Species Diversity Of Longicorn Beetles In Humid Warm Temperate

Biodiversity and Conservation 11: 1919–1937, 2002.  2002 Kluwer Academic Publishers. Printed in the Netherlands.

Species diversity of longicorn beetles in humid warmtemperate forests: the impact of forest management practices on old-growth forest species in southwestern Japan KAORU MAETO 1,3, *, SHIGEHO SATO 1 and HIROAKI MIYATA2 1

Shikoku Research Center, Forestry and Forest Products Research Institute, Asakura-Nishimachi, Kochi 780 -8077, Japan; 2 Kochi Prefectural Forest Technology Center, Ouhira, Tosayamada 782 -0078, Japan; 3 Current address: Entomological Laboratory, Faculty of Agriculture, Kobe University, Nada-ku, Kobe 657 -8501, Japan; * Author for correspondence (e-mail: maeto@ kobe-u.ac.jp; fax: 181 -78 -803 5871) Received 22 June 2001; accepted in revised form 26 November 2001

Key words: Attractant trap, Bio-indicator, Chronosequence, Coarse woody debris, Conifer plantation, Insect biodiversity, Pidonia, Pollinator, Saproxylic Coleoptera Abstract. In the humid warm-temperate zone of southwestern Japan, old-growth forests have been seriously fragmented to small remnants due to traditional agriculture and coppicing as well as recent rapid plantation with conifers. Assemblages of longicorn beetles (Coleoptera: Disteniidae and Cerambycidae) were compared among old-growth forests, second-growth forests and conifer plantations using collision traps baited with chemical attractants. Species richness of longicorn beetles was poorer in second-growth forests and conifer plantations than in old-growth forests. It was proved by multidimensional scaling (MDS) that the beetle assemblages of old-growth forests were distinct from those of conifer plantations, while those of second-growth forests were intermediate between them. Further analysis showed that a number of species, including many Pidonia spp., were specific to or closely associated with old-growth forests, and the results were largely supported by the indicator value (IndVal) approach. It is likely that many of such old-growth forest species in the larval and pupal stages require large broad-leaved trees standing or fallen with thick bark. At the same time, the flower-visiting adults would play an important role in pollinating various herbaceous and woody plants. Regional forest management for the conservation of insect biodiversity is also discussed.

Introduction Southwestern Japan was formerly widely covered by humid warm-temperate evergreen or deciduous broad-leaved forests often mixed with evergreen conifers, but such forests have been extensively exploited since ancient times (Shidei 1974). Almost all lowland forests have been converted to rice paddies, farmland and residential areas. In hilly and mountain areas, the forests have been severely altered through shifting cultivation, coppicing for manure and fuel wood production, and plantation with conifers. After World War II, remaining old-growth forests were rapidly cleared and usually converted to plantations of Japanese cedar and cypress (Japan FAO Association 1997). As a result, relatively unchanged forests cover less than 1% of the potential area of this forest type in Japan (Sasse 1998). NACS-J and

1920 WWF Japan (1996) reported that warm-temperate forests have been seriously affected by deforestation and plantation with conifers, and the remnants need urgent conservation. Although arthropod diversity may be deteriorated due to the forest conversion, only a little is known about the importance of old-growth forests in the maintenance of insect diversity in Japan (Maeto and Makihara 1999). For sustainable forest management in consideration of regional biodiversity, it is necessary not only to know the effects of forest conversion on insect species richness but also to understand the ecological requirements and functional interactions of the insects that depend on old-growth forests (e.g. Fisher 1998). Quite a few studies have been conducted on the influence of forest management practices on arthropod communities or assemblages in temperate and boreal forests, e.g., canopy arthropods (Schowalter 1989), ground beetles (Niemela¨ et al. 1993; Magura et al. 2000), boreal forest beetles (Martikainen et al. 2000). However, the implications of forest treatments for longicorn beetles are largely unknown, although they are principal components of forest ecosystems as herbivores and detritivores of woody plants, pollinators of herbaceous and woody plants, and prey of insectivorous animals (Linsley 1959; Ohbayashi et al. 1992; Hanks 1999). The purpose of this study was to compare the assemblage of longicorn beetles among old-growth forests without logging records, second-growth forests and conifer plantations in the humid warm-temperate zone of Japan, in order to identify the characteristic species of old-growth forests and understand their ecological requirements. Collision traps baited with chemical attractants (Maeto et al. 1995; Shibata et al. 1996) were used for a quantitative comparison of beetle assemblages.

Materials and methods Study sites The study focused on the Shimanto River Basin of Kochi Prefecture in Shikoku, the fourth largest island of the Japanese archipelago. The river basin is covered with lowland and hilly forests up to 1400 m in elevation, principally composed of conifer plantations (about 60% of total forest area), second-growth forests (about 40%), and remnants of old-growth forests without historical records of clearance (less than 1%). Old-growth forests are dominated by evergreen conifers (Abies firma, Tsuga sieboldii, Chamaecyparis obtusa) as well as by diverse evergreen broad-leaved trees (Quercus spp., Castanopsis cuspidata, Machilis japonica, Cleyera japonica, etc.). Starting at 800 m a.s.l., evergreen broad-leaved trees are gradually replaced by deciduous trees (Carpinus spp., Fagus spp., Betula grossa, Acer spp., etc.). Secondgrowth forests have been repeatedly cut at intervals of 30–80 years; coppices dominated by evergreen broad-leaved trees (C. cuspidata, Quercus glauca, etc.) are developed in lowlands, and mixed forests of red pine (Pinus densiflora) and deciduous broad-leaved trees (Q. serrata, Carpinus spp., etc.) are widespread in the hills. Conifer plantations are monocultures of Japanese cedar (Cryptomeria japonica) or Japanese cypress (C. obtusa); in appropriately managed plantations

