Nuri Nurlaila Setiawan · Endah Sulistyawati

Abstract Understanding the mechanisms and barriers to the restoration of degraded land, especially post agriculture,will help provide protocols on Effective ways of restoration into functional ecosystems. One of the barriers in early stages of forest restoration is the arrival and availability of propagules. Seed rain and factors aff ecting it (i.e., distance to the forest edge, species diversity and surrounding vegetation) were measured in a reforested post-agricultural f ield and in an adjacent secondary forest. Multivariate glm analysis was used on the seed rain community data and univariate lm analysis on the most abundant seed captured (i.e., Schima wallichii (DC.) Korth.). After 8 months of seed rain collection, there was a total of 3596 seeds from eight tree species.Seeds were more abundant and more diverse in the secondary forest (74.9%, 8 species) compared to the reforested f ield(24.1%, 2 species). There was a limitation on seed dispersal in reforested f ield from the adjacent forest. The abundance of S. wallichii seeds determined by the dominance of adult trees(136 trees/ha in the forest and 115 trees/ha in the reforested f ield). Our study suggests, that after 7 years of planting, the reforested f ield has received limited seed rain and has not yet recovered.

Keywords Seed rain · Montane tropical forest · Postagriculture succession · Forest edge · Multivariate analysis

Introduction

Climate change as a result of increased greenhouse gas emissions and deforestation has received global attention. Many countries have pledged to mitigate the impact of climate change by protecting forests and restoring degraded lands.As a country with the world’s third largest tropical forest,Indonesia has lost a signif icant amount of its primary forest(over 6.02 M ha in 2000-2012) with an increasing annual average loss of 47,600 ha (Margono et al. 2014). Without further action, this increasing loss will have a strong negative impact on biodiversity, climate regulation, and human well-being (Watson et al. 2018; Baccini et al. 2019; Brancalion et al. 2019). Forest restoration is considered as one of the most Effective strategies to mitigate the Effects of climate change (Bastin et al. 2019) and to ensure the availability of multiple ecosystem services (Chazdon and Brancalion 2019).

In the tropics, one of the main drivers of deforestation is land clearing for agriculture (Elgar et al. 2014). Due to government enforcement, these lands are often abandoned and left to regenerate themselves. In most cases, this passive strategy results in a slow recovery rate and arrested succession (Niering and Goodwin 1974). The disturbance caused by intensive agricultural activities may deplete the propagules and limit tree species recruitment (Arroyo-Rodriguez et al. 2015). Restoring the former agricultural land into forest poses a signif icant ecological challenge (Shoo et al.2016), since most tropical tree seeds are recalcitrant, i.e. do not survive drying or freezing (Kettle et al. 2011). Several studies mentioned that the lack of tree propagules (seeds or spores) dispersed to the degraded sites as one of the main barriers to forest restoration (Holl et al. 2000; Hooper et al.2005; Reid et al. 2015), especially because most rain forest species have extremely low seed dispersal (Clark et al. 1999;Terborgh et al. 2002). The arrival of tree propagules to an abandoned agricultural f ield becomes a focal point in succession, i.e., as a source of regeneration. Without this, the abandoned f ield may not be able to restore its late-successional vegetation cover. For this reason, one of the strategies to accelerate succession on abandoned f ields is by tree planting, preferably native species for their good adaptability to the local environment which will enhance seedlings survival and growth. Besides, introduced or exotic species for site restoration have the potential to become invasive (Norton 2009). Seedlings of native may mature to produce seeds and seedlings, changing the microclimate more favorable for tree propagules to grow. However, knowledge to propagate native tree species from the surrounding forest area is limited, mostly for its non-commercial value.

Understanding the seed rain community of tree species and its variation is important as an initial step towards forest recovery on abandoned areas (Bustamante-Sánchez and Armesto 2012). The seed rain community is a key factor in the dynamics of forests for providing ecosystem maintenance functions (Barbosa et al. 2012), i.e., as a blueprint ref lecting the possible future of the vegetation community(Levine and Murrell 2003). Several factors may aff ect the abundance, diversity, and community structure of seed rain arriving in an area, e.g., distance to the forest edge, season,timing, and surrounding vegetation (Souza et al. 2014).

