Moths comprise one of the most diverse insect groups in forest ecosystems. About 140,000 species have been identified throughout the world (New 2004) and more than 5000 species are listed in Japan (Sugi 2000). Such high diversity of moths may be maintained by plant diversity (e.g. Neuvonen & Niemelä 1981) and plant architecture (Lawton 1983). In addition to this impressive diversity, moths are easily collected with light traps that are widely recognized as the standard tool for sampling night-flying moths (Southwood & Henderson 2000).

In forest ecosystems, moths are mostly herbivores in their larval stage and are thus particularly sensitive to environmental changes that affect plant quality and quantity as their diets. On the other hand, moths are an important food resource for other animals and their abundance affects the population dynamics of animals in higher trophic levels. Because of their pivotal role in maintaining biodiversity in forest ecosystems, moths are generally regarded as useful indicator taxa for monitoring insect biodiversity and habitat disturbance caused naturally or artificially in tropical and temperate forests (Hammond & Miller 1998; Usher & Keiller 1998; Kitching et al. 2000; Summerville et al. 2004).

Seasonal patterns of larval moth assemblages have been investigated in relation to foliage quality, weather conditions and natural enemies (Feeny 1970; Niemelä & Haukioja 1982; Yoshida 1985; Butler & Strazanac 2000; Summerville et al. 2003; Murakami et al. 2005). However, only a few studies have focused on seasonal patterns of adult moth assemblages (Yoshida 1980; Yela & Herrera 1993; Butler et al. 2001). For example, Yoshida (1980) investigated seasonal fluctuations of species richness, abundance and diversity index of adult moth communities at four forest stands in Hokkaido, northern Japan. Yela & Herrera (1993) studied seasonal patterns of species richness and abundance of noctuid moths in Mediterranean mixed forests. However, few studies have investigated seasonal occurrence in each species except for pest species.

In temperate regions, moth assemblages consist of seasonal progressions of the occurrences of different species. The timing and duration of flight periods of adult moths depend on the various life history traits of each species, such as overwintering stage, voltinism (the number of generations per year), adult life span, and immature growth rate (Wolda 1988). These traits may be further influenced by weather conditions (e.g. temperature, precipitation and day-length) and availability of host-plants (e.g. quantity and quality). For example, Hunter & McNeil (1997) revealed effects of host-plant quality on diapause induction and subsequent voltinism in a tortricid moth.

In the present study, we investigated seasonal trends of night-flying moth assemblages in a cool-temperate region to contribute information about the regional moth diversity of Hokkaido. For this purpose, we sampled adult moths using portable light traps because light traps are useful tools to quantify the moth communities (Southwood & Henderson 2000). Using data collected from these traps, we estimated flight duration and voltinism of major species.

Materials and Methods

The study was conducted from April to December 2005, which covers almost the entire season of adult moth flight, in five stands of three different forest types located in Sapporo, central Hokkaido, northern Japan (42°53′–43°00′N, 141°22–26′E; 150–400 m a.s.l.). Study stands, which were always more than 10 ha in size and located more than 1 km apart, consisted of two deciduous broadleaved forests at Hitsujigaoka and Mizunenosawa, two larch (Larix kaempferi Carr.) plantations at Shimomitaki (42 years old) and Mt. Yagyu (47 years old), and a Todo-fir (Abies sachalinensis Masters) plantation at Mt. Yagyu (48 years old). We used these forest types because these are the most common forest types in the cool-temperate region in Hokkaido. The deciduous broadleaved forests were dominated by linden (Tilia japonica Simonkai), cucumber tree (Magnolia obovata Thunb.), oak (Quercus crispula Blume), and maple (Acer mono Maxim.). Forest floors in the study stands were more or less covered with bamboo grasses (Sasa kurilensis (Rupr.) Makino et shibata).

