Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Bark beetle diversity at different spatial scales

Bark beetle diversity at different spatial scales Pellonen, M., Heliovaara. K., Vaisanen. R. and Keronen. J. 1998. Bark beetle diversity at different spatial scales. - Ecograpliy 21: 510 517. To determine how the seale of observation affects ecological patterns we studied bark beetle (Coleoptera, Scolytidae) diversity in southern Finland. A block covering 160 X 160 m of a forest was delimited in four stands of different site types. Hach block was divided into 256 squares (10 x 10 m) in which the occurrence o\' bark beetle speeies was recorded. In addition, environmental variables describing site type, trees, and breeding material appropriate for bark beetles were measured. The species presence/absence data were combined at different scales of resolution (10 x 10 m. 20 X 20 m, 40 x 40 m, 80 x 80 m, 160 x 160 m)- At the finest scale a recently thinned pine stand showed relatively high diversity compared to other study stands due to a few evenly distributed and abundant speeies. However, the species diversity increased faster toward larger scales in mature spruce stands with several sporadically distributed species. According to logistic regression analyses, breeding material and site characteristics explained the occurrence of most beetle species. However, these variables did not explain the occurrence of the six most frcquenl species, probably because the factors regulating their distribution and occurrence operate at larger scales. M, Pclioncn (mikko.s.pellonen(ivjieisinki.fi), K. Helidvciarci and J. Keronen, Depi of Applied Zoology. P.O. Box 27.'FIN-00d}4 Univ. of Helsinki. Finland. R. Vaisanen, Finnish Forest and Park Service, Naiure Proieclion, P.O. Box 94, FIN-01301 Vanlaa, Finland. Ecological mechanisms may operate at different spatial, temporal, and organizational scales than the emerging patterns. Thus, linking information between different scale levels is a central problem in ecology (Wiens 1989, Kotliar and Wiens 1990, Levin 1992). The problem of pattern and scale has also a methodological aspect, because the scale of observation fundamentally affects the results of ecological studies (Hengeveld 1987, Dutilleul 1993). The family Coleoptera, Scolytidae is one of the most intensively studied insect groups, mainly due to the economical importance of bark beetles as forest pests. More than 6000 species of bark beetles have been named throughout the world wherever woody plants grow (Bright 1976). One fourth of these {1430 species) have been recorded from Central and North America {Wood 1982a), while 308 species have been listed from the central and west Palaearctic region {Pfeffer 1995). Lekander et al. (1977) listed 86 species from Denmark and Fennoscandia; they recorded 48 species in the ca 70 X 70 km square covering the present study area in southern Finland. A wide range of observational scales with analyses of related environmental variation is valuable for general understanding of patterns and the underlying ecological mechanisms (Levin 1992, Lavorel et al. 1993). Previously, spatial patterns of bark beetle diversity have been analysed at biogeographical and landscape scales (Vaisanen and Heliovaara 1994). The aim of this study is to describe and analyse patterns of bark beetle diversity at increasingly finer scales and to relate the species occurrence to environmental variation. Accepted 20 January 1998 Copyright © ECOGRAPHY 1998 ISSN 0906-7590 Printed in Ireland -- al! risjhts reserved ECOGRAPHY 2i:5 (1998) Table 1. Characteriziitioii of the study stands (160x160 tn blocks). MT = Mynillus site type, VT = Vaccinium site type, CT = Calluna site type. Block 1 Site type Number of trees Scots pine Noi^way spruce Birehes Others Meati age, years Dominant height of trees, m Cro\\n eoveragc % Number of dead trees Scots pine Norway spruce Number of branches and crowns Scots pine Norway spruce Number of pine stumps MT 284 2300 197 172 120 21 52 23 161 21 55 (1 Block 2 VT 2794 92 20 38 50 14 46 201 0 88 1 0 Block 3 MT 237 1602 102 37 100 23 67 4 49 39 86 0 Block 4 CT 729 2 5 0 80 16 30 1 0 2378 4 312 Table 2. The number of 10 x 10 m squares occupied by the species in the 160 x 160 m blocks of dilTerent site types. Block 1 Species on pine Tomicii.s pini/wrcla (L.) Hylaxicx brwmcus Erichson Pilyophlharm Uchiensieini (Ratzebiirg) Pilyogenes quadridcns (H:ir!ig) Piiyogenes hidenlaim (Hcrbst) Speeies on spruee Xylechinus pilosus (Ratzeburg) Hyhtrgopx ^hbrulus (Zettersledt) IlylasU'S ciiuicularius Hrichson Palygrctphus poligraphiis (L.) Diyocoi'Jcs sp. Cryplia/iis saliuurius Weise Pilyogenes chalcograptnis (L.) Ips lypographus (L.) Ips amitinus (Eichhoff) Speeies on pine and spruee HylwgopR paHiaiu.s (Gyllenhal) Crypiiirgiis sp. Trypock'iiclron lineaium (Olivier) OrrluiioiiiicN.s .siiinrali.s (Gyllenhal) OiilwKnuiiiis lariiis (Fabricius) Species oti birch ScalyliLs ralzcburgi Janson Trypoilcndfon signatum (Fabricius) 49 — \ 2 \ 53 6 3 65 26 31 12 — 2 34 38 33 — Block 2 251 4 23 50 — — — — — — — — 7 — 51 — Block 3 28 4 5 6 22 5 15 56 10 41 4 — 18 29 20 — Block 4 253 50 185 236 216 — — — — 4 — — 60 4 187 18 18 Material and methods Study stands The study was carried out in 1993 in Hyytiala, southern Finland (62°50'N, 24°19'E) in four stands of different site types. A block covering 160 x 160 m of the forest was delimited in each stand. The studied stands were described in detail (Table I). Each block was divided into 256 squares (10 x 10 m), in which the occurrence of ditTerent bark beetle species was recorded using presence/absence data. The fauna of logs, stumps, logECOGRAPHY 21:5 (I99S) ging residue and windthrown branches not older than two years as well as dying trees and broken tree crowns were investigated by collecting under the bark. The bark samples were removed from the breeding material using a knife. The surface area and the bark samples were visually examined in the field and systematically scanned by two specialists. Replication of data collection later at the same places was not possible