Evaluation of secondary forests as alternative habitats to primary forests for flower-visiting insects
- First Online:
- Cite this article as:
- Taki, H., Makihara, H., Matsumura, T. et al. J Insect Conserv (2013) 17: 549. doi:10.1007/s10841-012-9539-3
- 472 Downloads
Although primary forests are important for biological conservation, the value of secondary forests for forest-dependent organisms needs to be evaluated when habitat restoration is required. We examined whether flower-visiting insects can use secondary forests as alternative habitats to primary forests. In particular, we compared assemblages of bees (Anthophila) and flower longhorn beetles (Lepturinae: Cerambycidae) in young secondary, mature secondary, and primary forests. Our results showed that more bee species were captured in primary and mature secondary forests than in young secondary forests, and flower longhorn beetle species were captured more frequently in primary forests than in mature and young secondary forests. Ordination showed that the communities in the three forest types were not statistically identical, which indicated that secondary forests cannot provide an absolute alternative habitat to primary forests for bees and flower longhorn beetles. However, the results also suggest that as secondary forests mature, more primary forest species would be able to use secondary forests as habitats. This implies that restoration from other land uses, such as monoculture plantations, to secondary forests could help to promote the faunal biodiversity of primary forests.
KeywordsApoidea Cerambycidae Pollinator Hymenoptera Coleoptera Lepturinae
Increasing demands for timber have primarily been met by growing more timber in plantations (Hartley 2002), and more of the world’s commercial timber is produced in plantations than in natural and semi-natural forests (Sedjo and Botkin 1997). Significant transitions from deforestation to reforestation have occurred in many economically developed and developing countries. These processes lead to the replacement of natural primary forests with plantation forests, which creates challenges for forest management based on biodiversity conservation and restoration (Brockerhoff et al. 2008).
One candidate forest management technique in plantation-dominated landscapes involves natural regeneration and retention of original native trees (Hartley 2002; Lindenmayer and Hobbs 2004). However, primary forests may have unique value, even when compared with secondary forests, for various taxonomic groups (Barlow et al. 2007). In the case of Japan, where 68 % of the land is covered by forests and 42 % of the forests are plantations that mainly contain coniferous species (FAO 2007), public demand for restoration from planted coniferous species to native broadleaved species has been increasing (MAFF 2007). This trend in public demand may also occur in other developed and developing countries requiring afforestation. Consequently, restoration plans that transform monoculture plantations into naturally regenerated forests will be promoted if forest-dependent organisms can use secondary forests as alternative habitats to primary forests.
Bees and longhorn beetles mediate important ecosystem functions within forests by pollinating herbaceous and woody plants (Kevan and Baker 1983) and promoting decomposition processes of microorganisms on plants (Grove 2002). Community structure, including abundance and species richness, in both groups is sensitive to forest management practices (Maleque et al. 2006; Taki et al. 2010a). For instance, a previous study in deciduous forests indicated that the abundance and species richness of bees were positively correlated to the amount of surrounding forest cover (Taki et al. 2007). Other studies on longhorn beetles have shown that forest age and type affect community structure, including abundance and species richness (Maeto et al. 2002; Makino et al. 2007). Therefore, because these insect groups can demonstrate sensitive responses and provide early warning of environmental changes (Kremen et al. 1993), bees and longhorn beetles are excellent candidates for investigating the potential of secondary forests to serve as alternative habitats to primary forests for forest organisms. However, these two flower-visiting groups require different resources during their larval stages. Both adult and larval stages of all collected bee species depend on flower resources (Michener 2007). On the other hand, larval longhorn beetles feed on a variety of tree parts (wood, sapwood or inner bark of trunks, branches, and roots) under diverse conditions, such as decaying, dying, and living (Hanks 1999).
The main purpose of the present study was to compare bee and flower longhorn beetle assemblages between secondary and primary forests. In particular, we tested whether young or mature secondary forests could serve as alternative habitats to primary forests for bees and flower longhorn beetles. Also, the responses of bees and flower longhorn beetles were expected to be different because their larval stages require different resources. To examine these questions, we sampled bees and longhorn beetles using traps placed in young and mature secondary and primary forests in a forest-dominated landscape.
Materials and methods
Our study was conducted in Tadami, Fukushima, Japan (approximately 37°25′–37°40′N, 139°10′–139°25′E). The site was located at the southern edge of the Echigo Mountain Range, in a mountainous region where over 90 % of the land cover is dense forests (Tadami-machi Board of Education 2004). The region contains plantations and secondary forests as well as preserved areas of primary forests, represented by Siebold beech (Fagus crenata).
