ECOGRAPHY 22: 424-435. Copenhagen1999

Leaf litter and the small-scaledistribution of carabid l (Coleoptera,Carabidae) in the boreal forest i ! Matti Koivula, Pekka Punttila, Yriii Haila and Jari Niemelii

Koivula, M., Punttila, P., Haila, Y. and Niemeki, J. 1999. Leaf litter and the small-scale distribution ofcarabid beetles (Coleoptera, Carabidae) in the boreal forest - Ecography 22: 424-435.

Management practices favoring conifers at the expense of deciduous tree species, and the eradication of deciduous trees, especially aspen Populus tremula, from managed forests have resulted in population declines in several species in Fennoscandia. ln addition to species depending on decaying wood of deciduous trees, earlier evidence suggests that leaf litter, especially that of aspen, is favored by many carabid species. We ran a four-year experiment in order to compare carabid assemblages of un- changed forest floor with artificially created leaf-litter plots in central . A total of 18 plots (5 m in diameter) were established in three forest stands without aspen a few kilometers apart. Each stand had 3 litter plots (litter added) and 3 control plots. Pre-treatment samples were compared with those collected after litter addition. The litter addition affected the carabid-assemblage structure by increasing the catches of some species and decreasing the catch of one species. The number of carabid species was similar in control and litter plots. The litter elfect was smaller than variation among forest stands and year-to-year fluctuations. There was a strong "rich" "rich" temporal constancy among the plots: plots remained from year to year "poor" "poor". and similarly. plots remained The significant influence of leaf litter on carabid abundance can be attributable to both abiotic factors (microenvironmental conditions, especially humidity and temper- ature), and biotic ones (changes in niche structure, improved food supply). Leaf litter seems to have an effect on carabid distribution patterns, and deciduous trees scattered among conifers are likely to be of importance on carabid fauna in boreal IOrests.

M. Koiuula ([email protected]) and J. Niemeki, Dept of Ecology and Systemat- ics, Diu. of Population Biology, P.O.B. 17, FIN-00014 Unio. of Helsinki, Finland. - P Punttila, Finnish Enuironment Inst., Nature and Land Use Diuision., P.O.B. 140, FIN-00251 Helsinki, Finland. - Y. Haila, Dept of Regional Studies and Enuironmental Policy, Uniu. of Tampere, P.O.B. 607, FIN-33101 Tampere, Finland.

Boreal coniferous forests are heterogenousat several further, the suitable patches may be small and well spatialscales (Haila et al. 1994a,Niemeld et al. 1996, separated from each other. Esseenet al. 1997).Within the forest stands,variation Human-induced changes in the pattern of natural in fertility, humidity, microtopography, tree-species patchiness,however, may possessa seriousthreat to the compositionand heterogeneityof herb-layervegetation continued existence of several forest-inhabiting organ- result, in a microhabitat mosaic on the forest floor isms. In large parts of the boreal region, both environ- (Griim 1971, Niemela et al. 1992). This mosaic is mental heterogeneity and dynamics have changed as a reflectedin the distribution of the forest-floor fauna. consequence of intensive forest management (Hansson Often the microhabitats meeting the habitat require- 1992, Esseen et al. 19971.Qualitatively, certain forest mentsof individual speciesare patchily distributed and environments have become rare-and it is thus impossi-

Accepted27 February 1999 CopyrightO ECOGRAPHY 1999 rssN 0906-7590 Printed in Ireland - all rights reserved

424 ECOGRAPHY 22:4 (1999) ble for speciesrequiring theseenvironments to survive aspen leaf-litter plots and unchanged forest floor, as- and, quantitatively,spatial conflguration of populations suming that the leaf-litter microenvironment differs de- are affected(Haila and Kouki 1994).Fennoscandian cisively from the surrounding forest floor. examplesof increasedpatchiness of certain microhabi- In the present paper, we test whether 1) the abun- tat types include the destruction of spruce mires dances of individual species, 2) the number of species, through efficientditching, and eradicationof deciduous and 3) the carabid-assemblage structure are affected by trees (seee.g. Niemelii 1997,Esseen et al. 1997 ar'd the leaf litter. referencestherein). Large aspensscattered among conifersare important for many threatenedinsects, molluscs, lichens and poly- pores(Nuorteva 7987,Kuusinen 1994,Martikainen et at. 1994,Kotiranta and Niemelii 1996,Niemelii 1997, Material and methods Esseenet al. 1997).Inaddition to the wood material as The study design such, also the leaf-litter quality affects invertebrate distributions both directly and indirectly (Facelli and The studywas donein SeitseminenNational Park, Pickett 1991).Distributional data indicate that espe- CentralFinland (61'57'N, 23'24'E), on the borderbe- cially aspenlitter is important (e.g.Bossenbroek et al. tween southern and middle boreal zones (Ahti et al. 1977,Byers 1984,Niemele 1990,1997, Niemelii et al. 1968),in 1993-1996.We chosethree spruce-dominated 1992).Furthermore, experimental studies have shown Myrtillus-type forest stands(Cajander 1949) with no or that leaf litter is very important for microarthropods, very few deciduoustrees, but potentially suitable for slugsand snails,and spiders(Uetz 1979,Bultman and aspen.The standswere severalkilometers apart, previ- Uetz 1982, 1984, Seastedtet al. 1983,Reddy 1984, ously managedforests (the age of the dominant spruces Boag 1985,Bird and Chatarpaul1988). was 90-120 yr). The herb layer in the stands was Fennoscandianforest managementhas aimed at even dominatedby Vacciniummyrtillus and Vacciniumuitis- agedand sizedmonocultures of conifers,that is, Nor- idaea dwarf scrubs, and the moss layer by Dicranum, way sprucePicea abiesand Scots pine Pinus syluestris. Pleuroziumand Hylocomium species.The size of e.ach As a consequence,aspen has been eradicatedfrom the forest stand was several hectares.The surroundings managedforests during recent decadesbecause of its were different types of usually younger, spruce- or low economicvalue and its status as an intermediate pine-dominatedforests. host of the rust fiingus Melampsorapinitorqua which In eachforest stand, we establishedsix circularstudy attackspine (e.g.Kiirkkdinen l98l). Aspenis a promi- plotswith a diameterof 5 m (Fig. 1).Distances between nent componentin natural succession,and its removal the plots were 25-30 m. The minimum distancefrom altersthe successions. the edgeto anotherforest stand was 30 m to the nearest In this paper, we examine the effect of aspen-leaf study plot. litter on carabid-beetleassemblages by comparingartifi- The floristiccomposition of the plots was examined cial leaflitter plots with unchangedconiferous forest tn 1993prior to the adding of the litter by estimating floor. If the litter was found important for carabidsand the percentagecover of each of the moss and vascular affectedtheir distribution patterns, it is probable that plant species(scale 0-100%). Thesedata wereused in the leaf-litter microhabitat is important for many other evaluating the effect of vegetation variation on the taxa,too. We usedaspen-leaf litter becauseit is known carabid assemblages.Also the numbersof the wood to be important for many forest organisms,including ants (Formica rufa - group) were counted from the threatenedspecies of many taxa (Niemelzi 1997), and pitfall traps. We used these catchesas environmental partly becauseits leaves can easily be collected in data for carabidsbecause earlier work has shown that sufficientamount. wood ants have a profound impact on the carabid We chosecarabid beetles as a study object because fauna in borealforests (e.g. Niemelii et al. 1992,Punt- they are ecologicallydiverse and reliable indicators of trla 1994). many environmental characteristics and qualitative The effect of leaf-litter addition on carabid- changesof their environment(e.g. Langor er al. 1994). assemblageswas studied using "Before-after with con- Furthermore, they are well known both taxonomically trol-impact" - studydesign (BACI; Underwood1991). and ecologically.They are also easyto collect in large Here, the impact was the leaf-litteraddition. In 1993, enoughnumbers with pitfall traps. Carabidsmay move we studied the before-treatmentspecies composition of up to hundredsof metersby foot under some circum- the stands (pre-treatment trapping). In late autumn stances(e.g. Mascanzoniand Wallin 1986,den Boer 1993,litter was addedto nine randomly selectedplots 1990), and specieswith wings and flight capability (threein eachof the threestands, Fig. 1). Just after the obviouslymuch more. Thus, we assumethat the scale leaf-litter addition, the thicknessof the litter layer was we usedin this study is appropriatefor detectinghabi- 5 l0 cm, the total amountof litter per plot being 250 tat selection:carabids should be ableto choosebetween l. In spring 1994,the thicknesswas 3-5 cm. To mimic

