Robert Zomer^1,4, Chris Carpenter², and Susan L. Ustin³

1.Center for Spatial Technologies and Remote Sensing (CSTARS), Dept. of Land, Air and Water Resources, University of California, Davis, CA 95616

2. Wildland Studies Program, San Francisco State University, College of Extended Learning,

3 Mosswood Circle, Cazadero, CA 95616

3. Dept. of Land Air and Water Resources, University of California, Davis, CA 95616

4. Corresponding author email: rjzomer@ucdavis.edu
 
 

Tables and Figures:

Figure 1             Table 1
Figure 2             Table 2
Figure 3             Table 3
Figure 4             Table 4
Figure 5

Remnant low elevation forest vegetation is found along riparian corridors throughout the Middle Hills of eastern Nepal. These remnant forest patches provide important and essential habitat for an enormous diversity of plants, animals, insects and birds (Jackson et al, 1990). Networks of habitat patches, strung along riparian corridors and their tributaries, function as refugia for numerous rare and threatened species (Shrestha et al, 1990). These riparian networks provide both essential habitat and movement corridors for wildlife (Jackson et al, 1990) and other biota. River corridors and riparian vegetation are subject to disturbance associated with both subsistence and development activities. Within the primarily agricultural landscape matrix of the lower Middle Hills, riparian forests provide important resources for local subsistence agriculturalists, and are integral components of the complex and highly adapted mountain farming system (Ives and Messerli, 1989). Remnant patches are often highly utilized, semi-managed, or selectively cut for useful species. Fodder, fuelwood, medicinal plants and other useful plant materials are regularly harvested on all but the most inaccessible slopes. Domestic animals are also frequently grazed within nearby forest patches. Additionally, the dense cover of riparian forests provides important ecosystem and soil stabilizing functions (Ives and Messerli, 1989; Zomer and Menke, 1992), especially during the torrential downpours associated with the heavy summer monsoon.

Tropical zone forest throughout the Himalaya is threatened and has become increasingly limited in extent (Singh and Singh, 1987, 1992; Jackson et al, 1990). Only a few hectares of these tropical zone hill-variant ecotypes fall within currently protected areas (Hunter and Yonzon, 1993; ICIMOD, 1996). Along the upper Arun River, the Makalu-Barun Conservation Area, a buffer zone area created around the recently designated Makalu-Barun National Park, includes approximately 2000 hectares (Zomer et al, 1998) of remnant tropical zone forests. This current paper gives both a descriptive overview and a quantitative analysis of the community ecology of these forests.

Study Area

The Makalu-Barun Conservation Area (MBCA), established in 1991 by Dept. of National Parks and Wildlife Conservation, H.M.G. Nepal, has been designated as a development-assisted buffer area enhancing conservation efforts within the adjacent Makalu-Barun National Park. An estimated population of approximately 32,000 inhabitants, primarily subsistence agriculturalists, live within the 830 sq. km. of the MBCA. Implementation of the National Park and the MBCA are included in a series of reports and a comprehensive Management Plan (Makalu-Barun Conservation Project Task Force, 1990).

The study area comprises a narrow zone along the upper reaches of the Arun River, within the Makalu - Barun Conservation Area (MBCA) and its immediate environs (Fig. 1). The "tropical" valleys of the upper Arun River Basin (Dunsmore, 1988), referred to as the "world’s deepest valley" by Cronin (1979), lie along the bottom of the steep eastern slope of the Everest massif. Biota along this deep river corridor represent the lower extent of an especially long continuum of highly diverse vegetation and community types (Shrestha, 1989; Shrestha et al, 1990; Byers, 1996). Shrestha (1989) summarizes the bioclimatic zonation commonly used for the eastern Nepal Himalaya region. Within this taxonomy, elevations below 1000 m are generally referred to as the "tropical" bioclimatic zone. Although geographically outside the tropics, this zone is frost free with mean monthly temperature above 18° C throughout the year. Average annual precipitation within this region of eastern Nepal generally is high (4000 mm). However, within the complex mountainous terrain, there is a high spatial and temporal variability associated with the distribution of precipitation. Orographic effects produce significant topo-climatic differentiation throughout the landscape (Schweinfurth 1984,1992), with the tropical zone often receiving substantially less precipitation than zones further up the slope (Shrestha, 1989). In the dry season (November - March), precipitation is sparse and highly variable, with average monthly precipitation totals below 50 mm. Pre-monsoon drought stress within the tropical zone limits distribution of less tolerant plant species (Singh and Singh, 1987, 1992; Shrestha, 1989).

Forest communities of eastern Nepal are diverse and heterogeneous. Early descriptions of the vegetation of this region were recorded by Hooker (1848), in his well known Himalayan Journals and Flora of British India. More recently, forest vegetation of the Arun Basin have been surveyed and described by Hara (1966), Numata (1966,1983), Stainton (1972), Dobremez (1976), Shakya (1977), Ohsawa (1983), Shrestha (1989). Carpenter and Zomer (1996) provides a descriptive overview of the forest ecology of the Makalu-Barun National Park and Conservation Area. Several of these studies include a very limited sampling within the tropical zone of the MBCA, notably Shrestha (1990) and Sugita et al. (1994.)

