Search

Two morphological variants of Terminalia alata (Combretaceae) differed in leaf flushing phenology and spatial distribution in a Cambodian deciduous forest. The hairy-type trees displayed leaf exchange behavior in the middle of the dry season. The glabrous type flushed new leaves 3 months after the wet season started. The leafless period of the hairy type was estimated to be

Litter plays key roles in forest ecosystems, and forest degradation is likely to spur a further decline in leaf litterfall inputs to forest soils. However, the effects on physicochemical surface-soil properties remain largely unknown, especially in seasonal tropical forest ecosystems. We initiated a litterremoval manipulation experiment in a Cambodian lowland evergreen forest undergoing intensive selective logging. Litter removal performed for 2 and 4.4 years respectively triggered an increase in bulk density and decrease in surface-soil carbon (C) and nitrogen (N) contents to 67 and 73% of the original levels, respectively. After only 2 years of treatment, bulk density rose to very high value (>1.40 Mg m) likely preventing further soil compaction, while the C and N reduction effects lasted over 2 years. Greater soil compaction occurs in stands with a smaller initial bulk density. However, C (N)-rich soils did not necessarily lose a greater relative proportion of C (N) than C (N)-poor soils. Although N remained above C following the litter removal, conservative trends in the C:N ratio suggested a limited capacity for N retention. Together, our data suggest that shifts in leaf litter inputs in response to localized human disturbances may have rapid and lasting consequences on physicochemical surface-soil properties; possibly accelerated by a tropical climate. Moreover, a speedy recovery to an adequate litter supply, at least before reaching the upper soil compaction limit, is essential to conserve forest ecosystems. Discipline: Soils, fertilizers and plant nutrition Additional key words: bulk density, carbon, forest degradation, nitrogen, soil compaction

We studied the relationships among the soil C stock in tropical monsoon forests, the type of forest, and the environmental factors in the lower Mekong basin in Cambodia. We analyzed nine soil profiles in evergreen and deciduous forests growing over sedimentary rock and basalt. Evergreen forest soils tended to have a larger C stock than deciduous forest soils within geological formations. In evergreen and deciduous forest soils, carbon stocks were 56.9 ± 30.0 (mean ± SD) and 34.9 ± 23.5 Mg C ha-1, respectively, in the 0- to 30-cm depth range, and 108.7 ± 53.0 and 53.2 ± 30.4 Mg C ha-1, respectively, in the 0- to 100-cm depth range. Soil C stock was highly positively correlated with soil water content in the dry season, which is likely affected by the openness of the forest canopy and by soil clay content.

We examined non-destructive methodologies for practicalities in monitoring anthropogenic greenhouse gas (GHG) emissions from tropical dry-land forest under the influence of various forms of human intervention. Spaceborne SAR withstood comparison with Landsat ETM+ in land cover classification of degraded tropical forest. For measurement of carbon stock and GHG flux per unit land area, the gain-loss method requires both growth rate and removal rate of forest carbon stock. However, the latter has rarely been obtained in tropical forest. For the stock-difference method, permanent sampling plot data can be used to estimate mean carbon stock per unit land area of each forest type. For cyclic land use that includes a clear-cutting stage such as slash-and-burn agriculture, chronosequential changes in carbon stock can be predicted by determining the time and spatial-distribution of cleared land. Changes in forest biomass by logging, storm-damage, etc., may be identified by monitoring the presence and diameter of the crowns of overstory trees. We developed five equations containing the parameter for crown diameter for estimating tree biomass. Overstory height can be a parameter for estimating ecosystem carbon stock of various plant communities, and forest height can be measured by airborne and spaceborne sensors, etc. Generic equations containing the parameter for overstory height are available for estimating community biomass of tropical and subtropical forests. PALSAR has an advantage over other remote systems by enabling frequent sensing and semi-direct biomass estimation using backscattering coefficients. However, no reasonable remote sensing methods exist for monitoring the amount of carbon loss by forest conversion and logging in forests with high biomass. To compensate for the faults of the present PALSAR methodologies and to enable practical and frequent monitoring of all types of forests by humans, it is vital to devise a new methodology to detect changes in high-biomass forests.

The Reducing Emissions from Deforestation and Forest Degradation in Developing Countries (REDD) initiative requires accurate estimates of carbon stock changes in forested areas. However, estimating carbon emissions from stumps of various heights left by illegal loggers is difficult. To remedy this problem, we examined two methods of estimating diameter at breast height (DBH) from a reference diameter observation measured at any stump height. The one-reference diameter (OD) observation model estimates DBH from a single diameter observation using empirical coefficients derived mainly from emergent dipterocarp trees. The two-reference diameter (TD) observation model estimates DBH from two diameter observations and assumes a logarithmic relationship between diameter and height. Prediction data to establish the models were collected in Cambodian lowland evergreen forests that are undergoing intensive illegal logging of emergent dipterocarp trees for timber. The OD model performed better than the TD model in predicting DBH and is extremely practical, as it requires only a single diameter observation. Validation data previously collected in the Southeast Asian tropical forests established the general validity of the OD model. This study may improve the reliability of the REDD scheme by providing a reliable method to assess carbon emissions from Southeast Asian tropical forests.