ECOGRAPHY 22:4 (1999\ 425 natural leaf fall we added 80 1 of litter to each lit- Data analyses ter plot in late autumn 1994 and 1995. This did Differencesin beetlecatches between the two plot types not increasethe thickness of the litter layer the fol- wereexamined by repeatedmeasures analysis of variance lowing spring. The leaf litter began to decay quite (ANOVA). The data from 1993("before-impact") and rapidly in early summer 1994, approaching the from 1994,1995and 1996("after-impact") were trans- characteristicsof naturally fallen leaf litter. In 1995 formed to individuals per 100 trapping days. Normal and 1996, the decaying layers were very close to distribution of the data was achievedby log(x+ 1) natural conditions:the upper layer consistedof dry transformation.We used treatment (control/litter) and leaves, and the deeper layer was well decayed with stand (the three forest stands)as factors.The data from lots of fungi and even earthworms (Lumbricidae) 1993were used as covariablesin the ANOVA. (Koivula unpubl.). To study changesin the structure of carabid-beetle The leaf-litter material was raked from roads near assemblages,we calculatedShannon-Wiener and Simp- Seitseminenpark area. We examined random a sam- sondiversity indices (e.g. Krebs 1989).The indiceswere ple of 40 I of this litter material leaf by leaf using calculatedfor eachplot and the data wereanalysed with loupes (8 x ). In addition, we dried 20 I of litter in repeatedmeasures ANOVA, using treatmentand forest (see Berlese-Tullgrenfunnels e.g. Southwood 1978). standas factors,and data from 1993as covariables. No carabid-beetleadults or larvae were found in the Speciesrichness among the stands,sampling years and litter material. The majority of the invertebratesin the treatmentswas compared by calculatingthe expected the litter were mites (Acarina) (totally 345 individuals number of speciesfor standardizedsample size in each were found using both methods) and springtails plot type in eachstand (pooled catches ofthe threeplots (Collembola) (totally 83 individuals found). of eachtreatment) for the samplesfrom different years The beetlesamples were collectedwith pitfall traps usingrarefaction (Simberloff 1978). (diameter 65 mm, depth 88 mm), partly filled with We performedmultivariate analyses(e.g. Jongman et 30% ethylene glycol and detergent. The traps were al. 1995)using CANOCO 3.15 statisticalsoftware (ter covered with plexi roofs (10 x 10 cm) to protect Braak 1987and updatenotes) in order to explorethe them from litter and rain. In order to increase the floristic and faunistic variation among the plots and sampling efficiency,we used two perpendiculartrans- further, in order to relate the carabid-assemblagevarra- parent polycarbonatepanes (100x 20 cm each) to tion to the environmentalvariables. For the pre-treat- guide the beetles into the traps (Fig. 1). Three ment data (1993)of both vegetationand carabids,we groups of four traps were used in each study plot, applied a principal component analysis (PCA) to ex- for a total of 216 traps (3 standsx 6 plots x 12 plore the floristic and carabid-assemblagevariation traps). The trapping period covered most of the among stands and plots. We included all the plant growing seasonfrom mid-May through the beginning speciesin this analysis,and all carabid speciesoccurring of September,and the traps were emptied once a in more than one out of 18 samplessamples. Further, month each vear. 1993-7996. we usedredundancy analysis (RDA) to explorewhether

plotdiametor 5 m 12 pitfalltraps

.t- ....).F I -+?t I | '.1" IP l('' t- .L I:I rry- l

-tu(rl r3| A, B, C - brest stands O - controlplot a - lltterplot

Fig. 1. The design

426 ECOCRAPHY 22:4 (1999\ Table 1. The numbers of carabids caught before (1993) and after (1994 1996) the adding of the litter. C: control, L: litter plots. Nomenclature follows Lindroth (1985. 1986).

