Forest structure as influenced by different types of community forestry in a lower montane rainforest of Central Sulawesi, Indonesia

Forest structure as influenced by different types of community forestry in a lower montane rainforest of Central Sulawesi, Indonesia

Johannes Dietz^1,2*, Dirk Hölscher¹, Christoph Leuschner², Adam Malik³, M. Amran Amir³

¹ Tropical Silviculture, Institute of Silviculture, University of Göttingen, Büsgenweg 1, 37077 Göttingen, Germany

² Plant Ecology, Albrecht-von-Haller-Institute for Plant Sciences, University of Göttingen, Untere Karspüle 2, 37073 Göttingen, Germany

³ Institute for Forest Management, Faculty of Agriculture, Bumi Tondo Campus, Tadulako University, Palu 94118, Indonesia

Running title: Forest structure under different community forestry in Sulawesi

5516 (including authors, abstract, captions and references) / abstract 438 words

with 1 table and 5 figures

* Corresponding author

Email: jdietz@gwdg.de

Phone: +49 551 399556

Fax: +49 551 394019

Summary

Due to high deforestation rates, it is likely that in many regions tropical rainforests will survive in protected areas only. These refuges have to be surrounded by buffer zones where low impact forest management is permitted. In fact, increasing parts of the remaining rainforest areas are encroached by different types of low-intensity forest use. However, until now only few data exist on how different practices affect forest structure. The objectives of this study were to analyze differences in aboveground forest stand structure as caused by different types of forest utilization in the margin zone of a protected rainforest. The study was conducted in a lower montane rainforest region (800‑1140 m) in the vicinity of the village of Toro in Central Sulawesi, Indonesia. We assessed the structure of twelve forest plots (0.15 ha each) which represented four types of forest use which are widespread in the study region. These types were natural forest without major disturbance, forest extraction of small-diameter timber, forest with selective extraction of large timbers and cacao agroforest plantations under trees remaining from the natural forest. The tree basal area ranged from relatively high values in the natural forests (52.5 m² per ha) to 19.4 m² per ha in the agroforest, and was paralleled by a decrease in tree height. Stem density peaked in stands where large timbers had been extracted and abundant tree regeneration had established, and it was lowest in the agroforest stands. Canopy openness, as estimated from 30 hemispheric photographs per plot, was lowest in the natural forest after small timber extraction (7 %) and highest in cacao agroforest (15 %). Leaf area index (LAI), as estimated from the same photos, averaged 6.2 m² m^‑2 in the natural forest, 5.3 in the forests with small timber extraction, 5.0 in the forests with large timber extraction, and 5.3 on the agroforest plots. The latter result confirms earlier reports on relatively high leaf areas in the cacao agroforestry system. The relatively high LAI in forest stands after major timber extraction indicates a rapid recovery of leaf area after disturbance. We also analyzed the influence of stand structural parameters on rainfall partitioning into throughfall, stemflow and interception. In all forest stands, except for the agroforest plots, we observed a close negative correlation between mean throughfall and tree height (r² = 0.72) indicating a higher interception loss in taller stands. We conclude that local forest management at low to moderate intensities, as it is common in the margin zones of protected forests, has a significant and quantifiable impact on several forest structural parameters which in turn may influence ecosystem functions such as rainfall partitioning in the canopy.

Key words: forest use, leaf area index, rainfall partitioning, rainforest, stand structure, stem density

Introduction

Over the past decades, tropical rainforests have been suffering increasing losses through conversion to other land cover types. South East Asia and especially some regions on the Indonesian archipelago including Central Sulawesi were identified as areas of high deforestation rates (Achard et al. 1998, Drigo 2004, FAO 2001). However, more subtle forms of forest disturbance as through selective logging on conservation lands are difficult to identify via satellite and are assumed to be underestimated so far (Asner et al. 2005). Based on recent analyses for five states of the Brazilian Amazon, selective logging doubled previous estimates of the total amount of forest degraded by human activities (Asner et al. 2005).

