Rev.Bio1.Trop., 30.(1): 27-34, 1982.

Analysis of the leaf litterarthropod fauna of a lowland tropical evergreen forest site (LaSelv a, Costa Rica)

Susan Lieberman and Charles F. Dock Allan Hancock Foundation, University of Southern California. University Park, Los Angeles, California 9007.

(Received for publication August 13, 19.81)

Abstract: Leaf litter arthropod population changes were monitored by means of extensive can trapping over a thirteen month period at a lowland evergreen forest site in Costa Rica. Samples were taken within an undisturbed forest plot and an adjacent fallow cacao plantation. Comparisons of abundance were made between nocturna! and diurna! samples as well as between months. The dominant arthropod groups in the samples in order of abundance were the Formicidae, Apterygota, Orthoptera and Arachnida. Total numbers of individuals in all major groups were consistently greater in the initial stages of the rainy season and remained relatively constant throughout the remainder of the year. It is suggested that numbers in crease in response to the onset be the rainy season. Greater numbers of individuals were collected in the undisturbed forest than the fallow cacao plantation. Siptificant diel differepces in abundance were fo und for many of the major groups inc1uded in the study. Maximum arthropod abundimce corresponded to the time of greatest leaf litter depth within the study plots.

. ..:. Historically, studies of tropical arthropods trap samples. Additionally , differences in ar­ have been qualitative an d descriptive in nature thropod abundance between undisturbed forest (Allen, 1926 ; Davis, 1945 ; Strickland, 1945). and disturbed forest planted in cacao are More recently, insect abundance has been analyzed. The potential relationship between studied by many workers, concentrating on the arthropod numbers and rainfall patterns is also understory and canopy (J anzen, 1973 ; Mont­ exarnined. gomery and I..ubin, 1979; Buskirk and Buskirk, 1976). Though the litter floor has begun to MATERIAL AN D METHODS receive more attention (Anderson, 1975 ; Elton, 1973; 1975), most studies have concentrated The study area was Finca La Selva, Heredia on individual taxa, in particular beetles, mites, Province, Costa Rica . This area is a lowland and spiders (Meijer, 1974; Baars, 1979; evergreen wet forest (Holdridge , 1967) at an I Mitchell, 1963; Stanton, 1979 ; Uetz and altitude of 60 m, with a mean annual rainfall of Unzicker, 1976; Uetz, 1979; Hanlon and 3969 mm. Two dry months of J anuary-Febru­ Anderson, 1972). Bigger (1976) documented ary usually average less than 200 mm rainfall, the existence of oscillations in tropical insect while the wet months of March-December populations in general, and considere d preda­ average more than 200 mm. The study was tor-prey relations to be a causative factor. carried out at two sites: one in a relatively Wolda (1977 ; 1979) has shown seasonal fluctu­ undisturbed fo rest remnant, the other in an ations in tropical insect populations that may adjacent plantation that was once planted with relate to rainfall and food abundance. cacao trees for agricultural purposes but has In this report we evaluate seasonal and diel been fallow for at least 20 years. variation in arthropod abundance of tropical The arthropods were trapped using a series forest leaf litter, based on multiple monthly pit of one gallon pit traps. The pit traps were laid