1921 over 30 years old, forest floors are covered with rich species of shrubs after thinning, and diverse tree species grow along forest edges, paths and streams. Annual mean temperature and annual precipitation in 1991–2000 at Yusuhara (415 m a.s.l., Figure 1) averaged 13.4 8C and 2722 mm, respectively. Seven old-growth forests (without any records of clearance at least for 120 years), five second-growth forests (30–70 years old) and two appropriately managed conifer plantations (30–40 years old) were selected for sampling sites within a 50 3 50 km area of the river basin (338059–338309 N, 1328359–1338059 E; Figure 1). For each site, woody plant species observed in a plot of about 0.05 ha were recorded, and the diameter at breast height (DBH) of the trees (DBH .5 cm) with the canopy hanging over a 20 m randomly placed straight line was measured. Site location, forest type, approximate forest age, altitude, direction, slope, the number of woody plant species, and the maximum DBH of conifers and broad-leaved trees are listed in Table 1. Sampling and identification Specimens were collected with yellow or white collision traps, each baited with

Figure 1. Location of the study sites in Kochi Prefecture, Shikoku Island, Japan.

Irazu, H Tengu, H Yokogai, Y Tarohgawa, Y Kohnoshi, Y Takatori, Y Kubotani, Y Tsuzuragawa, T Ichinomata, T Ichinomata, T Nakaba, N Kuroson, N Kuroson, N Haraikawa, N

A B C D E F G H I J K L M N

OLD OLD SEC SEC CPL OLD OLD SEC OLD CPL SEC SEC OLD OLD

Forest type b Approx. area (ha)

150 100 .200 .200 .200 90 80 20 50 .200 .200 30 250 50

Age of forest (years)

.190 .120 40–50 30–40 30–35 .170 .200 40–50 .180 30–35 20–30 70 .120 .140

c

b

1050 1250 800 700 700 350 700 400 500 450 100 350 700 500

Altitude (m)

N S N W S N W W N N N S E N

Direction

15 40 20 40 25 20 45 40 40 35 45 40 35 40

Slope (8)

23 8f 24 11 17 27 24 36 22 23 21 15 15 16

Number of woody plant species d

99.8 (2) ,5.0 (0) 42.0 (6) 12.4 (5) 23.0 (11) 58.2 (2) 76.3 (4) 43.2 (2) 51.0 (2) 16.9 (9) ,5.0 (0) 37.6 (2) 116.6 (1) ,5.0 (0)

Conifers

33.1 (20) 26.8 (11) 13.5 (15) 25.3 (10) ,5.0 (0) 78.6 (19) 27.9 (5) 14.3 (14) 42.4 (12) ,5.0 (0) 21.7 (30) 30.6 (12) 34.4 (16) 39.4 (17)

Broad-leaved trees

Maximum DBH (cm) of trees e

c

H – Higashitsuno-mura; Y – Yusuhara-cho; T – Taishoh-cho; N – Nishitosa-mura. OLD – old-growth forest; SEC – second-growth forest; CPL – conifer plantation. e d Approximately estimated area of the same type, continuous forests. Number of woody plant species observed in a plot of about 0.05 ha. For all trees hanging over a f straight line of 20 m; numbers in parentheses indicate the number of trees with DBH .5 cm. Forest floor was poor in species, being covered with dwarf bamboo.

a

Location a

Site

Table 1. Location, forest type, approximate forest age and stand variables of the study sites.

1922

1923

Figure 2. Collision trap baited with benzyl acetate and ethyl alcohol.

benzyl acetate and ethyl alcohol. The collision traps consisted of a roof, two collision plates (about 25 3 20 cm) intersecting each other, and a bucket (Figure 2; Sankei Chemical Co., Ltd.; Maeto et al. 1995; Shibata et al. 1996). Benzyl acetate is one of the main components of floral fragrance, and it lures various flower-visiting beetles (Ikeda et al. 1993; Sakakibara et al. 1997b). Ethyl alcohol is known to attract xylophagous beetles for oviposition (Ikeda et al. 1980). Dispensers of the chemicals, each about 25 ml, were placed under the roof. Water containing a surface-active agent and sorbic acid was poured into the bucket to preserve the specimens. The traps were set at about 1.5 m above the ground. At each site, two white and two yellow traps were placed alternatively at intervals of about 50 m in line. Trapping was conducted from early April to mid-September in 1998. Every 2 weeks, chemicals were renewed and materials in the buckets were taken to the laboratory. Trap sampling was repeated in the same manner at five sites (A, C–F) in 1999, to confirm the invariability of species assemblages between years. Longicorn beetles (Coleoptera: Disteniidae and Cerambycidae) were pinned for identification. They were identified with Ohbayashi et al. (1992).Voucher specimens will be deposited in the Shikoku Research Center, Forestry and Forest Products Research Institute, Kochi, Japan. To evaluate the taxonomic bias in the trap sampling, the number of species collected in this study was compared with the number of species collected by general methods in the same region for each family and subfamily (Nakayama et al. 1994). They reported 148 species of longicorn beetles collected by hand, in beating,

1924 Table 2. Number of longicorn beetle species collected in white and yellow collision traps attached with benzyl acetate and ethanol in 1998, and of those collected with general methods (i.e. by hand, in beating, in light traps and in collision traps) over 6 years (1987–1993) in the same region. Family