The high rates of forest conversion together with the rapidly proceeding forest degradation have long since called upon conservation efforts. It has been emphasized that protection requires the interaction with the people that traditionally live in and around the forests (Schweithelm 2004). Contemporary policies advocate buffer zone management by a low impact forest management which, according to Schweithelm (2004), can be developed from the experience of traditional communities that regulate the access to land and forest resources. For the Lore Lindu National Park in Central Sulawesi, such traditional structures and regulations are found in the village of Ngata Toro (Helmi 2005). There, customary law has created a zonation of the forest into different use types which are distinguished by the local population. According to Helmi (2005), the “Wana” type of community forestry is predominantly a natural forest where only occasional rattan extraction takes place. In the “Pangale” small-diameter timbers can be extracted in the understorey, or alternatively, a restricted number of large stems can be selectively logged. Strong human intervention is characteristic for the “Pahawa Pongko” zone, which comprises cacao agroforests under a sparse canopy of large trees that remain from the natural forest (Helmi, 2005).
However, such definitions of land use are subjective and there is a need to quantify such categories by objective means (cf. Drigo 2004). Drigo (2004) suggests the translation from land use practice to land cover types where structural characteristics can be measured and monitored. Following this suggestion, the objective of our study was to identify and quantify the impact of different forest use practices on the stand structure of a lower montane rainforest in the region of the Toro village, Central Sulawesi. In a further study we analyzed the influence of stand structural parameters on rainfall partitioning into throughfall, stemflow and interception in order to relate stand structural characteristics to forest functions with respect to the hydrological cycle.

Materials and Methods

Study Site

This study was carried out in tropical rainforests of the sub-montane to lower montane elevation zone (800‑1140 m a.s.l.) in the Toro Valley on the western boundary of the Lore Lindu National Park, Central Sulawesi, Indonesia. Twelve sites of 1500 m² each were chosen for the study, evenly representing all four investigated forest use types (n = 3): natural forest, forest with extraction of small-diameter timbers (hereafter termed ‘small timber extraction’), forest with selective extraction of large timbers (hereafter termed ‘large timber extraction’), and cacao plantations under trees remaining from the natural forest (agroforest). The four forest use types (natural forest to agroforest) generally occurred in the vicinity of the Village of Toro along a gradient of accessibility with natural forest being most distant and agroforest closest to the settlements. During the investigation period no management was observed on the forested sites while the cacao agroforests received conventional treatment of weeding and pruning. All sites were situated on moderately to severely steep slopes from 17° to 39°. The tree species composition in these forest types was studied by Gradstein et al. (this volume). They found Meliaceae, Lauraceae, and Sapotaceae to be dominant in natural forests, which shifted to Rubiaceae, Fagaceae, and Myristicaceae after both, small and large timber extraction. Moraceae, Myristicaceae, and Melastomataceae dominated in the investigated agroforests. The tree species richness varied little between natural forest stands and after small timber extraction, but was lower in stands after large timber extraction and lowest in the agroforests (Gradstein et al. this volume).

Stand Inventory

On the forested sites, all trees with a diameter at breast height (dbh) of 10 cm or more were censused for dbh and height on the entire plot according to Kramer and Akça (1995). To account for smaller statured trees all individuals with dbh < 10 cm, but higher than 2 m, were assessed on three randomly located subplots of 5 x 5 m within the plot. In the agroforests, all stems taller than 2 m were censused. Measurement of dbh was done using a metric measuring tape, height measurements were taken with a Vertex III ultrasonic hypsometer (Haglöf, Långsele, Sweden). Stand basal area was corrected for edge and slope effects.

Hemispherical Photography

Data on canopy structure was derived from hemispherical photographs which were obtained with a Nikon Coolpix 900 digital camera / Nikon FC-E8 fisheye-converter of 2048 x 1536 pixel resolution mounted on a HemiView leveling device (Delta-T, Cambridge, UK). For reduction of horizon effects on the sloped sites, the device was consistently placed at 1.3 m above the ground. Thirty pictures per site and campaign were taken at randomly located points on several occasions during the study period from January 2004 until April 2005. All images were analyzed with the software WinScanopy 2004a (Régent Instruments Inc., Sainte-Foy, Canada) for canopy openness under an opening angle of 30°, and with CanEye V 3.6 (INRA, Avignon, France) for estimates of leaf area index (LAI).