27 28 REVISTA DE BIOLOGIATROPICAL

out at 10 m intervals in two transect lines, one Lepidoptera, Coleoptera, other identifiable lar­ in undisturbed forest, the other in the cacao vae , and unidentifiable larvae. The data base for plantation. f;ach trap had a top that could be the separation was too small, and no sienificant removed during trapping operations, and con­ differences between them were found. Conse­ tained pichc acid solution to fIx all specimens q�ently all larvaeare grouped together through­ at time of capture. Traps were operated at both out the paper. The miscellaneous category plots once a month for a period of 13 months, includes Ephemeroptera, Neuroptera, and from February 1974 to February 1975. Sam­ Lepidoptera. The unidentified category pling spanned a continuous four-day period in includes all arthropod individuals that could each of the 13 months. The traps were cleared not be accurately identified. every 6-10 hours day and night (mean satTlpling period: 9.02 ± 3.7 hours), the picric acid 'was RESULTS replaced, and the arthropods removed and preserved. A total of 21 ,944 individual arthropods we re In the laboratory, can trap samples were collected, of which 34.8% were ants, 22.7% evaluated. Each sample was recorded with Orthoptera, 18.0% Apterygota, 5.0% spiders, respect to numbers ofvarious arthropod groups and the remaining 19.5% was comprised of all and their respective sizes. Most arthropods were other groups combined. classifIed to order (Table 1). Ants we re sepa­ Figure 1 shows the average numbers of rated from other Hymenoptera ; arachnids were individuals collected per months for the major groups. Total monthly rainfall and rainfall during the sampling periods are also presented. Numbers for all major groups peaked in spring and subsequently declined in summer. With the exception of the Apterygota, numbers re­ mained relatively constant over the remainder � ! of the year. Apterygota exhibited a further .... decline in numbers in late fall and winter. 5 D. . More individuals were trapped du ring the � i day among the Apterygota, Diptera and the Miscellaneous group, than at night. More indi­ i � viduals were trapped at night than during the day among the ants, spiders, Isopoda, other arachnids, and Chilopoda and Diplopoda. Fur­ thermore , more individuals were found in the forest plot than in the cacao for the Or­ thoptera, spiders, and Homoptera. Table 1 lists the results of multiple analyses of variance for each arthropod group's abundance. The three factors in the analyses ofvariance were months of the study, forest versus cacao site, and day versus night. The purpose of such anovas is to test if there are significant effects due to any of these factors, and to test if there are any Fig. 1 Seasonal abundances of predominant arthropods and rainfall patterns for the same interactions between the factors.The table lists periodo the probability that there is a significant effect; probability values greater than .05 are not listed divided into spiders, ]l1ites, and other arachnids. and are not considered significant. The three­ Chilopoda and Diplopodawere placed together, way multiple anova used allowed for an adjust­ as were snails, worms, and leeches (all relatively ment for unequal sample sizes. The condition uncommon). Though snails, worms ana leeches ofhomoscedasticity was met. are not arthropods, they were the only other Significant effects exist for month of trap­ invertebrates found and were thus included. ping for all arthropod groups except larvae, The larvae were initially separated into Diptera, miscellaneous, and unidentifIed (all very rare). L1EBERMAN el DOCK: Arthropodfau na, La Selva, Costa Rica 29

Except fo r the Hemiptera (p = .006), Isoptera For ants, the average abundance in May is

(p = .03), and the Dennaptera (p = .007), these significantIy different from a11 other months; effects are significant at the .0001 level or March and April fonn a cluster, as do March. better. September, November, and the fo11owingJanu­ Interactions between sampling�month and ary. Sorne taxa have only two groups of months forest versus cacao occur in the following that cluster (e.g. 'spiders and mites) , while groups: Orthoptera, mites, Isopoda, Coleoptera, others evince a greater complexity, with up to Apterygota, Homoptera, Hemiptera, and five such clusters (Diptera, Apterygota). Isoptera. Interactions also occur between Figure 3 relates directly to Table 1. The month and time of day for ants, Orthoptera, abundances in the forest and cacao plots, Diptera, Isopoda, Coleoptera, Apterygota, separated into day and night collections, are Arachnida, Homoptera, Hemiptera, and tabulated. They are tabulated only for those Dermaptera. For many groups, the difference arthropods with significant effects dile to either between day-night abundances depends on the the forest-cacao factor, the day-night factor or month (and possibly the associated rainfall). both. For other groups, the diffe rences between Table 2 represents the Pearson produ ct­ abundances in the fo rest versus cacao plots also moment correlations between the numbers of depend on the month of the year. Three-way individuals of the arthropod groups. Those interactions between month, forest-cacao, and correlations significant at the .001 and .0001 day-night occur only for the Apterygota and levels are indicated. Such stríngent significance the Homoptera. levels were deemed necessary due to the large Figure 2 represents the results of Duncan's sample sizes. multiple-range a posteriori test. This test was DISCUSSION run o subsequent to the three-way multiple anovas on aH abundance variables for which Many studies have shown that pit traps (also there was a significant effect due to month. In caBed can traps or pitfall traps) are the most Figure 2, lines connect those months for which efficient way of coHecting litter arthropods and the means (average abundance of the group) are evaluating thcir abundances over time (Baars, not significantly different at the .05 level. 1979; Meijer, 1974; Mitchell, 1963; Uetz and Unzicker, 1976). Specifically,. Meijer (1974) .ITHIOII'OD MONTHS OooUl'S 1-_-. ,1,-, ,'1-.,,'1 -,_', I'�-, rI .�-"" I -"..' 'I-� " I-..--" I'M- , r1 ...-1"' 1....,- ,. ¡-.,, ', 1---1 and Denboer (1971) have founa pittraps to be .L-...L. , vtiry successful as measures of relative abundance and particularly useful in comparing populations in differe nt sites, even though trap efficiency varíes from species to species and habitat to habitat (Mitchell, 1963). Baars (1979) cautions that many traps are needed in micro-spatially heterogeneous habi­ tats. The large sample size in this study reflects this consideration. The numbers recorded in this study must be viewed comparatively; Le. sorne arthropod groups may be sampled more completely than others, hut this selectivity will presumably be the same for forest and cacao and fo r day and night. Previous studies (Baars, 1979; Meiier, 1974; Uetz and Unzicker, 1976) have used smaller pit traps (approxiinately 0.8 liter). It should be emphasized that this study used one gaBon (3.8 liters) pit traps, thus Fig. 2. Results 'of Duncan'sMultiple Range a posteriorl increasing the size of the sample, and pre­ test. Bars connect those months for which sumably the efficiency as well. there is no significant difference between the mean abundances of the particular arthropod Monthly differences : As seen ,in Figure 1, group. rainfall begins to increase in May and June and 30 REVISTADE BIOLOGIATROPICAL