We used portable light traps developed by Okochi (2002). This trap was equipped with a 6 W black (ultraviolet) light fluorescent tube powered by a 9 V alkaline battery made up of six 1.5 V cells. The light lasted about 12 h per night. Every month we set two light traps in each stand on a night around the new moon (nine nights in total). Trapping nights were chosen to avoid rain and snow. The two traps were hung at a height of approximately 1.5 m at fixed locations, which were located 100 m apart in the central part of each stand. Early the next morning, moths caught in the traps were killed with ethyl acetate, preserved in a refrigerator, and at a later date identified to species. We used scientific names of moths in Inoue et al. (1982) and changes after Inoue et al. (1982) followed Sugi (2000), Sugi & Jinbo (2004) and Jinbo (2004–2008). In the following analysis, individuals identified to species (67–85% of sampling individuals in each stand) were used. Unidentified individuals were those with either extensive wing wear or from microlepidopteran moths (e.g. Tortricidae). All voucher specimens in this study were deposited in the collection of Hokkaido Research Center, Forestry and Forest Products Research Institute.

For seasonal trends of moth assemblages, the number of species, the number of individuals, and the Pielou's index (J') for evenness were calculated for each trap and expressed as the averages of the two traps in each stand. Pielou's index (J') was calculated as follows:

where p_i is the proportion of individuals found in the ith species and S is the total number of species. Similarity indices (Cλ) between samples from successive months in each trap were also calculated and averaged in each stand to quantify the seasonal changes in species composition. A similarity index (Cλ) was calculated as follows:

where n_1i and n_2i are the numbers of individuals in the ith species and N₁ and N₂ are the total numbers of individuals in samples 1 and 2, which are collected from successive months, respectively. S is the total number of species. The value of C_λ is 0 when two samples have no common species and is nearly 1 when two samples are identical (Morisita 1959).

The seasonal occurrence of each moth species was examined for the species in which a total of at least 10 individuals were collected. The mean duration of occurrence in each species was calculated from the number of months in which each species was collected. This value indicates an occurrence of moths based on the assumption that the flight period of a moth species was one month for a moth species that was sampled in one trapping date since we conducted a monthly sampling. Voltinism (univoltine or multivoltine) of these species was estimated based on the following criteria. We assigned a species to multivoltine if the species was (1) collected in discontinuous sampling months, (2) collected in more than two sampling months and had two peaks in their occurrence or (3) collected in more than three sampling months. The other species were assigned to univoltine. The estimated voltinisms were compared with the voltinisms reported for the same species in eastern Hokkaido (Iijima 1990). Iijima (1990) assigned 12 species to partial bivoltine (basically univoltine with a partial second generation only in favorable seasons), but these species were excluded from the present analysis.

Results

Seasonal changes in species diversity and species composition. A total of 14,591 adult moths, consisting of 693 species in 29 families, was collected and identified in the five study stands. No moths were collected in December. The number of species was greatest in Geometridae followed by Noctuidae (Table 1). Seasonal changes in the mean number of species in each stand were unimodal with a peak in July or August (Fig. 1). Seasonal changes in the mean number of individuals in each stand were also unimodal with a peak in August, except in the larch plantation at Mt. Yagyu where a peak occurred in July (Fig. 2). This peak was due to the high abundance of two dominant arctiids, Ghoria collitoides Butler and Eilema cribrata (Staudinger) in this month. On the other hand, mean evenness (J') showed a weak declining trend as the season progressed, except in the larch plantation at Shimomitaki (Fig. 3). In the larch plantation at Mt. Yagyu, the lowest mean evenness was observed in July when two dominant noctuids, Aventiola pusilla (Butler) and Mimachrostia fasciata Sugi occupied 11.3% and 8.5% of the abundance, respectively. In the larch plantation at Shimomitaki, the lowest mean evenness was observed in August when two dominant arctiids, Ghoria collitoides and Eilema cribrata comprised 36.5% and 14.3% of the total abundance of moths, respectively.

Table 1.