Statistical Analysis

Data were statistically analyzed using the SAS 8.2 software package (SAS Institute Inc., Cary, NC, US). Multiple comparisons among group means were conducted by analyses of variance (Duncan’s test); a significance level of p < 0.05 was maintained throughout the analysis.

Results and Discussion

Stem Density

Koop et al. (1995) recognized the density of large trees as a good estimator for human influence on Sumatran rainforests. In our study region, the density of large stems with a diameter at breast height (dbh) larger than 50 cm also appeared as good indicator of the intensity of forest use (figure 1, solid bars). Highest values averaging at 78 and 75 stems per hectare were found in the natural forest and in forest after small timber extraction. Large timber extraction reduces the density to 48 stems per hectare. Agroforests contained an average of only 28 large trees per hectare (ranging between 6 and 51 stems per hectare, table 1). The very large variability in the density of large-diameter stems in the agroforestry systems may be related to the personal experience and preference of the plot owners. A similar structural diversity in agroforestry systems was previously documented for other tropical regions by Ruf and Schroth (2004). Small stems (< 10 cm in dbh) averaged slightly above 2000 stems per hectare in the natural forest, but reached higher values after large timber extraction (figure 1, hatched bars). In the gaps created by logging of big trees, abundant regrowth occurred with 3700 stems per hectare on the average. In the agroforest, the class of small stems included most of the planted cacao (Theobroma cacao) trees.
Few remaining large trees and a high proportion of small stems were characteristic for both forest stands after large timber extraction and agroforest stands. This contrasted with natural forest stands and stands after small timber extraction. In the latter, the dbh class < 20 cm contained less than 50 % of all stems, (figures 2 a and b), whereas after large timber extraction the proportion of trees in this diameter class reached values above 60 %. In the agroforest, the high proportion of thin stems mostly owed to planted crop trees (figures 2 c and d). Big trees were scarce in these two forest use types.

Basal Area and Stand Height

The abundance of large-diameter stems is known to be a good estimator of a stand’s basal area of stems ≥ 10 cm dbh (McElhinny et al. 2005). Average basal areas of 52.5 and 51.6 m² per hectare in the natural forest and in stands after small timber extraction, respectively, seem to be relatively high, especially when the peak value of 62.8 m² ha^-1 is considered However, a comparison with other studies in natural forests above 600 m asl in the South East Asian region reveals that the values for the natural forests in Toro are indeed slightly above the average, yet they are still within the reported range (figure 3). In our study region, the average basal area decreased to 38.4 m² ha^-1 in stands after large timber extraction and dropped to 19.4 m² ha^-1 in the agroforest. The decrease in tree basal area along the gradient of use intensity was paralleled by a decrease in tree height. The stand mid height, which is the height of a tree with the arithmetic mean of dbh (Kramer and Akça 1995), decreased from 24.4 m in the natural forest and from 20.5 m in stands after small timber extraction to 18.9 m in stands after large timber extraction and 18.8 m in the agroforest. The apparent drop in tree mean height between natural forest and forest after small timber extraction is statistically not significant. If existent, it may be caused by slight changes in environmental conditions between sites (cf. Takyu et al. 2002).

Leaf Area Index and Canopy Openness

Leaf area index (LAI) is a basic variable for modeling the energy and matter fluxes in forests; this parameter is also of great importance for vegetation models in the context of global change research (e.g. Asner et al. 2003). A variety of different assessment methods are used in the field (Breda 2003, Jonckheere et al. 2004). A consistent underestimation of LAI by indirect methods such as hemispheric photography was reported by Breda (2003) but their growing potential was acknowledged. On the global scale Asner et al. (2003) confirmed that indirect optical measurements of LAI were closer to the results from destructive assessments than interpolations derived from litter sampling or estimates obtained through allometric functions. The destructive sampling approach would have been very labor-intensive and would have conflicted with conservation policies in our forest plots which are located in a national park. Since hemispheric photography is a rapid, sufficiently reliable and non-destructive approach for the assessment of canopy structure (Jonckheere et al. 2004), it was thus chosen in our study.