he

N>D

A B N>D U N D A N e E

dO

Fig. 3.Tótal abundance of arthropod group s, separated into forest (F) and cacao (C) sites. Shaded porti ons are during the day. Qnly arthr opod groups with significant effects due to either the forest-cacao or day·riight fact ors, or both, are included. Those effects that are signifi€ant at the 0.5 level are indicated above the individual histograms. peaks fo r the sampling period in July and tane rain forest (also in Costa Rica), found ' August. Numbers of all major groupsof insects maximum insect abundance late in the dry reach a maximum between the onset of the season and early in the rainy season. Wolda rain y season and the period of peak rainfaÍl . {l979), working with light traps in Panama, While a causal relationshio cannot be proven , a also fo und a large decrease in insect ab undance correlation between the onset of the rainy during the dry season. Denlinger (1980), using season and insect abundance is indicated. Malaise traps in Kenya, foun d a complex The peak rainfall months of July and August relationship between rainfall and abundance, show no significant difference in iosect abun­ with the peak of abundance coincident wi th dance from subsequent drier months for many maximum rainfall. arthropod groups (ants, Orthoptera, spiders, for Several factors influencing the response of example). Notably , it is the non-arthop ods,the insect populations to the onset of the rainy snails, worms, and leeches, that have a signifi­ season have been postulated. The increase in cantly different abundance in August; these are insect populations may be due to the height­ the groups that are likely to do particularly well ened availability of plant fo od (Buskirk and in wet areas. The two Februaries share a Buskirk, 1976; Wo lda, 1979). Work in progress cornmon group of non-significantly different concerns various biotic and abiotic fa ctors means for all insect groups, suggesting that the which were measured concurrently with the can year studied is generalizable to other years. trap samples. These include temperature, rain­ Bigger (1976) has shown cyclical behavior of f all, and numerous vegetation variables. insect populations over several years. Though Following analyses of variance it was fou nd our data are insufficient to generalize an annual that among the vegetation variables, only leaf pattern, a striking monthly variation in litter depth varies significantly with month. arthropod abundances has been shown . Others Specifically, litter depth is greatest in May and have showo comparable responses in tropical least in October and November. Herbaceous forest ecosystems. Buskirk and Buskirk (1976), cover appears to be greatest in March and April, working with sweep samples in a lower moo- and least in November and December (though UEBERMAN & DOCK: Arthropod fauna, La Selva, Costa Rica. 31

TABLE 1

Three-way multipleanalyses oi varÚlnce. The threefa ctors are month (l3 1evels) .forf!st or cacao (2 1evels), and doy or nigh t (2 levels). Listed �f!low are the probabilities that the effects are significantfor tnese three fa ctors and their interactions (*). If p �.05, the probabl1ity is not li sted. FC is forest cacao, and DN is day-nigh t.

