Assessment of tree diversity and above-ground biomass in coffee agroforest dominated tropical landscape of India’s Central Western Ghats

Guddappa M. Devagiri · Anil Kumar Khaple · Hosuru B. Anithraj ·Cheppudira G. Kushalappa · Amaresh Kumar Krishnappa · Shashi Bhushan Mishra

Abstract The purpose of this study was to quantify the changes in tree diversity and above-ground biomass associated with six land-use types in Kodagu district of India’s Western Ghats. We collected data on species richness,composition and above-ground biomass (AGB) of trees,shrubs and herbs from 96 sample plots of 0.1 ha.Totals of 83 species from 26 families were recorded across the landuses.Tree species richness,diversity and composition were significantly higher in evergreen forest(EGF)than in other land-uses. Similarly, stem density and basal area were greater in EGF compared to other land-uses. Detrended correspondence analysis (DCA) yielded three distinct groups along the land-use intensities and rainfall gradient on the first and second axes, respectively. The first DCA axis accounted for 45% and second axis for 35% of the total variation in species composition. Together the first two axes accounted for over 2/3 of the variation in species composition across land-use types. Across the land-uses,AGB ranged from 58.6 Mg ha-1 in rubber plantation to 327.3 Mg ha-1 in evergreen forest.Our results showed that species diversity and AGB were negatively impacted by the land-use changes. We found that coffee agroforests resembled natural forest and mixed species plantation in terms of tree diversity and biomass production, suggesting that traditional coffee farms can help to protect tree species,sustain smallholder production and offer opportunities for conservation of biodiversity and climate change mitigation.

Keywords Western Ghats · Coffee agroforest · Land-use changes · Diversity · Above-ground biomass · Carbon

Introduction

India’s Western Ghats landscapes are unique mosaics of natural forests interspersed with agricultural lands, agroforests, coffee and tea plantations, mono-culture plantations and various other tree-based production systems which are known to be species diverse terrestrial ecosystems (Bhagwat et al. 2008). Forests of Western Ghats contain the diverse plant communities,with up to 350-400 tree and liana species coexisting in a single hectare (Murthy et al.2016).In addition to their rich biodiversity,these forests are also natural sinks of carbon, with sequestration potential of 80-150 Mg C ha-1(Devagiri et al. 2013).

Among the current environmental challenges, biodiversity loss and climate change due to habitat destruction and fragmentation are of great concern. Historically, the single most important factor for loss of biodiversity is change in land-use practices (Dembele et al. 2006; Pekin and Pijanowski 2012). Land-use changes may eliminate species locally and degrade the functioning of natural habitats and ecosystems,thereby affecting biodiversity and provision of ecosystem services (Gerstner et al. 2014; Barnes et al.2014;Phillips et al.2017).Gibson et al.(2011)showed,in a global meta-analysis, that the impact of tropical forest disturbance on biodiversity was more severe in Asia than in Africa, South America and Central America. Similarly,evidence of loss of ecosystem services owing to land-use changes is gradually accumulating,especially in the case of the carbon cycle. Since 1850, roughly 35% of anthropogenic CO2emissions resulted directly from land use changes (Turner et al. 2007; Carlson et al. 2013). Conversely, it has been argued that the future of tropical biodiversity depends on the successful integration of rural livelihoods with conservation efforts in human-modified landscapes such as plantation forests and coffee or cocoa based agroforests(Chazdon et al.2009).While state owned primary or secondary forests remain crucial for conserving tropical biodiversity, still a variety of species, plant communities and ecosystems occur outside protected areas, on lands under diverse land tenure systems and practices(Schroth et al. 2004). For example, CAFNET (2011)reported 280 tree species from shaded coffee agroforests in Western Ghats of India,more species richness than the 188 tree species in natural forest ecosystems. Similarly, other studies have established that agroforestry systems can increase the diversity of plants, vertebrates and arthropods(Clough et al. 2010; Dawson et al. 2013). Agroforests can share large proportions of species with adjacent natural forests(Bhagwat et al.2008; Hager et al.2015).Estimates of the potential of agroforests in Africa in terms of aboveground biomass carbon range from 1-18 Mg C ha-1(Nair and Nair 2014). In some coffee growing regions of Indonesia, shade coffee agroforests, depending on shade tree density and composition, can store up to 75% of the above-ground carbon stored in the adjacent remnant forests(Kessler et al. 2012). Understanding of the pattern of tree diversity and above-ground biomass (AGB) distribution in human modified landscapes can inform conservation planning and climate change mitigation strategies. However, very little is known about the species composition and their contribution to above-ground biomass across different land-uses such as natural forests, coffee, rubber,teak and other mixed species plantations in Western Ghats.Few studies of natural forests have reported the quantitative assessment of floristic diversity, biomass carbon and dynamics (Swamy et al. 2010; Rajashekar et al. 2018).

