Fauna of nocturnal moth species collected in a semi-natural grassland at Kanpu-zan in northern Japan

Abstract Semi-natural grasslands, which house species-rich ecosystems, have rapidly declined since the twentieth century due to land-use practices, such as agricultural intensification and abandonment. Owing to their diversity and known habitat associations, nocturnal moths are considered as one of the most suitable organisms to be studied for assessing the dynamics of species composition as a result of changes in landscape management of semi-natural grasslands. The present study provides the foremost description of nocturnal moth fauna of the semi-natural grassland at Kanpu-zan, northern Japan. Moth population data from 1987 were compared to the data collected in 2018 to evaluate the impact of decline in grasslands on species-richness. During the field sampling in 2018, a total of 226 nocturnal moth species were detected, which was nearly two-thirds of the number of species recorded in 1987, i.e. 396 species. The values obtained in 2018 were found to be nearly constant for different sites. For both periods, it was evident that moth fauna in Kanpu-zan mainly consisted of species that relied on woody plants. Amongst the species which were only recorded in 1987, 107 species were generalists that fed on plants that are commonly distributed in Kanpu-zan. No moth species were recorded that depended upon endangered or extinct plant food sources. Thus, it is unlikely that the decline in the number of moth species in Kanpu-zan was due to the loss in plant food sources. Our results suggest that environmental factors other than food plants may have caused decline and changes in nocturnal moth fauna. More studies on various organism fauna are needed for understanding the conservation of semi-natural grassland, considering that the loss of semi-natural grasslands is one of the major threats to biodiversity.


Introduction
Semi-natural grasslands house species-rich ecosystems including regional metacommunities by specialists of open habitats (Emanuelsson 2008, Pykälä et al. 2005, Alison et al. 2017. The persistence of semi-natural grassland depends on anthropogenic activities, such as mowing, grazing and burning (Suka et al. 2012, Dengler et al. 2014. Recent studies have shown that the grasslands may have persisted during the Holocene in a natural state due to natural disturbances and severe environmental conditions (Nakahama et al. 2018, Ohwaki 2018. However, these seminatural grasslands have shown a continuous and rapid decline since the twentieth century due to changes in land-use practices such as agricultural intensification and abandonment (Donald et al. 2001, Eriksson et al. 2002, Koyanagi and Furukawa 2013. As a result, many grassland species, including insects feeding on grassland plants and those inhabiting grassland habitat, are presently endangered due to the scarcity of natural grasslands in these regions (Maes andVan Dyck 2001, Habel et al. 2013). There is an urgent need to understand the changes in grassland fauna due to human activities.
Nocturnal moths are an ecologically important group of insects that play a key role in herbivory and pollination. They act as a food source for birds as well as potential indicators of ecosystem change across a wide variety of landscapes (Kitching et al. 2000, Summerville and Crist 2004, Macgregor et al. 2016. Declines in moth populations have been linked to habitat loss and agricultural intensification (Wenzel et al. 2006, Merckx et al. 2012). Owing to their diversity and known habitat associations, nocturnal moths are considered as one of the most suitable study taxa for assessing species richness against the changes in landscape management of semi-natural grasslands (Erhardt andThomas 1991, Wenzel et al. 2006;Alison et al. 2017).
Kanpu-zan, which is located in northern Japan, consists of semi-natural grasslands partly maintained by mowing and burning and accommodates many endangered grassland species (Akita Prefecture 2016). The region was a rich source of grassland resources as a livestock feed for farmers until the 1960s. However, the utilisation of the grassland, such as the grass harvest, gradually declined leading to succession by dense shrubs and trees from the 1970s as lifestyles changed. A study indicates that the grassland area, which was 319 ha in 1975, had decreased to 138 ha by 2014 (Masui et al. 2017). The prevalent plant community varied depending on the management frequency and plant richness was lower in abandoned sites with reduced or diminished grassland species (Masui et al. 2017). Documentation of moth species in the region was conducted in 1987 before the grassland had severely declined (Takahashi 1993). However, the present status of moth fauna is largely unknown. This study presents the first account of nocturnal moths in Kanpu-zan after the decline in grasslands and presents a model for understanding the changes in moth fauna related to altered vegetation in grassland areas caused by changes in human activity. Uchida and Ushimaru (2014) compared plant richness and butterfly and orthopteran richness and diversity amongst three different land use types in semi-natural grasslands: abandoned, traditional and intensified terraces and suggested that the number of individuals of most herbivorous species decreased randomly with respect to life-history traits, following a decline in plant richness reflecting disturbance frequency. Due to the present plant flora being revealed in Kanpu-zan, this site provides a promising opportunity to understand the relationship between moth fauna and the land use of the grassland vegetation.