MATERIALS AND METHODS:

Field Studies

Surveys and field work for this study were carried out between 1991 and 1994 (Carpenter and Zomer, 1996). Forest communities were sampled on a stratified random sampling basis within a variety of representative forest stands, contingent upon accessibility and the difficulties of the terrain.

1.) Structural analysis of the forest was done using 20 x 20 m quadrats (n=30), consistent with earlier work by Shrestha et al. (1990). Site characteristic and vegetation data were collected on forest quadrats below 1000 m along the west side of Arun River within the MBCA (Fig. 1). Both primary, disturbed and secondary forest communities were sampled. Quadrats were inventoried for all woody vegetation with stems greater than 10 cm diameter at breast height (DBH). Percent cover of the understory and canopy was estimated and characterized using ocular estimation, based upon a limited number of within-plot line-transect estimates. Canopy height was estimated using a clinometer. Specimen samples were pressed, and later identified with assistance of the National Plant Laboratory at Godavari i n Kathmandu.

2.) Site parameters including elevation, slope, aspect, and site description were recorded for each quadrat. Elevation was estimated by altimeter as the average of the higher and lower elevation of the quadrat . Slope was measured by clinometer, and aspect was measured by compass. Site description included an estimation of use and impacts by grazing, wood cutting, and fodder collection. The nature and extent of these landuse activities were categorized based upon observation and (when possible) informal interviews. Global Positioning System receivers were used to estimate geographic coordinates.

Community Analysis

1). TWINSPAN ordination (Hill 1979, Gauch and Whittaker 1981, Gauch, 1982) of species abundance data, as estimated by basal area (BA), was used to ordinate groupings of quadrats and species, and to develop a structured two-way table. Rare species (i.e. seldom occurring within the dataset) were not considered within community analysis, so as not to exert undue influences upon relative groupings and separation within ordination space. All identified species, including those seldom occurring, are enumerated in the overall quantitative results. Based upon the TWINSPAN analysis, forest formations and community-types are identified. Forest communities were classified into forest formation types based upon a taxonomy used by Singh and Singh (1987, 1992) for the central Himalaya.

Descriptive statistics, including both physiographic site variables and biophysical parameters, are reported as averaged for each of the community-types. The vegetation data were quantitatively analyzed for abundance, density and frequency (Curtis & McIntosh, 1950). Abundance is estimated as both basal area (BA) per hectare and as relative basal area (rel. BA), and is reported as averaged for all plots within the respective community (avg. BA and avg. rel. BA). An Importance Value Index (IVI) was calculated for all species within each of the TWINSPAN delineated groups, and is reported as averaged for each of the community-types. IVI's are calculated as the sum of relative density, relative basal area, and relative frequency of occurrence (Curtis, 1959), and allow estimation of importance for a particular species within a community not based solely upon dominance. Species diversity, as measured by the Shannon -Wiener Index (Shannon and Weaver, 1949), was estimated for all plots, and is reported as averaged for each of the respective TWINSPAN classification groups.

2.) Direct gradient analysis of community structure along the elevational gradient is approximated through a modeling of species distribution based upon a Gaussian species response model (Gauch & Whittaker 1972, 1976). Species abundance is represented as IVI for species within their respective communities, and plotted along an elevational gradient. Species abundance along an environmental gradient is assumed, for the purposes of modeling of community structure, to generally form a bell-shaped, unimodal distribution curve approximating the Gaussian (normal) curve (Gauch, 1982). The Gaussian equation is:

     Y = Y₀exp [-(x-m)²/2s²]

where m represents the mode, X represents the position along the environmental gradient, Y the species’ IVI, Y0 is the maximum value, and s is the dispersion in units of standard deviation. Gaussian distribution curves are fitted for the dominant and co-dominant species (as ranked by IVI) for each of the five TWINSPAN delineated community groups, and plotted along the elevational gradient.

3.) Population structure of tree species within forests can reveal information about stand dynamics (Metz, 1997) and regeneration status (Saxena and Singh, 1984). In the absence of age data for stands, population is approximated based upon analysis of tree size (DBH) classes. To estimate the population structure of each tree species, dbh classes were delineated at 10 cm intervals. Total number of individuals belonging to each of the classes is calculated for major species in each community type. Comparative size class distribution of dominant and co-dominant species (as ranked by IVI) within each of the five TWINSPAN delineated communities is presented as bar graph distributions.