This study was undertaken in Kodagu district to test the hypotheses that: (1) floristic diversity and AGB vary among six land-use types; (2) the highest diversity and AGB would be recorded in evergreen forests which enjoys greater protection; and (3) diversity and AGB would decline with increasing land-use intensity and disturbances in moist deciduous forest, coffee agroforests and tree plantations. To test these hypotheses we aimed at three important objectives to (1) compare variation in tree species diversity, assemblages, and dominance between natural forests and production landscapes such as coffee agroforests, mixed and monoculture plantations (2) quantify the differences in basal area and AGB between landuse types.

Materials and methods

Study site

This study was conducted in forest-coffee agroforest landscape mosaics of Kodagu district in the Central Western Ghats region (70°25′-76°14′E and 12°15′-12°45′N)covering an area of 4106 km2of which about 38%is under natural forest and tree plantations. Evergreen and moist deciduous forest types were predominant in the district.Mono-culture plantations dominated by Tectona grandis,rubber plantations dominated by Hevea brasiliensis and mixed species plantations were the predominant types of tree plantations in the region. Shade grown coffee plantations cover 33%of total area of the district complimenting the forested landscapes. The district has a steep West to East climatic gradient for temperature and rainfall from the edge of the ghats. The study area elevations ranged from 300 to 1300 m above sea level (a.s.l) and average annual rainfall ranged from 1500 to 3500 mm with maximum rainfall during monsoon season(June to September). April and May have the highest mean maximum temperature(32 °C), while December and January have the lowest mean minimum temperature (15 °C). Soils are lateritic to red loamy with a mature profile and the main rock formation is of the most ancient Archaean system with rock composed of peninsular gneiss, gneissic granites, and gneiss.

Data collection and sampling design

Six land-uses, namely evergreen forest (EGF), moist deciduous forest (MDF), coffee agroforests (CAF), rubber plantations (RP), teak monoculture plantation (TMP), and mixed tree species plantation (MSP) were sampled in this study and their locations within each land-use type are depicted in Fig. 1. Historically, evergreen and moist deciduous forests were selectively logged until the 1980s and thereafter commercial timber harvest was banned due to the state government designation of these forests as‘‘reserved forests’’ under the Indian Forest Conservation Act, 1980 (Kushalappa and Kushalappa 1998). However,these forests continued to undergo small-scale disturbances such as collection of fuel wood,livestock fodder,leaf litter collection, and illegal felling of small trees. No major wildfires were recorded in these forests. During the 1950s,secondary forests were clear felled and artificially planted to rubber, teak and other fast growing tree species. In the mid-19th century, most of the privately owned forests in this region were converted to coffee plantations while retaining many of the original trees of evergreen and moist deciduous forests as shade trees.

Fig. 1 Map of the study area showing the location of sample plots and land-use types of Kodagu district in India’s Western Ghats(MDF moist deciduous forest, EGF evergreen forest, CAF coffee agroforest, RP rubber plantation, TMP teak mono-culture plantation,MSP mixed-species plantation)

A nested two-stage sampling approach was adopted for collection of data on trees, herbs and shrubs (Fig. 2). Four super plots of 250 m × 250 m size were demarcated in each land-use type and four sample plots, each of 31.6 m × 31.6 m (0.1 ha) size, were demarcated within each super plot.Thus,the total sample size consisted of 24 super plots and 96 sample plots across six land-use types.In each 0.1 ha plot we counted all woody plants and identified each plant as far as possible in situ at species level using field keys of (Pascal and Ramesh 1987). Voucher specimens of species not identified in the field, werecollected for identification at College of Forestry, Ponnampet with the help of a taxonomist.Height and diameter at breast height(DBH)of all trees of DBH ≥10 cm in four sample plots within each super plot were measured using a Blume Leiss Hypsometer (based on the trigonometric method) and a digital tree caliper (Haglof, Sweden),respectively.