The LAI of evergreen forests may change seasonally with rainfall volumes and other climatic parameters (Asner et al. 2003, Breda 2003). Therefore, we analyzed most stands for their leaf area at three different points of time. However, in none of the stands, a distinct seasonality in LAI was observed (see figure 4a for the LAI of a natural forest). This may be due to the largely non-seasonal rainfall distribution in the study region. The following results are based on photos taken during a measurement campaign in April 2005.

The LAI as derived from 30 photos within a plot varied only moderately in space within a plot. In the natural forest, the coefficient of variation (CV) ranged between 15 and 24 % which is similar to results found in natural forests of the Brazilian Amazon using a different, yet quite comparable method (Licor LAI 2000; Aragão et al. 2005). In our study region, the CV values in forest stands after large timber extraction and the agroforest stands ranged between 24 and 32 %. This corresponds with CVs measured in Amazonian secondary forests (Aragão et al. 2005). The average LAI of the natural forest stands in Sulawesi ranged between 5.7 and 6.5 m² m^-2. In their global review Asner et al. (2003) reported a mean (± SD) LAI of 4.9 (2.0) for tropical evergreen broadleaf forests. Aragão et al. (2005) found values between 3.6 and 6.6 for natural forest in Amazonia, Trichon et al. (1998) reported values between 3.9 and 6.1 from rainforests in Sumatra, and similarly Kumagai et al. (2004) found values between 4.8 and 6.8 in Borneo. Roberts et al. (2004) collected LAI values ranging between 4.1 and 8.0 for tropical lowland rainforests with a tendency of higher values in Asia. Thus, the values obtained in Sulawesi are well within the range of other studies and they may also support slightly higher values in Asian tropical forests (Roberts et al. 2004). The average LAI of a given forest use type was 6.2 m² m^‑2 in the natural forest, 5.3 in forests with small timber extraction, 5.0 in forests with large timber extraction, and 5.3 in the agroforest. The variation among plots of the same use type was largest in the agroforestry systems, where values between 4.1 and 6.3 were observed. The large variability among the agroforest sites may be explained with the already mentioned densities of large and small stems that differ greatly under permanent human influence (Ruf and Schroth 2004). Our relatively high average LAI values for agroforest confirm studies claiming that plantations in general (Asner et al. 2003), and cacao in Central Sulawesi in particular (Falk 2004), can maintain a comparably large leaf area. The relatively high LAI values in managed stands of Sulawesi compare well with results from Amazonia where only little differences in LAI between natural forests and selectively logged sites were observed (Aragão et al. 2005). The high proportion of regrowth after large timber extraction may contribute substantially to such high LAI values in Sulawesi.

The average canopy openness, as determined from photographs taken at 1.3 m above the ground, was between 10 and 11 % per site in the natural forest; it was considerably higher in cacao agroforest (between 16 and 20 %; table 1). The values obtained in the forest stands after small and large timber extraction did not differ significantly from those in the natural forest. Trichton et al. (1998) found canopy openness values below 10 % in mature Sumatran rainforests but they report much larger values for gap phases. Tobón Marin et al. (2000) found canopy openness values from 9 to 17 % in natural rainforests of western Amazonia. In the agroforest systems of Sulawesi, many large trees have been removed arbitrarily and to different extents by the plantation owners who subsequently plant crop trees in different densities. Thus, in the agroforest, we found the highest average canopy openness which was associated with a relatively high spatial variation in this parameter in a given stand and among different stands (table 1).