4.) Canonical correspondence analysis (CCA), a direct ordination technique, was

used to explore the relationship of the vegetation and community structure to recorded environmental variables (ter Braak, 1987, 1990). Spatial relationships of both species and samples (quadrats) within a multi-dimensional ordination space are illustrated in relation to vectors representing statistically significant environmental variables. Environmental variables are tested for statistical significance within the CCA analysis based upon Monte Carlo statistical techniques and forward selection of variables (ter Braak 1990). Variables included in the CCA analysis are: elevation (ELEV), slope angle (SLOPE), slope position as three nominal variables (BOTTOM, MIDSLOPE, RIDGE), use intensity level (USE), and aspect. Aspect was converted to two continuous variables based upon a percent scale, for purposes of quantitative and statistical analysis. Sun-aspect (SUNASP) roughly represents relative solar exposure, assuming southern exposures receive the most solar radiation, and northern exposures the least. Likewise, east-aspect (EASTASP) roughly represents the influence of morning sun (or afternoon clouds), and is calculated in a similar manner, with eastern aspects (90°) hypothesized to receive more solar radiation (100%) than western exposures (0 %), due to cloudy afternoons associated with the orographic effects of adiabatic valley winds. Use level (USE) was subjectively rated on a one to three scale ranging from "least disturbed" to "most disturbed", based upon recorded site observations. Results of the CCA analysis are presented in biplots, showing ordination results for species and samples. Samples are designated by their TWINSPAN based classification group to facilitate the comparative analysis. Environmental vectors were scaled to allow a comparative presentation with sample and species biplot scores.

COMMUNITY ECOLOGY: TWINSPAN CLASSIFICATION

Forest vegetation of the Himalaya has been described by Singh and Singh (1987), and classified into eleven major formation types. Based upon this classification, three forest formations are identified within the tropical zone of the MBCA:

1.) Sub-Montane Semi-Deciduous Broadleaf Seasonal Rain Forest, dominated by the diptocarp Shorea robusta (Sal), is found intermixed with various associates, including palms, cycads, tree ferns, bananas, and other tropical elements, below 800 m .

2.) Low Montane Semi-Deciduous Needleleaf Forest, dominated by Pinus roxburghii (Chir Pine), is relatively scare in the Arun Basin (Shrestha, 1989). It is limited within the study area to a relatively xeric site above the confluence of the Sankuwa and Arun rivers.

3.) Low-Montane Evergreen Broadleaf Seasonal Rain Forest, is prevalent throughout the Arun Basin, and is characteristic of the subtropical bioclimatic zone. Lower ecotonal variants of this formation are found within the MBCA to well below 600 m and form an important component of the vegetation of the tropical zone.

Within these three formations, five forest community types are evident from the TWINSPAN ordination and community analysis (Table 1). Physiographic and biophysical characteristics associated with forest formation and community types, and a summary of IVI parameters for each species within each of the communities, are given in Table 2, Table 3, and Table 4 respectively, and are discussed below for each of the TWINSPAN delineated community types.

Sal Forests: (Sub-Montane Semi-Deciduous Broadleaf Seasonal Rain Forest)

Stands found within the Conservation Area are upper ecotonal, and exhibit a lower canopy dominance (avg. rel. BA. 64%) by Shorea robusta than the nearly monospecific stands of this forest type found lower in the Arun Basin (Carpenter and Zomer, 1996), or the Terai (Wesche, 1997). S. robusta within this formation tends to be an emergent canopy species, with locally varying subdominants. Common canopy associates include Schima wallichii, Sysguim cumimi, Lagerstromia parviflora, and several Termalia species. The tree layer and shrub layer consist of a mixture of evergreen and deciduous species. S. robusta continues up to approximately 1200 m, where it intergrades intoP. roxburghii dominated forest on drier south-facing slopes. Of the common associates, only S. wallichii displayed moderate IVI (42) and abundance within plots (avg. rel. BA = 13.7%). Understory cover tends to be sparse (23%), with canopy remaining fairly dense (48%), despite evidence of lopping. The middle tree layer is relatively less developed, and bamboos are generally absent. Species richness was fairly high, with 17 woody species. A moderately low Shannon Diversity Index of 1.05, as well as low IVI for associated species, reflects the dominance of S. robusta (IVI 139) within this community. These stands are generally heavily used, with only small patches of "natural" stands remaining, due to excessive tree cutting, wood collecting, and livestock grazing (Jackson et al., 1990).

Chir pine: (Low-Montane Seasonal Semi-Deciduous Needle-Leafed Forest)