Study site
Surveys of nocturnal moths were carried out at Kanpu-zan during May to October in 2018, twice a month (12 nights in total), on clear nights from sunset to midnight. Three sites that differ in vegetation were selected: 1) grassland maintained by mowing once or twice a year (grassland site), 2) site where grassland management had been abandoned leading to overgrowth of shrubs and trees (forest site) and 3) mixed vegetation with both grassland and woody plants (mixed site) (Fig. 1). In 1987, the moths were collected during May to September, once or twice a month (6 nights in total) at one site in a grassland dominated by tall grass such as Miscanthus sinensis, where shrubs and hedgerow trees were interspersed (Takahashi 1993).

Sampling methods
Moths were collected using UV light traps, equipped with a 20 W black (ultraviolet) light fluorescent tube. One light trap was set at each site. The same method was used in 1987 (Takahashi 1993). All collected specimens were dried at room temperature and mounted for morphological examination. All moths were identified to species level by M. Kamikura and K. Umetsu using descriptions and photos in the book "Standard of Moths in Japan" , Kishida 2011a, Kishida 2011b. Some of the family names (i.e. Arctiidae, Lymantriidae, Micronoctuidae) were updated to current taxonomy. All the specimens were preserved as a personal collection of M. Kamikura.

Data analyses
Owing to the annual variations in abundance of moth species (Highland et al. 2013), data analyses were based on presence-absence data of moths sampled at each site in each month (Table 1,Suppl. material 1). Species composition was compared between years and amongst sites. In addition, because micro-moths belonging to the families of Chrysopeleiidae, Cosmopterigidae, Oecophoridae and Depressariidae were not identified in the study in 1987 (Takahashi 1993), those specimens were excluded from the data in 2018, as well as from the following analysis. In order to test whether the community composition of moth species differed between 1987 and 2018, permutational multivariate analyses of variance (PERMANOVA) were performed. We included the year of the survey as an explanatory variable. Additionally, PERMANOVA was also performed to test whether the community composition of moth species differed amongst the three sites in 2018. Here each site was included as an explanatory variable. For these analyses, the ADONIS function in library VEGAN (Oksanen et al. 2016) in R. ver. 3.3.2 (R Development Core Team 2016) was used. Non-metric multidimensional scaling analysis (NMDS) on the Jaccard index was used to visually summarise the plant community compositions of each site at each month.   In order to explore the food habits of moth species at each site in both years, the moth species were classified into six feeding groups: 1) species that primarily feed on herbs, 2) species that primarily feed on shrubs and/or trees, 3) species that feed on both herbs and trees, 4) species that primarily feed on lichens, 5) species that primarily feed on dead leaves and 6) species whose food habit is unknown, referring to the guide "The Standard of Moths in Japan I-IV" (Kishida 2011a, Kishida 2011b. The collected moths were also classified into two groups based on the range of host plants: 1) specialist, defined as species that feed on only one plant family and 2) generalists, defined as species that feed on more than two plant families. We calculated the proportions of each feeding groups for each site in both years.

Results and Discussion
The number of taxa of moths recorded in 2018 at each site (excluding the 4 families and 10 species of micro-moths) was 14 families, 79 genera and 95 species for the grassland site; 16 families, 115 genera and 137 species for the forest site and 14 families, 103 genera and 123 species for the mixed site (Fig. 2). In total, 17 families, 181 genera and 226 species were recorded in 2018. This was nearly two-thirds of the species recorded in 1987, which comprised of 16 families, 291 genera and 396 species. Amongst all the moth species documented, only 118 species were observed in both periods. The 278 species that were recorded in 1987 were not recorded in 2018, while 115 species were newly recorded in 2018.