because of the collection method. Thus, the analyses rest on assumption of habitat saturation and equilibrium. The species recorded are listed in Table 2. Iterated sampling of the original data The effect of scale on the species catch was studied by picking random samples from the original data. Constant area (1600 m^) was sampled from each block using three different scales of resolution: one random 40 X 40 m square, four random 20 x 20 m squares and 16 random 10 x 10 m squares (Fig. 1). The sampling was iterated 20 times. The differences between mean species catch (number of species) in different blocks and resolutions were analysed with two-factor ANOVA model. The multiple comparisons between means were tested with Tukey's Studentized Range Test. / / - • -I" Block 12 3 4 12 3 4 12 3 4 12 3 4 A C A B 80 X 80 m n=4 160 X 16Um Species richness and environmental variation Logistic regression models were used to relate the environmental variation to the bark beetle presence/absence data. Only species with > 16 recordings in the data were included in the analyses. The following environmental variables were recorded in 10 x 10 m squares and used in the logistic regression models: forest site type according to Cajander (1949), with decreasing moisture and nutrient level: 1 = OMT {Oxalis-MyrtiHus site type), 2 = MT (MyrlUlus site type), 3 —VT {Vaccinium site type), 4 = CT {Calluna site type) and 5 — CIT (Cladonia site type); number of Scots pines Pinus sylvestris, Norway spruces Picea ahies and birches Betula pendula and B. pubescens\ crown canopy coverage (0, 5. 10...100'%)); mean height of trees (m); number of dead or injuried trees: number of broken branches and tree crowns including logging residue; number of stumps. Grouping B B C A Scale 10 s 10 m n=^256 B C C A 20 N 20 m A B A A 40 X 40 m n ^ 16 Fig. 2, The mean species richness ( + 1 SD) in different scales of resolution. The grouping of means is based on Tukey's Studentized Range Tests which are made for each scale separately. Means with the same letter are not significantly different (p>0.05). 160 m —" The logistic regression models were fitted for the entire data set (n = 1024). The models were fitted for each species by stepwise selection of variables. The significance criterion for a variable to enter or to remain in the model was p < 0.05. The final models were tested by the Hosmer-Lemeshow goodness-of-fit x' test, where the null hyphothesis is that no more variables are needed in the model. Thus, large p-values indicate good fit of the model. Spearman's rank correlation coefficients were used to analyse the spatial concordance of species richness (number of species in a square) and environmental variables at different spatial scales. Two new variables were formed from the variables measuring bark beetle breeding material: total amount of breeding material and breeding material diversity at site (number of breeding material types). / Om Results Spatial patterns of species richness 160 m 40 m \ 20 m' \ 10 m( 10 m 20 m 40 m 80 m Fig. 1. The scale:? of resolution used in the analyses. Overall, we recorded 21 species. The species richness, measured as the mean number of species in a square, was greatest in the recently thinned pine stand (block 4) at the two finest scales (10 x 10 m, 20 x 20 m). However, in the coarsest resolutions the species richness appeared to be greatest in the spruce-dominated sites (blocks 1 and 3). In the young and non-managed pine stand (block 2) species richness was low and it was the only site where all of the species found within 160 x 160 m block were recorded in a single 40 x 40 m square (Fig. 2). The spatial patterns of species diversity (for ECOGRAPHY 21:5 (1998) Block Block 2 Block 3 Block 4 lOx 10m 1-2 S o a, 20 X 20 ni 3-5 e 6-8 I 40 X 40 m 9-il 12-14 Fig. 3. The spatial patterns of species richness within different scales and blocks. Table 3. Spearman's rank correlation coefficients hctwccn species richness iind environmental variahles. Environmental variables having strongest rank correlation with species richness are listed for diflcrent spatial scales and hlocks. Only statistically significant coefficients (p<0.05) arc presented. Scale Block 1 Block 2 Pine branch 0.75 Br div 0.71 Br tot 0.68 Dead pine 0.43 Pine branch 0.81 Br tot 0.74 Br div 0.73 Dead pine 0.53 Biock 3 Br div 0.72 Br tot 0.71 Sprtice branch 0.61 Dead spruce 0.49 Br div 0.77 Br tot 0.71 Dead spruce 0.69 Spruce branch 0.41 Br tot 0.65 Br div 0.64 Dead spruce 0.52 Dead pine —0.48 Mean height 0.84 Spruce branch 0.82 Pine - 0 . 7 9 Site type -0.75 Block 4 Br tot 0.42 Pine branch 0.38 Br div 0.36 Pine stump 0.33 All blocks Br tot 0.83 Br div 0.81 Pine branch 0.66 Pine stump 0.62 Br tot 0.72 Br div 0.65 Pine stump 0.53 Pine branch 0.45 Dead spruce 0.59 Pine - 0 . 5 4 Spruce branch 0.49 10x10 m Br div* 0.77 Br tot** 0.74 Dead spruce 0.63 Spruce branch 0.41 20x20 m Br tot 0.59 Dead spruce 0.56 Spruce branch 0.55 Br div 0.53 4 0 x 4 0 m Spruce branch 0.70 Pine branch —0.67 Dead spruce 0.60 Br tot 0.50 80 X 80 m — Br div = breeding material diversity, ** Br tot = total amount of breeding material. instance, the location of the most species rich square), varied according to scale (Fig. 3). Most species had scattered spatial distribution and low occurrence frequency. Five species {Tomicus piniperda, Pityophthoru.s llchlensteifii, Pltyogcne.s qiiadrldens\ Pilyogenes hidenUitm and Trypodendron lincatum) were very frequent and evenly distributed in recently thinned pine stand (block 4). Tomicus plniperda was very common and evenly distributed in both pine dominated blocks 2 and 4 (Table 2). Distribution and assortment of dead wood varied considerably between blocks. In spruce dominated ECOGRAPHY 21:5 (199S) blocks (1 and 3) the breeding material was heterogeneous as well as relatively scarce and sporadically distributed. In pine dominated blocks (2 and 4) the breeding material was more homogeneous, abundant and evenly distributed over the space. Due to recent thinning, the breeding material of bark beetles in block 4 consisted of pine logging