Low-montane seasonal semi-deciduous needle-leaf forest is found all along the Himalaya. In the Arun Basin, this formation is dominated by Pinus roxburghii, a large conifer locally known as Chir Pine. P. roxburghii is a drought tolerant pioneering species which colonizes disturbed, burned or degraded sites, and not generally considered a climax forest formation type (Singh 1987). Within the Arun Basin, its distribution is discontinuous, possibly due to mesic conditions, with stands limited to the sunnier and drier slopes (Dobremez, 1976, Shrestha 1989). The ridge going up from the confluence of the Arun and Sankuwa rivers, at the southeastern corner of the MBCA, supports a contiguous stand of P. roxburghii in association with S. robusta, S. wallichii, and other evergreen subtropicals. This relatively xeric site has an elevation range from 700 to 1200 m and an average slope angle of 24 °. Canopy cover of the dominant overstory is relatively sparse with a moderately low density of stems (175 per ha.). P. roxburghii strongly dominates this community, with a high IVI (114) and an avg. rel. BA of 57%. S. robusta was found in all plots as a strong sub-dominant, with a avg. rel. BA of 20% . Canopy height for the P. roxburghii overstory averaged 23 to 28 m. Understory is fairly sparse (41%), especially in the shrub layer, with many of these sites having a thick mat of undecomposed pine needles covering the forest floor. Other sites within this stand, possibly more recently burned, are carpeted with bracken ferns. Use levels for this forest type were relatively low (1.37). Species richness was relatively low, with only 9 woody species, and a moderately low Shannon Diversity Index of 1.15. Associate species had very low IVI values, reflecting the strong dominance by P. roxburghii. Although spatially limited, this stand represents an important potential seed source for expansion of this forest type onto nearby disturbed sites within the Conservation Area.

Low-Montane Evergreen Broadleaf Seasonal Rain Forest

The low-montane evergreen broadleaf seasonal rain forest formation of eastern Nepal is commonly referred to as Schima-Castanopsis forest (Singh and Singh, 1987; Shrestha, 1989), Schima, or locally aschilaune . Characteristic of the subtropical bioclimatic zone, Schima forests are now difficult to find in a pristine state within the MBCA (Cronin 1979; Shrestha, 1989). These riverine subtropical evergreen forest types occur as narrow bands along the Arun and its major streams (Jackson et al 1990). Stands are linear, and are limited to the narrow corridors, ravines, gorges, and/or slopes in excess of 40°, which are not easily accessible to people. Although species rich, diversity tends to be higher in sites also inaccessible to livestock (Shrestha et al., 1990). Shrubs, nettles, climbers, and bamboo generally form a dense understory. Mixed stands of Schima, Engelhardtia, or Duabanga are found interspersed with Pandanus, Musa, Albizzia and Persea. These evergreen forests extend upward along mesic riparian corridors to the upper limit of the subtropical zone, and constitute important habitat and movement corridors through otherwise heavily-disturbed monotypic and cover-poor landscapes (Jackson et al., 1990).

Three sub-types of this formation are identified by TWINSPAN in the study area, and are designated below by the genera of their respective dominant species. Schima wallichii was not used as an indicator species, as its presence is ubiquitous within all the sampled communities

Engelhardia: (Low-Montane Evergreen Broadleaf Rain Forest)

Although characteristic for both the formation and this community type, S. wallichii exhibits only a moderate dominance (avg. rel. BA 22%, IVI 47) in this highly mixed assemblage. Associated co-dominants included Engelhardia spicata (avg. rel. BA 18%), Leucosceptrum cannum (avg. rel. BA 9%) and Castanopsis indica (avg. rel. BA 9%). In addition, several important fodder species are also present, e.g. several Ficus species. Alnus nepalensis (avg. rel. BA 7%) is an important pioneer species and generally indicative of disturbance. This forest type exhibited the highest species richness of the TWINSPAN groups, with 25 woody tree species (dbh 10 cm) found in more than one plot. The relatively high Shannon Diversity Index (1.53) reflects both the species richness and species evenness of this community type, without clear dominance. Understory and shrub cover is fairly thick where present (56%) with a clustered overstory canopy (avg. 43%), reflecting the rockiness of many of these sites. This community was found on the steepest slopes (avg. slope 35°).

Albizzia: (Low-Montane Evergreen Broadleaf Rain Forest)

Albizzia mollis predominates in these samples (avg. rel BA 27%), however, diversity is high. With an average of 11 species per plot, and 15 species in all plots, both species richness and evenness are moderately high. Important associates include S. wallachii (avg. rel. BA 12%), Persea odoratissima (avg. rel BA 12%) and Sapium insigne (avg. rel BA 6%). Lindera pulcherrima (avg. rel BA 3%), Bombax ceiba (avg. rel BA 1%), and Adina cordifolia (avg. rel BA 1%) are present as minor species. This community type was found to have the highest Shannon Diversity Index within our study (1.88, s.d. 0.22), and had the highest biomass of the TWINSPAN groups, as approximated by avg. BA (47 m² /ha).

Duabanga: (Low-Montane Evergreen Broadleaf Rain Forest)

This inner riparian zone community type is strongly dominated by Duabanga grandiflora (avg. rel. BA 76 %, IVI 159). Associate species include Albizzia spp. (avg. rel. BA 14 %), Terminalia myriocarpa (avg. rel BA 3.0), and C. indica (avg. rel. BA 1%). Understory cover was dense with a thick shrub layer, despite a relatively high avg. canopy cover (53%). This community is generally found along the bottom of moist ravines and gullies, and shadier sites.