Species 1993 1994 t995 1996 t994 1996

(c) (L) C Total '78'7 Calathus microp terus (Dft.) 168 137 136 183 437 609 2t4 29s 1087 2179 P t ero st ichus oblongopunc tatus (F .) 67 3t 81 165 206 251 15 22 302 438 838 biguttatus (F.) 25 40 34 47 34 )L 60 52 128 131 324 Cychrus caraboides L. 28 10 18 z) 22 43 t2 20 52 86 176 Carabus glabratus Payk. 10 510 2 t2 10 10 9 32 21 68 Amara brunnea (Gyll.) 8 12 6 4 2 2418 t2 38 Carabus hortensis L. / 235 5 2 2l10824 (Hellw. 2 Leistus rcrminarzs in Pz.) 5 I 1 l-) 2919 Dromius agilis (F.) I 4l2 -1 i39 Dromius fenestratus (F.) I I -12 P terostichus strenuus (Pz.) 1 I la Car abus can c e llatus lll. I _I a1 Trechus secalis (Payk.) I I Trechus rubens (F.) I -11 Harpalus quadripunctatus Dej. 1 -l P t er os t i chus di lig ens (Slurm) 1 1-1 No. of individuals 319 nt 2s: 440 i21 953 3t7 40; 1333 1801 3685 No. of species t2 8 8 11 8 11 9 ttl l0 14 16 No. of trapping days 109 110 106 106 322 431 the variation in the beetle assemblages was related to sequenttesting using Monte Carlo permutations(n : the floristic variation in the pre-treatment year 1993. 99 permutations for repeated measurements, We included log(x + l)-transformed catch of wood ants conditioned on covariables:permutation classeswere in these environmental data. In this analysis, we in- determined by block and time-BACI design in cluded again only carabid species occurring in more CANOCO). The data were centered by speciesand than one plot. We applied forward selection of environ- log(x* l) transformed. mental variables by first examining which single vari- able explained the species variation best. The statistical significance of the relation between this variable and the species data was subsequently tested using Monte Results Carlo permutations (n : 99, permutations conditioned Species richness and abundance on blocks, i.e. the stands), and if the relation was found statistically significant (critical p:0.05) the variable The carabid-beetledata consistedof 3685 individuals was added into the explaining model. The selection was representing16 species(Table l). The most numerous continued in the same manner by examining the rela- speciesboth in the litter and control plots were Ca- tion between the remaining variables and the residual lathus micropterus,Pterostichus oblongopunctatus, No- variation of the species data. Redundancy analysis was tiophilusbiguttatus and Cychruscaraboides. These four also utilized in testing the successof randomization in speciesmade tp 95.4ohof the total catch.Three species both floristic and carabid data ofthe pre-treatment year were found exclusively in the litter-plot samples in "imag- 1993.This was done by testing the effect of the 1994-1996,and one speciesexclusively in the control- inary" environmental variable, litter treatment. In the plot samplesin the sameperiod. above multivariate analyses, we applied centering by The numbersof individuals caught varied a lot from speciesand all the data were log(x+ 1) transformed. year to year. The mean catchespeaked in 1995being The effect of stand (block) was partialled out by defin- clearly higher then than in the other years (Fig. 2). ing the stands as covariables. Pterostichusoblongopunctatus catch was lowest it 1996 We applied redundancy analysis also to test the effect (only 37 individuals)and highestin 1995(457 individu- of the litter treatment on the structure of the carabid als). Also C. miuopteru,sand C. caraboidespeaked in assemblage in the BACI data (1993-1996). In this 1995,whereas N. biguttatuspeaked one year later (Fig. analysis, we included species which occurred in more 3). The total carabid catcheswere consiitentlymuch than two samples (out of 72 samples). The effects of higher in stand C than in the other stands,and after the stand (block), year (time) and plot were partialled out start of the experiment,consistently somewhat higher in by making them covariables. The litter treatment and the litter plots than in the control. Also the mean the log(xA l)-transformed catch of wood ants were number of species'caughtwas consistentlyhigher in used as environmental variables. These environmental stand C than in the other standsthroughout the study variables were subjected to forward selection with sub- period,but no clearlitter effectwas detected. However,

ECOGRAPHY 22:4 (1999\ 427 CABABIDBEETLES Vegetation structure and carabid assemblages before the treatment Principal component analysis(PCA) of the vegetation data before the litter treatmentindicated that the three o ;e standswere rather clearly separatedfrom eachother in 3 100 the ordination spacedetermined by the first two princi- + pal components. These floristic differencesprobably E75 reflectedsubtle differences in the fertility and shadingof the stands.The three standsrepresented typical varia- o c50 tion within mature sprucedominated Myrtillus type, i.e. c o the mosslayer was characterizedby high cover of either o Pleurozium (especiallyin the most well-lit stand B), 25 Hylocomium(the most fertile stand C) or Dicranum(lhe most shady stand A) mosses.In the partial PCA, a large proportion of the floristic variation 1993 1994 1995 1996 was explained by the covariables:29.6ok of the total floristic variation Fig.2. Themean numbers ofcarabid beetles per 100 trapping was among daysin controland litter plots in 1993-1996. the stands.As amongthe stands,the gradi- ents in fertility were detectablealso among the plots within the standsin the ordination space,determined the meannumber of speciesincreased in the litter plots by the two first principal components.This ordination of all three standsand decreasedsimilarly in respective spacecontained 31.0% of the total floristicvariation in controlplots in the flrst yearafter litter wasadded. This addition to the proportion already explained by the was at least partly attributable to the increased sample covariables.A partial redundancyanalysis testing the size in litter plots. "imaginary" When the effect of samplesize was effect of the litter treatment verified our removed using rarefaction, the expected number of successin randomizing the litter treatment among the speciesin samplesstandardized to 30 individualswas plots. There were no systematic floristic differences similar both among the stands and betweenthe plot between the control and litter plots before the litter types.The only statisticallysigniflcant differences were addition(Monte Carlo permutationtest; n:99 permu- in the control plots of standA: the expectednumber of tationsconditioned on blocks,F:0.65, p:0.800). specieswas higher in 1993than in the subsequentyears. Principal component analysis of the carabid data The slight, albeit non-significant,decrease in the ex- before the litter treatment revealedsimilar pattern as pectednumber of speciesin all standsand both plot observedin the floristic data. The three stands were typesin 1995was due to the peakin the catchesof the again separatedfrom eachother, the greatestdifference few dominant species. being betweenthe most fertile stand C and the two less fertile stands (the main gradient). Catchesof most of the specieswere highestin the stand C, whereaslarge catchesof Notiophilusbiguttatus in the stand B was the

6 main causeof the slight separationof the somewhat 3o more shadystand A from the more well-lit stand B (the

;4 secondarygradient). Partial PCA revealedthat a very large proportion of variation was explained by the 36 covariables:59.7% of the total variationwas among the stands,and the ordination spacedetermined by the two first principal components explained an additional 22.0%of the total variation. The two main gradientsof variation were characterized by the catch of C. caraboides(and also the catchesof most other species) increasing to one direction, and by the catch of N. biguttatusincreasing to the perpendiculardirection (i.e. the main gradientswere again rather similar as among the stands).As with the floristic data, partial redun- dancy analysisshowed that the effect of the imaginary litter treatment was not significant (Monte Carlo per- Fig. 3. The mean numbers of the four most numerous carabid mutation test; n:99 permutationsconditioned species per 100 trapping days in 1993-1996. Note dilferent on scales in the two upper and the two lower figures in the blocks,F :1.62, p:0.180). Thisverifled our successin vertrcal axes. randomizing the litter treatment among the plots also

428 ECOGRAPHY 22:4\1999) as regards to beetle assemblages.A partial RDA of the carabid data with forward selection of environmental A Stand-A, control variables revealed that the catch of the wood ants was a Stand-A,mer tr Stand-8, control the best variable to explain the beetle-speciesvariation. I Stand-B, litler Only one additional variable, the cover of the moss o Stand-C, control Hylocomium splendens, improved the fit between the a Stand'C, litter model and the carabid data significantly. These two df, to' A variables explained 1l.9ok of the total variation in the -!n-A (in N carabid data addition to the 59.7o/oexplained by the p r covariables; above). The result showed that the catches 4 of most species (the a rc only exceptions were try'.biguttatus A a d atd Amara brunnea) were negatively correlated with the o I a catch of the wood ants (a gradient of increasing inter- ference and/or predation). Additionally, the catches of especially N. biguttatus, Leistus terminatus and C. caraboides were positively correlated, and those of Carabus hortensis and C. micropterus negatively corre- lated with the cover of the moss Hylocomium (pre- (a) sumably a gradient of increasing fertility; gradient perpendicular to the wood-ant gradient).