Outlook – the relationship between stand structure and ecosystem functions

The aboveground structure of tropical forests has been shown to influence many ecosystem properties that control energy and matter fluxes including radiation transmission, the temperature regime, the atmospheric saturation deficit (Leigh 1975, Montgomery 2004, Parton et al. 1996, Trichon et al. 1998), soil temperature, fine root biomass and turnover (Hertel et al. this volume), as well as the diversity of the soil microfauna (Migge-Kleian et al. this volume) and of butterflies (Fermon et al. 2005). Forest structure is also considered to be an important parameter in the forest hydrological cycle of forests (e.g. Bigelow 2001, Bruijnzeel et al. 1993, Chappell et al. 2001, Hall 2003, Hölscher et al. 2004).
In our study plots in Central Sulawesi, we monitored throughfall with 30 funnel collectors each on all 12 plots, stemflow with 10 collectors per plot, and gross rainfall with rain gauges adjacent to each plot. The mean throughfall was 66 % of incident precipitation in natural forest, 81 % in forest with small timber extraction, and 82 % in forest with large timber extraction and in the agroforest. Stemflow was estimated with less than 1 % in all studied forest use types. Among the stand structural parameters that correlate with rainfall partitioning we found the best correlation between throughfall and tree top height (height of the 10 % thickest trees; ≥ 10 cm dbh) for rainfall events of less than 10 mm. The mean throughfall decreased significantly with tree height. When all 12 plots are considered, the coefficient of determination (r²) was 0.43 and increased to 0.72 when the three agroforest plots are omitted from the analysis (figure 5). We conclude that tree top height is an easy-to-measure parameter that allows to estimate the throughfall volume along the disturbance gradient from natural forest to forest after large timber extraction. In contrast, conversion to agroforest seems to introduce further, so far unexplained, variation in the hydrological processes. In other studies, LAI has often been used for modeling the partitioning of rainfall into throughfall, stemflow and interception (Gash 1979, Gash et al. 1995), and this parameter has been described as an estimator for throughfall volumes (Hall 2003, Holwerda et al. 2005). In contrast, LAI was not significantly correlated to throughfall (or rainfall interception) in our study. A possible explanation is that the total LAI and the associated canopy water storage differed only slightly among the studied stands while the vertical distribution of the foliage varied greatly. Taller trees presumably increase the aerodynamic roughness length of the stand, resulting in a more effective turbulent energy exchange with the atmosphere. As a consequence, the canopy dries up more often between rainfall events in tall stands, while it remains wet for longer periods in stands with lower-statured trees and with a more clumped distribution of the leaf area.

Conclusions

We conclude that the different types of community forestry in Central Sulawesi changed stand structural characteristics considerably. We found strong effects on stem density, basal area and stand height, whereas the leaf area index seemed to be less affected. The changes in forest structure were associated with differences in the forest hydrological cycle as the volume of net precipitation increased with decreasing stand height.

Acknowledgements

This study was conducted in the framework of the joint Indonesian-German research project ‘Stability of Tropical Rainforest Margins, Indonesia (STORMA)’ funded by the German Science Foundation (SFB 552). We thank the village of Toro for their friendly help and cooperation and André Twele for substantial support in processing the hemispherical photographs.

References

Aiba SI, Kitayama K (2002) Effects of the 1997‑98 El Niño drought on rain forests of Mount Kinabalu, Borneo. J Trop Ecol 18:215‑230

Aragão LEOC, Shimabukuro YE, Espírito Santo FDB, Williams M (2005) Landscape pattern and spatial variability of leaf area index in Eastern Amazonia. For Ecol Manage 211:240‑256

Archard F, Eva H, Glinni A, Mayaoux P, Richards T, Stibig HJ (1998) Identification of deforestation hot spot areas in the humid tropics. TREES Publication Series B4, European Commission, Luxembourg

Asner GP, Knapp DE, Broadbent EN, Oliveira PJC, Keller M, Silva JN (2005) Selective logging in the Brazilian Amazon. Science 310:480‑482

Asner GP, Scurlock JMO, Hicke JA (2003) Global synthesis of leaf area index observations: implications for ecological and remote sensing studies. Global Ecol Biogeogr 12:191‑205

Aylward B (2004) Land use, hydrological function and economic valuation. In Bonnell M and Bruijnzeel LA (eds.) Forests, water and people in the humid tropics. Cambridge University Press, Cambridge, pp 99‑120