Subfamily

Disteniidae Cerambycidae

Prioninae Spondylinae Lepturinae Necydalinae Cerambycinae Lamiinae

Total number of species

Number of species collected in the traps in 1998 a WT

YT

Total (T)

0 2 1 34 0 17 16 70

2 2 1 13 0 17 18 53

2 3 1 34 0 19 23 82

Number of species collected with general methods b (O)

T/O

2 4 4 36 1 36 65 148

1.00 0.75 0.25 0.94 0.00 0.53 0.35 0.55

a

WT – white traps; YT – yellow traps; both were baited with benzyl acetate and ethanol. b At Mt. Ohnakayama (Figure 1), Yusuhara-cho, Kochi Prefecture (Nakayama et al. 1994).

in light traps and in various bait traps for 7 years on Mt. Ohnakayama, Kochi Prefecture, up to about 800 m in elevation (Figure 1). Vegetation was composed of second-growth forests, plantations of Japanese cypress, plantations of Q. acutissima, and old-growth remnants. Data analyses All the specimens captured in the two white and two yellow traps at each site every year were pooled together for the following analyses. Two species richness indices, i.e. the number of species and the Margalef index (Magurran 1988), were compared among three forest types and among four directions by the Kruskal–Wallis test based on the samples of 1998. Kendall’s coefficient of rank correlation (Kendall’s tau, t ) between the species richness indices and the stand variables (altitude, slope, the number of woody plant species, and the maximum DBH of conifers and broad-leaved trees) were tested. The Margalef index was calculated as Margalef index5(S21) / lnN where S is the number of species and N the number of individuals. Similarity of assemblages between each pair of samples was computed from abundance data using the rank correlation coefficient (Kendall’s tau, t ) (Ghent 1963; Huhta 1979). From the matrix of similarity (t ), the samples were plotted in a two-dimensional space by multidimensional scaling (MDS). First, we performed two-dimensional MDS for all samples collected in 1998 (14 sites) and 1999 (5 sites). The samples were clustered with the group average method based on the Euclidean distance between each pair of them in the MDS space. After we confirmed that the samples of 1998 and 1999 collected at every site were fairly close to each other in the dendrogram, we performed two-dimensional MDS for the

1925 samples of 1998 (14 sites) again. For each of the two MDS dimensions, the differences in site score among forest types and among directions were tested by the Kruskal–Wallis test. Kendall’s coefficient of rank correlation was tested between the MDS site scores and the stand variables. For the species collected at three or more sites, Kendall’s coefficient was tested between the MDS scores and the number of individuals. All statistical analyses were performed with STATISTICA (StatSoft Inc. 1998). ˆ and Legendre We also applied the indicator value method proposed by Dufrene (1997) to identify the most characteristic species of each forest type for the samples of 1998. This method combines a species relative abundance with its relative frequency of occurrence in a particular habitat type to obtain the indicator value for the habitat type. The species indicator value (IndVal) is the maximum indicator value over all habitat types (three forest types in our study). Statistical significance of IndVal was evaluated with a Monte Carlo test (McCune and Mefford 1997).

Results

Taxonomic bias in trap sampling A total of 25 115 individuals belonging to 82 species of longicorn beetles were collected in 1998. As compared with the fauna of longicorn beetles in the same region (Nakayama et al. 1994), our trapping system collected all species of the family Disteniidae, most of the subfamily Lepturinae, but a rather small portion of the subfamilies Cerambycinae and Lamiinae (Table 2). The lepturine species were all captured in the white traps, while the disteniid species and some cerambycine and lamiine species were captured only in the yellow traps.

Figure 3. Species richness indices of longicorn beetles in relation to the number of individuals collected in 1998. For forest type symbols, see Figure 1.

1926

Figure 4. Margalef index in relation to the number of woody plant species (a), and to the maximum DBH of broad-leaved trees (b) (cf. Table 4). For forest type symbols, see Figure 1.

Table 3. Median and range (in parentheses) of the species richness indices and MDS site scores for longicorn beetle assemblages of three forest types in 1998. Type of forest

Species richness Number of species Margalef index MDS site scores First dimension Second dimension

H

Old-growth forest n 5 7

Second-growth forest n 5 5

Conifer plantation n 52

27 (15–33) 3.79 (3.03–4.27)

22 (9–24) 2.97 (1.67–3.32)

21 (20–22) 2.41 (2.03–2.79)

2.61 8.14 *

20.60 (21.05 to 20.04) 0.03 (21.34–0.74)

0.33 (20.24–0.82) 20.07 (20.44–1.21)

1.45 (1.35–1.55) 20.38 (20.98–0.22)

10.13 ** 0.56

Differences were examined by the Kruskal–Wallis test. * P , 0.05;

**

P , 0.01.

Table 4. Rank correlation coefficient (t) between the species richness indices and MDS site scores for longicorn beetle assemblages of 1998 and the stand variables of study sites (n 5 14 sites). Altitude Species richness indices Number of species 20.036 Margalef index 0.265 MDS site scores First dimension 20.219 Second dimension 20.357 ms *P , 0.05;

ms

0.05 , P , 0.1.

Slope

Number of woody plant species

Maximum DBH of conifers

Maximum DBH of broad-leaved trees

20.217 20.061

0.268 0.156

0.430* 0.380 ms

0.380 ms 0.420*

0.012 0.209

0.045 20.022

20.313 20.067

20.508* 0.331

1927

Figure 5. Cluster analysis of the longicorn beetle assemblages from 1998 to 1999. For site abbreviations, see Figure 1. Boldfaced letters and italic letters indicate the samples of 1998 and 1999, respectively.