We studied the seed rain on abandoned reforested f ield and the adjacent forest in Mount Papandayan Nature Reserve, West Java, Indonesia. Part of the secondary forest in this nature reserve had been cleared, converted into agricultural f ields and then abandoned. Through community and government initiatives, the abandoned f ields were planted with native speciesSchima wallichii(DC.) Korth andDistylium stellareKuntze and left to undergo succession. Through this study, we aimed: (1) to determine whether tree seeds from the adjacent forest (seed source)had reached a reforested f ield; (2) to investigate the parameters aff ecting seed arrival at a specif ic location; and, (3)to investigate the parameters aff ecting the abundance ofS. wallichii,the most dominant seed species found. We hypothesized that: (1) there will be more abundant and more diverse seed rain in the adjacent forest compared with the reforested f ield; and, (2) the distance to the forest will impact the arrival of seeds in reforested f ield; i.e.,the closer to the forest edge, the more diverse and higher vegetation cover will result in a more complex seed rain community. The results of our study will help identify actions needed to accelerate succession in reforested postagricultural f ields.

Materials and methods

Study site

The secondary forest site (Fig. 1 ) is located in the outer part of the Mount Papandayan Nature Reserve (MPNR),West Java, Indonesia, with elevations of 1200 m a.s.l. at 7° 17ʹ 53.15 S and 107°44ʹ 46.33 E. The climate is type B according to the Schmidt and Ferguson’s climate classif ication system, with average rainfall 2500-3000 mm/year, evaporation 76-85 mm/month, humidity 70-80%,and mean annual temperatures of 10-25 °C (BKSDA Jabar II 2003). In the 1990s, areas of the Reserve were encroached upon and some of the natural secondary forest converted into agricultural f ields. Following the Government’s reclaiming action in 2003, agricultural activity ended and some abandoned f ields reforested through community-initiatives and government programs. In the study area, two native tree species raised in local nurseries were planted,Schima wallichiiandDistylium stellare.Seven years later, annual grasses and tall shrubs, mostlyEupatorium inulifoliumKunth andEupatorium adenophorumSpreng., had colonized the area and trees from the initial planting were mostlyS. wallichii.

Experimental design

Seed rain was collected by installing 60 (50 × 50 cm wide)nylon-mesh traps (Appendix Fig. S1) over 8 months from the beginning of March until the end of November 2010(i.e., 40 weeks). Traps were placed along three, 200 m parallel transects, each consisting of 20 traps placed every 10 m extending from the centre of the forest into the centre of reforested f ield. Seeds were collected every 2 weeks and identif ied to the species level. The vegetation around each trap was analyzed by: (1) identifying the species and measuring diameter at breast height and height of trees inside a circular plot with a 5-m radius (Barbosa and Pizo 2006); and,(2) measuring total vegetation cover using a hemispherical photograph (Ishida 2004) and further analyzed withRGBf isheyesoftware. Seed and tree species identif ication were based on identif ication books (Backer and van den Brink 1968; van Steenis 2006), comparison with samples from previous studies (Setiawan and Sulistyawati 2008; Sulistyawati et al. 2012), and information from local informants.

Fig. 1 Study location in Mount Papandayan Nature Reserve,Indonesia

Data analysis

All analyses were done in R version 3.6.1 (R Core Team 2019); graphs were made with the R packageggplot2(Wickham 2009). One-way ANOVA was used to check whether there were signif icant diff erences between the monthly seed rain and the vegetation cover. Due to diff erences in site conditions between the forest and the reforested f ield, the analyses were done separately. The main aim was to describe the relationship between seed rain collected in the forest and in the reforested f ield with several related environmental parameters. This was done using the absolute abundance of diff erent seed species collected as a multivariate response variable. Since most of the seeds collected throughout the study period wereS. wallichii, an extra analysis was carried out using the abundance ofS. wallichiiseeds as a univariate response variable. Each response variable used in the analysis was the seed absolute abundance from each trap.

Based on the study design and vegetation analysis data,we gathered 14 explanatory variables (9 for multivariate analysis and 14 for univariate analysis; Table 1 for details)from four group variables (i.e., trap location, surrounding tree diversity, vegetation cover and S.wallichiitrees).These explanatory variables were used to explain the variations in seeds collected in each trap. Explanatory variables belonging to the same groups showed signs of collinearity(Dormann et al. 2012) with Spearman correlation values above 0.5 (Appendix Table S1 for more details). Therefore,one variable was selected from each group for each response variable prior to analysis (Castagneyrol et al. 2013; Setiawan et al. 2016). For each response variable, a series of univariate regression models was f itted using the diff erent explanatory variables with the lowest AIC (Akaike information criterion; Appendix Table S2 and S3).