residue and stumps. According to rank correlation coefficients, spatial patterns of species richness were strongly associated with the amount and diversity of breeding material (Table 3). Mean height of trees, abundace of pines, and site type showed strong correlation with species richness 513 Table 4. Mean species catclies of the different sampling scales in (lie four studied stands obtained witii 20 riuulom iterations. Scale 16 10x10 m 4 20 X 20 m ] 40 X 40 m Mean Block 1 9.20 8.05 8.20 8.48 Block 2 4.55 4.25 4.20 4.33 Block 3 8.33 6.95 7.20 7.48 Block 4 7,90 7.10 7.15 7.38 Mean 7,49 6,59 6.69 at the coarsest scale (80 x 8(3 m) when data from all blocks were combined. Discussion Kotliar and Wiens 1990) in the present study was the size of a block (2.54 ha),, whieh corresponds quite well with the average stand size of intensively managed forests in the study region. Wiihin blocks, we scaled the lower limit of resolution ("grain". Wiens 1989. Kotliar and Wiens 1990) to reveal the spatial patterns of bark beetle diversity at stand level. It is generally acknowledged that species richness is highly dependent on sample size. This is often described with species-area curves (e.g. Smith ct ai. 1985. Condit et al, 1996). The present study demonstrates the scaledependency of speeies richness: uneven distribution of species leads to relative underestimation of species richness at small sample sizes. Thus, assuming similar distributions in comparisons of species richness between different areas is not appropriate, not even when there are only minor differences in species composition and habitat characteristics. The distribution of most bark beetle species was scattered and occurrence frequency low. Thus, higher resolution in insect sampling increases the species catch. i.e. many independent small samples gave a better estimate of the species diversity than few large samples. Viiisanen et al, (1993) compared the sub-cortical fauna of primeval and managed forest, and the relatively high diversity in managed forest appeared to be largely attributable to bark beetles. None of the stands included in the present study can be classified as primeval forest, since they all have been subjected to silvicultural practices, at least to some extent. Though the occurrence frequency of bark beetles was lower in mature spruce stands, the species diversity was considerably higher than in recently thinned pine sland. The Species catch of iterated sampling The mean species catch o\' the random sampling depended on the block and the scale of resolution (Table 4). In two-factor ANOVA statistically significant differences were obtained between scales and blocks, but the interaction between these two factors was nonsignificant (Table 5). The multiple comparisons between the ranges of the species catch showed that the species catch was greatest when 16 random H) x 10 m squares were used in sampling. Significant difference was not found between 4 20 x 20 m and I 40 x 40 m sampling scales (Fig. 4). The species catch was the greatest in block I (mature spruce stand) and the smallest in block 2 (pine stand). Significant difference was not found between species catches of blocks 3 (mature spruce stand) and 4 (thinned pine stand) (Fig. 5). Logistic regression models According to the logistic regression analyses the presently used environmental variables explained bark beetle incidence with varying success depending on the species (Table 6). The forest site type, the amount of suitable breeding material, the mean height of trees, the density of pine and birch, and the crown canopy coverage were sufficient variables to predict the occurrence of species associated with spruce. However, goodnessof-fit tests showed that the used variables alone could not give significant predictions on the occurrence of four species assoeiated with pine (7". piniperda., P. lichtensleini, P. quadrickns, P. bidentaltis) and two polyphagous species {Hylurgops paUiatus, T. lineatum). Table 5. Two-factor ANOVA table of species calches of iterated random sampling. The factors arc block and sampling scale. DF = degrees of freedom. MS = mean square. F = F-ratio. p = corresponding statistical value. Source Block Scale Interaction DF r-J 16* 10 X 10 m MS 193.34 19.47 1.13 2,73 70.90 7.14 0.42 0.0001 0.0010 0.8683 4 * 20 X 20 m 1 * 40 X 40 m Error Fig. 4. Multiple comparisons of species catehes in different sampling seales based on Tukey's Studentized Range Test (DF = 228. MSB = 2.730, n = 240), The arrow points to the scale having bigger species cateli (p<(l,05). eCOGRAE'HY : i : 5 (I'WS) Block Block 2 Block 3 Fig. 5. Muliiple comparisons of species catches in different blocks based on Tukey's Studentized Range Test (DF = 228. MSE = 2.730, n = 240). The arrow points to the block having bigger species eateh (p < 0.05). amount of breeding material does not necessarily cotitribute to high species diversity. Also the diversity of breeding material is itiiporlant, because microhabitat demands (e.g. dimension of bark and wood, age, moisture etc.) vary between species {Rudinsky 1962, Lekander et al. 1977). The present analysis of spatial concordance between species richness and environmental features support this conclusion. In general, the breeding material of scolytids is scarce and sporadically distributed over the space (Sjodin et al. 1989, Lindelow et al. 1992). but site characteristics and human activities strongly affect the availability and distribution of wooden substrate appropriate for particular bark beetle species. In our study the amount and distribution of breeding material differed between stands partly due to different forest tnanagemcnt and partly because of site characteristics. The importance of suitable breeding material for the species occurrence can be seen in the logistic regression models, though other variables are also tieeded in the models. However, the set of variables used at the present local stand scale was not adequate to explain the occurrence of the six most frequent species. Four of these species are associated with the Scots pine {T. pimperda, P. lichtensleini\ P. quadridens and P. bidcnlatus) and two are polyphagous species [H. palliatus and T. Uneatum). Only these six species were found frotn each of the four studied stands. In a previous study at landscape scale the assessment of bark beetle occurrence was based on satellite imagery and field measurements (Vaisiincn and Heliovaara 1994). Environmental variables (tree species, soil type, forest site type, stand characteristics and forest management history) predicted well the occurrence of monophagous species associated with pine, while predictions were much less accurate for species associated with spruce (e.g. Xylechinus pilosus. Dryocoetes sp.). In the present analyses at local stand scale, the results were the opposite: the environmental variables (site characteristics and availability of breeding material) predicted the occurrence of bark beetle species associated with spruce more accurately than the occurrence RCOGRAPHY 2I;5 of species associated with pine (e.g. T. piniperda, P. quadriden.s. P. hidentatus). These results indicate that factors regulating the habitat use of the scolytid species associated with spruce are operating at finer scales than those of the species associated with pine. The habitat selection of an individual species may respond either to microhabitat or macrohabitat variation depending on the species" physiology, morphology, and life strategies (Morris 1987). There are different ways that bark beetles may find the breeding material: random Right and landing, localization of suitable host by compounds volatilizing from wood, and aggregation pheromoncs released by other bark beetle specimens (Wood 1982b). Species have different colonization behaviours and this may be reason for differences in the ecological scale of resource exploitation. Pheromones have an important role in the colonization behaviour of several scolytids associated with spruce, while pine associated T. piniperda orientates by ethanol and acetaldehyde released from fresh wood (Sjodin et al. 1989, Lindelow et al. 1992). Trypodendron Uneatum is an ambrosia beetle which carries a fungus and feeds on it. The selected environmental variables in the present study as well as in the previous study at landscape scale (Vaisanen and Heliovaara 1994) did not explain the occurrence o^ T. Uneatum. Probably some other factors determine the occurrence of the ambrosia fungus and thus, also the occurrence of T. Uneatum. At landscape level, there are factors that may affect strongly the occurrence of bark beetle species, e.g. forest fragmentation and associated edge effect (Peltoncn et al. 1997. Peltonen and Heliovaara in press). In this study the blocks were delimited in large and relatively homogeneous stands to eliminate possible edge effect. 0kland et al. (1996) studied the diversity of saproxylic beetles at different scale levels in relation to ecological variables. According to their results, the smallest scale (40 x 40 tn) was too small for the study of the relationship between the ecological variables and the fauna of saproxyhc beetles. The extent and grain of their study were larger than ours, and they used window traps instead of direct collection. Differences between our results and those of 0kland et al. (1996) demonstrate the significance of scale in interpreting the ecological patterns. Our present results on bark beetles indicate that scaling is essential for the evaluation of species diversity even within a local stand. Relatively high species diversity at finer scales may be due to the even distribution of a few common species, but still, diversity at larger scales may remain low. Conversely, low species diversity at finer scales may be due to sporadical distribution of several less frequent species, resulting in increasing species diversity towards larger scales of resolution. 515 Table 6. Vanables of logistic regression models with t;orrespondiiig odds ratios [95% confidence limits], p refers to goodness-offit tests. Term 1 Species on pine Tomicus Site type 3,48 [1.88, piniperda Hylasles Site type brunneus 8.00 [3.61, Pityophthorus Site type lichlensteini 4.48 [2,66, Pityogenes Site type quadridem 2.34 [1,37, Pityogenes Site type bidenlatus 4.74 [2,95, OrthoWmicus Site type suluralis 7.09 [2,80, OrthoXomicus Site type laricis 4.55 [1.63, Species on spruce Xylechinus Site type pliosus 0.43 [0.23, Polygrapims Pine poligraphus 0,78 [0.65, Dryncoeles sp. Site type 0.26 [0.13, Site type Cryphalus sahuarius 0.43 [0.25, Pine Pityogenes chakographm 0.81 [0,68, Species on pine and spruce Hylurgops Dead pine patliaius 1.51 [1.15, Cryplurgus sp. Pine 0.74 [0.60, Trypodendron Dead pine Hneatum 5.08 [3.70, Term 2 Pine 1.59 [1.39. 1.81] Pine stump 17.86] 2.34 [1.63, 3.34] Pine branch 7.52] 1.27 [1-17, 1.37] Dead pine 3.98] 1.44 [1.08, 1.91] Pine 7.58] 1.!! [1.06, 1.17] Pine branch 17,86] 1.08 [I.OI. 1.15] Height 12.66] 0.84 [0.76. 0.94] 6.45] Pine 0.82] 0.78 [0.61. 0.99] Height 0.94] 1.16 [1.09, 1.24] Birch 0.55] 1.60 [1.13, 2.27] Dead spruce 0.74] 3.26 [2.34, 4.53] Birch 0.97] 0.44 [0.25, 0.80] Term 3 Height 0,93 [0.88, 0.99] Pine stump !.78 [1.20, 2,65] Pine branch 3.39 [2.48, 4.65] Pine branch 1.48 [1.32, 1.67] Spruee branch 5.32 [1.18, 23.81] Pine stump 1.99 [1.22, 3.26] Height 1,16 [1,07, 1.26] Dead spruce 5.88 [4,10, 8.48] Height l.il [1.02, 1,20] Spruce branch 1.86 [1.00, 3,45] Crown coverage 1.03 [1.15, 1.05] Dead spruce 6.94 [4.70, 10.31] Spruce branch 90.91 [37.04. 200,00] Spruce branch 12.05 [6.80. 21,74] Term 4 Dead pine 6.58 [2.45. 17.86] Term 5 Pine branch 1.65 [1.20, 2.28] P 0.0002 0.7395 0.0001 0.0294 0.0018 0.5494 0.9654 ns ns ns ns Dead spruce Spruee branch Pine stump 2.18 [1.67. 2.87] 2.27 [1.45, 3.55] 2.73 [2.13, 3,51] Dead spruce Spruee branch 0.90] 2.14 [1.58, 2.89] 6.80 [4.25, 10.99] Dead spruce Spruce branch 7.00] 2.59 [1.96, 3.44] 2.39 [1.47, 3.91] 1,97] ns ns Acknowledgements - Sincere thanks to John Wiens, http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Ecography Wiley

Bark beetle diversity at different spatial scales

Download PDF
9 pages

Loading next page...