Species richness The number of longicorn beetle species rose as the number of individuals collected increased (Figure 3a; t 5 0.412, n 5 14, P , 0.05). It was not significantly different among forest types, whereas the median value was largest in old-growth forests (Table 3). The Margalef index was independent of the number of individuals (Figure 3b; t 5 20.121, n 5 14, P . 0.5). It was significantly different among forest types, and the median value decreased from old-growth forests to secondgrowth forests to conifer plantations (Table 3). Neither value was significantly different among directions (Kruskal–Wallis test; H 5 1.36, P . 0.5, for the number of species; H 5 0.79, P . 0.5, for the Margalef index). While not correlated with altitude, slope or the number of woody plant species, they were positively correlated with the maximum DBH of conifers and broad-leaved trees, either significantly or marginally (Table 4). Figure 4 shows the relationship of the Margalef index to the number of woody plant species, and to the maximum DBH of broad-leaved trees. Site ordination As shown in the dendrogram in Figure 5, the longicorn beetle samples of 1998 and 1999 collected at every site (A, C–F) were close to each other, indicating that the beetle assemblages were stable and did not change considerably from 1998 to 1999. Figure 6 shows the final two-dimensional configuration of study sites by MDS based on the rank correlation coefficients between them calculated from the samples of 1998. The site score for the first dimension was significantly different among forest types, while the score for the second dimension was not (Table 3). The former

1928

Figure 6. Two-dimensional ordination of study sites by the MDS based on the degree of similarity between each pair of sites measured by means of the rank correlation coefficient (Kendall’s tau, t ) on the abundance of longicorn beetle species collected in 1998. For site abbreviations, see Figure 1.

Figure 7. Relationships between the site score on the first MDS dimension and the maximum DBH of broad-leaved trees (a), and between that on the second MDS dimension and altitude (b) (cf. Table 4). For forest type symbols, see Figure 1.

increased from old-growth forests to second-growth forests to conifer plantations (Figure 6). The site score for the first dimension was significantly correlated with the maximum DBH of broad-leaved trees, whereas the score for the second dimension was marginally correlated with altitude (Table 4, Figure 7). The site scores were not different among directions (Kruskal–Wallis test; H 5 0.99, P . 0.5, for the first dimension; H 5 4.61, P . 0.2, for the second dimension). Neither were

1929

Figure 8. Number of longicorn beetle species collected in 1998 in relation to the site score on the first MDS dimension.

they correlated with the slope, the number of woody plants and the maximum DBH of conifers (Table 4). Table 6 shows the rank correlation coefficients between the abundance and the MDS site scores for 40 species collected at three or more sites. Negative correlation with the site score for the first dimension was significant in 12 species, including six species of the genus Pidonia. This shows that they are specific to or closely associated with old-growth forests. On the other hand, positive correlation was significant in four species, e.g. Parastrangalis spp., suggesting that they are absent or uncommon in old-growth forests. For the other 24 species, no significant correlation with the site score was shown for the first dimension. Three species showed a positive correlation with the site score for the second MDS dimension, which was somewhat correlated with altitude (Table 4). For all eight species of Pidonia listed in Table 5, the correlation between the abundance and the site score for the first MDS dimension was negative if not significant at the 5% level. Eleven species of Pidonia, including three infrequent species (P. yamato, P. neglecta, P. chujoi), were collected in 1998. The number of Pidonia species was highly and negatively correlated with the site score for the first MDS dimension (Figure 8; t 5 20.716, n 5 14, P , 0.001), although the number of the other species was not correlated with it (Figure 8; t 5 0.012, n 5 14, P . 0.5). Also the number of Pidonia species was significantly different among three forest types (Kruskal–Wallis test; H 5 6.65, P , 0.05). Species indicator values Species indicator values (IndVal) were computed for 40 species collected at three or more sites in 1998. The values over 40% are shown in Table 5. Large IndVal for old-growth forests were indicated in Dinoptera minuta, many Pidonia species, and Pseudalosterna misella. Many species, including Parastrangalia spp., exhibited

1930

Table 5. Rank correlation coefficient (t ) between the number of individuals and the site scores in the MDS ordination (n 5 14 sites) for the species collected at three or more sites in 1998, and the species indicator value (IndVal .40%) for the species. Family or subfamily species

Disteniidae D. gracilis Prioninae Psephactus remiger Lepturinae Lemula japonica Di. minuta Pidonia mutata P. signifera P. grallatrix P. aegrota P. puziloi P. approximata P. amentata P. simillima Ps. misella Anoploderomorpha excavata Anastrangalia sequensi Leptura ochraceofasciata Parastrangalis lesnei Pa. shikokensis Idiostrangalia contracta Cerambycinae Allotraeus sphaerioninus Stenodryas clavigera Ceresium holophaeum Cleomenes takiguchii Chloridolum viride Callidiellum rufipenne Xylotrechus emaciatus X. pyrrhoderus X. cuneipennis X. grayii Demonax transilis Lamiinae Asaperda rufipes Pterolophia tsurugiana Acalolepta fraudatrix A. sejuncta Uraecha bimaculata Xenicotela pardalina Rhodopina integripennis Pareutetrapha simulans Praolia citrinipes Glenea relicta a

Correlation coefficient (t) between abundance and MDS site score First dimension

Second dimension

20.42*

0.13

0.04

0.01

0.02 20.66** 20.54** 20.46* 20.51* 20.45* 20.64** 20.38 20.12 20.76** 20.44* 20.49* 0.19 0.15 0.43* 0.62** 0.43*