Diff erent analyses were used for the two response variables, i.e., multivariateglm(manyglm) with negative binomial error distribution for seed rain abundances with the packagemvabund(Wang et al. 2012), and linear regressions for theS. wallichiiseed abundance. Themvabundpackage with a multivariatemanyglmstructure has higher statistical power over the typical distance-based analysis since it accounts for mean-variance relationships in multivariate datasets. For both analyses, a full model was f irst f itted with the three explanatory variables, i.e., seed trap location, surrounding tree diversity, and surrounding vegetation cover, and interaction between variables. For the second analysis, one more explanatory variable was added, the surroundingS. wallichiitrees (Table 1). The best-f it model with backward selection was then selected based on likelihood ratio tests (Zuur et al.2009). The residuals of all f inal models were inspected andan additional test of overdispersion for the f inal model of the second analysis was carried out by comparing the residual deviance with the residual degrees of freedom.

Table 1 The four explanatory variable groups used in the study

Results

Seed rain abundance variations

After 8 months, a total 3596 seeds were collected from eight tree species (Table 2). Based on the dispersal mode,the seeds were classif ied in three categories, i.e., anemochory or wind-dispersed (Engelhardia spicataLesch. ex Bl.,Lithocarpus dolichocarpaRehder, Schima wallichii; 3318 seeds), autuchory, dispersal without help from an external vector (Astronia spectabilisBl., Distylium stellare, Omalanthus populneus(Geiseler) Pax, Villebrunea rubescens(Bl.) Bl.; 187 seeds), and zoochory, dispersal by animals(Acronodia punctataBl.; 31 seeds). The size of the seeds was between 5-10 mm. Seed abundance was higher in the forest (74.9%) than in the reforested f ield (24.1%). Of the total eight seed species, all were found in the forest and only two in the reforested f ield. Among all seed species,S. wallichiidominated both sites with 89.7% of the total.

Seed captured each month ranged from 237 to 637(Fig. 2). Overall, the highest number of seeds captured were in the months of June (544 seeds) and August (637 seeds);the lowest number was in October (237 seeds). However, the average number of seeds captured in each trap in each month did not diff er signif icantly (one-way ANOVA,p= 0.172).Seed ofS. wallichiidominated the number of seeds captured every month (13-34 seeds/m 2 ; 84.8%-95.8%). In addition toS. wallichii, three other seed species were found over most of the months of observation,O. populneus(< 1-3 seeds/m 2 , found every month),E. spicata(< 1-2 seeds/m 2 , found over 8 months), andA. punctata(< 1 seeds/m 2 , found over 8 months).

On the forest site, monthly captured seeds were relatively constant (Fig. 2), except for the highest abundance in August (466 seeds) and November (418 seeds). In the reforested f ield, the highest abundance occurred between June to August (171-271 seeds). From overall number of seeds captured every month, the proportion of the seed in the forest ranged between 50-90% and 10-50% in the reforested f ield.In general, seeds found monthly in the forest were higher than in the reforested f ield; however, the proportion of seeds was similar on both sites in June. In each month of observation, the seed species richness in the forest site was always higher than the reforested f ield.

Over the 8 months, most seeds were found near to the forest edge (5-35 m into the forest area and 5 m outwards to the reforested f ield (Fig. 3). The high number of seeds capturedin these segments was mostly due to the abundance ofS.wallichii.The total number of seeds captured in the forest decreased with the distance from the forest edge. The number ofS. wallichiiseeds decreased in segments further than 35 m from the forest edge. In the last segment (95 m from the forest edge), there was a high number ofO. populneus. In the reforested f ield, the number of seeds captured decreased dramatically in segments further than 5 m from the forest edge. The number ofS. wallichiiseeds was constantly low in the 15-95 m segments, except for segment 65 m where the number increased.

Table 2 Seed species trapped on 60 (0.5 m × 0.5 m) seed traps in Forest (F) and Reforested f ield (RF) during 8 months of study

Surrounding vegetation

From the 60 circular plots (78.5 m 2 ), 158 trees (335 trees/hectare) were taller than 1.5 m, 28 species belonging to 21 families (Table 3). Species richness in the forest (23 species) was higher than in the reforested f ield (13 species). The most dominant species with the highest relative density wasS. wallichii(31.7%), followed byL. dolichocarpa(17.8%)andE. spicata(12.9%).Schima wallichiitrees was present in every segment of the transect. Seeds from the three most dominant species composed the seed rain in this study.

