Acta arachnol., 46 (1): 5-18, July 30, 1997 Ecological Studies on Wolf Spiders (Araneae: Lycosidae) in a Northwest Area of Kanto Plain, Central Japan: Diel Activity and Habitat Preference Observed by Pitfall Trapping Yasuhiro FuJIII> 藤井靖浩1):関 東平野北西部におけるコモ リグモ類 (クモ目:コ モリグモ科)の 生態学的研究: 落:し穴 トラップで調査された日周期活動 と生活場所 Abstract Interspecific differences in die! activity and habitat preference of wolf spiders (Araneae: Lycosidae) were observed by pitfall trapping in a northwest area of Kanto Plain, Japan. Pardosa agraria, P.astrigera and P.graminea were all collected only in daytime, and Pirata yaginumai and Pirata piratoides showed slightly diurnal and nocturnal tendency, respectively. Arctosa fujiii, Pirata clercki, Pirata procurvus, Pirata tanakai, and Tricca japonica can be regarded as aperiodic. Lycosid habitats were classified by combining qualities or grades of three environmental elements of substratum (B, bare soil; L, live plants; D, dead plants), light condition (s, sunshine; d, dark or shade) and water condition (0, rain alone; 1, standing water; 2, running water). Though every lycosid species occurred in three or more habitat classes, and seldom on upper parts of live plants, the highest frequency was observed in Bs0 in Pardosa astrigera, Ls0 in Pardosa graminea and Trochosa ruricola, Lsl in Pirata clercki, Ls2 in Pardosa agraria, Pirata piratoides and Pirata yaginumai, Ds0 in Pirata procurvus and Pirata tanakai, Dsl in Arctosa ebicha and Tricca japonica, and in Dd0 in Arctosa fujiii. Other seven lycosid species (Alopecosa virgata, Arctosa depectinata, Arctosa subamylacea, Lycosa coelestis, Pardosa pseudoannulata, Pardosa yaginumai and Pirata subpiraticus) in the study area were not or seldom trapped because of their low mobility or low density at the trapping sites. Differences between stages or sexes were obscure both in die! activity and habitat preference. Introduction Most wolf spiders (Araneae; Lycosidae) are hunters wandering near the ground surface. They exhibit, however, large Interspecific differences in ecological properties such as die! activity, habitat preference and life cycles. In an area on the northwest border of Kanto Plain, central Japan, I found a total of 19 lycosid species and observed large interspecific variation in their mobility and maternal care (FUJIi, 1974, 1976, 1978, 1980, 1983). This difference may be the result of adaptation especially to each lycosid 1) Department of Biology, The Nippon Dental Uiversity, 1-9-20, Fujimi, Chiyoda-ku, Tokyo 102, Japan T 1021-9-20 Accepted April 25, 1997 6 Y. FUJII habitat. But little is known about the ecological properties in these lycosids except Pardosa agraria and Trochosa ruricola, whose populations were well studied in Hokkaido, northern Japan (SUwA,1986; MATSUDA & SHIBATA,1995), and in Europe (HACKMANN, 1957; ENGELHARDT,1964), respectively. As my study area was narrow mosaic of heterogeneous habitats, it was easy to compare lycosid populations in many different habitats with each other, without consid- eration of climatic difference, and the populations were investigated by pitfall trapping and hand-sorting from 1981 to 1987. This paper shows interspecific differences of diel activity and habitat preference observed in pitfall samples. Some lycosids change diel activity (DONDALE et al., 1972) or habitat preference (EDGAR, 1971; HOLLANDER & L OF, 1972; DONDALE, 1977; SUWA, 1986; EUBANKS & MILLER, 1992) after their maturation and/or ovipositions. These changes were also examined. Use of pitfall traps for arthropods on the ground surface is effective in estimating relative abundance or activity of particular species or instar between different habitats (GIST & GROSSLEY,1973; UETz,1977; ORAZE et al., 1988; TOUYAMA et al., 1991), but not in estimating that between different species or seasons (TOPPING & SUNDERLAND, 1992). Species composition and seasonal changes in abundance are thus