MONTH MONTH MONTH 1 VARIABLES MONTH FC DN *FC *DN FC*DN FC*DN

Total .0001 .0001 .0004 .0001 Ants .0001 .0001 .0:395 Orthoptera .0001 .0001 .000 1 .0001 Spiders .0001 .0001 .0073 Mites .0001 .0208 .0204 Hymenoptera .0001 Diptera .0001 .0001 .0001 1 sopada .0001 .0183 .005 1 .0039 .0005 Coleoptera .0001 .0006 .0273 Apterygota .0001 .000 1 .0001 .0001 .0001 .000 1 Chilopoda Arachnida .0001 .0232 .000 1 .0016 Larvae Homoptera .0001 .0004 .0196 .0001 .0430 Hemiptera .0056 .0156 .0009 Isoptera .0324 .0111 Dermaptera .0007 .0003 Miscellaneous .0147 .0013 Snails, Worms .0001 Unidentified

Chilopoda, aboye, includes both Chilopoda and Diplopoda. Arachnida includes Arachnid� !?ther ' aH than spiders. Snails. Wonns includes snails, wonns, and lee ches. The miscellaneous category is explained in the texto not at the .05 level of significance). Thus, we Orthoptera, spiders, and Homoptera. These find that insect populations are greatest when groups may be particularIy sensitive to the there is more food available, just befare and at effects of environmental perturbations. In the the onset of the rainy season. case of mites, more individuals were found in The ability of insect populations to peak in the cacao than in the fo rest. Perhaps mites are response to the beginning of the rainy season more capable of exploiting pertu rb ed eco­ (whether the response is specifically to the systems. While other groups have more indi­ vegetation, the rainfall, both, or sorne other viduals in the cacao than the forest (ants, other factors) has been postulated to be due to arachnids, Chilopoda and Diplopoda), these developmental arrest, with environmental para­ differences are not large enough to be statis­ meters triggering the termination of diapause tically significant. (Denlinger, 1980). This may be happening in the leaf litter at La Selva as well. In the groups Further study is recomrnendedto determine studied, it seems far less likely that individuals if these groups, as well as the mites, are either are migrating in and out of the area. related ecologically or are (as a group) better capable o futili zing the resources in a disturbed Forest versus cacao : Mixed cacao plantations ecosystem. Conversely,study is also recomrnend­ are believed to harbor sorne of the most ed to determine if those groups more abun­ complex insect communities (Bigger, 1976). dant in the forest are indeed less capable of However, more individuals were found in the exploiting perturbed ecosysterns. Indeed, this forest plot than in the cacao for the shift in composition of the leaf litter arthropod 32 REVISTA DE BIOLOGIATROPICAL

TABLE 2

Pearson product-momen.tco"e lations

2 4 5 6 9 lO 11 12 13 14 15 )6 17 18 19 20

1- Formicidae X *+ *+ *+ *+ *+ *+ *+ *+ *+ *+ *+ 2- Orthoptera X *+ *+ *+ *+ *+ *+ *+ 3· Spiders X *+ *+ *+ *+ + *+ + *+ 4· Mites X 5· Hymenoptera X 6· Diptera X *+ *+ *+ *+ *+ *+ *+ 7· Isopoda X *+ *+ *+ 8· Coleop tera X *+ *+ *+ *+ *+ + 9· Apterygota X *+ *+ 10· Chilopoda X *+ *+ 11· Arachnids X + + *+ 12· Larvae X 13· Homoptera X *+ *+ *+ *+ 14· Hemiptera X + + *+ 15· !soptera X 16· Oermaptera X *+ *+ 17· Mise. X 18· Snails, Worms X 19· Unidentified X 20· Total X

The variables, above, are lhe numbers of individu als colleetcd per can trap (n= 138). The symbols used above are : *+: If p < .0001 +: If p < .001 In all cases, r >0.

