Acta arachnol., 47 (1): 7-19, July 30, 1998 Ecological Studies on Wolf Spiders (Araneae: Lycosidae) in a Northwest Area of Kanto Plain, Central Japan: Habitat Preference Observed by Hand-sorting Yasuhiro Fujii1 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:ハ ンドソーティングで 調査された生活場所選好性 Abstract Wolf spiders (Araneae: Lycosidae) of 19 species were collected by hand- sorting in a northwest area of Kanto Plain, Japan, and their habitat preferences were inferred from their frequency in 10 habitats classified by three environmental elements (substrata, light, water). In addition to the 12 species surveyed by trapping (Fujii 1997), the hand-sorting enabled to examine the other seven species, Alopecosa virgata, Arctosa depectinata, A. subamylacea, Lycosa coelestis, Pardosa pseudoannulata, P. yaginumai, and Pirata subpiraticus, and to know more accurate preference of Pardosa agraria, P. astrigera, Pirata clercki, and P. piratoides. The interspecific differences were large within each genus, but were not found between Arctosa ebicha and A. fujiii and between Pirata procurvus and P. tanakai. Partial overlap was also detected among the preferences of other several Pardosa or Pirata species. The differences among the developmental stages were not obvious. Introduction Most wolf spiders (Araneae: Lycosidae) are common hunters wandering near the ground surface. I found large interspecific variations in maternal care (Fujii 1976) and life cycles among 19 lycosid species dwelling in a northwest area (35°54'N, 139°23'E) of Kanto Plain, central Japan. It seems that these variations had differentiated correspond- r ingly to environmental conditions or their stability of each lycosid habitat. As little was known for habitats and life cycles in these lycosids, I collected them by pitfall-trapping and hand-sorting from 1981 to 1987, and examined their frequency in various habitats and seasonal changes in body size or developmental stages. For comparison of habitat preferences, lycosid habitats were classified by qualities or grades in three environmental elements, substratum, light condition, and water condition. The trapping was effective in forests and tussocks but not in open habitats, and was inapplicable especially to low-mobile species in wet land (Fujii 1997). In contrast, hand-sorting was applicable to most habitats except dense tussocks and also to low-mobile species or stages. The 1) Department of Biology, The Nippon Dental University, 1-9-20, Fujimi, Chiyoda-ku, Tokyo 102 -8159 , Japan q 7 ~ T 102-8159 *,5; m i± 1-9-20 Accepted April 16, 1998 ~ 8 Y. Fuj ii present paper shows the frequencies of all the 19 species collected by the hand-sorting, then discusses their habitat preferences inferred from both the trapping and hand-sorting and differences in stability or mildness of environmental conditions among the habitat classes. Sampling Sites Sampling by hand-sorting was carried out at 28 sites of Subarea A---E (Fig. lc). Fifteen sites were located in subareas on a northern flat plateau (Figs. la, lc), and the other 13 sites were located in Subarea B (Figs. lb, lc) in a southern lowland with a shallow pebbly river (Koaze-gawa R.), a small spring, and paddy fields. Habitats of these sites were classified into any one of the 12 classes (Table 1). There was no site for hand-sorting in classes of Ddl and Dd2. Characteristics of the sampling sites except those already described in Fujii (1997) were as follows. [Ab 1, Cb, Eb, Bk2] Sunny bare ground. Ab 1 and Eb, cropland; Cb, an athletic field; Bk2, an edge of a pond artificially isolated from the river at the end of 1983. The soil was sandy, clayey, and pebbly, respectively. Bk2 was reclaimed in the autumn of 1984. [Ab2, Bk3, Bl, Bm2, Bn, Bk15, Bml] Meadows with poor litter. Ab2, a fallow upland field contiguous to Ab l; Bk3, a riverbank ridge; B 1 and Bn, fallow paddy fields; Bm2, a levee of a paddy field; Bk 15 and Bml, O.5- l m-wide zones along the river and an irrigation ditch, respectively. The soil was sandy at Ab2, pebbly at Bk15, and clayey at the others. That of B 1 and Bn was covered with standing water from May to September. [Ca, Ec, Bk13, Dm] Sunny ground with thick litter. Ca and Ec, forest edges; Bkl3 Fig. 1. Maps of the study area in 1984. Locations of Subarea A' -E (c); vegetation and 22 sampling sites in Subarea A, E (a), and B (b). The old course of Koaze- gawa River in Subarea B is indicated with broken lines (b, bottom). The other six sampling sites (Bh of Subarea B; Ca, Cb and Cf of Subarea C; Dh and Dm of Subarea D) are not shown here. Habitats of Wolf Spiders 9 Table 1. Lycosid habitat classes an d sampling sites. and Dm, edges of a tiny stream originating from a small spring in a tussock (Bk 13) or a forest (Dm). The litter of Bk 13 and Dm was usually soaked. [Ah, Bh, Cf] Dark ground with thick litter. Ah and