The effect of litter addition on carabids Noti bigu . In the partial RDA of the BACI beetle data and the environmental variables (litter treatment and the wood- ant catch), the covariables again explained most of the Cala micr variation among the samples (Fig. a). Of the total o .9 variation, 77.4ok was attributable ... a Leisterm to the covariables 0.5 Dromagil AM TDruN (stands alone 40.lok, time alone 16.8oh and plots alone . Caraglab 60.6% when the variation was decomposed for each covariable class). Litter treatment was found to have a . Cychcara significant effect on the beetle assemblage (no. of per- . Pteroblo mutations: 99, F :4.15, p :0.010), whereas the wood-ant catch did not significantly explain the resid- ual variation (no. of permutations : 99, F : | .32, p : .0,5 0.170). The litter treatment, however, explained only (b) Axis 1 1.7"k of the total variation in the species data. The Fig. 4. a) Partial RDA ordination of the sampling plots catches of especially C. caraboides, L. terminatus and C. according to their carabid samples 1993-1996 (the effects of micropterus correlated positively, and those of Dromius stand, time and plot removed by using these as covariables). agilis and N. biguttatus negatively with the litter treat- Key for the plot types and the stands is given in the figure. b) Partial RDA ordination of the carbid species in the sampling ment (Fig. 4a). In addition to the litter-treatment effect, plots (parallel to a). The active environmental variable (litter there was another rather clear gradient in the ordina- treatrrlent) and the passive one (wood-ant catch, log(Formica)) tion of the samples in the ordination space determined in the analysis are shown as arrows. by the two first RDA axes (the first axis was con- strained to be the litter-treatment axis). This gradient included in the covariables (plots) in the present RDA. was due to large catches of N. biguttatas in one direc- The ordination also suggests that the litter addition tion and to large catches of P. oblongopunctatus and C. may have lowered the catches of the wood ants, al- caraboides in approximately the opposite direction. The though such effect was not found signiflcant in the wood-ant catch did not contribute statistically signifi- ANOVA (see below). cantly to the species variation, but when we fitted this The repeated measures ANOVA results for the wood variable as a passive environmental variable to the ants, total carabid catches and the four most common ordination, the direction of the wood-ant vector closely carabid species are presented in Table 2. The effect of matched the direction of the gradient observed in the the factor Stand was statistically significant C. ordination of the samples. As the wood ants are territo- caraboides, the factor Treatment in the total carabid rial, it is likely that their effect on the carabids (see the catchesand three out of four most common species(C. RDA results for the pre-treatment data) is already micropterus, P. oblongopunctatus and C. caraboides).

ECOGRAPHY 22:4 (1999\ 429 The interaction term Stand * Treatment was Significant Wiener diversity index values(H') varied between0.25 in C. caraboides catches, and the covariable (catch in alnd1.37 rn 1993,and between0.22 and 1.53in 1994- the pre-treatment year 1993) in the wood ants, total 1996among the plots. The mean H' valuesin control carabid catch and three out of four most common plotswere the same in 1993and in 1994-1996(0.91). In carabid species, with C. caraboides an exception. The the litter plots, the valueswere 0.96 and 0.92,respec- statistical signiflcance of the covariable indicates a tively. The Simpson'sevenness index values(D) varied strong temporal constancy among the study plots, i.e. between0.10 and 0.68 in 1993,and between0.08 and "rich" plots with large beetle catch in 1993yielded large 0.73in 1994 1996.In 1993,the meanvalues of D were "poor" catches also in 1994-1996 and, similarly, plots 0.49 in control and 0.48 in litter plots. After the leafJit- "poor". remained Overall, the litter-plot samples har- ter addition(i.e. 1994-1996),the meanvalues were 0.48 bored signiflcantly more carabid beetles than the con- and 0.47, respectively.Thus, there were no treatment trol-plot samples(Table 2, Frg.2). Also C. micropterus, effects on the values of the diversity and evenness P. oblongopunctatus and C. caraboides were more nu- indices.The forest stand C had higher valuesof Simp- son's evennessindices than merous in the litter-plot samples as compared with the the other stands,indicating that the proportion of "numerous" specieswas higher control-plot samples (Fig. 3). Notiophilus biguttatus did than in the two other study forest stands. Thus, the not respond to the litter addition, and there were no leaf-litter addition seemedto have no effect on the differences among the forest stands. Forest stand was carabid-assemblagestructure of the studied forest an important factor in explaining the observed differ- stands. ences in the C. caraboides catches among the samples. Interaction between treatment and forest stand was significant in C. caraboides numbers, indicating that the litter effect was dependent on the stand. To study the Discussion possible effect of leaf-litter addition on carabid-assem- blage structure, we calculated two different diversity Our results indicated that 1) both the total catchesof indices for the study plots (Table 3). The Shannon- carabid beetlesand the catchesof three out of four

Table 2. The results of the repeated measures ANOVA (F) for the total catch of wood ants (Formica rufa group), carabids and the four most common carabid species. Stand : forest stand (A, B, C), Treatment (control, litter) and N1993 : no. of individuals caught in 1993 (covariate).

Group/species Source SS

Wood ants Stand 0.052 2 0.026 0.231 Ns Treatment 0.099 I 0.099 0.886 Ns Stand * treatment 0.471 2 0.238 2.128 Ns N1993 1.579 I 1.5'79 14,097 0.003 Error t.232 ll 0.112 Carabid beetles Stand 0.153 2 0.016 t.479 Ns Treatment 0.658 I 0.6s8 12.'752 0.004 Stand x treatment 0.113 2 0.056 1.091 Ns N1993 0.436 1 0.436 8.456 0.014 Error 0.568 11 0.052 C. micropterus Stand 0.030 2 0.015 0.015 Ns Treatment 0.507 I 0.507 6.561 0.026 Stand x treatment 0.067 2 0.034 0.436 Ns N1993 0.471 I 0.471 6.097 0.031 Error 0.849 11 0.07'7 P. oblongopunctatus Stand 0.347 0.114 2.94'7 Ns Treatment 0.570 0.570 9.674 0.010 Stand * treatment 0.348 0.174 2.948 Ns N1993 1.035 1.035 17.s62 0.002 Error 0.649 l1 0.059 N. biguttatus Stand 0.0'72 2 0.036 0.345 Ns Treatment 0.149 I 0.149 1.438 Ns Stand x treatment 0.134 2 0.067 0.645 Ns N1993 1.045 I 1.045 10.083 0.009 Error 1.t40 l1 0.104 '1,.512 C. caraboides Stand 2 0.156 20.569 < 0.001 Treatment 0.342 I 0.342 9.307 0.011 Stand * treatment 0.334 2 0.16'7 4.5s1 0.036 N1993 0.005 I 0.005 0.12'7 Ns Error 0.404 l1 0.037