Based on f ield observations, the forest site has a more closed canopy dominated by trees. In contrast, the reforested f ield has a more open canopy due to low tree density and scattered trees. Vegetation in the reforested f ield was dominated by annual grasses and tall herbs such asE.adenophorum and E. inulifolium. Using the hemispherical photograph taken from 50 cm above ground, the vegetation cover in every seed trap area ranged from 17.3 to 94.6% and diff ered signif icantly (one-way ANOVA,p< 0.001) between forest site and the reforested f ield.

Seed rain community composition

The multivariateglmanalyses of seed species abundance as a response variable showed that distance and surrounding vegetation signif icantly aff ected seed rain composition in both sites (Table 4). In the forest, distance to the edge, surrounding tree species and surrounding vegetation cover (tree height) signif icantly aff ected the seed rain community. In the reforested f ield, distance to the forest edge and the surrounding vegetation (mean tree height) were signif icant factors.

Schima wallichii seeds

The univariate linear regression model ofS. wallichiiseeds showed that the surrounding tree species richness and vegetation cover signif icantly aff ected the abundance ofS. wallichiiseeds on both sites (Table 5). In the forest site, the abundance ofS. wallichiiincreased with the increase in surrounding vegetation cover (Table 5, Fig. 4), decreased with the increase in distance to the forest edge, and increased with the surrounding species richness (Table 5, Fig. 4). In the reforested f ield,S.wallichiiseeds increased with an increase in the surrounding vegetation cover (see Table 5, Fig. 4), and decreased with the increase in the surrounding species richness expressed by Shannon index (Table 5, Fig. 4).

Fig. 2 Abundance of seed species captured monthly in all sites, forests, and reforested f ields

Discussion

The consistently higher abundance and richness of seed rain captured in the forest compared with the reforested f ield was expected. However, with wind-dispersed or anemochory seeds, the low number found in the reforested f ield was most likely due to low seed production and/or low density of mature trees than to poor dispersal (Levine and Murrell 2003). The high abundance of wind dispersed seeds (e.g.,S. wallichii) found in forest edge was possibly an indication that the reforested f ield received anemochory seed from the forest as well as from the reforested f ield itself.

Seed rain variations: seasonal and distance to forest edge

The dominance of wind-dispersed seed in this study has also noted by Augspurger and Franson ( 1988), Cubiña and Aide ( 2001) and Martínez-Garza et al. ( 2011). The number of seeds captured monthly on both sites and the reforested field coincided with the frequency of monthly rain(see Table S4 for related weather data). There were more seeds captured in months with less frequent rainy days(i.e., June-August). When precipitation and the frequency of rainy days increased in September, the data showed a significant decrease in the number of seeds captured. The temporal variability of seed rain related to rainfall has been reported by other studies (Holl 1999; Barbosa and Pizo 2006; Perini et al. 2019). Research by Holl ( 1999)in Costa Rica specifically noted that seed dispersed byanimals increased in rainy months, while anemochory seed rain increased in dry months. The results in this study agree with the latter that anemochory seed rain (S.wallichii,O. populneus) increased in dry months. The information gained from this is especially relevant for seed collection in the peak period in the future.

Fig. 3 Abundance of seed species captured based on their distance to the forest edge

Table 3 Tree species density per hectare found in 60 circular plots(d = 10 m; area size = 78.5 m 2 ) placed in the forest (F) and reforested f ield (RF) with seed trap as the center of the plots

Table 3 (continued)

Table 4 The best-f it generalized linear model ( manyglm) with negative binomial error distribution for multivariate seed rain abundances found on the seed traps in forest and reforested f ield. Signif icance was based on log-likelihood ratio test and calculated with 999 resampling iterations

Table 5 The best-f it linear models for the log-transformed Schima wallichii (SW) seed abundances found on the seed traps in forest and reforested f ield

Fig. 4 Signif icant relationship between the log-transformed Schima wallichii (SW) seed abundance and diff erent explanatory variables for forest:a distance to the forest edge, b surrounding tree richness, c surrounding vegetation cover, and reforested f ield: d surrounding tree richness, e mean height of surrounding S. wallichii trees).The dots represent the data points; the lines depict model predictions; and the grey area are the 95% conf idence interval