Study sites in Tadami, Fukushima, Japan
Area (m2) of forest types within 100 m radius
Mean GBH (±SE, range) (cm)
Major tree species
No. of trees
25,980 (83 %)
55.1 (±0.49, 15.3–394.4)
Chengiopanax sciadophylloides, Acer japonicum, Fagus crenata
28,522 (91 %)
71.7 (±0.65, 15.5–275.6)
Fagus crenata, Acer japonicum, Magnolia hypoleuca
29,431 (94 %)
71.0 (±0.66, 15.8–299.2)
Fagus crenata, Chengiopanax sciadophylloides, Acer nipponicum
27,357 (87 %)
56.8 (±0.53, 20.8–140.0)
Quercus serrata, Aria alnifolia, Acer pictum
26,367 (84 %)
41.8 (±0.38, 15.3–123.1)
Quercus serrata, Aria alnifolia, Chengiopanax sciadophylloides
29,262 (93 %)
69.3 (±0.53, 17.0–110.3)
Quercus serrata, Quercus crispula, Pinus densiflora
22,825 (73 %)
21.1 (±0.31, 15.0–44.9)
Padus grayana, Cerasus leveilleana, Tilia japonica
29,334 (93 %)
23.5 (±0.36, 15.4–53.3)
Quercus crispula, Castanea crenata, Fagus crenata
29,836 (95 %)
20.1 (±0.36, 15.0–41.1)
Fagus crenata, Magnolia hypoleuca, Padus grayana
We used collision traps that consisted of a roof, two intersecting collision plates, and a bucket (Sankei Chemical Co., Ltd., Kagoshima, Japan) (Maeto et al. 1995; Yamaura et al. 2011). We used benzyl acetate (Akane-call BA: Sankei Chemical Co., Ltd.) as an attractant, which is one of the main components of floral fragrances and lures various longhorn beetles (Ikeda et al. 1993; Shibata et al. 1996; Sakakibara et al. 1997). Dispensers of the chemical were placed under the trap roof. Propylene glycol was poured into the bucket to preserve the specimens. The traps were set at about 1.5 m above the ground. In each of the nine forest sites, four traps with benzyl acetate as an attractant, two yellow and two white, were set. Although no reports have shown that benzyl acetate is an attractant for bees, pan trapping with yellow and white bowls or buckets is a standardized and commonly used method for collecting bees (Leong and Thorp 1999; Cane et al. 2000). The four traps at each dense forest site were set at the corners of a rectangle 20 m long on each side. We set the traps between May 12 and August 29 in 2008. During this period, we collected insects within the collision traps three times, once per month in June, July, and August.
To evaluate the effects of forest type at the forest stand level, we collected bees (Colletidae, Andrenidae, Halictidae, Megachilidae, and Apidae) and flower longhorn beetles (Lepturinae: Cerambycidae). Among the bee species, we excluded eusocial bees, bumblebees (Bombus spp.) and honeybees (Apis spp.), although we captured 380 bumblebee and 4 honeybee individuals. Their known foraging ranges (Dyer and Seeley 1991; Westphal et al. 2006) were too wide for evaluating effects at the forest stand level. For flower longhorn beetles, we used all of the captured species of Lepturinae in our analyses. A previous study reported that the mean influential spatial scale for 27 species of longhorn beetles was 342 m (Holland et al. 2005).
We analyzed total numbers of collected individuals and species for bees and flower longhorn beetles among primary, mature secondary, and young secondary forests using generalized linear models (GLMs) with a Poisson error distribution; the analyses were conducted using R version 2.10.0 (R Development Core Team 2009). A model with three classes of the explanatory variable (primary, mature secondary, and young secondary forests) was used. Also, models in which one of the three classes was combined with a null model were created to compare bee individuals, bee species, flower longhorn individuals, and flower longhorn species. Instead of Akaike’s information criterion (AIC), we used the second-order AIC (AICc), which is appropriate for small-sized samples (Burnham and Anderson 2002), to compare all of the models. AICc values, for which lower values indicate better model fit, were calculated using the MuMIn package in R (Barton Barton 2010).
Non-metric multidimensional scaling (NMS) on a Bray-Curtis dissimilarity index was used for the ordination of both bee and flower longhorn beetle communities (Mather 1976; Kruskal 1964). Autopilot mode in PC-ORD ver. 4 (McCune and Mefford 2006) was used in the NMS to avoid arbitrary decisions. The multi-response permutation procedure (MRPP) on the Bray-Curtis dissimilarity index was used to test the null hypothesis that there were no differences among the three forest types (primary, mature secondary, and young secondary forests) in the bee and flower longhorn beetle communities that were identified by NMS (Mielke 1984). Indicator species analysis (Dufrêne and Legendre 1997) was used to identify representative bee and flower longhorn beetle species for the forest types that were identified by NMS. This analysis produced indicator values for each species in each forest type, and statistically significant differences in these values were subsequently identified using a Monte Carlo technique. MRPP and indicator analysis were also performed using PC-ORD ver. 4 (McCune and Mefford 2006).