Numbers of species and individuals of forest moths collected in five stands in Sapporo in 2005.

Fig. 1.

Seasonal changes in the number of adult moth species in five stands in deciduous broadleaved forests (Hitsujigaoka, Mizunenosawa), larch plantations (Shimomitaki, Mt. Yagyu 1), and a Todo-fir plantation (Mt. Yagyu 2) in Sapporo, 2005. Values show the averages of two traps in each stand.

Fig. 2.

Seasonal changes in the number of adult moth individuals in five stands in deciduous broadleaved forests (Hitsujigaoka, Mizunenosawa), larch plantations (Shimomitaki, Mt. Yagyu 1), and a Todo-fir plantation (Mt. Yagyu 2) in Sapporo, 2005. Values show the averages of two traps in each stand.

Mean similarity indices (C_λ) between samples from successive months in each stand were always low (less than 0.2) (Fig. 4), although the indices fluctuated differently in each stand. The low similarity indices indicate that species composition changed greatly between successive months.

Fig. 3.

Seasonal changes in evenness (Pielou's J') of adult moth assemblages in five stands in deciduous broadleaved forests (Hitsujigaoka, Mizunenosawa), larch plantations (Shimomitaki, Mt. Yagyu 1), and a Todo-fir plantation (Mt. Yagyu 2) in Sapporo, 2005. Values show the averages of two traps in each stand.

Fig. 4.

Seasonal changes in the value of similarity index (C_λ) of adult moth assemblages between samples from successive months in five stands in deciduous broadleaved forests (Hitsujigaoka, Mizunenosawa), larch plantations (Shimomitaki, Mt. Yagyu 1), and a Todo-fir plantation (Mt. Yagyu 2) in Sapporo, 2005. Values show the averages of two traps in each stand.

Seasonal occurrence of each moth species. At least 10 individuals were collected from 248 species in five stands (Appendix 1). These species were listed in order of the increasing mean occurrence period, which was calculated from the average of sampling months weighted by the number of individuals collected in each month. Each species was collected within a short period. The mean duration of occurrence in each species was 1.8 months (range 1–4 months), and 90 species (36.3%) were collected in only one month. Among the 248 species, 14 species (5.6%) were collected only in spring (from April to June; Appendix 1). On the other hand, 25 species (10.0%) occurred only in autumn (from September to November; Appendix 1). Especially, some geometrid winter moths occurred only in November (Appendix 1).

Of the 248 species, 227 species (91.5%) were assigned to univoltine and only 21 species (8.5%) were assigned to multivoltine (Appendix 1). Eleven of the multivoltine species were geometrids. Voltinisms estimated for 156 species (90.2%) were similar to the voltinisms reported for the same species in Iijima (1990), in which 148 species were classified as univoltine and 25 species were classified as bivoltine. There were only a few differences in the two studies: 14 species estimated to be univoltine were bivoltine in Iijima (1990), and three species estimated to be multivoltine were univoltine in Iijima (1990).

Discussion

The present study shows that seasonal changes in the numbers of species and individuals of adult moths were markedly unimodal with a peak in summer (July or August) (Fig.1, 2). Yoshida (1980) also found that the numbers of species and individuals were greatest in July or August in a deciduous broadleaved natural forest and a larch plantation in the Tomakomai Experimental Forest of Hokkaido University, which is located about 60 km from Sapporo. However, Yoshida (1980) showed multimodal seasonal patterns with peaks in summer and autumn in the number of individuals in a deciduous broadleaved secondary forest and a Todo-fir plantation. The peak abundance in autumn is probably caused by some dominant species that were extremely abundant that autumn (Yoshida 1980). Hirao et al. (2006) collected adult moths in a deciduous broadleaved forest at the same locality as Yoshida (1980) and reported that July possessed the highest number of species and individuals. These studies suggest that a unimodal seasonal pattern with a peak in summer (July or August) is a general trend for species richness and abundance in adult moth assemblages in the cool-temperate region of central Hokkaido. This information is critical to determine efficient sampling periods in a future study and provides a useful basis for comparative studies with other regions.