POPULATION STRUCTURE AND FOREST DYNAMICS

Within the Chir Pine community, P. roxburghii exhibits a DBH size-class distribution (Fig. 2) indicative of an established pioneer, successional community, with fewer small trees, and a markedly greater number of larger diameter trees. Although more recently reproduction may be tapering off, or perhaps inhibited by disturbance, e.g. grazing or fire, the relatively even size-class distribution generally indicates on-going reproduction. Nevertheless, it is assumed that this community, which is confined to the southeastern portion of the Conservation Area, was most likely was established fairly recently as a result of a disturbance, probably a fire. Chaturvedi and Singh (1987) report that it may take 60 to 80 years to reach its current canopy height. However, selective cutting of poles may account for an absence of stems between 30 and 40 cm dbh. Unlike most pine, P. roxburghii has a certain ability to coppice (Jackson, 1987), which may also affect the size distribution. The co-dominant S. robusta displays a size class distribution indicative of a competitive successional species, with a few established larger trees and many smaller diameter trees. Excluding mitigating circumstances, i.e., heavy use, grazing and frequent fires, it might be expected that this community will continue increasing the proportion of broadleaf species, particularly S. robusta.

In contrast, within the Sal community, S. robusta displays a size-class distribution representative of a competitive dominant, within an established climax community, likely to continue as a strong dominant in the future. A "reverse J" distribution, with few trees distributed among the larger size classes, and many smaller diameter trees, suggests shade tolerance and continuous recruitment (Sugita et al, 1994). However, as this community is generally subject to heavy use, this distribution may also display evidence of selective pressure (Sundriyal, et al., 1994) on this multipurpose species, particularly lopping of larger trees. A similar distribution for the co-dominant S. wallichii suggests that this mixed broadleaf combination is likely to continue to co-dominate this community, with relatively the same proportions in the future. An absence of larger stems suggests S. wallichii may benefit from canopy gaps or more recent disturbance of the S. robusta overstory, to establish emergent specimens. Growth of middle story trees was also likely inhibited by a formerly denser overstory canopy.

Likewise, S. wallichii, within the Engelhardia community of the subtropical evergreen hill forest formation, exhibits a smooth decline of the number of trees within size classes with an increasing diameter size. Although not a "reverse J" distribution, it appears likely that S. wallichii will remain a dominant within this mixed broadleaf community. Engelhardia spicata, with an even-aged size class distribution, appears to maintain a relatively steady low frequency occurrence as co-dominant. Albizia mollis, as the dominant species of the Albizia community, shows a even-sized distribution with at least one particularly large specimen, indicating that this is a long established community. Reproductive success is indicated as infrequent but successful over time. The co-dominant S. wallichii appears to have been established as a cohort, perhaps a result of gap phase dynamics.

A strong dominance by Duabanga grandiflora is likely to continue within the Duabanga community type. Establishment by the co-dominant, Albizia procera, appears infrequent. Its size class distribution suggests an increasing importance of this species. However, D. grandiflora appears to be particularly adapted to this mesic inner riparian habitat, developing both dominance and constancy.

ENVIRONMENTAL DETERMINANTS OF COMMUNITY COMPOSITION

Although microclimate and habitat differentiation strongly mitigate elevational influence (Schweinfurth, 1984), vertical zonation of plant communities along an altitudinal gradient can be strongly evident in steep terrain (Troll, 1958; Beals, 1969). Distributional patterns for dominant and co-dominant species of the five described community types are modeled along an elevational gradient, based upon a Gaussian distribution of species IVI values (Fig. 3). Below 600 m, mixed Sal forests dominate. However, the elevational range for S. robusta continues to 1200 m, well up into the subtropical. High importance values for P. roxburghii are confined to a relatively narrow zone between 600 m to 900m, although this species can also be found up to 1200 m intergraded with S. robusta and S. wallichii. Both D. grandiflora and E. spicata are found at the upper portion of the elevational range of the tropical zone, indicating their range likely continues up into the subtropical. Castanopsis indica is present, but appears to be at the lower extent of it’s potential range. A strongly individualistic species’ response to environmental gradients is apparent, as is continuous change in community composition along the elevational gradient. However, elevation alone is clearly insufficient to explain distribution of assemblages of co-occurring species.

Direct ordination (CCA) of species composition clearly separated the TWINSPAN delineated communities along measured topo-climatic gradients (Fig 4). Five measured environmental variables display significant relationships (p _ 0.05) with community species composition, based upon Monte Carlo estimation (ter Braak, 1987,1990). The summary of Monte Carlo test results showed significance both for the first canonical axis (F-ratio = 3.95; p-value _ 0.01) and the overall test (F-ratio = 2.53, p _ 0.01) of the ordination. As to be expected in steep topography, elevation exerted the strongest influence (ELEV, p _ 0.01), both on community composition and individual species. abundance. However, both southern aspect (SUNASP, p _ 0.01) and eastern aspect (EASP, p _ 0.05) are important determinants mitigating the pronounced elevational gradient. The effect of increasing EASTASP is inverse to increasing elevation. Slope position showed a strong spatial correspondence with distribution of forest types within the ordination biplot, particularly separating out P. roxburghii forest along ridge tops (RIDGE, p. _ 0.0X) and riparian forest (BOTTOM, p. _ 0.01), with the remaining forest types associated with the mid-slope (MIDSLOPE, p. _ 0.01) position. The percent slope showed no significance in relation to distribution, reflecting the generally steep terrain associated with remaining forest (average slope for all communities 23°). In addition, disturbance (USE, p. _ 0.03) is shown within the ordination to exert a strong influence on species composition. Both the Sal forest and the Duabanga communities are positively associated with the higher use levels, while the Chir pine is negatively associated.