Seed rain on both sites was higher along the forest edge, also observed by other studies (McDonnell and Stiles 1983; Willson and Crome 1989; de Melo et al.2006). The abundance of seeds dispersed by wind or animals may differ along the edge areas due to changes in abiotic conditions (e.g., resistance to wind flow), and/or to behavior and abundance of dispersal agents (Vespa et al. 2014). Along the forest edge, there is a transition in vegetation and in wind patterns between the closed canopy of tall trees with vertical wind and the reforested field which is more open with horizontal winds. A possible explanation for the high number of wind-dispersed seed on the forest edge of this study may be due to differences in wind patterns between the forest and reforested field, causing edge areas to receive wind-dispersed seed rain input from both areas.

The low abundance and diversity of seed captured in the reforested area shows that there was a limitation on seed dispersal from the adjacent forest. Similar results by other studies for seed dispersal from forests adjacent to pasture found that seeds were mostly anemochory and limited to 10 m distance (Aide and Cavelier 1994), 5 m(Holl 1999), and 4 m (Zimmerman et al. 2000; Cubiña and Aide 2001) from the forest edge. Because there were only anemochory seeds found on the reforested field shows the importance of attracting animal dispersal agents to disperse zoochory seeds from forest to reforested field,for instance by installing bird perches (Aide and Cavelier 1994; Zwiener et al. 2014).

Impact of surrounding vegetation on seed rain community

The seed rain community in this study was signif icantly aff ected by the distance to the forest edge, the surrounding tree diversity (only in the forest area), and the surrounding tree heights. The forest area with a more complex structure provides diverse mature trees which, in turn, shape a more complex seed rain community compared with the reforested f ield. This f inding highlights the importance of protecting forest areas as a seed source for the recovery of reforested f ields. The more diverse and abundant seed sources from the forest represents a higher chance that multi-species seeds might disperse to reforested f ields. The fact that the seed rain community in both sites was aff ected by the surrounding tree heights shows that taller trees may produce more seeds, and especially trees producing anemochory seeds which have a higher chance to disperse more seeds (Thomson et al. 2011).Our f indings, which show that the seed rain community was aff ected by the distance to the forest edge, implies that there is potential for seeds from the forest to disperse to the reforested f ield. However, some intervention might be needed to enhance this process.

It is important to note that the seed rain community data in this study was collected over 8 months, and it is possible that the results do not fully ref lect the actual seed rain community. This may be due to the time constraint and to the patchiness of seed rain, especially those dispersed by vertebrates (Muller-Landau et al. 2008; Bustamante-Sánchez and Armesto 2012). Another long-term study in the tropical forests of Barro Colorado Island, Panama, showed this patchiness of the seed rain. Over 19 years of seed collection,a third of the species with mature trees failed to hit the traps in a given year (Terborgh et al. 2002; Muller-Landau et al.2008). Another study in the Amazonian plain forest, Peru,showed that, from hundreds of mature tree species, only seeds of an average 3-4 species reaching the trap (Terborgh et al. 2019). In addition to these reasons, there is also a possibility that some seeds have been eaten by animals before the collection and dispersed elsewhere outside the traps (i.e.,secondary dispersal). Therefore, our study results should be interpreted cautiously.

Does planting Schima wallichii facilitate succession?

Besides being a native species and originating from the surrounding forest,S. wallichiiwas selected for reforestation in this area because of its pioneer characteristics, e.g.,its ability to grow in open areas, its high growth rate and survival, its dense crown, and producing a large amount of wind-dispersed seeds. Wind-dispersed seeds are considered an important character for species selection since they have a higher range of dispersal in open areas. In addition, the high growth rate ofS. wallichiiwas expected to provide shade to eliminate light-demanding weed grasses and a more suitable microclimate for tree seedlings to grow. In other parts of Indonesia,S. wallichiihas been commonly used for reforestation in catchment areas, and as a cover crop to provide shade in agroforestry systems with coff ee,Pinus merkusii,andAgathis dammara(Orwa et al. 2009). According to the plant database in Asia,S. wallichiimay produce seeds after 4 years (Orwa et al. 2009). In fact, 7 years after planting,S.wallichiiin the reforested f ield are producing viable seeds.