In contrast to the unimodal seasonal pattern reported here, seasonal changes in the number of species of adult moths in a mixed deciduous forest in southern Korea were bimodal with peaks in June and August (Choi 2008). In mixed hardwood forests in North America, no clear peak was detected in the number of species, although the number of individuals was highest between March and June (Landau et al. 1999) or in July and early August (Butler et al. 2001). In Mediterranean mixed forests, abundance and diversity of adult noctuid moths were markedly bimodal with two distinct peaks, in early summer (mid-July) and in early autumn (late September to early October) (Yela & Herrera 1993). These studies demonstrate that seasonal trends of adult moth assemblages vary widely among regions. This difference may reflect different climate conditions among regions because climate variables are often important factors influencing moth abundance and diversity (Yela & Herrera 1993).

Seasonal changes of larval moth communities in forests have also been studied as an alternative measure of moth activity in forest ecosystems (Yoshida 1985; Yela & Herrera 1993; Butler & Strazanac 2000). Yoshida (1985) showed that the number of species and individuals of macrolepidopterous larvae on oak trees had two peaks, in June (spring) and August (summer). Butler & Strazanac (2000) sampled lepidopteran larvae at oak-dominated Appalachian forests from May to mid-August and showed that the numbers of species and individuals of larvae were higher in May and August. Yela & Herrera (1993) showed that the frequency of occurrence of noctuid larvae exhibited a distinct peak in the first half of June. Yela & Herrera (1993) also suggested that the duration of the pupal stage might affect the difference in the occurrence pattern between larvae and adults as a result of life history strategies adapted to the hot and dry summer season in Mediterranean habitats. This suggests that seasonal occurrence patterns of adults do not correspond to those of larvae.

The value of evenness (J') showed a decreasing trend as the season progressed except in the larch plantation at Shimomitaki (Fig. 3). In central Hokkaido, Yoshida (1980) also found that a relative diversity index, which is equivalent to evenness, decreased with season in deciduous broadleaved forests, a larch plantation, and a Todo-fir plantation, although the value of the index fluctuated considerably. These findings suggest either that dominant species occupy a relatively larger part of moth assemblages or that there are a large number of rare species in autumn.

Most (90.2%) of the voltinisms estimated in the present study were consistent with the voltinisms recorded in eastern Hokkaido (Appendix 1; Iijima 1990). In addition, more than 90% of the species were assigned to univoltine, suggesting that the univoltine life cycle is predominant for moths in Hokkaido. This may be caused by the relatively short growth period in the cool-temperate region in Hokkaido. For the species estimated to be univoltine in the present study but bivoltine by Iijima (1990), it is possible that we were simply unable to detect multivoltinism using our monthly sampling intervals. For the species estimated to be multivoltine in the present study but univoltine by Iijima (1990), these species may be univoltine only in eastern Hokkaido where temperatures are relatively low in comparison to central Hokkaido. Further studies are needed to clarify the voltinism of these species.

Low values of the similarity index (C_λ) between samples from successive months (Fig. 4) show that species composition changed greatly between successive months. In a deciduous broadleaved forest in central Hokkaido, Hirao et al. (2006) also found that species composition of adult moth communities was distinctly partitioned into each month. These high species turnovers were due to the short occurrence period (1.8 months on average in the present study) of each species. For moth species to have short occurrence periods, species are likely to have a univoltine life history, synchronized adult eclosion, and short life spans for adult moths. In cool-temperate regions, relatively shorter growing season may result in one generation per year. General life expectancy of adult moths is assumed to be from one to three weeks (Zborowski & Edwards 2007), although we could not find any field studies that examined life span of adult moths. Similar life history traits seem to be common in moths because many moths have well-defined and characteristic periods of seasonal activity and, particularly in temperate regions, may fly for only short periods (New 2004).