Bigelow S (2001) Evapotranspiration modeled from stands of three broad-leaved tropical trees in Costa Rica. Hydrol Proc 15:2779‑2796

Breda NJJ (2003) Ground-based measurements of leaf area index: a review of methods, instruments and current controversies. J Exp Bot 54:2403‑2417

Brodbeck F (2004) Structure and Processes in Traditional Forest Gardens of Central Sulawesi, Indonesia. Cuvillier Verlag, Göttingen

Bruijnzeel LA, Waterloo MJ, Proctor J, Kuiters AT, Kotterink B (1993) Hydrological observations in montane rainforests on Gunung Silam, Sabah; Malaysia, with special reference to the ‘Massenerhebung’ effect. J Hydrol 81:145‑167

Chappell NA, Bidin K, Tych W (2001) Modelling rainfall and canopy controls on net-precipitation beneath selectively-logged tropical forest. Plant Ecol 153:215‑229

Drigo R (2004) Trends and patterns of tropical land use change. In Bonnell M and Bruijnzeel LA (eds.) Forests, water and people in the humid tropics. Cambridge University Press, Cambridge, pp 9‑39

Falk U (2004) Turbulent fluxes of CO₂, H₂O and energy in the atmospheric boundary layer above tropical vegetation investigated by Eddy covariance measurements. PhD thesis, University of Göttingen

FAO (2001) Global forest resources assessment 2000. Main Report. FAO Forestry Paper 140

Fermon H, Waltert M, Vane-Wright RI, Mühlenberg M (2005) Forest use and vertical stratification in fruit-feeding butterflies of Sulawesi, Indonesia: impacts for conservation. Biodiv Conserv 14:333‑350

Gash JHC (1979) An analytical model of rainfall interception of forests. Quart J Roy Met Soc 105:43‑55

Gash JHC, Lloyd CR, Lachaud G (1995) Estimating sparse forest rainfall interception with an analytical model. J Hydrol 170:79‑86

Gradstein SR, Kessler M, Pitopang R (2006) Tree species diversity relative to human land uses in tropical rain forest margins in Central Sulawesi. In: Tscharntke T, Leuschner C, Guhardja E, Zeller M (eds.) The stability of tropical rainforest margins: Linking ecological, economic and social constraints of land use and conservation. Springer, Berlin Heidelberg New York, this issue

Hall RL (2003) Interception loss as a function of rainfall and forest types: stochastic modelling for tropical canopies revisited. J Hydrol 280:1‑12

Hamman A, Barbon EB, Curio E, Madulid DA (1999) A botanical inventory of a submontane tropical rainforest on Negros Island, Philippines. Biodiv Conserv 8:1017‑1031

Hart MJ (1928) Stamtal en dunning. Dissertation, University of Wageningen

Helmi I (2005) Local people are well-placed to develop zonation plans in Indonesia’s Lore Lindu National Park. Tropical Forest Update 15:31‑32

Hertel D, Leuschner C, Harteveld M, Wiens M (2006) Fine root mass, distribution and activity in disturbed primary forests and secondary forests of the moist tropics. In: Tscharntke T, Leuschner C, Guhardja E, Zeller M (eds.) The stability of tropical rainforest margins: Linking ecological, economic and social constraints of land use and conservation. Springer, Berlin Heidelberg New York, this issue

Hölscher D, Mackensen J, Roberts JM (2004) Forest recovery in the humid tropics: changes in vegetation structure, nutrient pools and the hydrological cycle. In Bonnell M and Bruijnzeel LA (eds.) Forests, water and people in the humid tropics. Cambridge University Press, Cambridge, pp 598‑621

Holwerda F, Scatena FN, Bruijnzeel LA. Throughfall in a Puerto Rican lower montane rain forest: a comparison of sampling strategies. J Hydrol, in press

Jonckheere I, Fleck S, Nackaerts K, Muys B, Coppin P, Weiss M, Baret F (2004) Review of methods for in situ leaf area index determination Part I. Theories, sensors and hemispherical photography. Agri Forest Meteor 121:19‑35