 
/lp/wiley/bark-beetle-diversity-at-different-spatial-scales-zBly4pDAYA

References (23)

Publisher
Wiley
Copyright
Copyright © 1998 Wiley Subscription Services, Inc., A Wiley Company
ISSN
0906-7590
eISSN
1600-0587
DOI
10.1111/j.1600-0587.1998.tb00442.x
Publisher site
See Article on Publisher Site

Abstract

Pellonen, M., Heliovaara. K., Vaisanen. R. and Keronen. J. 1998. Bark beetle diversity at different spatial scales. - Ecograpliy 21: 510 517. To determine how the seale of observation affects ecological patterns we studied bark beetle (Coleoptera, Scolytidae) diversity in southern Finland. A block covering 160 X 160 m of a forest was delimited in four stands of different site types. Hach block was divided into 256 squares (10 x 10 m) in which the occurrence o\' bark beetle speeies was recorded. In addition, environmental variables describing site type, trees, and breeding material appropriate for bark beetles were measured. The species presence/absence data were combined at different scales of resolution (10 x 10 m. 20 X 20 m, 40 x 40 m, 80 x 80 m, 160 x 160 m)- At the finest scale a recently thinned pine stand showed relatively high diversity compared to other study stands due to a few evenly distributed and abundant speeies. However, the species diversity increased faster toward larger scales in mature spruce stands with several sporadically distributed species. According to logistic regression analyses, breeding material and site characteristics explained the occurrence of most beetle species. However, these variables did not explain the occurrence of the six most frcquenl species, probably because the factors regulating their distribution and occurrence operate at larger scales. M, Pclioncn (mikko.s.pellonen(ivjieisinki.fi), K. Helidvciarci and J. Keronen, Depi of Applied Zoology. P.O. Box 27.'FIN-00d}4 Univ. of Helsinki. Finland. R. Vaisanen, Finnish Forest and Park Service, Naiure Proieclion, P.O. Box 94, FIN-01301 Vanlaa, Finland. Ecological mechanisms may operate at different spatial, temporal, and organizational scales than the emerging patterns. Thus, linking information between different scale levels is a central problem in ecology (Wiens 1989, Kotliar and Wiens 1990, Levin 1992). The problem of pattern and scale has also a methodological aspect, because the scale of observation fundamentally affects the results of ecological studies (Hengeveld 1987, Dutilleul 1993). The family Coleoptera, Scolytidae is one of the most intensively studied insect groups, mainly due to the economical importance of bark beetles as forest pests. More than 6000 species of bark beetles have been named throughout the world wherever woody plants grow (Bright 1976). One fourth of these {1430 species) have been recorded from Central and North America {Wood 1982a), while 308 species have been listed from the central and west Palaearctic region {Pfeffer 1995). Lekander et al. (1977) listed 86 species from Denmark and Fennoscandia; they recorded 48 species in the ca 70 X 70 km square covering the present study area in southern Finland. A wide range of observational scales with analyses of related environmental variation is valuable for general understanding of patterns and the underlying ecological mechanisms (Levin 1992, Lavorel et al. 1993). Previously, spatial patterns of bark beetle diversity have been analysed at biogeographical and landscape scales (Vaisanen and Heliovaara 1994). The aim of this study is to describe and analyse patterns of bark beetle diversity at increasingly finer scales and to relate the species occurrence to environmental variation. Accepted 20 January 1998 Copyright © ECOGRAPHY 1998 ISSN 0906-7590 Printed in Ireland -- al! risjhts reserved ECOGRAPHY 2i:5 (1998) Table 1. Characteriziitioii of the study stands (160x160 tn blocks). MT = Mynillus site type, VT = Vaccinium site type, CT = Calluna site type. Block 1 Site type Number of trees Scots pine Noi^way spruce Birehes Others Meati age, years Dominant height of trees, m Cro\\n eoveragc % Number of dead trees Scots pine Norway spruce Number of branches and crowns Scots pine Norway spruce Number of pine stumps MT 284 2300 197 172 120 21 52 23 161 21 55 (1 Block 2 VT 2794 92 20 38 50 14 46 201 0 88 1 0 Block 3 MT 237 1602 102 37 100 23 67 4 49 39 86 0 Block 4 CT 729 2 5 0 80 16 30 1 0 2378 4 312 Table 2. The number of 10 x 10 m squares occupied by the species in the 160 x 160 m blocks of dilTerent site types. Block 1 Species on pine Tomicii.s pini/wrcla (L.) Hylaxicx brwmcus Erichson Pilyophlharm Uchiensieini (Ratzebiirg) Pilyogenes quadridcns (H:ir!ig) Piiyogenes hidenlaim (Hcrbst) Speeies on spruee Xylechinus pilosus (Ratzeburg) Hyhtrgopx ^hbrulus (Zettersledt) IlylasU'S ciiuicularius Hrichson Palygrctphus poligraphiis (L.) Diyocoi'Jcs sp. Cryplia/iis saliuurius Weise Pilyogenes chalcograptnis (L.) Ips lypographus (L.) Ips amitinus (Eichhoff) Speeies on pine and spruee HylwgopR paHiaiu.s (Gyllenhal) Crypiiirgiis sp. Trypock'iiclron lineaium (Olivier) OrrluiioiiiicN.s .siiinrali.s (Gyllenhal) OiilwKnuiiiis lariiis (Fabricius) Species oti birch ScalyliLs ralzcburgi Janson Trypoilcndfon signatum (Fabricius) 49 — \ 2 \ 53 6 3 65 26 31 12 — 2 34 38 33 — Block 2 251 4 23 50 — — — — — — — — 7 — 51 — Block 3 28 4 5 6 22 5 15 56 10 41 4 — 18 29 20 — Block 4 253 50 185 236 216 — — — — 4 — — 60 4 187 18 18 Material and methods Study stands The study was carried out in 1993 in Hyytiala, southern Finland (62°50'N, 24°19'E) in four stands of different site types. A block covering 160 x 160 m of the forest was delimited in each stand. The studied stands were described in detail (Table I). Each block was divided into 256 squares (10 x 10 m), in which the occurrence of ditTerent bark beetle species was recorded using presence/absence data. The fauna of logs, stumps, logECOGRAPHY 21:5 (I99S) ging residue and windthrown branches not older than two years as well as dying trees and broken tree crowns were investigated by collecting under the bark. The bark samples were removed from the breeding material using a knife. The surface area and the bark samples were visually