Bh, evergreen forests of white oaks, red pines, cedars, and cypresses with poor undergrowth; Cf, a deciduous forest of konara oaks with dwarf bamboos. A hygrophyte, Polygonum thunbergii, dominated at B 1, while many herbaceous plants were found and seasonally alternated at every other site in open land. In winter, floors of deciduous forests were exposed to sunrays, and water was drained from paddy fields. But the classes of these habitats were decided by conditions in the other seasons because most lycosids retreated into litter or soil crevices in winter. Methods Sampling procedure Hand-sort sampling was carried out 368 times on 233 calm days by 'time-sampling' or 'zone-sampling'. In each sampling by both methods, lycosids were individually driven into 15 X 100 mm glass tubes and then confined with cork plugs until 50-- 100 individuals were collected. Soaked cotton was previously placed at the bottom of the tubes to prevent the lycosids from drying. In time-sampling, I walked at a slow constant pace and crouched down when capturing lycosids until the collection was attained, then recorded the total time of sampling for calculation of lycosid frequencies. When no lycosid was added for 10 or more minutes (this case was frequent in winter), the sampling was ended even before attaining of the collection. This sampling was employed as the main method 273 times (7,414 min.) in the daytime, and 13 times (507 min.) in the nighttime aided with a head lamp. These samples were called 'time-samples'. In zone-sampling, I cautiously searched for lycosids from outer parts to inner parts within a 2--4 m2 zone, and selected additional zones until the collection was attained, then recorded the total area of searching for the calculation. A 50 X 50X 20 cm aluminum quadrat was applied to sites with thick litter layers, and the area in the quadrat was regarded as the zone. At a site with low lycosid density, all the lycosids found in the site were collected. This method was employed 82 times (8,336 m2) only in the daytime to obtain more precise frequencies and compare them with those in time-sampling. These samples were called 'zone-samples'. 10 Y. Fujii At one or two fixed sites of every class except Bsl and Dsl, time-sampling was carried out almost monthly. At those of LsO, DsO, and DdO, zone-sampling was also performed. Lycosids at the other sites were collected from March to November mainly by time-sampling. Most of these lycosids were released at the center of each sampling site soon after examinations in the laboratory to minimize disturbance by sampling. Categories in stages and sexes Youngs, males, free females, females with egg cocoons, and females with pulli are used in this paper as categories of stages and sexes. Pulli on their mothers were not counted. Measuring temperature For an examination of difference in environmental conditions among the habitat classes, a maximum-minimum thermometer shaded with a 10 X 25 X 1.5 cm plate of aluminum and styrofoam was held at 3 cm above the ground of each center of neibouring three sites, Ab2 of LsO, Ae of DsO, and Afl of DdO. These thermometers were read and reset weekly in 1986 and 1987. Temperatures of ground surfaces or water were also measured at the beginning and the end of every sampling. Results Of all 21,308 individuals collected by time- and zone-sampling, 12 individuals were those of unknown Pardosa species. Four species, Arctosa fujiii, Pardosa agraria, P. astrigera, and P. graminea, occupied 65% (10, 12, 25, and 18%, respectively) of the remnants. On the other hand, individuals of Alopecosa virgata, Lycosa coelestis, Pardosa yaginumai, Tricca japonica, and Trochosa ruricola were less than 0 .5%. Percentage for each stage or sex was 74 in youngs, 8 in males, 10 in free females, 6 in females with egg cocoons, and 0.5 in females with pulli. Similar ratios were obtained in each species except 25-v43% in youngs of Arctosa depectinata, Pirata procurvus , and P. tanakai, 18.19% in females with egg cocoons of the latter two species, and 0 .09% in females with egg cocoons of Arctosa fujiii. The 545 individuals in time-samples were collected at night mainly from Ab 1 and Bbl. Their 50, 34, and 7% were Arctosa depectinata, Pardosa astrigera , and P. graminea, respectively. The percentage was lower in P. graminea but slightly higher in P. astrigera and remarkably higher in A. depectinata than that in the daytime. The 85% J 411 ivv a.a41 Iltla 1114) Jwl J41 ALLY V~J V4t 1\VY LW J0.ll 1'GU lY1Q1 Ak11 1Y10.y JLLII JUl C1Ub' Fig. 2. Maximum and minimum temperature in a week at 3 cm above the ground in three contiguous sampling sites of different habitat classes, DdO, DsO, and LsO. Habitats of Wolf Spiders 11 of A, depectinata was found at night. The weekly ranges of ground temperature in LsO, DsO, and DdO were 46, 35, and 35°C at the maximum (February