Kessler M, Keßler PJA, Gradstein SR, Bach K, Schmull M, Pitopang R (2005) Tree diversity in primary forests and different land use systems in Central Sulawesi, Indonesia. Biodiv Conserv 14:547‑560

Kitamura S, Suzuki S, Yumoto T, Chuailua P, Plongmai K, Poonswad P, Noma N, Maruhashi T, Suckasam C (2005) A botanical inventory of a tropical seasonal forest in Khao Yai National Park, Thailand: implications for fruit–frugivore interactions. Biodiv Conserv 14:1241‑1262

Koop H, Rijksen HD, Wind J (1995) Tools to diagnose forest integrity: an appraisal method substantiated by Silvi-Star assessment of diversity and forest structure. In: Boyle TJB, Boontawee B (eds.) Measuring and Monitoring Biodiversity in Tropical and Temperate Forests. CIFOR, Bogor, Indonesia, pp 309‑333

Kramer H, Akça A (2002) Leitfaden zur Waldmeßlehre. Sauerländer Verlag, Frankfurt

Kumagai T, Katul GG, Saitoh TM, Sato Y. Manfroi OJ, Morooka T, Ichie T, Kuraji K, Suzuki M, Porporato A (2004) Water cycling in a Bornean tropical rain forest under current and projected precipitation scenarios. Water Resour Res 40:W01104

Leigh EG (1975) Structure and Climate in Tropical Rain Forest. Ann Rev Ecol Syst 6:67‑86

McElhinny C, Gibbons P, Brack C, Bauhus J (2005) Forest and woodland stand structural complexity: Its definition and measurement. For Ecol Manage 218:1‑24

Migge-Kleian M, Woltmann L, Schaefer M (2006) Impact of forest disturbance and land use change on soil invertebrate fauna in tropical rainforest margins. In: Tscharntke T, Leuschner C, Guhardja E, Zeller M (eds.) The stability of tropical rainforest margins: Linking ecological, economic and social constraints of land use and conservation. Springer, Berlin Heidelberg New York, this issue

Montgomery RA (2004) Effects of understorey foliage on patterns of light attenuation near the forest floor. Biotropica 36:33‑39

Parton WJ, Haxeltine A, Thornton P, Anne R, Hartman M (1996) Ecosystem sensitivity to land-surface models and leaf area index. Global and Planet Change 13:89-98

Roberts JM, Gash JHC, Tani M, Bruijnzeel LA (2004) Controls on evaporation in lowland tropical rainforest. In Bonnell M and Bruijnzeel LA (eds.) Forests, water and people in the humid tropics. Cambridge University Press, Cambridge, pp 287‑313

Ruf F, Schroth G (2004) Chocolate forests and monocultures: A historical review of cocoa growing and its conflicting role in tropical deforestation and forest conservation. In: Schroth G, da Fonseca GAB, Harvey CA, Gascon C, Vasconcelos HL, Izac AMN (eds.) Agroforestry and biodiversity conservation in tropical landscapes. Island Press, Washington, pp. 107-134

Schweithelm J (2004) Forest people and changing tropical forestland use in tropical Asia. In Bonnell M and Bruijnzeel LA (eds.) Forests, water and people in the humid tropics. Cambridge University Press, Cambridge, pp 66‑74

Sri-Gernyuang K, Kanzaki M, Mizuno T, Noguchi H, Teejuntuk S, Sungpalee C, Hara M, Yamakura T, Sahunalu P, Dhanmanonda P, Bunyavejchewin S (2003) Habitat differentiation of Lauraceae species in a tropical lower montane forest in northern Thailand. Ecol Res 18:1‑14

Takyu M, Aiba SI, Kitayama K (2002) Effects of topography on tropical lower montane forests under different geological conditions on Mount Kinabalu, Borneo. Plant Ecol 159:35‑49

Tobón Marin C, Bouten W, Sevink J (2000) Gross rainfall and its partitioning into throughfall, stemflow and evaporation of intercepted water in four forest ecosystems in western Amazonia. J Hydrol 237:40‑57