0.37 20.17 20.01 0.04 0.31 0.22 20.06 0.36 0.53** 20.09 0.23 0.24 0.27 20.02 20.34 0.05 20.11

20.41* 20.08 20.02 0.04 20.37 0.44* 0.11 20.07 0.12 0.19 0.16

0.00 20.03 0.41* 0.19 0.40* 20.02 20.24 0.31 20.31 0.09 0.05

20.03 20.48* 0.06 20.12 20.26 0.07 20.30 0.21 20.36 20.31

0.36 20.34 20.04 20.35 0.28 0.27 20.13 0.08 0.13 0.16

IndVal (%)

Total number of individuals/sites of occurrence a Old-growth forests (7 sites)

42.9

82.3* 71.4* 42.9 70.8* 62.5 90.9** 81.4*

42.5 93.3*

70.0* 80.6* 48.3 42.9 61.7*

62.9

Boldfaced numbers show the data set for the IndVal. *P , 0.05; **P , 0.01.

Conifer plantations (2 sites)

8/3

3/1

3/3 8/5

57.1 54.3 48.6 83.0* 47.8 92.2* 56.5

Second-growth forests (5 sites)

3/2 4/4 49/5 8/4 43/6 30/6 422/7 23/6 19/3 34/6 12/5 3/3 4/2 86/6

1/1 11/2 1/1 1/1 12/3 18/3 6/3 2/1 1/1

2/1 3/1 1/1

3/2 86/5 3/2 1/1 1/1

2/1 143/2 2/2 4/2 3/2

18/5 17/3 4/2 4/3 58/4 1/1 6/4 6/3 5/4 3/1 5612/7

6/3 25/3

5/1 2/1 1/1

4118/5

4/1 2/2 13488/2

42/6 4/3 13/6 9/4 190/7 4/3 5/4 4/1 31/5 2/2

35/5

3/1

19/5 7/4 111/5 3/3 2/2 2/2 3/2 1/1

1/1 1/1 38/2

1/1

2/2 11/2 2/2 5/2 3/3

4/2 5/1

1/1

1931 Table 6. Host condition, larval host tissue and host trees for the indicator species with IndVal. Host condition Indicator species for old-growth forests D. gracilis Dying or humid dead tree Di. minuta Dead twig P. mutata P. signifera Dead branch P. grallatrix P. aegrota Humid dead tree and branch P. puziloi P. approximata P. simillima Dead branch Ps. misella Standing tree An. excavata Dead tree Ch. viride Dead branch Pt. tsurugiana Dead branch Pr. citrinipes Dead tree and branch Indicator species for second-growth forests As. rufipes Dead twig Ac. fraudatrix Dead tree and branch Indicator species for conifer plantations L. ochraceofasciata Dead tree Pa. lesnei Dried dead branch Pa. shikokensis Dead tree Dead vine I. contracta Ca. rufipenne Dead or dying tree X. grayii Dead tree De. transilis Dead tree and branch

Larval host tissue a

Host trees b

IB / SW

BL / C BL

IB

BL BL BL

IB

IB IB / SW/ R W IB / SW

BL BL BL BL / C BL BL

W IB / W

BL BL / C

W IB / SW

C / BL C / BL BL BL C BL BL / C

IB / SW W

Data sources are Kiyosawa et al. (1981), Kojima and Nakamura (1986), Kuboki (1987). a IB – inner bark; SW – sap wood; W – wood; R – root. b BL – broad-leaved tree; C – conifer.

high indicator values for conifer plantations. Acalolepta fraudatrix was also characteristic of second-growth forests. The results of the IndVal approach agreed essentially with those from the MDS ordination analyses. According to literature (Kiyosawa et al. 1981; Kojima and Nakamura 1986; Kuboki 1987), host condition, larval host tissue and host trees for the indicator species with IndVal .40% are compiled in Table 6.

Discussion Limitations of trap sampling As previously reported by Ikeda et al. (1993), Shibata et al. (1996) and Sakakibara et al (1997a, b, 1998), the white collision trap baited with benzyl acetate to mimic wild flowers is efficient for collecting flower-visiting species of the subfamilies Lepturinae and Cerambycinae. On the other hand, the species without flower-

1932 visiting habits (Lamiinae, Prioninae and Spondylinae) are probably lured to ethyl alcohol (Ikeda et al. 1980; Shibata et al. 1996; Sakakibara et al. 1997a). While yellow traps are generally less attractive than white traps, certain species (e.g. Disteniidae) have been largely collected in them (Sakakibara et al. 1997a). Thus, the taxonomic bias observed in our trap sampling is fundamentally consistent with these previous reports. The present combination of white and yellow collision traps baited with benzyl acetate and ethyl alcohol appears to be useful for sampling diverse longicorn beetles, although the collection is rather poor for non-flower-visiting species. Although the attractant traps lure the longicorn beetles moving just around them (T. Ikeda, personal communication), they can catch those that fly from adjacent stands as in other sampling methods. For example, adults of a lepturine species (Anaglyptus subfasciatus) were caught in the traps located at a distance of 30–50 m from the stands of emergence (Makihara 1992). Thus, the results may be biased by the contamination with strong-flying species from other habitat types, especially when the sampling site is small in area. However, such biases would be not serious because our results show distinctive ordination of sampling sites corresponding to forest types. Malaise traps are also useful for the investigation of beetle assemblages (Maeto and Makihara 1999). They can catch more non-flower-visiting species than the collision traps with attractants, but they cost much more to operate than do collision traps. To compare beetle assemblages among many sites at a time, collision traps with attractants would be more practical than Malaise traps. Although abundance data obtained from attractive traps are not direct measures of real abundance of each species, the rank of abundance might change with the alternation of species between sampling sites. Therefore, we used the rank correlation coefficient (Kendall’s t ) as the similarity measure between assemblages (Ghent 1963). According to Huhta (1979), it is one of the best indices to measure the species alternation of spiders and beetles in succession after clear-cutting. Our results demonstrate that site ordination by MDS based on rank correlation coefficients is practical for analyzing the changes in insect assemblage using abundance data obtained from attractant traps. It is generally known that some herbivorous insects (moths, sawflies, bark beetles, etc.) exhibit wide density fluctuations over years (e.g. Varley et al. 1973). If the relative abundance of species varies greatly between years, 1-year sampling would be insufficient to identify any indicator species being characteristic of certain habitat types. However, the assemblages of longicorn beetles collected in two consecutive years (1998 and 1999) were very close to each other for every site, so that the general pattern of the beetle assemblages may be discussed based on a single year sampling. Decline of species richness An obvious decrease in the Margalef index, the number of species adjusted for the number of individuals, indicates that the conversion of old-growth forests into