430 ECOGRAPHY 22:4 (1999) Table 3. The result of the repeated measures ANOVA (F) for Shannon-Wiener and Simpson diversity indices. The Tukey's post-hoc test was done using pooled data 1994-1996 and without covariables. The test indicates which plot type (control, litter) or stands (A, B, C) differed from the others. For example, C>A, B indicates that the sample sizesin stand C were higher than in stands A and B (which were thus similar)

Shannon-Wiener's H' Source SS DF Post-hoc test

Treat 0.002 1 0.002 0.354 Ns Stand 0.025 2 0.013 1.955 Ns Treat x Stand 0.066 2 0.033 5.098 0.027 N1993 2.00'7 I 2.007 310.569 < 0.001 Error 0.071 l1 0.006

Simpson'sD Source SS DF Post-hoc test

Treat 0.000 I 0.000 0.007 Ns Stand 0.027 2 0.014 8.802 0.005 C>A, B Treat x Stand 0.017 2 0.009 5.529 0.022 N1993 0.620 1 0.620 398.655 < 0.001 Error 0.017 u 0.002

most common speciesincreased in the litter plots as litter plots. Additionally, the trapping efficiency may comparedwith the control ones, but 2) we did not have been different in different plot types (see e.g. detect any litter effects on the number of speciesor Southwood1978). diversity and evennessindices, whereas 3) the carabid- Some important characteristicsof the eight most assemblagestructure was affected owing to different common carabid speciesin Finnish coniferous forests responseof speciesto the litter addition. We discuss are presentedin Table 4. Many species-specificcharaa- thesefindings in relation to speciescharacteristics, habi- teristics, such as feeding behavior, may be related to tat and microhabitatvariation" and soecies'interactions speciesdistribution in associationto leaf litter. Cara- below. bids are mainly predators,but nearly every specieshas quite diverse diet consistingof invertebrates,carrions and plant material(Hengeveld 1980a, b, c). Of the two most common forest-carabidspecies in mature taiga Species characteristics explaining the distribution (Niemeliiet al. 1988),P. oblongopunctatusis a generalist patterns but C. micropterus seemsto be strictly carnivorous Carabidsare often non-randomlydistributed on the (Lindroth 1986,see also Hengeveld1980c). The repre- forestfloor, althoughthe environmentmay seemto be sentativesof genus Amara are plant eaters feeding homogeneousto the human eye (Griim 1971,Niemeld mainly on seedsbut also on other parts of plants(e.g. et al. 1986,1992,L]uff 1987,Halla et al. 1994b).Cara- Lindroth 1986).In contrast,N. biguttatuseats mainly bid populationsconsist of decision-makingindividuals springtails(Collembola) (Hengeveld 1980a, b, c, Lin- (den Boer 1979),which activelyselect the most favor- droth 1985),and C. caraboldesis specializedon preying able microhabitatin their environment(Mossakowski upon molluscs (Thiele 1977).The amount of food is 1979, Niemelii et al. 1992).Humidity, temperature, one of the most important biotic factors affectingcara- light, and physical and chemical qualities of soil can bid distributions(Thiele 1977, Niemelii 1993a, Sergeeva have an effect on their distribution patterns (Thiele 1994).In the soil samplestaken from the humus layer 1977, Lindroth 1985).Also interspecificinteractions of our study plots, there were more springtailsin the such as competition, predation and parasitismcan litter than in the control plots, whereasthe numbersof greatlyinfluence the distribution of carabid beetlesand mites did not show differencesbetween the plot types food availableto them (Sergeeva1994). However, abi- (Koivula 1996).Further, the numbersof carabidscorre- otic factors are probably more important than biotic latedpositively with the numbersof springtails,but not ones in determining distribution patterns (e.g. Thiele with thoseof mites.A similar positivecorrelation be- 1964).The highercatches of carabidsin the litter plots tween pitfall catchesof carabids and springtails was in our data may be due to several factors. Firstly, reportedby Niemeliiet al. (1986). carabid individuals may have actively selectedthis mi- Despite the higher numbers of potential food items, crohabitat,as e.g.Niemelii et al. (1992)have proposed. springtails, the springtail specialistN. biguttatus was Alternatively, demographicprocesses may have differed equally numerousin both plot types.This may indicate betweenthe litter and the control plots. The mortality that the amount of food is not a crucial factor explain- rate may have beenlower or the birth rate higher in the ing our resultsfor this species.On the other hand, the

ECOGRAPHY 22:4 (1999\ 431 speciesmay not find the increaseof springtail numbers ronment has been suggestedto offer more stable hu- sufficient enough to move to or to survive better in midity and temperature conditions, thus providing litter plots. Alternatively,this speciesdoes not prey favorableconditions for reproduction(Uetz 1979,Bult- upon springtailswithin the litter layer but insteadhunts man and Uetz 1982, 1984).This may be the casefor them on the top of the litter. The increasednumbers of carabid beetles as well. In our study, the leaf-litter springtailswere recordedwithin the humus layer, and material was moist evenduring the driest months in the we haveno data on the springtailnumbers from the top summer in contrast to very dry moss layer in the of the litter (Koivula 1996).Cychrus caraboides is spe- aontrol plots (Koivula unpubl.). cialized on preying upon terrestrial land snails (Lin- Some speciesmay respond negatively to leaf litter. droth 1985).ln 1994,the pitfall catchesof slugswere For instance, N. biguttatus is sunJoving, day-active significantlyhigher in the litter than in the control plots speciesliving in quite dry habitats (Lindroth 1985). (Koivula unpubl.). This may indicate that the litter Although we have no data on the shadinessof the plots are favorable for terrestrial land snails, too, and plots, the shadinessmay have been a crucial factor for further, they might attract C. caraboidesindividuals to N. biguttatus.Indeed, the catchesof N. biguttans were the litter plots. at their highest at the well-lit stand B. In contrast, ,4. Carabids also differ in their dispersalability. Long- brunneaand L. terminatustetd to live in microhabitats winged speciesare usually more successfulcolonizers with much leaf litter (Lindroth 1985,1986). In a study (Ranta and As 1982).However, many of the most with only one forest stand,Niemelii et al. (1992)got the numerousspecies, e.g. C. miuopterus,P. oblongopunc- highest catchesof C. miuopterus at placeswith much tatus and C. caraboides,are poor dispersers(den Boer needlelitter. In our data, however,the catchesin- 1977).Notiophilus biguttatr.rspopulations may include creasedin the leaf-litter plots. Thus, the litter itself, or both long- and short-winged individuals (Lindroth the amount, regardlessthe quality, might be important 1985).The most numerousspecies in our samplesare for the species.In the same study, the catchesof P. likely to be poor dispersers,with N. biguttatusas a oblongopunctatuswere highest at placeswith much as- possibleexception. However, the speciesthat increased penJeaf litter. However, the catchesincreased in the in the litter-plot sampleswere the most commonones. litter plots in our study,too. Additionally,the distribu- The pool of mature-forestspecies is rather small (e.g. tion pattern of C. caraboides,which increasedin litter- Niemelii 1993a);thus the colonization of the litter plots plot samplesin our study, is affectedby the amount of by any further speciesis not likely. humusand soil moisture(Lindroth 1985.Niemeld et al. 7992).Generally, our resultsregarding the relationship betweenforest-floor vegetation and occurrenceof these four carabid speciessupport the earlier studies. The importance of the physical environment Variety of niches may favor a diverse community Higher beetlecatches in the litter-plot samplesmay be (Giller 1984).Leaf litter may offer new niches,thus attributable to abiotic factors such as humidity and enablingmore speciesto coexistin a given stand.The temperature,and to some physical and chemicalprop- diversity and evennessindices and speciesrichness, ertiesof the soil which all are important environmental however,did not differ betweenthe leafJitter and con- factors for carabid beetles(Thiele 1964,1977, Lindroth trol plots although the overall numbers of carabid 1985).Leaf-litter material may offer shelter against individuals in the samplesincreased after the litter was predatorsand evaporation.For spiders,leaf-litter envi- added.Our resultsmav. however.be attributable to the