not shown in this paper. For comparison of habitat preferences, we need to extract main elements of lycosid habitats and to classify the habitats by using these elements. As main lycosid sense organs are eyes and tactile receptors (FOELIx, 1982; YOSHIKURA, 1987), light intensity and substrata may be important elements. Lycosids are rather sedentary but they begin to move when some elements exceed their tolerable range, seeking for other preferable sites (NORGAARD, 1951; EDGAR, 1969; FUJII, 1974; YEARGAN, 1975; CADY, 1984). Density and thickness of cover vegetation that have large influence on micro-climate of habitat must be selected by lycosids (MILLER, 1984), and both can be represented by light intensity. For the substrata, bare soil, live plants, dead plants (or litter), and their water contents are probably selected (NORGAARD,1951; VLIJM & KESSLER-GESCHIERE, 1967; SCHAEFER, 1972; ROVNER & KNOST, 1974; DONDALE, 1977; UETZ, 1977; GREENSTONE, 1980). Specific composition of plants or animals was excluded from the elements of this classification, because there are no obvious relationships between lycosids and any particular species of plants or food animals (KUENZLER, 1958; MIYASHITA, 1968; EDGAR, 1969; HALLANDER, 1970; YEARGAN, 1975; DONDALE & BINNS, 1977; SUWA, 1986; UETz et al., 1992). TER BRAAK (1986), by his new statistical technique, extracted six elements from 26 elements, which had been measured by VAN der AART & SMEENK-ENSERINK (1975) to examine habitat preferences of 12 lycosid species in a Dutch dune area. Though these six elements agree with those above mentioned and the difference of habitat preference was successfully revealed, this analysis demands laborious measure of the elements. On the other hand, HOLLANDER & LOF (1972) classified a Dutch meadow habitat into four types by two elements of cover vegetation with each two grades. Their classification of habitats is much simpler. The present study applies a similar method, classifying the habitats of the study sites by substrata, light and water conditions respectively with three, two and three qualities or grades. Activities and Habitats of Wolf Spiders 7 Study Area and Sampling Sites The study area of c. 150 ha lies on a boundary zone (35°54'N, l39°23'E) of three adjacent cities, Kawagoe, Hidaka and Tsurugashima, near to Kanto Mountains (Fig.la). The study was carried out in Subarea A and D on a northern flat plateau (53 and 50m a.s.l. at the west and east end), and in Subarea B on a southern lowland (48 and 44m as.!. at the west and east end). In the middle of Subarea B, there were paddy fields and a stream (Koaze-gawa R.) with a small spring. This stream was pebbly and shallow, usually 0.2---lm deep and 3--'6m wide, but it often flooded and eroded the surrounding cropland during heavy rains in the summer of 1983. These erosions caused public works to straighten and concrete banks at the largest turn in the following winter and spring (Fig. lb, bottom, broken lines). Other parts of the banks were also reinforced with concrete blocks after the summer of 1985. A total of 18 sampling sites was used (Fig. lb). Site Ab3 (on a roadside), Bbl, Bbl (in cropland), Bd (on the spring edge), and Bk 11, Bk 14, Bk21 (on stream edges) were open areas with poor vegetation. Other 11 sites were in tussocks, shrubs, woods or forests, and their vegetation can be outlined as follows. [Ae, Bell Shrubs of forest edges comprised dwarf bamboos (Pleioblastus chino, Pseudosasa humilis). Spiny plants (Chaenomeles japonica, Rubus microphyllus, R. parvifolius), viny plants (Pueraria thunbergiana), and stout grass (Miscanthus sinensis) were also found at Site Ae. [Be2] A dense tussock of grass, forbs and viny herbaceous plants between a small wood and cropland. [Afl, Bf, Df] Deciduous broad-leaved forests. Konara oak (Quercus serrata) and chestnut (Castanea crenata) in canopy trees, snowbell (Styrax japonica), aucuba (Aucuba japonica), torch azalea (Rhododendron kaempferi) and the dwarf bamboos in undergrowth were common. The dwarf bamboos maintained dark floors with thick litter layers and poor herbaceous vegetation throughout all seasons. [Ac 1 ] A young plantation of cypress (Chamaecyparis obtusa). Many herbaceous plants invaded but they were mown by the owner before thickening, and a sunny floor was kept. [Ai, Bi] Mature cypress plantations. Undergrowth comprised mainly cypress sprigs and eurya (Eurya japonica). As floors were always dark and covered with thick litter, herbaceous plants could hardly grow. [Bj] A white oak (Quercus myrsinaefolia) wood by the stream. The outline was almost the same as that of Site Ai or Bi. [Dh] A red pine (Pinus densiflora) wood. Canopy layer was fairly thin, and richness both in undergrowth and herb vegetation was intermediate. Methods Habitat classification of the sampling sites Each of the three habitat elements was divided into the following qualities or grades. [Substrata] bare soil (B), live plants (L), dead plants (D) [Light condition] sunshine (s), dark or shade (d) [Water condition] rain alone (0), standing water (1), running water (2) A patch of direct sunrays and main water resources were used as markers of each 8 Y. FUJII Activities and Habitats of Wolf Spiders 9 grade in light and water conditions, respectively. Lycosid habitat can be divided into 18 (3 x 2 X 3) classes, and they were called BsO, Bsl, Bs2, etc. by combining the abbrevia- tions. However, six classes of BdO~-2 (bare soil in dark) and LdO- -2 (undergrowth in dark) were eliminated because lycosids were extremely scarce in them. Each class of the sampling site was determined by the chracteristics of the habitat elements within a distance of 2m horizontally from each trap (Table 1). This distance was determined by lycosid usual motilities. Good places for the trapping site could not be found in habitat classes of Bsl, Ds2 and Ddl. Traps Traps of Type A and B were used (Fig. 2). Each comprised three parts, a cylinder embeded in the ground, a bottle (Type A) or a cup (Type B) put into the cylinder, and a square transparent cover. VLIJM & KESSLER-GESCHIERE (1967) put their traps in the ground and vegetation layers to capture meadow-living lycosids. However, all traps of this study were placed only in the ground because the lycosids of this area were found on tiny plants but seldom on upper parts of tall plants. Traps of Type A These traps were prepared to examine both diel activities and habitat preferences of lycosids (Figs. 2a' 2c). The trap cover had a floor plate with a 25-mm-diam. hole in its center, and a 100-m1 bottle with a hollowed cap was fixed to the floor plate beneath the hole. A body of the bottle was exchanged for another one at sunrise and sunset. Thin detergent solution was previously poured into the bottles to prevent escape and cannibalism of captured lycosids, and a little of folmalin as preservative was added to the solution in the bottles after detaching them. At each of seven sampling sites, five traps of this type were placed at intervals of c. 3m, and used 14 times of fine weather from May 9 to July 12, 1981 (Figs. 1b, 3). All these sites were set close in Subarea B to finish the exchange of 35 bottles within 30 minutes of each dusk period. Two sites, in shrubs similar to one another, were dealt as a site, Bel. Some of the traps were pushed up by subterranean mammals, and their samples were thus excluded from counting. Traps of Type B These traps were prepared for long-term samplings (Figs. 2d---2f). They were tougher and larger than traps of Type A, and were used from 1983 to 1986 (Mar. 12 -Jul . 31, 1983; Mar. 31-Oct. 27,1984; May l2'--Nov. 22,1985; Jan. 15'---Aug. 17,1986) Table 1. Lycosid habitat classes and trapping sites. 