1933 second-growth forests and conifer plantations has diminished the species richness of longicorn beetles. Shibata et al. (1996) mentioned that the number of longicorn beetle species tended to increase in proportion to the number of woody plant species. However, except for a cool-temperate beech forest, no correlation was shown between the number of beetle species and that of woody plant species in the warm-temperate forests in their study (Figure 3 of Shibata et al. 1996). This agrees with our results, and it is not likely that forest conversion has reduced the species richness of longicorn beetles through the decrease of woody plant species. Our study suggests that the species richness of longicorn beetles increases with the maximum diameter of trees. However, it is not surprising that old-growth forests have large trees. Further examination will require more information about the biology and natural history of the species closely associated with old-growth forests. Ecological requirements and functions of the old-growth forest species It was proved by the MDS ordination that the longicorn beetle assemblages of old-growth forests were distinct from those of conifer plantations, while those of second-growth forests were intermediate between them. Further analysis showed that 12 species were specific to or closely associated with old-growth forests. Of the 12 species, six belong to the genus Pidonia of the subfamily Lepturinae. Furthermore, the total number of Pidonia species has definitely increased in old-growth forests along the first MDS dimension. These findings indicate that most species of Pidonia require some ecological conditions particular to old-growth forests, and were supported by the results from the indicator value (IndVal) approach. While it is widely distributed in the Holarctic Region, Pidonia is most highly diversified in humid temperate forests of East Asia (Kuboki 1981). Larvae of longicorn beetles feed on various parts (wood, sapwood or inner bark of trunks, branches or roots) of woody plants in different conditions (decaying, dying or barely living) (Hanks 1999). As shown in Table 6, specialization to inner bark is a peculiar habit of Pidonia (Kuboki 1987). The larvae of this genus are found under the thick bark of comparatively large, dead or living, trees and branches of various broadleaved species (Aceraceae, Araliaceae, Betulaceae, Cercidiphyllaceae, Fagaceae, Rosaceae, Salicaceae) or, rarely, conifers (Pinaceae) (Kiyosawa et al. 1981; Kojima and Nakamura 1986; Kuboki 1987). They pupate under the bark (never in wood) or drop into humus for pupation. The humidity is unstable under thin bark, and probably so they need thick bark under which they spend 1 or 2 years for growing up. On the other hand, a thick, stable and moist humus layer of old-growth forests may be also crucial to the larvae and pupae of Pidonia, which often live in humus or under the bark of fallen logs lying on the forest floor (Kuboki 1987). An additional species associated with old-growth forests, Distenia gracilis, also prefers dead humid logs with thick bark to dried logs, having a small preference to particular tree species (Kiyosawa et al. 1981). Another species, Ps. misella, is known as a bark or root borer of standing trees of Salicaceae (Ohbayashi et al. 1992). Therefore, it is most likely that the presence of large, dead or living, broad-leaved

1934 trees with thick bark is essential for the longicorn beetles that depend on warmtemperate old-growth forests. This is not inconsistent with our finding that the beetle assemblages change with the increase of the trunk size of broad-leaved trees. We did not measure the volume and size of dead wood, while both the size of trees and amount of dead wood usually increase together with forest age (e.g. Spetich et al. 1999; Siitonen et al. 2000). In addition to decomposing woody materials by the larvae, Pidonia and other lepturine beetles may be playing an important role in pollinating herbaceous and woody plants, since they are the most dominant beetles visiting wild flowers in temperate forests (Kato et al. 1990; Sakakibara et al. 1997b). According to Kuboki and Shimamoto (1979) and Kuboki (1980), Pidonia species have been found on wild flowers of more than 15 families (e.g. Saxifragaceae, Rosaceae, Umbelliferae, Caprifoliaceae). Moreover, some species-specific relationships have been observed between Pidonia and flowers (Kuboki 1980). It must be noted that the old-growth forest beetles, depending on large full-grown trees, will support the reproduction and genetic diversity of other plant species. They might also be major prey for predatory insects, spiders and birds visiting wild flowers during late spring and early summer. At any rate, further investigations are necessary to understand effects of the decline in insect diversity of old-growth forests upon the forest ecosystem.