Comparison among models. P, M, and Y represent primary, mature secondary, and young secondary forests, respectively
Flower longhorn beetle
(P + M + Y)
(P), (M + Y)
(P + M), (Y)
(P + Y), (M)
(P), (M), (Y)
The number of sampled bees was relatively low, and the bees showed the high proportion of singletons and doubletons in the samples (14 out of 32 species). However, the findings for bees suggest that mature secondary forests and primary forests provide habitats for abundant unique bee assemblages compared with young secondary forests. No indicator species were identified for each of the forest types. In fact, some bee species do not inhabit forests in temperate regions and many of the collected Andrena and Lasioglossum species nest in the ground. One individual of Megachile sculpturalis, which nests in trees and twigs and uses resin as a nest resource, was sampled in primary forests. However, this species, as large as most bumble bee workers, may have too wide foraging range to discuss the effects at the forest stand level (Greenleaf et al. 2007). Although Taki et al. (2007) found a positive relationship between forest cover in a landscape and the abundance and species richness of spring-occurring bees in deciduous forests in temperate regions of North America, Winfree et al. (2007) showed that bee abundance and species richness, for collections in spring and summer, in forest habitats decreased with increasing forest cover. Also, a study on cavity-nesting bees in Japan found that relatively young forests housed large numbers of species that preferred early forest successional stages (Makino et al. 2006). Bees may be largely present in forest gaps and edges, such as meadows, rocky barrens and large tree falls, and populations of many bee species may be enhanced by small disturbances. Our samplings made from May 12, when the canopy trees started leafing out. It would be interesting to examine the same effects on bees during pre-leaf period in the deciduous forests, where understory herbaceous flowering occurs, for the future study.
The pattern in the number of species of flower longhorn beetles clearly showed the importance of primary forests as habitats required to maintain species richness and uniqueness compared with mature and young secondary forests. Two species, P. grallatrix and P. misella, were identified as indicator species of primary forests. Although many of the traits of these flower longhorn beetle species are still uncertain, their larval stages feed on a variety of tree parts (wood, sapwood or inner bark of trunks, branches, and roots) under diverse conditions (i.e., decaying, dying, and living trees) (Hanks 1999). In contrast, both the adult and larval stages of all of the collected bee species depend on flower resources. However, the larvae of many species of flower longhorns (subfamily Lepturinae), especially those in Pidonia, to which the majority of collected flower longhorn species belonged, are found under the thick bark of large trees and the branches of broadleaved tree species (Kuboki 1981). In addition, solid, stable, and moist humus is required by larvae and pupae of flower longhorn species (Kuboki 1987). These ecological characteristics of flower longhorn species might have influenced our results. Similar trends for flower longhorns were observed in a study that focused on different forest management practices and wider groups of longhorn beetles (Maeto et al. 2002).
The ordination results for bees and flower longhorn beetles indicated that secondary forests cannot provide complete alternative habitats to primary forests. Although the mean girth at breast height for trees was 65.9 cm in primary forests and 60.0 cm in mature secondary forests, the maximum ranges were rather different between the two forest types (Table 1). Additionally, dead trees should be more abundant in primary forests. Large dead trees would contribute to various faunal assemblages, such as flower longhorn beetles. Nevertheless, our results also suggest that as secondary forests mature, more primary forest species will be able to use secondary forests as habitats, even if the dominant broadleaved tree species are different. Temporal changes in plant communities can affect forest faunal assemblages. Previous studies have shown that in Japanese temperate regions, the ages of both natural and planted forests influences arthropod assemblages (Inoue 2003; Maleque et al. 2010).
The results suggest that as secondary forests mature, more primary forest species would be able to use secondary forests as habitats. Demonstrating that native bee and longhorn beetle species would recolonize restored forest habitats can help support efforts to restore existing evergreen plantation forests to native deciduous forests. In Japan, 42 % of forests are monoculture conifer plantations (Yamaura et al. 2012). There is public demand for restoration from planted coniferous species to native broadleaved species. Studies on longhorn beetles have shown that conifer plantations and secondary forests have significantly different community structures, differing in abundance and species richness (Maeto et al. 2002; Makino et al. 2007). Similarly, Taki et al. (2010b), sampled moth assemblages in forests of different ages in both secondary broadleaved forests and conifer plantations and found that unique moth communities existed among forest types, even in similar age classes. These results suggest that faunal biodiversity in primary forests would be better aided by a policy of restoring monoculture conifer plantations to naturally regenerated secondary forests than by leaving conifer plantations as they are.
We thank T Kitajima for caring for insect samples. We also thank K Maeto, N Tanaka, K Sugimura and H Masuya for assistance during field sampling of insects, and S Sugiura and Y Yamaura for comments on an earlier version of the manuscript. This study was supported by research funds of Forestry and Forest Products Research Institute and the Global Environment Research Funds (E-0801 and S-9) of the Ministry of the Environment, Japan.