10 Y. FUJII Fig. 2. Pitfall traps of Type A (used in 1981) and Type B (used in l983-l986). Trap covers, bottles or cups (a, d); cylinders (b, e); traps in the ground (c, f). The cover consisted of a roof plate and two rectangularly crossing wall plates. The roof kept out rainwater, and the walls held the roof and guided passing lycosids into the bottle or the cup. They were all made of plastic materials. (Figs. lb. 3). The cups contained ethylene glycol of c. 100ml that prevented samples from decaying, and were exchanged for new cups every one or two weeks. In order to minimize the influence of the trappings on hand-sort samplings in the same sites, the number of traps at each site was restricted to one in 1983 and three thereafter. Calculation of lycosid frequency As the type of traps, their operative numbers (t) and days (d) differed between sites or periods, the number of lycosids in each sample (n) was converted to frequency (f) when they were compared with one another. The frequency was defined as n per trap of Type B per 100 days merely because the value per day or ten days was too small. A value of n in Type A was assumed to be 75% of that in Type B because it probably changes in proportion to length of wall plates of the trap cover. To obtain a value of f directly from n, quantity of sampling (q) and conversion ratio (c) were calculated by following equations: q=0.75~tAi' dAi+ L~tBidBi, c= 100/q, f = c' n, where tAi and dAi, tBi and dB, are t and d of Type A and B in the i-th period, respectively. A value of f in a class was also calculated from c and n in the class, not by averaging values of f in sites of the class. Categories in stage and sex The categories used in this study were youngs (pulli, nymphs and sub-females), sub-males, males, females with egg cocoons, and females without egg cocoons (free females and females with pulli), since it was difficult to separate pulli or sub-females from Activities and Habitats of Wolf Spiders 11 nymphs, or females with pulli from free females in trap samples. Results and Discussion The sampling quantity was 2,703.5 in total, 89.6 in daytime, 55.9 in nighttime (Fig. 3). Samplings at every site in open habitat (BsO, Bs2, Ls0~2) and those at Site Dh and Df in forests were stopped by July of 1983 and April of 1986, respectively, because of frequent disturbance by passersby or floods after heavy rains. A total of 3,099 lycosid individuals were collected with the traps of Type A and B. They comprised 1,088 youngs (35%),111 sub-males (4%), 1,455 males (47%), 413 females without egg cocoons (13%), and 32 females with egg cocoons (1%). The females with pulli may be fewer than 1 %, because a pulli-carrying period is much shorter than a cocoon-carrying period (FUJIi, 1976, 1978). Fig. 3. Sampling periods (rectangles) and sampling quantity (figures) in each of the trapping sites and years. For abbreviations and calculation of the quantity, refer to Fig.lb, Table 1 and text. 12 Y. Fu II The 2,975 individuals (96%) were identified as any of the following 14 lycosid species; Arctosa depectinata, A.ebicha, A. fujiii, Pardosa agraria, P.astrigera (P. T insignita), P.graminea, P.pseudoannulata (Lycosa pseudoannulata), Pirata clercki, P. piratoides, P.procurvus, P.tanakai, P.yaginumai, Tricca japonica (Arctosa japonica) and Trochosa ruricola. The remnants were one sub-male of unknown species at Site Ae, 123 pulli at Site Af 1 and Dh. The pulli must be on females of Pirata procurvus and/or Pirata tanakai collected together. Arctosa depectinata, Pardosa astrigera, and P.agraria, P.pseudoannulata were seldom trapped, though the former two and the latter two were abundant in the field around Site Ab3 or Bb2, and in the paddy fields contiguous to Site Bd or Bb2, respective- ly. Other five species of Alopecosa virgata, Arctosa subamylacea, Lycosa coelestis, Pardosa yaginumai, and Pirata subpiraticus, were also found in the study area, but none of them were trapped. Alopecosa virgata, Lycosa coelestis, and Pardosa yaginumai were very