The univariate model showed that in both sites the abundance ofS. wallichiiseeds was negatively aff ected by the surrounding tree richness, but positively aff ected by the surrounding vegetation. However, the magnitude of these relations in the forest was lower than in the reforested f ield. This may be due to the more complex structure of the forest compared with the reforested f ield. In the forest area, distance to the forest edge also negatively aff ected seed abundance.This indicates that the abundance ofS. wallichiiseeds was directly related to the dominance ofS. wallichiitrees. With more species in the surrounding vegetation,S.wallichiiwas less dominant. In the reforested f ield, the larger the area of surroundingS. wallichii,the more dominant the species and the more seeds produced.

Based on the f ield observations, most ofS. wallichiiseeds failed to germinate and establish into mature trees in the reforested f ield. The plantedS. wallichiiproduced dense crowns but there was as yet no canopy closure. There was a high cover ofEupatoriumshrubs and grasses. The planting ofS. wallichii,7 years prior to the study, did not accelerate succession in the reforested f ield.

Besides the lack of propagules, one of the barriers to seedling establishment in the reforested area was possibly the abundance ofEupatoriumshrubs. Previous research on vegetation structure in this area showed a high dominance ofE. ripariumandE. inulifolium(Setiawan and Sulistyawati 2008).Eupatoriumshrubs are considered invasive and emit allelopathic compounds to the soil environment, inhibiting other organisms (e.g., plants and soil microbes) to thrive(Tripathi et al. 1981, 2006; Rai and Tripathi 1984; Sharma et al. 1998). The dominance ofEupatoriumshrubs was due to their characteristics, e.g., fast-growing, wide-range habitat, and orthodox seeds, ones that survive drying and/or freezing (Sharma et al. 1998). Thapa et al. ( 2017) demonstrated the allelopathic Effect ofE. adenophorumlitter on growth ofS. wallichiiseedlings. However, this Effect seems to only persist in low density S.wallichii. This means that plantingS. wallichiiat high density may counter the allelopathic Effects and the invasiveness ofEupatoriumshrubs.Another solution would be to eliminate the shrubs in the early seedling establishment period (Ammondt et al. 2013;Rezende and Vieira 2019).

Implications for management

Based on these findings, the reforested field has yet to develop into a diverse tree species community after 7 years of planting nativeS. wallichiiandD. stellare. The combination of a lack of seed source and competition with woody shrubs might partly explain this as shown by other studies (Ferguson et al. 2003; Hooper et al. 2005; Elgar et al.2014). Several steps are necessary to enhance the recovery rate and facilitate succession: (1) plant more diverse native seedlings in a tree island, with a priority to animal-dispersed species; (2) eliminateEupatoriumshrubs; (3) install bird perches; and, (4) control weed-grass cover in the f irst weeks of planting. Although it may be diffi cult to f ind a source of native tree seedlings, using native, local species is necessary.Apart from being adapted to local conditions, native species reduces potential negative outcomes related to associated organisms, e.g., the emergence of bud burst prior to the emergence of herbivores, interactions with pollinators and pathogens (Broadhurst and Boshler 2014).

Planting tree islands is widely known for the benef its of attracting related biodiversity such as birds, which in turn will help to disperse large seeds (Cole et al. 2010). Besides that, planting tree islands instead of a conventional uniform plantation design costs relatively lower (Zahawi and Augspurger 2006). Hopefully, a tree island can accelerate the succession by providing suitable microclimate for seedling establishment, because, in the end, seed dispersal is meaningless if the seeds cannot survive and establish into mature trees (Reid and Holl 2013).

Conclusions

Seven years after planting seedlings of native tree species on the post-agricultural area, the f ield has yet to recover into a diverse tree species forest. This study of the seed rain in the reforested f ield and the adjacent secondary forest showed that the f ield received limited seed rain, and only anemochory or wind-dispersed seeds. To accelerate succession,management actions involve planting more diverse tree seedlings in tree islands and introducing structures that will attract vertebrate seed-dispersers.

Acknowledgements We appreciate the help of Avniar Noviantini and Nusa Mashita with the f ieldwork; Ipin Suryana and Ujang with establishing transect and traps; Arief Hamidi with tree identif ication;Mahendra Primajati with the hemispherical photograph; and Theo Syamuda with the map of the study location. The helpful comments from two anonymous reviewers and language revision from Ronald D.Ayling are greatly acknowledged.

Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affi liations.