Trichton V, Walter JMN, Laumonier Y (1998) Identifying spatial patterns in the tropical rain forest structure using hemispherical photographs. Plant Ecol 137:227‑244

Willott SJ, Lim DC, Compton SG, Sutton SL (2000) Effects of selective logging on the butterflies of a Bornean rainforest. Conserv Biol 14:1055-1065

Table 1 Aboveground forest structure in stands influenced by different types of community forestry in the region of Toro, Central Sulawesi. The different forest use types are: natural forest (NF), forest after small timber extraction (STE), forest after large timber extraction (LTE) and agroforest (AF). Significantly different means between forest types (n = 3) are indicated by different small letters (analysis of variance, Duncan’s test, p < 0.05).

Forest Use Type	Elevation [m asl]	Stem Density [n ha-1]			Basal Area [m² ha-1]		DBH mean [cm]	Top Height [m]	Mid Height [m]	Leaf Area Index [m2 m-2]		Canopy Openness [%]
	plot center	height ≥ 2 m	dbh ≥ 10 cm	dbh > 50 cm	height ≥ 2 m	dbh ≥ 10 cm	dbh ≥ 10 cm	top 10 %	mean dbh	mean (n = 30)	cv (n = 30)		mean (n = 30)	cv (n = 30)
NF 1	948	2272	472	103	68.6	62.8	30.3	42.9	25.1	5.7ab	20.7		9.7	53.9
NF 4	1042	1806	406	58	50.0	48.1	31.5	47.6	26.7	6.4a	23.9		9.3	52.5
NF 5	1130	3455	655	71	51.1	46.6	26.7	35.8	21.5	6.5a	14.9		10.9	58.7
NF mean	1040	2511a	511a	78a	56.6a	52.5a	29.5ab	42.1a	24.4a	6.2a	19.9		10.0ab	55.0
Std Dev		850	129	23	10.4	9.0	2.5	5.9	2.7	0.4	4.6		0.8	3.2
STE 2	873	2020	620	61	55.7	43.5	26.8	35.1	20.7	4.3de	19.5		7.7	49.2
STE 3	1078	3855	655	91	67.0	62.7	29.5	37.5	22.5	6.3a	18.9		7.0	50.9
STE 4	982	2420	620	72	41.4	48.7	23.2	35.3	18.4	5.2bc	20.6		6.4	51.7
STE mean	978	2765a	632ab	75a	54.7a	51.6a	26.5ab	36.0ab	20.5ab	5.3a	19.7		7.0a	50.6
Std Dev		965	20	15	12.8	9.9	3.2	1.3	2.1	1.0	0.9		0.7	1.3
LTE 1	974	5495	695	25	41.1	34.9	20.4	33.2	16.8	4.9bcd	30.2		12.9	56.6
LTE 3	827	3740	740	64	53.6	50.1	25.9	37.0	21.3	4.9cd	24.4		8.2	48.5
LTE 4	959	4052	652	55	34.6	30.2	22.1	27.1	18.7	5.2bc	31.4		9.3	66.7
LTE mean	920	4429b	695b	48ab	43.1ab	38.4ab	22.8b	32.4b	18.9b	5.0a	28.7		10.1ab	57.2
Std Dev		936	44	21	9.6	10.4	2.8	5.0	2.3	0.2	3.7		2.4	9.1
AF 2	952	1706	125	6	8.6	8.5	25.6	28.4	15.9	5.7ab	31.6		19.7	85.7
AF 3	832	2705	237	25	23.7	23.4	32.7	35.7	20.6	6.3a	26.3		15.9	64.7
AF 4	806	2612	247	51	26.5	26.3	38.1	35.0	19.9	4.1e	26.1		9.8	86.7
AF mean	863	2341a	203c	28b	19.6b	19.4b	32.2a	33.0b	18.8b	5.3a	28.0		15.1b	79.0
Std Dev		552	68	23	9.6	9.6	6.3	4.0	2.5	1.1	3.1		5.0	12.4