Implications for the management of forest landscape Conversion of old-growth forests into young forests results in considerable changes in the diversity of various arthropod guilds, e.g., canopy arthropods (Schowalter 1989), ground beetles (Niemela¨ et al. 1993), and xeric insects (Lattin 1993), but there is general agreement that saproxylic arthropods, depending on large pieces of dead wood, are most threatened by the long-term reduction of temperate old-growth forests (e.g. Warren and Key 1991; Lattin 1993; Maeto and Makihara 1999; Martikainen et al. 2000; Thunes et al. 2000). While the larvae occasionally bore into living tress, Pidonia species are also saproxylic in a broad sense since they depend on large woody materials that are dead or dying. To enhance the diversity of saproxylic insects, extended rotation as well as leaving old trees, snags and dead wood in clear-cuts is recommended (Hansen et al. 1991; Martikainen et al. 2000). Our results suggest that size increase of living, and thus dead or dying broad-leaved trees will augment the diversity of old-growth forest beetles like Pidonia. Extended rotation or postponement of cutting is no doubt most important to increase large trees in second-growth forests, and thinning may be also effective in accelerating the growth of remaining trees. Thinning in secondgrowth forests and plantations should be recommended since it also enhances the diversity of ground insects (Magura et al. 2000) as well as woody and herbaceous plants. After thinning, felled trees should be retained in the forests to be later available for saproxylic species. As suggested by Siitonen et al. (2000), it would be the most efficient short-term management strategy for the increase of structural diversity and old-growth attributes in managed forests to retain the old-growth

1935 characteristics (i.e. large living trees, snags and logs) that already exist in mature stands. In southwestern Japan, old-growth forests of the warm-temperate zone have been seriously fragmented due to traditional agriculture and coppicing as well as recent rapid plantation with conifers (e.g. Shidei 1974; Sasse 1998). On Shikoku Island, only a dozen small remnants of old-growth forests (each at most 300 ha) are found in lowlands and hills up to about 1000 m in elevation, and conifer plantations and young second-growth forests distantly separate them. For the conservation of regional forest biodiversity, restoration of old forests from young second-growth or man-made forests surrounding old-growth remnants is necessary to secure the habitat area of old-growth forest species. It is also essential to re-establish a belt or stepping-stones of old second-growth forests connecting old-growth remnants within the region. Pidonia and other longicorn beetles closely associated with old-growth forests, which can be easily monitored with simple traps, may be valuable indicators for the progress of such forest restoration.

Acknowledgements We would like to thank Ryuichi Tabuchi, Takeshi Sakai, Shigeo Kuramoto and Atsushi Sakai for the survey of site vegetation. Our thanks are also due to Toshihiko Yamasaki for his help in the field, Tsuyoshi Yamada for showing us meteorological data at Yusuhara, and Mariko Takeuchi for the preparation of insect specimens. We also thank Shikoku Regional Forest Office, Kochi Prefecture and Yusuhara Town for the permission to do field work in their forests. This work was partly funded by the International Collaborative Research Programme of the Ministry of Agriculture, Forestry and Fisheries.

References ˆ M. and Legendre P. 1997. Species assemblages and indicator species: the need for a flexible Dufrene asymmetrical approach. Ecological Monographs 67: 345–366. Fisher B.L. 1998. Insect behavior and ecology in conservation: preserving functional species interactions. Annals of the Entomological Society of America 91: 155–158. Ghent A.W. 1963. Kendall’s Tau coefficient as an index of similarity in comparisons of plant or animal communities. The Canadian Entomologist 95: 568–575. Hanks L.M. 1999. Influence of the larval host plant on reproductive strategies of cerambycid beetles. Annual Review of Entomology 44: 483–505. Hansen A.J., Spies T.A., Swanson F.J. and Ohmann J.L. 1991. Conserving biodiversity in managed forests: lessons from natural forests. BioScience 41: 382–392. Huhta V. 1979. Evaluation of different similarity indices as measures of succession in arthropod communities of the forest floor after clear-cutting. Oecologia 41: 11–23. Ikeda T., Enda N., Yamane A., Oda K. and Toyoda T. 1980. Attractants for the Japanese pine sawyer, Monochamus alternatus Hope (Coleoptera: Cerambycidae). Applied Entomology and Zoology 15: 358–361. Ikeda T., Ohya E., Makihara H., Nakashima T., Saitoh A., Tate K. et al. 1993. Olfactory responses of