Table 4. The eight most common taiga carabid species (according to NiemelZi et al. 1988, 1994, 1996) and some of their ecological characteristics.Nomenclature follows Lindroth (1985, 1986). Wing length: B : brachypterous, M: macropterous. Activity: N: night-active, D: day-active species.Guild: I: litter surface, III: within the litter and II: between these two. Overwintering stage:L: larval, A: adult. Food preference:C: carnivorous, G: generalist,S: specialistand P: plant eater. Size,wing length, activity and overwintering habits according to Lindroth 1985, 1986. Guild according to Sergeeva1994. Food preference according to Lindroth 1985, 1986, Hengeveld 1980a, b, c. Habitat preference according to Lindroth 1985, 1986, Niemel2i eI al. 1988, 1992.

Species Size (mm) Wings Activity Guild Winter Food Habitat pref

Calathus micropterus (Dfl.) 6.5,8.8 BNIIILC needlelitter P t eros tic hus oblongopunct atus (F .) 9.5-12.6 M* N(D) rr A G leaf litter N ot iop hilus b igut t atus (F .) 5-6 B+MDIIIACS dry Cychrus caraboides L. 14-t9 BNILGShumus, moist Carabus glabratus Payk. 22-30 BDIAGmosses Amara brunnea (Gyll.) 5.2-6.8 MNIIILP mosses,leaf litter Carabus hortensis L. 22 28 BNILGhumus, dry Leistus terminarus (Hellw. in Pz.) 6-8 BNIIILG leaf litter

* no flight observations (Lindroth 1986)

432 ECOGRAPHY 22t4 (1999) rather small pool of carabid speciesinhabiting ma- turbed by the wood ants than the night-active spe- ture mesic taiga. LeafJitter microhabitat seems to cies. be favorable for some species(e.9. C. micropterus, P. oblongopunctatusand C. caraboides),but not for (e.g. some others N. biguttatus).At a forest-stand The levels of variation and conclusions level, this probably leads to different aggregation patterns of different carabid species. Our data revealedtwo levelsof variation. First, both floristic and, especially,carabid-assemblage variation was considerableamong the stands.Further, for cara- bids this variation was consistent throughout the study years.Second, a similar kind of variation was Species' interactions found also at the smaller scale, among the plots Interspeciflc competition may affect carabid distri- within the stands, in both vegetation and carabid bution patterns, although this may not be impor- data, and this variation was again consistentthrough- tant in the entire carabid communities (e.g. out the study years in the carabid data. The effect of Dennison and Hodkinson 1984), and the impor- the experimentallitter addition was detectablein the tance of competition remains largely unclear carabid assemblage,although not similarly in the dif- (Niemelii 1993b, but see Currie et al. 1996). In a ferent forest stands or among the species.Our data forest ecosystem, it has been suggested that only suggestthat the magnitude of the litter-addition effect the most numerous species compete (Loreau 1992), may depend on the fertility of the stand in the first but that there are many differences in these spe- place. The litter effect was largest in the least fertile cies' ecology (e.g. food preferences, phenology) that stand B, and weakestin the most fertile stand C. make interspeciflc competition unlikely in taiga Thus, in forests with poor soil and humus layer, the forests. Niemeld (1993a) also suggested that in a scattered deciduous trees are likely to be especially predictably unfavorable habitat, such as taiga important for forest-floor fauna. The existence of forest, the severity might keep the competitors out. such a phenomenon,however, awaits further research In carabids, however, competition at the larval with larger data set.If litter aggregationshave a posi- stage may be more important than at the adult tive effect on common, generalizedcarabid species,it stage, and competition could exist between e.g. is likely that they favor scarcerspecies, too. The im- staphylinid beetles and carabid larvae (e.g. Thiele portance of microhabitat differences, such as litter 1977, Spence 1979). Nevertheless, interspecific com- depth, was found to be important in explaining cara- petition may be one factor in keeping carabid den- bid abundance in a field experiment in Canada sities quite low in the taiga forests. Sergeeva (1994) (Niemeld et at. 1997). Thus, studies on litter effects separated three guilds of carabid species living in on other beetle groups, spiders and molluscs, would different layers of litter (Table 4). The increase in be fruitful as well as studying the litter quality in the catches of some species (e.g. P. oblongopunc- more detail. Such data are neededin order to draw tatus, C. caraboides and C. micropterus) may be more generalconclusions on the importance of decid- due to altered forest-floor structure, i.e. the leaflit- uous trees for forest-floor invertebrates.Also further ter microenvironment may provide more room in experimentalwork on the effect on the litter quality, vertical direction for carabid adults and larvae, e.g. different fertilizing value and pH of various types thus enabling more individuals to live simulta- of litter and artificial litter without any fertilizing neously in a given microhabitat. value, providing only shelter, might shed some light Our RDA analyses (data 1993 and 1993-1996) on the reasons for the aggregateddistribution pat- indicated that also the wood ants have an effect ternsof carabids. on the carabid catches. There is evidence that t/. The removal of aspensfrom managed forests has biguttatus copes better in high wood-ant densities causedpopulation declines of e.g. many insectspecies than e.g. P. oblongopunctatus and C. caraboides (Siitonen and Martikainen 1994), epiphytic lichens (Niemel6 et a\. 1992). Punttila (1994) has suggested (Kuusinen 1994) and polypores (Kotiranta and that such a difference is likely to be a result from Niemelii 1996)living on aspentrunks. Our results in- the different susceptibility of night- and day-active dicated that also the litter of aspen has a positive carabid species to interference and/or predation effect on invertebrate numbers, and this may further caused by the wood ants. The wood ants hunt be reflectedin the aggregateddistribution patterns on (mostly) in daytime and may thus find the night- the forest-stand level. Thus, we warmly recommend active species from their resting sites e.g. within forest owners and managersto leave more deciduous the litter. In contrast, the vividly moving, day-ac- trees in their forests, and favor mixed forests mimick- tive I/. biguttatus should be more capable of ing natural tree-speciescomposition instead of mono- avoiding the wood-ant predation and be less dis- cultures of conifers.

