Table 3).

DISCUSSION
Antlion and wormlion larvae are sand-dwelling insects, predominantly found in arid and semiarid regions, occupying fine sand or loose soil. Wormlions and many antlion species are highly specialized trap-building predators, whose foraging behaviour is not simply guided by a few factors but is much more complex, influenced by a number of variables (e.g. Arnett & Gotelli, 2001;Scharf & Ovadia, 2006;Scharf et al., 2011;Dor et al., 2014). Antlions occur in "antlion zones" (Gotelli, 1993), characterized by a specific combination of habitat variables such as sand particle size, sand density, density of plants, shade, and shelter (Scharf et al., 2008b;Devetak et al., 2012). Topoff (1977) believed that substrate particle size is less important because antlions can build pits in a wide range of substrate particle sizes. He therefore proposed shade as a major factor because M. immaculatus, the subject of his research, inhabits shaded habitats. Yet our findings suggest that sand particle size in determining habitat is also important for antlions and wormlions.
In natural habitats, larvae will experience sand as a mixture of material with different substrate particle sizes. Substrates in different geographical areas differ according to their substrate particle sizes. In Israel, for example, sands along the Mediterranean coast are coarser than those in the desert (Scharf et al., 2008a). M. hyalinus occurs in both habitats there, i.e. in coastal and desert sands (Scharf et al., 2008a), and it fits to our finding that it is a relatively generalist antlion in its habitat preference. The generalistic nature of M. hyalinus was previously demonstrated by Barkae et al. (2012). In experimental conditions, when two antlion species were exposed to two different substrates, the prey capture success of the habitat generalist M. hyalinus was higher than that of the habitat specialist C. lineosa (Barkae et al., 2012).
For antlions and wormlions sand particle size appears to be of key importance for efficient sit-and-wait and sitand-pursue predatory strategies. In natural habitats, pitconstructing antlion larvae are capable of discriminating between areas of substrate differing in particle size, choosing fine to medium sands and avoiding coarser sands (Lucas, 1986;Loiterton & Magrath, 1996;Botz et al., 2003;Devetak et al., 2005). The physical characters of substrate influence predation strategies in antlions, including selection of convenient pit-building sites, pit construction, prey recognition, etc. (Kitching, 1984;Allen & Croft, 1985;Klokočovnik et al., 2012).
In accordance with previous studies (Youthed & Moran, 1969;Kitching, 1984;Botz et al., 2003;Devetak et al., 2005;Ábrahám, 2006), we have shown that sand particle size plays an important role in pit-building decisions in two of our investigated antlion species (E. nostras and M. immaculatus). E. nostras larvae readily build pits in a fraction with a particle size of 0.23-0.54 mm, but in a fraction with a particle size of 1-1.54 mm only occasionally (Devetak et al., 2005). No pits are built in a coarser sand fraction (1.54-2.2 mm). This finding is in accordance with the present study where eight sand fractions were offered to E. nostras larvae. In African and Australian Myrmeleon species, the majority of pits were constructed in medium sand fractions (0.2-1 mm) (Youthed & Moran, 1969;Kitching, 1984).
The hypothesis suggested by Devetak et al. (2005) that larger sand particles are probably more difficult to handle for larvae, and thus they require more time to construct pits, was later rejected (Klokočovnik et al., 2012). In a previous study, the duration of three stages of pit construction -i.e. excavation of the initial furrow, deepening of a pit, and finishing the pit -was the longest in the most convenient particle size, i.e. medium sand fraction (0.23-0.54 mm), but not in coarser sands (> 1 mm) (Klokočovnik et al., 2012). Furthermore, substrate particle size has an effect on the frequency of jerks produced during sand tossing; the frequency was lower in coarser sand and increased with decreasing sand particle size (Klokočovnik et al., 2012). Probably the morphology of antlion species also influences the preference for a certain substrate particle-size fraction. As it is shown here in the choice experiment, M. hyalinus prefers the finest sand fraction and the possible explanation is that this preference is a consequence of its slender head. The ratio head capsule width/head capsule length (HW/ HL) in M. hyalinus is 0.75, in contrast to HW/HL ranging from 0.81 to 0.89 in the other Mediterranean Myrmeleontini species (Badano & Pantaleoni, 2014).