Individual species associated with the respective forest types were distributed within the ordination biplot of species scores (Fig. 5) in a manner generally similar to the distribution of the TWINSPAN groups (i.e. the sample quadrats as delineated by their respective TWINSPAN classification). Both of the drier low montane evergreen forest types i.e. those dominated by S. wallichii or Albizzia spp., are found positively associated along the sun-aspect ordination gradient. Both the Sal and the Duabanga (riparian low montane) community were strongly positively associated with east-aspect. Individual species, many of which occur in more than one forest type, differed markedly in the strength of association with the environmental gradients. S. robusta for instance is substantially less strongly associated with east-aspect then the Sal forest type. Eurya acuminata, on the other hand, showed a definitively strong association with east-aspect, as well as being strongly associated with elevation and use.

DISCUSSION

Separation of communities along the significant environmental gradients within the CCA ordination illustrates a general agreement with the TWINSPAN delineated community groupings. Although both TWINSPAN and CCA are calculated using a similar divisive algorithm (Gauch, 1982, ter Braak, 1987), a clear separation of the TWINSPAN communities along the constrained environmental axes of the CCA validates the dependence of community composition on the measured environmental variables. Thus, the limited number of environmental variables used within the analysis were shown to be sufficient (at the scale of the current study) to explain the distribution of these communities, both within the ordination and the landscape. Elevation, although by far the major controlling factor, was highly mitigated by both slope-position and aspect. Although various authors have suggested that aspect was less important within the warmer tropical zone (Shrestha, 1989), a clear separation of communities along both the sun-aspect and east-aspect axes is evident. Sal forest in particular are associated with the sunnier eastern and south-eastern aspects. The drought tolerance of S. robusta allows Sal forest to compete and thrive on these sunnier aspects prone to pre-monsoon drought stress. The Duabanga community show a similar affinity for sunny aspects, however is strongly associated with the more mesic inner riparian zone of the bottom slope position. Chir pine, on the contrary, is associated with both sunny aspects and the xeric ridge slope position.

Anthropogenic disturbance, as measured along the disturbance gradient, is clearly affecting species composition of many of these communities. Use levels have a significant effect, both on species composition and age structure, and contributes to a highly heterogeneous mosaic of successional states within these forests. Disturbance tolerant species, such as Eurya acuminata or Schima wallichii, often become relatively more abundant within highly fragmented landscapes. Particularly useful, or vulnerable, species can become atypically rare if heavily impacted (Laurance et al., 1997), e.g. Castanopsis indica. Grazing has most likely had particularly strong effects on species composition, as browsing livestock tend to favor more palatable species, or tender saplings. Lopping for fodder and firewood, or cutting for timber, appears to impact most heavily on dominant, or late successional species, i.e. individuals with high biomass. These practices have had significant effect on canopy cover and crown densities within all of the delineated communities. Lower and middle stories within the study zone tend to be relatively dense as a result of the increased light levels associated with canopy disturbance. Gaps in canopy cover also tend to facilitate the growth of early successional or shrub species in the understory. In addition, fragmentation and a small average patch size (Zomer et al., 1998) is an important component affecting species composition and community structure (Laurance et al., 1997). Patches are often very small, and margins adjoining either agricultural or scrub lands can be abrupt, leaving a high ratio of exposed edge and very little core area.

Tropical zone forest communities found along the upper Arun were found to be both diverse, and regionally unique, assemblages of high species and local diversity. When compared with studies from other nearby regions of both the Himalaya in general (Singh and Singh, 1987) and eastern Nepal in particular (Dobremez, 1976, Sugita et al. 1994), forest communities within the tropical zone of the MBCA differ significantly, in terms of both species composition and community structure. The high regional diversity associated with the physiographically complex and diverse landscape of the eastern Himalaya (Dobremez, 1976) emphasizes both the botanical uniqueness and conservation value of these remnant forests. Beyond their important role as repositories of a highly threatened biological diversity, these forests have essential functional roles within the landscape ecology of the tropical riparian corridor. In addition, remnant tropical zone forests throughout the eastern Mid-Hills contribute substantially to agricultural and rural welfare. Preservation of these forests, both within the MBCA, and within the Himalayan region more generally, presents a unique and pressing conservation challenge.