This study of a water cycle was conducted in an evergreen forest located in the Mekong River Basin in central Cambodia. At the observation site, we measured the dynamics of the spatial distribution of groundwater levels. The groundwater movement was analyzed two-dimensionally using boundary conditions and parameters that had been observed in the field. The climate in the research area is dominated by two seasons, which occur annually: a rainy and a dry season. The groundwater levels are generally high during the rainy season and low during the dry season. Groundwater levels were measured along a stream, which flowed through the study site. The streambed was visible at the head of the stream in January. At the next downriver well point, the streambed appeared in March. Finally, it became visible at all well points in April, meaning that surface runoff had disappeared temporarily and instead flowed underground during the ensuing dry period. Groundwater levels of the studied lateral flow perpendicular to the stream that seeped and infiltrated into the stream were 1.2–2.5 m deep (in April), which was the lowest level recorded for the year. During that period, the depth of the groundwater of the studied lateral flow fell by as much as 56 mm per month. In addition, the lateral flow groundwater infiltrated into groundwater of the stream during that period. The groundwater level fluctuation was estimated based on a two-dimensional analysis of lateral flow perpendicular to the stream using a numerical simulation model with soil physical parameters and observed boundary conditions. The observations of ground water fluctuations were well reproduced. Deep seepage of groundwater was estimated using a uniform boundary condition that allowed efflux through the bottom, estimated as being approximately 30 mm per year. The simulated deep seepage rate was considered plausible considering other hydrological components such as soil water storage fluctuation.

A methodological characteristic of the REDD+ scheme is that it attempts to reduce deforestation by rewarding communities that change problematic land use practices. This has led to discussions on benefit sharing. This article focuses on incentives for alternative land use practices among village members living in frontier areas, especially in relation to support for sustainable land use and people’s livelihoods, and clarifies the issues that REDD+ projects are likely to face in this context. Although some documents regarding REDD+ projects have mentioned support to encourage such incentives, insufficient consideration has been given to the realities of the changes in frontier communities. REDD+ projects are unlikely to motivate members to embrace alternative land use practices if support or benefit sharing does not match members’ expectations. Here, we examine the changes in household (HH) income and structure, as well as in livelihood activities, experienced by Cambodian frontier villagers living at the site of a planned REDD+ project. During the nine years compared in this study, the frontier villages experienced broad and imbalanced changes in HH income owing to the rapid expansion of the cultivation of cash crops. Our results indicate that benefit sharing or support inevitably becomes more difficult and challenging in frontier areas than in areas where subsistence production systems still predominate, although such frontiers could, in theory, yield maximum returns with regard to forest carbon balance if the REDD+ projects addressed benefit sharing and support and came to fruition.

We enumerated all trees 10 cm or more in DBH with respect to DBH, height, and species identity in 29 circular plots of 20-m radius from Kampong Thom, Kratie, and Mondolkiri Provinces, Cambodia. The composition data were analyzed using cluster analysis with group-averaging protocol, and Sorensen’s similarity index based on basal area data and the resulting clusters were also described with respect to height structure and indicator species. We found four main clusters corresponding to traditional qualitative forest types known as evergreen forest, deciduous forest, hill evergreen forest, and swamp forest. The evergreen cluster was further divided into two stand types of dry evergreen forest and two stand types of secondary evergreen forest. The deciduous forest cluster was divided into three stand types of deciduous dipterocarp forest and a mixed deciduous forest. We describe the correspondence between the forest stand types of our study and the many regional names previously used for the different forest types in varying classification systems. Some of the stand types, for example, an evergreen forest overtopped by deciduous dipterocarp (Dipterocarpus intricatus) or by a pine (Pinus merksii), and a D. obtusifolius stand on seasonally waterlogged habitat, seemed to be unique in Cambodia. The application of this method and the needs of regional forest mapping are discussed.

Forest ecosystem parameters related to the amount of evapotranspiration and rain interception are key elements to successful hydrological modeling. Thus, we evaluated ASTER (Advanced Spaceborne Thermal Emission and Reflection radiometer) reflectance bands and optical indices for qualitative and quantitative estimation of various characteristics of tropical seasonal forests. Ground conditions were measured in 14 sites in Kampong Thom Province, Cambodia, representing six major tropical seasonal forest types: dry evergreen forest, mixed evergreen-deciduous forest, dry deciduous forest, regrowth of dry evergreen forest, moist evergreen forest, and swamp forest. We performed a discriminant analysis to classify forest types using ASTER reflectance bands and optical indices. We used Visible and near infrared Radiometer (VNIR) and Shortwave Length Infrared Radiometer (SWIR) surface reflectance, four vegetation indices: NDVI (Normalized Difference Vegetation Index); SR (Simple Ratio); DVI (Difference Vegetation Index), and MSAVI2 (Second Modified Soil Adjustment Vegetation Index), and three water content indices: SRWI (Simple Ratio Water Index); NDWI (Normalized Difference Water Index); and LWCI (Leaf Water Content Index), for the discriminant analysis. ASTER image products were acquired on January 12, 2002 in the dry season. We also performed regression analyses to identify an optical index closely correlated with forest qualitative characteristics such as tree density, tree height, basal area, and leaf area index (LAI). Each forest type showed a distinctive pattern in reflectance bands, demonstrating that satellite images can potentially be used for regional forest type classification.