1936 Anaglyptus subfasciatus Pic and Demonax transilis Bates (Coleoptera: Cerambycidae) to flower scents. Journal of the Japanese Forestry Society 75: 108–112. Japan FAO Association 1997. Forests and Forestry in Japan. 2nd edn. Japan FAO Association, Tokyo. Kato M., Kakutani T., Inoue T. and Itino T. 1990. Insect–flower relationship in the primary beech forest of Ashu, Kyoto: an overview of the flower phenology and the seasonal pattern of insect visits. Contribution to the Biological Laboratory of Kyoto University 27: 309–375. Kiyosawa H., Hayakawa H., Furuhata T., Hori K. and Kobayashi Y. 1981. Illustrated Guide to the Longicorn Beetles of Nagano Prefecture. Nihon-Minzoku-Shiryokan, Tokyo-Horei Shuppan, Matsumoto, Nagano (in Japanese). Kojima K. and Nakamura S. 1986. Food Plants of Cerembycid Beetles (Cerambycidae, Coleoptera) in Japan. Hiba Society of Natural History, Shobara, Hiroshima (in Japanese). Kuboki M. 1980. Studies on the flower visiting habits of the cerambycid genus Pidonia. Japanese Journal of Ecology 30 (in Japanese with English summary): 133–143. Kuboki M. 1981. Study on the lepturine genus Pidonia Mulsant (Coleoptera, Cerambycidae). Kontyu, Tokyo 49: 525–541. Kuboki M. 1987. Insects of Japan. Vol. 5: Cerambycid Genus Pidonia. Bun’ichi-Sogo Shuppan, Tokyo (in Japanese). Kuboki M. and Shimamoto R. 1979. Studies on the vertical distribution of Pidonia species (Coleoptera: Cerambycidae) in Shikoku, Japan. Kontyu, Tokyo 47: 196–203. Lattin J.D. 1993. Arthropod diversity and conservation in old-growth Northwest forests. American Zoologist 33: 578–587. Linsley E.G. 1959. Ecology of Cerambycidae. Annual Review of Entomology 4: 99–138. Maeto K. and Makihara H. 1999. Changes in insect assemblages with secondary succession of temperate deciduous forests after clear-cutting. Japanese Journal of Entomology (N.S.) 2 (in Japanese with English summary): 11–26. Maeto K., Fukuyama K., Sajap A.S. and Wahab Y.A. 1995. Selective attraction of flower-visiting beetles (Coleoptera) to floral fragrance chemicals in a tropical rain forest. Japanese Journal of Entomology 63: 851–859. ´ ´ ´ B. and Bordan ´ Z.S. 2000. Effects of nature management practice on carabid Magura T., Tothmeresz assemblages (Coleoptera: Carabidae) in a non-native plantation. Biological Conservation 93: 95–102. Magurran A.E. 1988. Ecological Diversity and Its Measurement. Croom Helm, London. Makihara H. 1992. Dispersal of the cryptomeria twig borer (Anaglyptus subfasciatus) from sugi plantations. Forest Pests 41 (in Japanese): 112–115. Martikainen P., Siitonen J., Punttila P., Kaila L. and Rauh J. 2000. Species richness of Coleoptera in mature managed and old-growth boreal forests in southern Finland. Biological Conservation 94: 199–209. McCune B. and Mefford M.J. 1997. PC-ORD. Multivariate Analysis of Ecological Data. Version 3.0. MjM Software Design, Gleneden Beach, Oregon. NACS-J and WWF Japan 1996. Red Data Book of Plant Communities in Japan. Abokkusha, Yokohama (in Japanese with English summary). Nakayama K., Yoshinaga K. and Inui K. 1994. Cerambycid beetles of Mt. Ohnakayama, Kochi. Gensei 65 (in Japanese): 17–26. Niemela¨ J., Langor D. and Spence J.R. 1993. Effects of clear-cut harvesting on boreal ground-beetle assemblages (Coleoptera: Carabidae) in western Canada. Conservation Biology 7: 551–561. Ohbayashi N., Satoˆ M. and Kojima K. 1992. An Illustrated Guide to Identification of Longicorn Beetles of Japan. Tokai University Press, Tokyo (in Japanese). Sakakibara Y., Iwata R. and Yamane A. 1997a. Performance of traps with attracting chemicals in faunal monitoring longicorn beetles (Coleoptera: Disteniidae and Cerambycidae) in a broad-leaved deciduous forest. Japanese Journal of Environmental Entomology and Zoology 8: 154–163. Sakakibara Y., Yamane A., Iwara R. and Yamada F. 1997b. Evaluation of beetle capture in traps as compared with manual capture on flowers in a long-term investigation in a beech forest. Journal of Forest Research 2: 233–236. Sakakibara Y., Kimura A., Iwata R. and Yamane A. 1998. Performances of four chemicals with floral

1937 scents as attractants for longicorn beetles (Coleoptera: Cerambycidae) in a broadleaved forest. Journal of Forest Research 3: 221–224. Sasse J. 1998. The Forests of Japan. Japan Forest Technical Association, Tokyo. Schowalter T.D. 1989. Canopy arthropod community structure and herbivory in old-growth and regenerating forests in western Oregon. Canadian Journal of Forest Research 19: 318–322. Shibata E., Sato S., Sakuratani Y., Sugimoto T., Kimura F. and Ito F. 1996. Cerambycid beetles (Coleoptera) lured to chemicals in forests of Nara Prefecture, central Japan. Annals of the Entomological Society of America 89: 835–842. Shidei T. 1974. Forest vegetation zones. In: Numata M. (ed.), The Flora and Vegetation of Japan. Kodansha, Tokyo, and Elsevier Scientific Publishing Company, Amsterdam, the Netherlands, pp. 87–124. Siitonen J., Martikainen P., Punttila P. and Rauh J. 2000. Coarse woody debris and stand characteristics in mature managed and old-growth boreal mesic forests in southern Finland. Forest Ecology and Management 128: 211–225. Spetich M.A., Shifley S.R. and Parker G.R. 1999. Regional distribution and dynamics of coarse woody debris in midwestern old-growth forests. Forest Science 45: 302–313. StatSoft Inc. 1998. STATISTICA for Windows. Release 5.1. StatSoft Inc., Tulsa, Oklahoma. Thunes K.H., Midtgaard F. and Gjerde I. 2000. Diversity of coleoptera of the bracket fungus Fomitopsis pinicola in a Norwegian spruce forest. Biodiversity and Conservation 9: 833–852. Warren M.S. and Key R.S. 1991. Woodlands: past, present and potential for insects. In: Collins N.M. and Thomas J.A. (eds), The Conservation of Insects and Their Habitats. Academic Press, London, pp. 155–211. Varley G.C., Gradwell G.R. and Hassell M.P. 1973. Insect Population Ecology: An Analytical Approach. Blackwell Scientific Publications, Oxford, UK.