ACKNOWLEDGMENTS

We wish to extend thanks to the many people who assisted with our field work, especially scientists and staff affiliated with the Makalu-Barun National Park and Conservation Area and The Mountain Institute (formerly the Woodlands Mountain Institute) in Nepal: Dr. Nanda Joshi, Narayan Poudrel, Ang Rita Sherpa, Dr. Alton Byers, Dr. Gabriel Campbell and Robert Davis have provided invaluable advice and logistical support both in Kathmandu and in the field.

Teaching Assistants from Tribhuvan University have accompanied us in the field and have proved themselves a major asset to our research effort. Special thanks to Suresh Ghimire, Birendra Karna, Ek Raj Sigdel, Udhav Khadka and Kiran Dongol. Nabin Acharya of the National Herbarium, Godavari and Dr. Thirta Bahadur Shrestha at IUCN in Kathmandu have given their time to share with us their botanical knowledge and insights into conservation in Nepal.

Our work would not have been possible without the tireless and dedicated help of our support team in the field. Special thanks to Laxmi Dewan and the staff at Nilgiri Treks for the enthusiasm they have shown for our project, their willingness to share their knowledge of biological diversity and the efficiency and skill they bring to the field work. We acknowledge a great debt of gratitude to all of the students who have worked with us on the Wildland Studies Nepal Program during the past two years, sunshine and rain. Many thanks to Marcel Rejmanek and Jack Ives for their advice and manuscript review.

References:

Beals, E. 1969. Vegetational change along altitudinal gradients. Science 165: 981-985.

Byers, A. 1996. Historical and Contemporary Human Disturbance in the Upper Barun Valley, Makalu-Barun National Park and Conservation Area, East Nepal. Mountain Research and Development 16 (3): 1996.

Carlson, A., Chapel, M., Colborn, A., et al. 1991. Review of Literature Addressing Wildlife and Fish Habitat Relationships in Riparian and Stream Habitats. Tahoe National Forest, Old Forest and Riparian Habitat Planning Project..

Chaturvedi, O.P. and Singh, J.S. 1987. The Structure and Function of Pine Forest in Central Himalaya. I. Dry Matter Dynamics. Annals of Botany 60: 237-252.

Carpenter , C. and Zomer, R. 1996. Forest Ecology of the Makalu-Barun National Park and Conservation Area, Nepal. Mountain Research and Development. 16 (2): 1996.

Cronin, E.W. 1979. The Arun: A Natural History of the World’s Deepest Valley. Houghton Mifflin Co., Boston.

Curtis, J.T. 1959. The Vegetation of Wisconsin, An Ordination of Plant Communities. University of Wisconsin Press, Madison.

Curtis, J.T. and McIntosh, R.P. 1950. The inter-relations of certain analytic and synthetic phytosociological characters. Ecology 31: 476-496.

Dobremez, J.F. 1976. Le Népal: Écologie et Biogéographie. Centre National de la Recherche Scientifique, Paris.

Dunsmore, J.R., 1988. Mountain Environmental Management in the Arun River Basin of Nepal. ICIMOD Occasional Paper No. 9. Kathmandu.

Gauch, H.G. 1982. Multivariate Analysis in Community Ecology. Cambridge Studies in Ecology. Cambridge University Press. New York.

Gauch, H.G. and Whittaker, R.H. 1972. Coenocline simulation. Ecology 53: 446-451.

Gauch, H.G. and Whittaker, R.H. 1976. Simulation of community patterns. Vegetatio. 33: 13-16.

Gauch, H.G. and Whittaker, R.H. 1981. Heirarchical classification of community data. Journal of Ecology . 69: 135-152.

Hara, Hiroshi, 1966: The Flora of the Eastern Himalaya: Results if the botantical expedition to the Eastern Himalaya, organized by the University of Tokyo Press.

Hill, M.O. 1979. TWINSPAN - A FORTRAN Program for Arranging Multivariate Data in an Ordered Two-Way Table by Classification of the Individuals and Attributes. Cornell University, Ithaca, N.Y.

Hooker, J.D. 1852. On the climate and vegetation of the temperate and cold regions of East Nepal and the Sikkim mountains. J. Hort. Soc. London 7: 69-131.

Hunter, M. L. Jr. and Yonzon, P. 1993. Altitudinal Distributions of Birds, Mammals, People, Forests, and Parks in Nepal. Conservation Biology 7(2): 420-423.

ICIMOD . 1996. GIS Database of Key Indicators of Sustaninable Mountain Development in Nepal. International Center for Integrated Mountain Development (ICIMOD), Kathmandu.

Ives, J.D. and Messerli, B. 1989. The Himalayan Dilemma: Reconciling development and conservation. United Nations University. Routledge, London.

Jackson, J.K., 1987. Manual of Afforestation in Nepal. Nepal-United Kingdom Forestry Research Project Forest Survey and Research Offic. Dept. of Forest, Kathmandu, Nepal.