examined in the field and systematically scanned by two specialists. Replication of data collection later at the same places was not possible because of the collection method. Thus, the analyses rest on assumption of habitat saturation and equilibrium. The species recorded are listed in Table 2. Iterated sampling of the original data The effect of scale on the species catch was studied by picking random samples from the original data. Constant area (1600 m^) was sampled from each block using three different scales of resolution: one random 40 X 40 m square, four random 20 x 20 m squares and 16 random 10 x 10 m squares (Fig. 1). The sampling was iterated 20 times. The differences between mean species catch (number of species) in different blocks and resolutions were analysed with two-factor ANOVA model. The multiple comparisons between means were tested with Tukey's Studentized Range Test. / / - • -I" Block 12 3 4 12 3 4 12 3 4 12 3 4 A C A B 80 X 80 m n=4 160 X 16Um Species richness and environmental variation Logistic regression models were used to relate the environmental variation to the bark beetle presence/absence data. Only species with > 16 recordings in the data were included in the analyses. The following environmental variables were recorded in 10 x 10 m squares and used in the logistic regression models: forest site type according to Cajander (1949), with decreasing moisture and nutrient level: 1 = OMT {Oxalis-MyrtiHus site type), 2 = MT (MyrlUlus site type), 3 —VT {Vaccinium site type), 4 = CT {Calluna site type) and 5 — CIT (Cladonia site type); number of Scots pines Pinus sylvestris, Norway spruces Picea ahies and birches Betula pendula and B. pubescens\ crown canopy coverage (0, 5. 10...100'%)); mean height of trees (m); number of dead or injuried trees: number of broken branches and tree crowns including logging residue; number of stumps. Grouping B B C A Scale 10 s 10 m n=^256 B C C A 20 N 20 m A B A A 40 X 40 m n ^ 16 Fig. 2, The mean species richness ( + 1 SD) in different scales of resolution. The grouping of means is based on Tukey's Studentized Range Tests which are made for each scale separately. Means with the same letter are not significantly different (p>0.05). 160 m —" The logistic regression models were fitted for the entire data set (n = 1024). The models were fitted for each species by stepwise selection of variables. The significance criterion for a variable to enter or to remain in the model was p < 0.05. The final models were tested by the Hosmer-Lemeshow goodness-of-fit x' test, where the null hyphothesis is that no more variables are needed in the model. Thus, large p-values indicate good fit of the model. Spearman's rank correlation coefficients were used to analyse the spatial concordance of species richness (number of species in a square) and environmental variables at different spatial scales. Two new variables were formed from the variables measuring bark beetle breeding material: total amount of breeding material and breeding material diversity at site (number of breeding material types). / Om Results Spatial patterns of species richness 160 m 40 m \ 20 m' \ 10 m( 10 m 20 m 40 m 80 m Fig. 1. The scale:? of resolution used in the analyses. Overall, we recorded 21 species. The species richness, measured as the mean number of species in a square, was greatest in the recently thinned pine stand (block 4) at the two finest scales (10 x 10 m, 20 x 20 m). However, in the coarsest resolutions the species richness appeared to be greatest in the spruce-dominated sites (blocks 1 and 3). In the young and non-managed pine stand (block 2) species richness was low and it was the only site where all of the species found within 160 x 160 m block were recorded in a single 40 x 40 m square (Fig. 2). The spatial patterns of species diversity (for ECOGRAPHY 21:5 (1998) Block Block 2 Block 3 Block 4 lOx 10m 1-2 S o a, 20 X 20 ni 3-5 e 6-8 I 40 X 40 m 9-il 12-14 Fig. 3. The spatial patterns of species richness within different scales and blocks. Table 3. Spearman's rank correlation coefficients hctwccn species richness iind environmental variahles. Environmental variables having strongest rank correlation with species richness are listed for diflcrent spatial scales and hlocks. Only statistically significant coefficients (p<0.05) arc presented. Scale Block 1 Block 2 Pine branch 0.75 Br div 0.71 Br tot 0.68 Dead pine 0.43 Pine branch 0.81 Br tot 0.74 Br div 0.73 Dead pine 0.53 Biock 3 Br div 0.72 Br tot 0.71 Sprtice branch 0.61 Dead spruce 0.49 Br div 0.77 Br tot 0.71 Dead spruce 0.69 Spruce branch 0.41 Br tot 0.65 Br div 0.64 Dead spruce 0.52 Dead pine —0.48 Mean height 0.84 Spruce branch 0.82 Pine - 0 . 7 9 Site type -0.75 Block 4 Br tot 0.42 Pine branch 0.38 Br div 0.36 Pine stump 0.33 All blocks Br tot 0.83 Br div 0.81 Pine branch 0.66 Pine stump 0.62 Br tot 0.72 Br div 0.65 Pine stump 0.53 Pine branch 0.45 Dead spruce 0.59 Pine - 0 . 5 4 Spruce branch 0.49 10x10 m Br div* 0.77 Br tot** 0.74 Dead spruce 0.63 Spruce branch 0.41 20x20 m Br tot 0.59 Dead spruce 0.56 Spruce branch 0.55 Br div 0.53 4 0 x 4 0 m Spruce branch 0.70 Pine branch —0.67 Dead spruce 0.60 Br tot 0.50 80 X 80 m — Br div = breeding material diversity, ** Br tot = total amount of breeding material. instance, the location of the most species rich square), varied according to scale (Fig. 3). Most species had scattered spatial distribution and low occurrence frequency. Five species {Tomicus piniperda, Pityophthoru.s llchlensteifii, Pltyogcne.s qiiadrldens\ Pilyogenes hidenUitm and Trypodendron lincatum) were very frequent and evenly distributed in recently thinned pine stand (block 4). Tomicus plniperda was very common and evenly distributed in both pine dominated blocks 2 and 4 (Table 2). Distribution and assortment of dead wood varied considerably between blocks. In spruce dominated ECOGRAPHY 21:5 (199S) blocks (1 and 3) the breeding material was heterogeneous as well as relatively scarce and sporadically distributed. In pine dominated blocks (2 and 4) the breeding material was more homogeneous, abundant and evenly distributed over