ECOGRAPHY 22:4 (1999\ 43J Acknowledgements - Several colleagues and others helped in Haila, Y. et al. 1994b. Metsiitalouden ekologiset vaikutukset the field: S. Finnilii, H. Kinnunen, H. Koivula, R. Lumiaro, boreaalisessa havumetsdssii: tutkimustuloksista kiiytiinnon M. Nylund, T. Seppii, N. Wahlberg and Y. Haila's home- suosituksiin. In: Haila, Y., Niemelii, P. and Kouki, J. troops. We thank R. Forsman,.T. Hyv6nen, S. Lehvdvirta, T. (eds), Effects of management on the ecological diversity of Seppii, L. Sundstrom and M. Ost, H. Rita and J. Kouki for boreal forests. Metsiintutkimuslaitoksen tiedonantoja 482, constructive comments. We also thank the staff of Seitseminen Finnish For. Res. Inst. Finland, pp.7-17, in Finnish. National Park, Satakunta Environmental Research Center of Hansson, L. 1992. Landscape ecology of boreal forests. - University of Turku, and Hdme Park Area of Finnish Forest Trends Ecol. Evol. 7: 299-302. and Park Service. The Academy of Finland financed this work Hengeveld, R. 1980a. Qualitative and quantitative aspects of (project 10134485). the food of ground beetles (Coleoptera: Carabidae): a review. - Neth. J. Zool. 30:555-563. Hengeveld, R. 1980b.Polyphagy, oligophagy and food special- izalion in ground beetles (Coleoptera, Carabidae). Neth. J. Zool.30: 564-584. Flengeveld, R. 1980c. Food specialization in ground beetles: an References ecological or a phylogenetic process? (Coleoptera: Cara- bidae). - Neth. J. Zool.30:585-594. Ahti, T., Himet-Ahti, L. and Jalas, J. 1968. Vegetation zones Jongman, R. H. G., ter Braak, C. J. F. and van Tongeren, O. - and their sections in northwestern Eurooe. Ann. Bot. F. R. (eds) 1995. Data analysis in community and land- Fenn. 5: 169 2ll. scape ecology. - Cambridge Univ. Press. Bird, G. A. and Chatarpaul, L. 1988. Effect of forest harvest Kiirkkiiinen, M. 1981. Happa- ja poppelilajien (Populus) on decomposition and colonization of maple leaf litter by keiytto. - Silva Fennica 15: 156-119, in Finnish. - soil microarthropods. Can. J. Soil. 68:29-40. Koivula, M. 1996. Lehtikarikkeen lisiiiimisen vaikutus MT- Boag, D. A. 1985. Microdistribution of three genera of small metsiin maakiitiijiiisten runsauteen. - M.Sc. thesis, Dept of terrestrial snails'(Sylommatophora: Pulmonata). - Can. J. Ecology and Systematics, Univ. of Helsinki, in Finnish. Zool. 63: 1089-1095. Kotiranta, H. and Niemell, T. 1996. Threatened polypores rn Bossenbroek, P. H. et al. 19'77. The significance of plant - "shelter" Finland. 2nd revised ed. Finnish Environ. Inst., Helsinki, growth-forms as a for terrestrial . J. in Finnish with English summary. (Lond.) Zool. 182: 1 6. Krebs, C. J. 1989. Ecological methodology. Harper and Bultman. T. L. and Uetz. G. W. 1982. Abundance ano uom- Row. munity structure of forest floor spiders following litter Kuusinen, M. 1994. Epiphytic lichen flora and diversity on - manipulation. Oecologia 55: 34 41. Populus tremula in old-growth and managed forests of Bultman. T. L. and Uetz. G. W. 1984. Effect of structure and southem and middle boreal Finland. Ann. Bot. Fenn. nutritional quality of litter on abundances of litter-dwelling 31: 245-260. - . Am. Midl. Nat. 111: 165-l'72. Langor, D. et al. 1994. biodiversity in the boreal forests Byers, J. A. 1984. Electronic multiprobe thermometer and of Alberta, Canada. - In: Haila, Y., Niemelii, P. and multiplexer for recording temperatures of microenviron- Kouki, J. (eds), Effects of management on the ecological ments in the forest litter habitat of bark beetles diversity of boreal forests. Metsintutkimuslaitoksen tiedo- (Coleoptera: Scolytidae). - Environ. Entomol. 13: 863- nantoja 482: 25 31. Finnish For. Res. Inst., Finland, in 867. Finnish. Cajander, A. K. 1949. Forest types and their significance. Lindroth, C. H. 1985. The Carabidae (Coleoptera) of Acta For. Fenn. 56: I 71. Fennoscandia and . Vol. 15. oart 1. Fauna Currie, C. R., Spence,J. R. and Niemeld, J. 1996. Competi- Entomol. Skandinavica.Leiden. Copenhagen. tion, cannibalism and intraguild predation among ground Lindroth, C. H. i986. The Carabidae (Coleoptera) of beetles (Coleoptera: Carabidae): a laboratory study. Fennoscandia and Denmark. Vol. 15" oart 2. Fauna Coleopt. Bull. 50: 135 148. Entomol. Skandinavica.Leiden. Copenhagen. den Boer, P. J. 19'77.Dispersal power and survival. Carabids Loreau, M. 1992. Species abundance patterns and the struc- in a cultivated countryside. - Misc. Pap., Landb. ture of ground-beetle communities. Ann. Zool. Fenn. 28: Hogeschool Wageningen 14. 49-56. den Boer, P. J. 1979. Populations of carabid beetles and Luff, M. L. 1987. Biology of polyphagous ground beetles in individual behaviour: general aspects.- In: den Boer, P. J., agriculture. - Agricult. Zool. Rev. 2:237-218. Thiele, H. U. and Weber, F. (eds), On the evolution of Martikainen, P. et al. 1994. The elfect of forest management behaviour in carabid beetles. Misc. Pao. 18. H. Veenman on the distribution of saproxylic beetles in Finnish and and Zonen B. V., Wageningen, , pp. 145-150. Russian Karelia. - In: Haila, Y., Niemel2i, P. and Kouki, den Boer, P. J. 1990. Density limits and survival of local J. (eds), Effects of management on the ecological diversity populations in 64 carabid species with different powers of of boreal forests. MetsZintutkimuslaitoksen tiedonantoja dispersal. - J. Evol. Ecol. 3: 19-48. 482, Finnish For. Res. Inst., Finland, pp. 97 104, in Dennison, D. F. and Hodkinson, I. D. 1984. Structure of the Finnish with English summary. predatory beetle community in a woodland soil ecosystem. Mascanzoni, D. and Wallin, H. 1986. The harmonic radar: a V. Summary and conclusions.- Pedobiologia 26: l7l,l17 . new method of tracing in the field. Ecol. Ento- Esseen, P.-E. eI al. 1997. Boreal forests. - Ecol. Bull. 46: mol. l1: 387 390. 16 47. Mossakowski, D. 1979. Evolution of habitat preference illus- Facelli, J. M. and Pickett, S. T. A. 1991. Plant litter: light trated by the phylogeny of Chrysocarabus. - In: den Boer, interception and effects on an old-field plant community. - P. J., Thiele, H. U. and Weber, F. (eds), On the evolution Ecology'72: 1024-1031. of behaviour in carabid beetles. Misc. pap. 18, H. Veenman Giller, P. S. 1984. Community structure and the niche. - andZonen B. V., Wageningen, Netherlands, pp. 103 112. Chapman and Hall. Niemelii, J. 1990. Spatial distribution of carabid beetles in the Griim, L. 1971. Remarks on the differentiation in Carabidae southern Finnish taiga: a question of scale.- In: Stork, N. mobility. Ekol. Pol. 19: 47 56. (ed.), Ground beetles: their role in ecological and environ- Haila, Y. and Kouki, J. 1994.The phenomenon of biodiversity mental studies. Intercept Publ., Andover, England, pp. in conservation biology. Ann. Zool. Fenn. 31: 5-18. 143-155. Haila, Y. et al. 1994a. Forestry and the boreal fauna: match- Niemelii, J. 1993a. Mystery of the missing species: species- ing management with natural forest dynamics. Ann. abundance distribution of boreal sround beetles. - Ann. Zool. Fenn. 31: 187-202. Zool. Fenn. 30: 169 172.