April), 15, 7, and 6°C at the minimum (June September), and 27.5, 18.8, and 15.6°C on the average, respectively (Fig. 2). Ground temperatures measured at sampling were 7--30°C in DdO, whereas they were 6-'-41°C in the other classes. The highest temperature observed in running water was 26, 25, 24, and 14°C at Bml, Bkl1, Bkl3, and Dm, respectively, while that in standing water was 28°C at B1 and 2TC at Bk2. Differences between time-samples and zone-samples All 19 species were found in time-samples of 15,319 individuals from 26 sampling sites of the 10 classes. Their frequencies in the number per h (f) were calculated in each sampling site (Table 2). In zone-samples of 5,989 individuals from 10 sites of the seven classes, the 17 species were found. Their frequencies in the number per 25 m2 (f) were also calculated (Table 3). Ab2 and Ab3 were here treated as one site, since they neighbored on each other (Fig. 1 a) and were classified into the same class of LsO. The sites common to both of the samples were thus nine (the sites except Ae in Table 3). Rare two species, Lycosa coelestis and Pardosa yaginumai, did not occur in zone- samples. In above two species and other four species, Alopecosa virgata, Arctosa depectinata, Tricca japonica, and Trochosa ruricola, the values off and f. did not exceed 1 and most of them were 0 at every nine sites. Therefore, the relationship between ft and fz was examined in the 117 pairs from the remnant 13 species of the nine sites, and significantly positive correlation (r=0.624, P<0.001) and a ratio of 0.241 for ft/ fz were obtained. They did not largely change even after removing 63 pairs lower than 0.5 both in f and fz (r = 0.574, P <0.001, f / f = 0.209). Furthermore, the highest ft and fz were observed at the same site in each 11 species. In four species of them, the coincidence occurred also in the second highest. Such coincidence was not obtained in Pardosa graminea and Pirata tanakai, but similar tendencies were observed. Percentage for each stage or sex in time- and zone-samples was 71 and 84 in youngs, 10 and 5 in males, 12 and 7 in free females, 7 and 3 in females with egg cocoons, and 0.7 and 0.2 in females with pulli, respectively. They were similar in each category though youngs' percentage in time-samples was lower and the others' were higher than those in zone-samples. Frequency in habitat classes Difference among species Lycosid frequency was compared between habitat classes except that of Lycosa coelestis with only one individual in each of BsO and DdO (Fig. 3). Frequency in a class was calculated not by averaging ones in the class but by the whole individuals and sampling quantity (in minutes or m2) in the class. The nine species of Arctosa subamylacea, Pardosa agraria, l? pseudoannulata, P. yaginumai, Pirata clercki, P. piratoides, P, subpiraticus, P, yaginumai, and Tricca japonica were found almost only in habitats around water (the upper nine in Fig. 3). Arctosa subamylacea and Pirata piratoides were frequent in Bsl, Bs2, Lsl, and Ls2, and Pardosa agraria showed high frequency in Lsl, Ls2, and Dsl. Classes with obviously high frequency were three or four in these three species, but were only two or one in the others. They were Lsl and Ls2 in Pardosa pseudoannulata, Bsl and Lsl in Pirata subpiraticus, Bsl and Bs2 in Pirata yaginumai, and Ds2 in Pirata clercki. 12 Y. Fujii Habitats of Wolf Spiders 13 14 Y. Fujii Fig. 3. Frequencies of the 18 lycosid species in each habitat class. The upper nine, species frequent in wet habitats; the lower nine, species frequent in dry habitats. Solid rectangles, frequencies in n/25 m2 (zone-samples); stippled ones, those in n/ h (time-samples); asterisks, frequencies lower than 0.5 but not 0. Pardosa yaginumai also concentrated on Bs2 though it was very rare. Tricca japonica was also rare but occurred in various classes of wet habitats. The remnant nine species, Alopecosa virgata, Arctosa depectinata, A. ebicha, A. fujiii, Pardosa astrigera, P. graminea, Pirata procurvus, P, tanakai, and Trochosa ruricola, were more frequent in drier habitats though Alopecosa virgata, Arctosa ebicha, Habitats of Wolf Spiders 15 Fig. 4. Stage- or sex-composition of the 11 lycosid species in the habitat classes. The illustration is omitted for the other eight species, which had one or no class with 30 or more individuals. and Trochosa ruricola were rare (the lower nine in Fig. 3). The classes with higher frequency were Bs0 and Ls0 in Alopecosa virgata, Arctosa depectinata, Pardosa astrigera, and Trochosa ruricola, Ds0 or Dsl in Pardosa graminea and Pirata procur- vus, and Dd0 in Arctosa ebicha, A. fujiii, and Pirata tanakai. Difference among stages or sexes In order to detect stage- or sex-specific habitat selection, percentage for each stage or sex was calculated in samples of each habitat class (Fig. 