the space. Due to recent thinning, the breeding material of bark beetles in block 4 consisted of pine logging residue and stumps. According to rank correlation coefficients, spatial patterns of species richness were strongly associated with the amount and diversity of breeding material (Table 3). Mean height of trees, abundace of pines, and site type showed strong correlation with species richness 513 Table 4. Mean species catclies of the different sampling scales in (lie four studied stands obtained witii 20 riuulom iterations. Scale 16 10x10 m 4 20 X 20 m ] 40 X 40 m Mean Block 1 9.20 8.05 8.20 8.48 Block 2 4.55 4.25 4.20 4.33 Block 3 8.33 6.95 7.20 7.48 Block 4 7,90 7.10 7.15 7.38 Mean 7,49 6,59 6.69 at the coarsest scale (80 x 8(3 m) when data from all blocks were combined. Discussion Kotliar and Wiens 1990) in the present study was the size of a block (2.54 ha),, whieh corresponds quite well with the average stand size of intensively managed forests in the study region. Wiihin blocks, we scaled the lower limit of resolution ("grain". Wiens 1989. Kotliar and Wiens 1990) to reveal the spatial patterns of bark beetle diversity at stand level. It is generally acknowledged that species richness is highly dependent on sample size. This is often described with species-area curves (e.g. Smith ct ai. 1985. Condit et al, 1996). The present study demonstrates the scaledependency of speeies richness: uneven distribution of species leads to relative underestimation of species richness at small sample sizes. Thus, assuming similar distributions in comparisons of species richness between different areas is not appropriate, not even when there are only minor differences in species composition and habitat characteristics. The distribution of most bark beetle species was scattered and occurrence frequency low. Thus, higher resolution in insect sampling increases the species catch. i.e. many independent small samples gave a better estimate of the species diversity than few large samples. Viiisanen et al, (1993) compared the sub-cortical fauna of primeval and managed forest, and the relatively high diversity in managed forest appeared to be largely attributable to bark beetles. None of the stands included in the present study can be classified as primeval forest, since they all have been subjected to silvicultural practices, at least to some extent. Though the occurrence frequency of bark beetles was lower in mature spruce stands, the species diversity was considerably higher than in recently thinned pine sland. The Species catch of iterated sampling The mean species catch o\' the random sampling depended on the block and the scale of resolution (Table 4). In two-factor ANOVA statistically significant differences were obtained between scales and blocks, but the interaction between these two factors was nonsignificant (Table 5). The multiple comparisons between the ranges of the species catch showed that the species catch was greatest when 16 random H) x 10 m squares were used in sampling. Significant difference was not found between 4 20 x 20 m and I 40 x 40 m sampling scales (Fig. 4). The species catch was the greatest in block I (mature spruce stand) and the smallest in block 2 (pine stand). Significant difference was not found between species catches of blocks 3 (mature spruce stand) and 4 (thinned pine stand) (Fig. 5). Logistic regression models According to the logistic regression analyses the presently used environmental variables explained bark beetle incidence with varying success depending on the species (Table 6). The forest site type, the amount of suitable breeding material, the mean height of trees, the density of pine and birch, and the crown canopy coverage were sufficient variables to predict the occurrence of species associated with spruce. However, goodnessof-fit tests showed that the used variables alone could not give significant predictions on the occurrence of four species assoeiated with pine (7". piniperda., P. lichtensleini, P. quadrickns, P. bidentaltis) and two polyphagous species {Hylurgops paUiatus, T. lineatum). Table 5. Two-factor ANOVA table of species calches of iterated random sampling. The factors arc block and sampling scale. DF = degrees of freedom. MS = mean square. F = F-ratio. p = corresponding statistical value. Source Block Scale Interaction DF r-J 16* 10 X 10 m MS 193.34 19.47 1.13 2,73 70.90 7.14 0.42 0.0001 0.0010 0.8683 4 * 20 X 20 m 1 * 40 X 40 m Error Fig. 4. Multiple comparisons of species catehes in different sampling seales based on Tukey's Studentized Range Test (DF = 228. MSB = 2.730, n = 240), The arrow points to the scale having bigger species cateli (p<(l,05). eCOGRAE'HY : i : 5 (I'WS) Block Block 2 Block 3 Fig. 5. Muliiple comparisons of species catches in different blocks based on Tukey's Studentized Range Test (DF = 228. MSE = 2.730, n = 240). The arrow points to the block having bigger species eateh (p < 0.05). amount of breeding material does not necessarily cotitribute to high species diversity. Also the diversity of breeding material is itiiporlant, because microhabitat demands (e.g. dimension of bark and wood, age, moisture etc.) vary between species {Rudinsky 1962, Lekander et al. 1977). The present analysis of spatial concordance between species richness and environmental features support this conclusion. In general, the breeding material of scolytids is scarce and sporadically distributed over the space (Sjodin et al. 1989, Lindelow et al. 1992). but site characteristics and human activities strongly affect the availability and distribution of wooden substrate appropriate for particular bark beetle species. In our study the amount and distribution of breeding material differed between stands partly due to different forest tnanagemcnt and partly because of site characteristics. The importance of suitable breeding material for the species occurrence can be seen in the logistic regression models, though other variables are also tieeded in the models. However, the set of variables used at the present local stand scale was not adequate to explain the occurrence of the