434 ECOCRAPHY 22.4(1999) Niemelii, J. 1993b. Interspecific competition in ground-beetle Seastedt, T. R. et al. 1983. A two-year study of leaf litter assemblages(Carabidae): what have we learned? - Oikos decomposition as related to maoroclimatic lactors and 66: 325-335. microarthropod abundance in the southern Appalachians. NiemelZi,J. 1997. Invertebrates and boreal forest management. - Holarct. Ecol. 6: 11-16. - Conserv.Biol. 11: 601 610. Sergeeva, T. K. 1994. Seasonal dynamics of interspecific Niemeld, J., Haila, Y. and Ranta, E. 1986. Spatial heterogene- trophic relations in a carabid beetle assemblage. - In: ity of carabid beetle dispersion in uniform forests on the Desender K. et al. (eds), Carabid beetles: ecology and Aland Islands,SW Finland. - Ann. Zool. Fenn.23:289- evolution. Kluwer, pp. 367 370. 296. Siitonen, J. and Martikainen, P. 1994. Occurrence of rare and Niemelii, J.,Haila, Y. and Halme, E. 1988. Carabid beetleson threatened insects living on decaying Populus tremula: a isolated Baltic islands and on the adjacent Aland marn- comparison between Finnish and Russian Karelia. land: variation in colonization suciess.- Ann. Zool. Fenn. Scand.J. For. Res. 9: 185-191. 25:133 143. Simberloff, D. 1978. Use of rarefaction and related methods in Niemelii, J. et al. 1992. Small-scale heterogeneity in the spatial ecology. - In: Dickson, K. L., Garins, J. Jr and Liv- distribution of carabid beetles in the southern Finnish ingston, R. J. (eds), Biological data in water pollution taiga. J. Biogeogr. 19: 173-181. assessment: quantitative and statistical analysis. Am. Soc. Niemeld, J., Tukia, H. and Halme, E. 1994. Patterns of for Testing and Materials STP 652: 150 165 carabid diversity in Finnish mature taiga. Ann. Zool. Southwood, T. R. E. 1978. Ecological methods. 2nd ed. Fenn. 31: 123 129. Chapman and Hall. Niemelii, J., Haija, Y. and Punttila,P. 1996.The imporrance Spence, J. R. 1979. Riparian carabid guilds-a spontaneous of small-scaleheterogeneity in boreal forests: var;ation in question generator. In: Erwin, T. L., Ball, G. E. and diversity in forest-floor invertebrates across the succession Whitehead, D. R. (eds), Carabid beetles: their evolution, gradient. - Ecography 19:352-368. natural history, and classification. Junk, Den Haag, pp. Niemeld, J., Spence,J. R. and C6rcamo, H. 1997. Establish- s25- 537. ment and interactions of carabid populations: an experi- terBraak, C. J. F. 1987.CANOCO - A FORTRANprogram ment with native and introduced species. Ecography 20: for ganonical gommunity ordination by [partial] [de- 643-652. trended] [canonical] correspondence analysis, principal Nuorteva, M. 1987. For?indringar i insektsfaunan i components analysis and redundancy analysis (ver. 2.0). skogar. - Entomol. Medd. 55: 125 128, in Swedish. TNO Inst. of Applied Computer Sci., Statistics Dept Wa- Punttila, P. 1994. Keomuurahaiset ja niveljalkaisyhteisojen geningen, Wageningen, The Netherlands. rakenne. - In: Haila, Y., Niemelii, P. and Kouki, J. (eds), Thiele, H.-U. 1964. Experimentelle Untersuchungen iiber die Effects of management on the ecological diversity of boreal Ursaclsn der Biotopbindnung bei Carabiden. - Z. Mor- forests. MetsZintutkimuslaitoksen tiedonantoja 482: 47 - 5B. phol. Okol. Tiere 58: 355-372. Finnish For. $es. Inst., Finland, in Finnish. Thiele, H. -U. 197'7.Carabid beetlesin their environments. - Ranta, E. and As, S. 1982. Non-random colonization of Springer. habitat islands by carabid beetles.- Ann. Zool. Fenn. 19: Uetz, G. W. 1979. The influence of variation in litter habitats 175-181. on spider communities, - Oecologia 40: 29 42. Reddy, M. V. 1984. Seasonal fluctuation of different edaphic Underwood, A. J. 1991. Beyond BACI: experimental designs microarthropod population densities in relation to soil for detecting human environmental impacts on temporal moisture and temperature in a pine, Pinus kesiyct P.oyle variations in natural populations. - Aust. J. Mar. Fresh- plantation ecosystem.- Int. J. Biometeorol. 28: 55-59. water Res. 42:569-587.

ECOGRAPHY 22:4 (1999\ 435