4). Because the compositions in time- and zone-samples were similar to each other, the individuals from both samples were summed, and the percentages were calculated only in the classes where the sum reached to 30 or more. The calculation was omitted in eight species since the class was one or no. Youngs of seven species, Arctosa fujiii, A. subamylacea, Pardosa astrigera, P. graminea, P. pseudoannulata, Pirata clercki, and P. subpiraticus, dominated others in every class. Adults of Pardosa agraria and Pirata piratoides increased their percent- ages in Bs2 and Lsl. In Pirata procurvus and P. tanakai, percentage of youngs was generally low but youngs of the former were dominant in DdO. However, these large changes were found mainly in classes with small individuals (Bs2 with 47 P. agraria and Dd0 with 89 P. procurvus). Discussion Usefulness of time-sampling Trapping was effective in collecting small lycosids (Pirata procurvus and P. tana- kai) at tussocks and forest floors with thick litter, or rare but active males (Arctosa ebicha, Tricca japonica, and Trochosa ruricola), but was not in collecting lycosids of 16 Y. Fujii low-mobile species or stages (youngs and females with egg cocoons) (Fujii 1997). I observed in the laboratory that two Arctosa lycosids (A. depectinata and A. subamylacea) and two Pirata lycosids (P, piratoides and P. subpiraticus) made their nests below and above the soil surface, respectively, and seldom walked out of the nests. Hand-sorting collected these low-mobile lycosids especially in open habitats. Females with egg cocoons of the small Pirata species were easily found even in dark forests because their egg cocoons are white and conspicuous. Since females of Arctosa ebicha and A. fujiii do not carry their pulli (Fujii 1976, 1983) and the latter female retreats in an ovipositing room spun in litter, neither female with pulli could be found and the latter females with egg cocoons were seldom collected even by hand-sorting. Zone-sampling gave more precise population density and composition of species or stages but demanded much more time than time-sampling, and it was hardly applicable to many sites. In time-sampling, time required for capturing one lycosid was only 5 seconds on the average, and that for 100 lycosids may be less than 20 minutes in high-density populations. Rough estimation of the density will be also possible by using f and f / fz. In this study, 10 in f corresponded to 42 in fz, thus to 2 per m2. However, the ratios were remarkably higher at Ab2, Ab3, and Ad. The much bare ground in these sites probably raised f and/or depressed fz as lycosids on bare ground were easily found and captured but were not abundant. Habitat classification Though most species in one site occurred also in the others of the same class, the frequencies in several species were different among these sites probably because of differences in some qualities of soil or litter, or in distances from water. Pirata yaginumai occurred only on pebbly soil beside water (pebbles were rare around stand- ing water but abundant at Bk2). Arctosa depectinata was found almost only on soft sandy soil. Immigrants of Pardosa pseudoannulata at Ca and Cb, which were both far from water, survived at least for two months because of puddles elongated by the clayey soil. Pirata procurvus and P. tanakai exhibited high frequencies at sites with litter containing red pine or cedar leaves. These leaves may raise survival rates of these small lycosids, offering them a lot of narrow retreat. Pardosa agraria and Pirata piratoides were generally rare in Bs0 and LsO, but many individuals were collected at Bbl or Bk3 probably because both sites were located nearer to water than the others. The habitat classification did not adequately correspond with above lycosid fre- quencies because of its simplicity, but it may be very convenient and useful to examine the difference between lycosid habitat preferences: Habitat preference In habitats with an unfavorable or intolerable condition, lycosids probably show extremely low frequency in all of the trap-, time-, and zone-samples, and in the others, they may show high or moderate frequency in any of these samples. Thus, preferable or tolerable habitats of the lycosids in the study area may be as follows (Fig. 5). [Sunny open habitats] BsO; Arctosa depectinata: Bs0-LsO; Alopecosa virgata: Lsl; Pirata subpiraticus: Bs2; Pardosa yaginumai: Lsl -Ls2; Pardosa pseudoannulata: Bs2-Ls2; Pirata yaginumai (Bk2 was treated here as a river edge): Bsl -Bs2-Lsl -Ls2; Arctosa subamylacea, Pirata piratoides: all of these classes; Pardosa astrigera. [Forest floors or their edges] Ds2; Pirata clercki: DsO-Dsl-Ds2-Dd0-Ddl-Dd2; Arctosa ebicha, A, fufii, Pardosa graminea, Pirata procurvus, P. tanakai. [Ecotones between meadows and forests] Ls0-Lsl-DsO-Dsl; Trochosa ruricola: Lsl-