fauna may have implications both in succession fore st. Several' workers have noted a compara­ and resource management in the tropics. ble dorninahce andfor preponde:rance of ants in It should be noted that total abundance is tropical forest litters and sOlls (Strickland, also greater in the forest. This is to be expected 1945; Wolda, 1977 ; Collins, 1980). Bigger in light of the concept that disturbed eco­ (1976) found 250 species of ants in a cacao systems, particularly in the tropics, have few er foresto Collins (1980), working in Sarawak on available niches, less environmental predict­ the soil macrofauna, found terrnites (62.2%) ability and gre ater variability , and sub sequent and ants (19.7%) to be the most abundant in a reduced abundance and species richness within lowland site . Qur finding of more ants than any and among very different taxa. Research is in other group (34.8%) and only 0.l9% terrnites p rogre ss to assess the ecological differences may in part be due to a soil-Ieaf litter differ­ between disturbed and undisturbed forest eco­ ence, or possibly an interesting Old World-New systems for leaf litter amphibians and reptiles World difference. This study fu rther supports that prey upon the arthropods in this study . the contention that ants are ecological domi­ nants. Diel differences: There were significantly more ants, Isopoda, spiders, and other Though not statistically si gnificant , the arachnids at night than during the day, and greater abundance of ants in the cacao may be more Apterygota and Diptera during the day due to the presence of one or more species than at night. It seems probable that niche which thrive in disturbed habitats. For exam­ Wasmannia auropuncta partitioning exists among the arthropods in the pIe, is present in both afeas, but does better in disturbed habitats, is forest leaf litter, with respect to time of day. more flexible in its requirements, and is a very aggressive species (Roy Snelling, pers. comm.). Ants: Ants are by far the most abundant There are more ants active at night than during macro-invertebrates in the leaf litter of this the day. It should be noted that day-night LlEBERMA N & DOCK: Arthropod fa una, La Selva, Costa Rica 33

differences may relate in part to the possibility James Talbot, ando R. Wayne Vandevender for that·arthropods may better see and avoid traps their invaluable help in collection and identifi­ during the day. Ants collected during the day cation of the specimens in Costa Rica. We also are larger on the avemge than those ,taken at want to thank Roy Snelling for reading and night. Furthermore, no difference exi�ts in ant commenting on the manuscript. size between the forest and cacao, su�esting that the same groups are utilizing both habitats. RESUMEN Different groups appear to be foraging at different times of the day. As seen in Table 2, Los grupos de artrópodos más fre cuen­ the abundance of ants corre lates highly with temente capturados con trampas cilíndricas de the abundances of most other groups, with the un galón (3 .78 litros) en ambientes de suelo exception of mites, other Hymenoptera, and vegetal en un bosque húmedo de tierras .bajas de Apterygota (and other rarer groops). Ant abun­ Costa Rica fueron hormigas, ortópteros y co­ dance is seen as an excellent predictor of overall lémbolbs. Las muestras indican variaciones esta­ arthropod abundance. Average ant abundance cionales significativas en la abundancia de los peaks in April and May, with a sharp drop froin grupos mayores. Los números de artrópodos May to June. As se en in Figure 2, March, April, fueron, mayores entre la iniciación de la tempo­ and May are significantly different from the rada lluviosa y el máximo de precipitación.Los other months. As in all other arthropod groups, colémbolos, moscas y grupos menos comunes the February values for the two years are not misceláne os se capturaron en mayor cantidad significantly different. durante el día, mientras que las hormigas, arañ as, milpiés y cienpiés se colectaron en Orthoptera: Orthoptera are the second most mayor cantidad durante la noche. se capturó abundant leaf litter group. The sampling una mayor cantidad de artrópodos en los method used may underestimate very mobile, ambientes no perturbados que en los habitats hopping and flying groups such as Orthoptera. modificados · de cultivo de cacao. Los grupos While they may be underestimated, their num­ más comunes en ambientes no perturbados bers are great enough for meaningful com­ fueron los ortópteros, arañas y homópteros. parisons. Though there is no difference in Los mayores números de artrópodos correspon­ Orthoptera abundance between day and night, den a los espesores máximos de cubierta vegetal � there is a substantial difference between the durante los trece meses de muestreq. forest and cacao. Orthopteran abundance is reduced in the disturbed ecosystem (Fig. 1). Unlike ants and Apterygota, Orthoptera may be LlTERATURE CI TED less able to tolerate environmental perturba­ tion . Orthoptera al so peak in April and May, Allen, W.C. 1926. Distribution of animals in a tropical just prior to the rain iest months. rain-forest with relation to environmental fa ctors. Ecology, 7: 44 5-468. Apterygota: Th e third most abundant group Anderson, J.M. 1975. Succession, diversity, and is the Apterygota. Their abundance in tra ps trophic relationships of sorne soil anirnals in varies greatly, from 0-36 in December­ decomposing leaf litter. J. Anim. E col. , 44: February, to 899-1011 in April-June. 475-495. Lieberman (1982) has shown Apterygota to be Baars, M.A. 1979. Catches in pitfall traps in relation to a rare prey item for Amphibians and Reptiles, mean den sitie s of carabid beetles. -Oecologia which may be related to this un predictable (Berl.), 41: 25-46. abundance. Apterygota are significantly more abundant during the day than at night, while no Bigger, M. 1976. Oscillations of tropical insectpo pula­ 259 : 207-209. forest-cacao difference in abundance exists. tions. Nature,