Soil moisture conditions were observed in four types of forest in central Cambodia, where dry evergreen forests are distributed widely, to investigate differences of soil moisture in each forest and to clarify relationships between forest types and soil moisture conditions. Observations revealed that soil water contents were high during the rainy season in dry deciduous forest (DDF) and mixed forest (MF), which contained both evergreen and deciduous trees. Those areas have thinner tree crown density and less stand biomass than a dry evergreen forest (DEF). In contrast, during the dry season, water content was low in DDF and MF. That difference is attributable to the disparate evapotranspiration rates of forests caused by the tree crown density and stand biomass. Moreover, soil temperatures were affected by the type of forest. In areas with DDF forests, the temperatures were high in the months of April and May but were lower in MF, DEF, and DEFlog forests. Those differences were caused by inhibition of temperatures through shading effects of tree crowns and evapotranspiration by trees. Based on those observed data, this study clarified a relationship between forest stand type and soil moisture conditions in Kampong Thom forest area.

Changes of soil water conditions in a soil profile were observed and estimated using a one-dimensional vertical soil water movement model for a dry evergreen forest area of Kampong Thom Province, Cambodia. The research site was in a dry evergreen forest where a meteorological observation tower had been established. Soil water matric potentials were measured at 20-, 50-, 100-, 150-, 200-, and 250-cm depths in an observation plot. Groundwater levels were observed at the site. Soil water matric potentials at each observation depth in a soil profile were simulated using a one-dimensional water movement model that was based on Richards’ equation. Results of observations and simulation revealed the following. (1) The site’s water-saturated zone was close to the ground surface during the rainy season. Water conditions in the unsaturated zone, which was above the groundwater level, were influenced strongly by groundwater. The groundwater level was 400 cm deep even in the dry season. The entire soil profile, from the surface to the bottom, never dried completely. (2) At the beginning of the rainy season, at the surface and subsurface depths of 20 cm and 50 cm, respectively, soil matric potentials were increased by rainfall events, which often supply water to the ground surface. Meanwhile, matric potentials at 200 and 250 cm depths were influenced directly by groundwater; they retained high potential values even during the dry season. In the middle zone, at depths of 100 and 150 cm, the soil was quite dried; the minimum matric potentials in the rainy season resembled those of the surface zone. (3) At the beginning of the dry season, matric potentials at 20- and 50-cm depths decreased because of soil water loss by transpiration of trees and evaporation from the ground surface.

Leaf area index (LAI) is a key biophysical variable in most process-based models of forest ecosystems and water cycles. We compared the LAI of two types of tropical seasonal forest in Kampong Thom Province, Cambodia. The two forest types are extremes of crown-cover density, i.e., closed dry evergreen forest (DEF) and open dry deciduous forest (DDF), suggesting marked spatial variation in forest site conditions such as soil moisture. Monthly changes in LAI were estimated indirectly using a plant canopy analyzer and hemispherical photographs. Both methods of LAI estimation showed instrument errors, i.e., low reproducibility in the plant canopy analyzer data and LAI-saturation in hemispherical photograph data; nevertheless, LAI values differed between DEF and DDF. The average LAI from three years of measurements was about 4.6 times higher in DEF than in DDF. DDF exhibited much greater seasonality than DEF. The annual minimum LAI averaged 76% and 84% of the annual maximum LAI for DDF and DEF, respectively. LAI showed high peaks in the rainy season and decreased in the dry season. However, in DEF, LAI decreased twice annually, at the beginning of the dry season and the beginning of the rainy season. Seasonal changes in LAI could be approximated using a third-degree Fourier-series equation.

Some regions of the Mekong River basin still have considerable forest resources, but the pressure for exploiting these resources is very high. Changes of forest cover may strongly affect the water circulation of this region. Therefore, it is important to monitor changes of land cover of this region. In this study, we tested an objectoriented classification method to create a land cover classification map in Cambodia. A commercial object-oriented image analysis software package (eCognition) was used in this analysis. In an object-oriented classification method, the success of classification depends largely on the result of image segmentation. In this study, we overcame the difficulty in image segmentation by combining temporal images acquired in the early and late dry season. The overall accuracy was 0.70, and the Khat statistics value was 0.60. Although the accuracy was moderate, the discrimination between evergreen and deciduous forest types was good. However the mixed or the degraded land cover types were still hard to distinguish from each other. Using images taken in different phenological stages made it possible to both segment the images accurately and classify objects appropriately in an object-oriented classification process.