rare, but Arctosa subamylacea and Pirata subpiraticus were very common in Fig. 4. Comparison of ~~ lycosid frequencies between daytime and VV nighttime and those among categories in stage or sex. Each rectangle or figure represents the frequency (n/trap/100 days). Activities and Habitats of Wolf Spiders 13 paddy fields. These results suggest that the trapping method is suitable for estimating habitat preference or mobility and not density. The traps of Type A captured 134 individuals, but Arctosa depectinata, A.ebicha, Pardosa pseudoannulata and Trochosa ruricola were absent. The composition of stage and sex (23% youngs, 8% sub-males, 53% males, 11% females without egg cocoons, 3% females with egg cocoons) was similar to that of all samples. Diet activities Total frequency was 148.5 in daytime and 90.0 in nighttime, and the ratio of the latter to the former was 0.61. The ratio was similar in every stage or sex (0.53 in youngs, 0.60 in sub-males, 0.68 in males, 0.50 in females without egg cocoons) except 0.33 in females with egg cocoons. On the contrary, the ratio largely differed among species (Fig. 4). The three lycosids of Pardosa were collected only in daytime. They may be diurnal like the Pardosa lycosids in previous papers (DONDALE et al., 1972; YEARGAN, 1975; SuwA, 1986). Pirata yaginumai and Pirata piratoides showed slightly diurnal tanakai, and nocturnal tendency, respectively, but Arctosa fujiii, Pirata procurvus and P, and probably also Pirata clercki and Tricca japonica, can be regarded as aperiodic. Changes in diel activity after maturation were not obvious in these species though youngs of Pirata piratoides were more active in nighttime. These results suggest that samplings in daytime are useful at least for these ten species. Habitat preference Difference among species Table 2 shows frequency of the 14 lycosids at each site. Arctosa fujiii occurred at all sites, and was frequent at Afl, Ai, and Bi of DdO. Pardosa graminea, Pirata procurvus and P, tanakai were also trapped at almost all sites, and they were especially frequent at Ab3, Bi and Afl, respectively, though P.procurvus showed generally low frequency. The other species were trapped at more or less restricted sites, and Arctosa depectinata and Pardosa pseudoannulata were trapped only at Bb2 and Bk 11, respec- tively. Frequencies of a species should be more similar to one another in each class, but some of them were considerably low or high, e.g., 8 at Df of DdO in Arctosa fujiii, 1 at Dh of DsO in Pardosa graminea, 42 at Bi of DdO in Pirata procurvus, and 4 at Adl of DsO or 120 at Afl of DdO in Pirata tanakai. The total frequency at Site Acl, Bf and Df was much lower than the others in DsO or DdO. Trap types, trap fluids and intervals of their exchange differed among the trapping sites. This must have affected the trapping efficiency to some extent (TOPPING & LUFF,1995), but the above low or high frequencies seem to derive mainly from shortage of the sampling quantity. Figure 5 shows frequencies of the 12 species in the nine habitat classes. Pardosa agraria, Pirata piratoides and P.yaginumai showed their highest frequencies in Ls2, and Pirata clercki and Tricca japonica showed them in Lsl and Dsl, respectively. Though a few individuals of Pirata clercki, P.piratoides and Tricca japonica were also trapped in dryer habitats (BsO, DdO), these five lycosids seem to prefer the sites around standing or running waters. It was noted that Pardosa agraria, Pirata piratoides and P. yaginumai were obviously abundant in sunny habitats (Bs2, Ls], Ls2), Pirata yaginumai was found only in traps at stream edges (Bs2, Ls2, Dd2), and Pardosa agraria was absent in dark and litter-rich habitat (Dd2). The remnant seven lycosids were also abundant in dryer habitats, and showed their highest frequencies in BsO in Pardosa astrigera, LsO in P.graminea and Trochosa ruricola, DsO in Pirata procurvus 14 Y. FUJII