Buskirk, R., & W. Buskirk. 1976. Changes in ACKNOWLEDGEMENTS arthropod abundance in a highland Costa Rican Forest. Amer. Midl. Nat., 95: 288-298. This study was undertaken as part of NSF Collins, N.M. 1980. The distribution of soil macro­ Grant #BMS 7301619 A01, to Jay M. Savage fauna on the West Ridge of Gunung (Mount) Mulu, and Ian R. Straughan. We thank Cad Ueb, Sarawak,Oecolo gia (Berl.), 44: 263-275. 34 REVISTA DE BIOLOGIATROPICAL

Davis, D.E. 1945. Theannual cycle of plants, mosqui­ Meijer, J .1974. A comparativestudy of the immigration tos, birds and mammals in two Brazilian forests. of carabids (Coleoptera: Carabidae) into a new Eco!' Monogr., 15: 243-295. . polder. Oecologia (Ber!.), 16: 18-208.

Denboer, P.J. 1971. Stabilization of ani¡nal numbers Mitchell, B. 1963. Ecology of two carabid beetles. 11. and the heterogeneity of the envhonment, p. Studies on Populations of ádults in the field with 77-97. In Proc. Adv. Study Inst. Dynamics. Num. special reference to the technique of pitfall Popu!. (Oosterbeck 1970). Wageningen:Pudoc. trapping. J. Anim. Eco!. , 32: 377-392.

Denlinger, D. 1980. Seasonal and annual variation of Montgomery, G., & Y. Lubin. 1979. Abundance and insect abundance in the Nairobi National Park, diversity of arthropods in the canopy of a moist Kenya. Biotropica, 12: 100-106. tropical foresto SmithsolÚan Contr. to Zoo!. (In Press). Elton, C. 1973. The structure of invertebrate populations inside neotropical rain foresto 1. Anim. Stanton, N. 1979. Piltterns of species diversity in E col. , 42: 55-104. temperate and tropical litter mites. Ecology, 60: 295-304. Elton, C. 1975. Conservation and the low population density of invertebrates inside neo tropical rain Strickland, A.H. 1945. A survey of the arthropod soil foresto Bio. Conservation, 7: 3- 15. and litter fauna of sorne fo rest reserves and cacao estates in Trinidad, British W. Indies. J. Anim. Hanlon, R., & J. Anderson. 1972. The Effe cts of Eco!., 14: 1-11. Collembola grazing on microbial activity in decomposing. leaf litter. Oecologia (Ber!.) , 39: Uetz, G.W. 1979. The influence of variation in litter 93-99. habitats on spider communities. Oecologia (Ber!.), 40: 29-42. Holdridge,¡ L.R: 1967. Life zone ecology. Tropical Science Center. San José, Costa Rica. Uetz, G.W., & J.D. Uilzicker. 1976. Pitfall trapping in e cological studies of wandering spiders. J. Arachnology, A3: 101-1 11. Janzen, D. 1973. Sweep samples of tropical foliage insects. Ecology, 54: 687-708. Wolda, H. 1978. Fluctuations in abundan ce of tropical insects. Amer. Nat., 112: 1017-1045. Lieberman, S. 1982. The ecology of the leaf litter herpetofauna of a Neotropical rain forest: La Wolda, H. 1979. Seasonal fluctuations in rainfall, Selva, Costa Rica. Ph.D. Díssertation. University of food, and abu ndance of tropical insects. J. Anim. Southern California. Eco!., 47: 369-382.