Jackson, R., Nepali, H.S., and Sherpa, A. R. 1990. Aspects of Wildlife Protection and Utilization in the Makalu-Baru n Conservation Area. Report 11, Working Paper Publication Series, Makalu-Barun Conservation Project, Dept. of National Parks & Wildlife Conservation, H.M.G. Katrhmandu, Nepal.

Laurance, W.R., Bierregaard, R.O. Jr., Gascon, C., et al. 1997. Tropical Forest Fragmentation: Synthesis of a Diverse and Dynamic Discipline. In: Laurance, W.R. and Bierregaard, R.O. Jr. (eds.) Tropical Forest Remnants: Ecology, Management, and Conservation of Fragmented Communities. University of Chicago Press. Chicago.

Makalu-Barun Conservation Project Task Force: 1990 The Makalu-Baurn Conservation Project Managment Plan. Makalu-Barun Conservation Project, Kathamndu .

Metz, J. J. 1997. Vegetation Dynamics of Several Little Disturbed Temperate Forests in East Central Nepal. Mountain Research and Development. 17 (4): 333-351.

Numata, M. 1966. Vegetation and Conservation in Eastern Nepal. Jour. of College of Arts and Sciences, Chiba University 4: 559-569.

Numata, M. 1983. Ecological Studies in the Arun Valley, East Nepal and Mountaineering of Mt. Baruntse, 1981. Himalayan Comm. Chiba University.

Oshawa, M., 1983. Distribution, Structure and Regeneration of Forest Communities in Eastern Nepal. Chiba University, Japan. pp. 89-120.

Saxena, A.K. and Singh, J.S. 1982. A phytosociological analysis of woody species in forest commnities of a part of Kumaun Himalaya. Vegetatio . 50 : 3-22.

Schweinfurth, U. 1984, The Himalaya: Complexity of a Mountain System Manifested by Its Vegetation . Mountain Research and Development, 4 (4): 339-344.

Schweinfurth, U. 1992. Mapping Mountains: Vegetation in the Himalaya. GeoJournal 27.1: 73-83.

Shakya, P.R. 1977. Vegetation Study in East Nepal. Report of Ecological Studies of Mt. Makalu II. Dept. of Medicinal Plants, H.M.G. Nepal, Kathmandu.

Shannon, C.E. and Weaver, W. 1949. The Mathematical Theroy of Communication. University Illinois Press, Urbana, Illinois.

Shrestha, T.B. 1989. Development Ecology of the Arun River Basin in Nepal. International Centre for Integrated Mountain Development, Kathmandu .

Shrestha, T.B., Shakya, P.R., and H.S. Nepali. 1990. Scientific Report on 1989 Field Survey: General and Phyto_Ecology. Report 8, Working Paper Publication Series, Makalu-Barun Conservation Project, Dept. of National Parks & Wildlife Conservation, H.M.G. Katrhmandu, Nepal.

Singh, J.S. and Singh, S.P. 1987. Forest Vegetation of the Himalaya. The Botanical Review. 53(1): 80-192.

Singh, J.S. and Singh, S.P. 1992. Forests of the Himalaya: Structure, Functioning and Impact of Man. Gyanodaya Prakashan, Nainital, India. Fine Art Press. Delhi.
 
Stainton, J.D.A. 1972. Forest of Nepal. John Murray and Co., London.

Sugita, H., Subedi, M.N., Yonebayashi, C., Takayam, H., and Minaki, M. 1994. Floristic Composition and Structures of Forests in the Arun Valley, Eastern Nepal. Ecological Review. 23 (1): 17-42.

Sundriyal, R.C., Sharma, E., Rai, L.K. and Rai, S.C. 1994. Tree structure, regeneration, and woody biomass removal in a sub-tropical forest of Mamlay watershed in the Sikkim Himalaya.

ter Braak, C.J.F. 1987. Ordination. In: Jongman, RHG., ter Braak, C.J.F. and van Tongeren, O.F.R. (eds.) Data Analysis in Community and Landscape Ecology. PUDOC, Wageningen, 91 - 173.

ter Braak, C.J.F. 1990. CANOCO - a FORTRAN program for Canonical Community Ordination: Update Notes. Agricultural Mathematics Group, Wageningen, T%he Netherlands.

Troll, C. 1958. Tropical mountain vegetation. Proc. 9th Pac. Sci. Cong. 20: 37-45.

Wesche, K. 1997. A classification of a tropical Shorea robusta forest stand in southern Nepal. Phytocoenologia. 27 (1): 103-118.

Zomer , R. and Menke, J. 1992. Site Index and Biomass Productivity Estimates for Himalayan Alder-Large Cardamom Plantations: A Model Agroforestry System of the Middle Hills of Eastern Nepal. Mountain Research and Development 13(3): 235-255.

Zomer, R., Ustin, S. and Ives. 1998. Land Cover Change Along Tropical and Subtropical Riparian Corridors Within the Makalu-Barun Conservation Area, Nepal. Mountain Research and Development. Accepted