Species composition of moths between years and amongst sites
The results of the PERMANOVA showed that the species composition of the moths significantly differed between years ( Fig. 3a; R = 0.06, P = 0.02). This suggests that species composition at the three sites in 2018 differed from that in 1987. On the other hand, when the data of 1987 was excluded from the analysis, it did not differ amongst sites in 2018 ( Fig. 3b; R = 0.08, P = 0.99). This may be because the three sites in 2018 were located in a geographically adjacent area (an average of 400 m between sites). Studies that detected differences in moth species composition amongst different vegetation and management were sites that were geographically separated by more than 1 km (Öckinger andSmith 2006, Pöyry et al. 2009). As some moth species are known to show high mobility (i.e. a flight range larger than 500 m in radius) (van der Meulen and Groenendijk 2005, Merckx and Macdonald 2015), species composition of moths may not differ in a small spatial scale. Furthermore, because the present data included only one replicate for each type of vegetation, more replicates from multiple sites would be useful in confirming 2 2 Figure 2.
Number of moths collected in each site in both years in each month.
Fauna of nocturnal moth species collected in a semi-natural grassland at ... that the difference in vegetation reflecting the management did not affect the moth species composition in the semi-natural grassland in Kanpu-zan.

Food habits
The number of species that may rely on woody plants comprised more than half of the total number of species at all sites in both years (Fig. 4a). This includes species that feed only on woody plants; species that feed on both herbs and woody plants; species that feed on lichens and mosses; and species that feed on dead leaves. These species were known to consume woody plants that belong to the families of Fagaceae, Rosaceae, Fabaceae and Caprifoliaceae. Trees and shrubs that belong to Fagaceae and Rosaceae were dominant in the forest site and Fabaceae shrubs and vines were scattered across the grassland site, while the mixed site had a mixture of vegetation of both sites. Moreover, trees and shrubs such as Cerasus jamasakura H.Ohba and Quercus serrata Murray were present in grasslands in 1987 (Takahashi 1993, K. Umetsu, personal observation). These suggest that, in both years, moth fauna in Kanpu-zan was mainly comprised of species that may feed on woody plants.  For the 118 species that were recorded in both in 1987 and 2018, the composition rate of species that may use both herbs and woody plants was 19%, which was higher compared to that for the species that were only recorded in either 1987 or 2018, at 10.1% and 11.2%, respectively (Fig. 4b). This may be because the species that can feed on both herbs and woody species are adapted to habitats with various vegetations and were tolerant to the environmental change between 1987 and 2018 caused by decline in grasslands. Species that primarily feed on herbs such as Deilephila askoldensis, which is known to inhabit volcanic grasslands (Yano 2011), consisted of 15-20% in all three sites in 2018. The seminatural grasslands in Kanpu-zan may be a valuable habitat for the larvae of these moths. On the other hand, the composition rate of species that feed on lichens was relatively low in all sites and especially low at the grassland site in 2018 (Fig. 4a). This is consistent with the result in 1987, which identified that it is one of the characteristics in grasslands that lichen feeding moths are scarce (Takahashi 1993).
The composition rate of specialist and generalist moths was stable across sites including the year 1987. Generalists consisted of 43.8±1.5% (mean±SE) of the total number of moth species in both years (Fig. 5). Out of the 280 species of moths that were only recorded in 1987, 107 species were generalists that feed on plants that are commonly distributed in Kanpu-zan (Masui et al. 2017).

Conclusion
The present study suggests the species composition changed between 1987 and 2018 and the number of moth species largely declined in the last 30 years. Wenzel et al. (2006) also showed that the number of butterflies and burnet moths had declined between 1972 and 2001 in calcareous grassland remnants in south-western Germany. They suggested that species which require structured habitats, species with low mobility, species which require more than 16 ha of habitat and specialist feeders were especially affected by the decline. As we did not find any moths that depend on endangered plants or whose food plant had become extinct, it is unlikely that the decline in the moths in Kanpu-zan was due to the loss in food plants. Other factors such as plant height, flowering species richness and habitat connectivity (Pöyry et al. 2009) may have caused the decline. Continuing landscape management such as mowing and burning may be important, not only for endangered species, but also for common moth species and to maintain the nocturnal moth community in semi-natural grasslands.
A previous study in Japan covering 31 agricultural areas, including semi-natural grasslands, demonstrated that declines in herbivorous insects in both abandoned and intensified use of agricultural landscapes are also explained by multiple factors (Uchida and Ushimaru 2014). It has been reported that there are large differences with respect to the preferred management intensity amongst species groups in the grassland habitats (Pöyry et al. 2006). More studies on other organism fauna are needed for understanding the conservation of semi-natural grassland, considering that the loss of semi-natural grasslands is one of the major threats to biodiversity.