six most frequent species. Four of these species are associated with the Scots pine {T. pimperda, P. lichtensleini\ P. quadridens and P. bidcnlatus) and two are polyphagous species [H. palliatus and T. Uneatum). Only these six species were found frotn each of the four studied stands. In a previous study at landscape scale the assessment of bark beetle occurrence was based on satellite imagery and field measurements (Vaisiincn and Heliovaara 1994). Environmental variables (tree species, soil type, forest site type, stand characteristics and forest management history) predicted well the occurrence of monophagous species associated with pine, while predictions were much less accurate for species associated with spruce (e.g. Xylechinus pilosus. Dryocoetes sp.). In the present analyses at local stand scale, the results were the opposite: the environmental variables (site characteristics and availability of breeding material) predicted the occurrence of bark beetle species associated with spruce more accurately than the occurrence RCOGRAPHY 2I;5 of species associated with pine (e.g. T. piniperda, P. quadriden.s. P. hidentatus). These results indicate that factors regulating the habitat use of the scolytid species associated with spruce are operating at finer scales than those of the species associated with pine. The habitat selection of an individual species may respond either to microhabitat or macrohabitat variation depending on the species" physiology, morphology, and life strategies (Morris 1987). There are different ways that bark beetles may find the breeding material: random Right and landing, localization of suitable host by compounds volatilizing from wood, and aggregation pheromoncs released by other bark beetle specimens (Wood 1982b). Species have different colonization behaviours and this may be reason for differences in the ecological scale of resource exploitation. Pheromones have an important role in the colonization behaviour of several scolytids associated with spruce, while pine associated T. piniperda orientates by ethanol and acetaldehyde released from fresh wood (Sjodin et al. 1989, Lindelow et al. 1992). Trypodendron Uneatum is an ambrosia beetle which carries a fungus and feeds on it. The selected environmental variables in the present study as well as in the previous study at landscape scale (Vaisanen and Heliovaara 1994) did not explain the occurrence o^ T. Uneatum. Probably some other factors determine the occurrence of the ambrosia fungus and thus, also the occurrence of T. Uneatum. At landscape level, there are factors that may affect strongly the occurrence of bark beetle species, e.g. forest fragmentation and associated edge effect (Peltoncn et al. 1997. Peltonen and Heliovaara in press). In this study the blocks were delimited in large and relatively homogeneous stands to eliminate possible edge effect. 0kland et al. (1996) studied the diversity of saproxylic beetles at different scale levels in relation to ecological variables. According to their results, the smallest scale (40 x 40 tn) was too small for the study of the relationship between the ecological variables and the fauna of saproxyhc beetles. The extent and grain of their study were larger than ours, and they used window traps instead of direct collection. Differences between our results and those of 0kland et al. (1996) demonstrate the significance of scale in interpreting the ecological patterns. Our present results on bark beetles indicate that scaling is essential for the evaluation of species diversity even within a local stand. Relatively high species diversity at finer scales may be due to the even distribution of a few common species, but still, diversity at larger scales may remain low. Conversely, low species diversity at finer scales may be due to sporadical distribution of several less frequent species, resulting in increasing species diversity towards larger scales of resolution. 515 Table 6. Vanables of logistic regression models with t;orrespondiiig odds ratios [95% confidence limits], p refers to goodness-offit tests. Term 1 Species on pine Tomicus Site type 3,48 [1.88, piniperda Hylasles Site type brunneus 8.00 [3.61, Pityophthorus Site type lichlensteini 4.48 [2,66, Pityogenes Site type quadridem 2.34 [1,37, Pityogenes Site type bidenlatus 4.74 [2,95, OrthoWmicus Site type suluralis 7.09 [2,80, OrthoXomicus Site type laricis 4.55 [1.63, Species on spruce Xylechinus Site type pliosus 0.43 [0.23, Polygrapims Pine poligraphus 0,78 [0.65, Dryncoeles sp. Site type 0.26 [0.13, Site type Cryphalus sahuarius 0.43 [0.25, Pine Pityogenes chakographm 0.81 [0,68, Species on pine and spruce Hylurgops Dead pine patliaius 1.51 [1.15, Cryplurgus sp. Pine 0.74 [0.60, Trypodendron Dead pine Hneatum 5.08 [3.70, Term 2 Pine 1.59 [1.39. 1.81] Pine stump 17.86] 2.34 [1.63, 3.34] Pine branch 7.52] 1.27 [1-17, 1.37] Dead pine 3.98] 1.44 [1.08, 1.91] Pine 7.58] 1.!! [1.06, 1.17] Pine branch 17,86] 1.08 [I.OI. 1.15] Height 12.66] 0.84 [0.76. 0.94] 6.45] Pine 0.82] 0.78 [0.61. 0.99] Height 0.94] 1.16 [1.09, 1.24] Birch 0.55] 1.60 [1.13, 2.27] Dead spruce 0.74] 3.26 [2.34, 4.53] Birch 0.97] 0.44 [0.25, 0.80] Term 3 Height 0,93 [0.88, 0.99] Pine stump !.78 [1.20, 2,65] Pine branch 3.39 [2.48, 4.65] Pine branch 1.48 [1.32, 1.67] Spruee branch 5.32 [1.18, 23.81] Pine stump 1.99 [1.22, 3.26] Height 1,16 [1,07, 1.26] Dead spruce 5.88 [4,10, 8.48] Height l.il [1.02, 1,20] Spruce branch 1.86 [1.00, 3,45] Crown coverage 1.03 [1.15, 1.05] Dead spruce 6.94 [4.70, 10.31] Spruce branch 90.91 [37.04. 200,00] Spruce branch 12.05 [6.80. 21,74] Term 4 Dead pine 6.58 [2.45. 17.86] Term 5 Pine branch 1.65 [1.20, 2.28] P 0.0002 0.7395 0.0001 0.0294 0.0018 0.5494 0.9654 ns ns ns ns Dead spruce Spruee branch Pine stump 2.18 [1.67. 2.87] 2.27 [1.45, 3.55] 2.73 [2.13, 3,51] Dead spruce Spruee branch 0.90] 2.14 [1.58, 2.89] 6.80 [4.25, 10.99] Dead spruce Spruce branch 7.00] 2.59 [1.96, 3.44] 2.39 [1.47, 3.91] 1,97] ns ns Acknowledgements - Sincere thanks to John Wiens,

Journal

Ecography – Wiley

Published: Oct 1, 1998

There are no references for this article.