Analysing some of the physical properties of substrate vibrations and measuring behavioural reactions, Devetak et al. (2007) showed that larvae of the antlion E. nostras detect their arthropod prey by sensing the vibrations that arthropods generate during locomotory activity. Frequency of prey signals and sand particle size both influence the propagation properties of vibratory signals. Vibrations in finer sand are attenuated more strongly than in coarser sand Table 3. Chi-square test of substrate particle-size preference in four antlion species. The left lower side shows Chi-square results and the right upper side provides the respective P values. d.f. = 7. ns = not significant. and, consequently, an antlion detects its prey at a shorter distance . The most convenient sands for prey detection for all species are considered to be mediumsized particles.
Little is known about the role substrate particle size may play in the behaviour of animals, although it is known that substrate particle size influences pit construction behaviour of E. nostras larvae (Klokočovnik et al., 2012). They found that different sand particle-size fractions changed behaviour of antlions in terms of the occurrence and duration of particular behavioural stages. In sand composed of larger particles the larvae occasionally constructed irregular traps with a figure eight shape, and they crossed the centre of the cone during deepening. In the coarsest substrate the antlions did not build pits (Klokočovnik et al., 2012).
In this study, in samples from natural antlion habitats, we show that thirteen antlion and one wormlion species occupy a variety of substrates, differing according to their densities (Table 1). Each antlion species clearly preferred a substrate of certain particle size regardless of its predatory strategy. The question arises whether each antlion species prefers a substrate similar to its own habitat of origin. From this and previous studies (Botz et al., 2003;Devetak et al., 2005) it is evident that, in laboratory conditions, particle size affects pit-building decision and pit size in pit-building antlion larvae. Samples of antlion habitat showed that antlions are found naturally in coarser substrates than they chose in choice experiments, although Neuroleon microstenus and Distoleon chose substrate of similar course quality in laboratory conditions. Neuroleon differs significantly in prey capturing strategy from pit-builders and shows a plasticity in behaviour (Klokočovnik & Devetak, 2014). This could be a reason that Neuroleon prefers to forage for prey in coarser substrates.
Among psammophilous insects, wormlions are highly specialized for the finest substrates, consequently their larvae often occur in powders (Wheeler, 1930;Devetak, 2008). But why do antlion larvae in natural habitats build pits in substrates with an extreme range of particle sizes? Besides the fact that the density of psammophilous larvae plays important role, a probable explanation is that females lay eggs in the only substrate available that is appropriate for the larvae. Antlion larvae are sedentary predators that rarely move great distances. Matsura et al. (2005) investigated substrate selection based on particle-size composition in M. bore in Japan, and found that the selection of oviposition substrates by adult antlion females was the primary factor governing spatial distribution of individuals in the field, and that larval movement would not be very important.
Particle-size composition of substrate of the pit and its surroundings is very complex. Ábrahám (2006) studied particle size composition of substrates in Hungary in antlion zones of four pit-building antlion species (E. nostras, M. bore, M. formicarius and M. inconspicuus) and found that the structure of the substrate depends on the position of a sample regarding the pit. The finest substrate was found inside the pit, and the sizes of the particles collected from the edges of the pit were bigger than the ones taken from the pit slope but smaller than the ones from a point 3 cm away from the pit edge (Ábrahám, 2006). The larvae move the finer particles to line the pit and they throw out the coarser particles (Lucas, 1982;Gepp, 2010;Ábrahám, 2006).
It is also known that sand particle size affects pit morphology. In the substrate with fine particles, antlions build pits with the highest pit angle (Botz et al., 2003). Pits constructed in fine sands are more effective because escape of potential prey is hindered due to the steeper slope of the pit wall (Lucas, 1982 ;Botz et al., 2003;Fertin & Casas, 2006;Devetak et al., 2012). Pitfall traps are effective when slopes are steep enough to guide prey to predator's jaws without any attack or with small effort of the antlion, and shallow enough to avoid the likelihood of avalanches typical of crater angles (Fertin & Casas, 2006). On the other hand, for small sand-dwelling insects it is also inconvenient to dig and move in substrate with extremely large particles.
In summary, we have found that for sand-dwelling antlion and wormlion larvae the composition of substrates in natural habitats and the effect of substrate particle size on site selection are important. In natural habitats, different antlion and wormlion species occupy places with different substrate particle sizes, and fall into four categories: (i) Larvae preferring fine substrates are the wormlion Lampromyia and antlion Myrmeleon hyalinus originating from desert habitats. (ii) Larvae preferring fine to medium or (iii) medium substrates belong to antlion genera Cueta, Euroleon, Myrmeleon, Nophis and Synclisis. (iv) Larvae preferring coarse substrates are Distoleon and Neuroleon. In the choice experiment, four species of antlion primarily chose medium grained substrates, except for M. hyalinus, which preferred fine grains, similar to the grains found in its natural habitat.