Fig. 2 Nested two-stage sampling method used for each land-use type(250 m × 250 m), trees(31.6 m × 31.6 m), shrubs(5 m × 5 m) and herbs(1 m × 1 m)

Data analysis

Measures of diversity and species assemblage

Species richness (SR) was estimated by counting individuals of different tree species per unit area and plotting species area accumulation curves as suggested by Chazdon et al. (1999). Species diversity (Shannon-Wiener diversity index-H’) and dominance (Simpson’s index-D) were calculated as per Magurran (1988). Vegetation structure was characterized using DBH and total tree height classes.Importance Value Index (IVI) for each species was computed and expressed as the sum of relative density,relative dominance and relative frequency of the species within and among plots(Curtis and McIntosh 1959).Based on the IVI values,we identified the top ten tree species for estimation of density (stems ha-1) and basal area m2ha-1and their contribution to AGB (Mg ha-1). Detrended Correspondence Analysis (DCA) was performed on the species-plots matrix for identifying the major tree species assemblages across land-uses and underlying environmental and disturbance gradients.Ordination was done using the‘‘vegan’’community ecology package (Kindt and Coe 2005) in R-3.0.2 (R Development Core Team 2013). The species drawn in the ordination diagram were abbreviated into eight characters i.e. first four letters of genus and species name.

Estimation of AGB

The strata considered for the estimation of AGB were trees,shrubs and herbs. Tree biomass was estimated indirectly,using stem volume and wood density while, biomass of herbs and shrubs was estimated using destructive method.Previous studies have also estimated the above-ground biomass indirectly using wood density and stem volume(Chave et al. 2003; Vashum and Jayakumar 2012). The data collected on tree parameters such as dbh (＞10 cm)and height were used for volume estimations using volume equations published by Forest Survey of India (FSI 2006).We used local as well as regional volume equations depending on the availability for each species and mostly these regression equations are in the form of general linear models (V = a + bD + D2or V = a + bD2H, with DBH ranging from ＞10 to 180 cm). The regional volume equation, V = 0.16948-1.85075D + 10.63682D2H was used for estimating the volume of tree species for which the species-specific volume equations were not available (FSI 2006).Tree biomass was estimated by multiplying volume with wood density values of particular species obtained from Forest Research Institute (FRI 1996). All shrubs and herbs occurring in the sample plot of 5 m × 5 m and 1 m × 1 m, respectively were harvested and oven-dry weight was estimated. Biomass estimates for different strata were summed to calculate total AGB, which is expressed here as Mg-dry wt ha-1. Biomass of coffee bushes was not included in the calculation of total AGB of coffee agroforest because it was not possible to destructively harvest the coffee bushes in sample plots.

Analysis

The differences in tree diversity (SR, H’ and D) and vegetation parameters (basal area, height, dbh, density and AGB) between the land-use types were evaluated using analysis of variance (ANOVA) after verification of the assumption of normality. Where parameters differed among land-use types, means were contrasted with post hoc Tukey HSD test at P ＜0.05 level of significance (Zar 1999). All statistical analyses were performed using R version 3.0.2(R Development Core Team 2013).Size(dbh and height) class distribution of the species and contribution of size classes to AGB were analyzed in MS-EXCEL using aggregated data of the four super plots in each landuse type.

Results

Species diversity and dominance

Species richness (F5,75= 20.96, P ＜0.05), diversity(F5,75= 27.81, P ＜0.05) and dominance (F5,75= 9.55,P ＜0.05) varied by land-use (Table 1). We recorded 83 tree species representing 26 families in six land-uses and the highest number of species (49) was recorded in evergreen forest (Fig. 3). Mean species richness of evergreen forest differed significantly from other land use types,while coffee agroforest,moist deciduous forest,and mixed species plantation had similar species richness. Diversity index (H’) was highest in evergreen forest (mean ± SE =2.9 ± 0.06). In moist deciduous forest, coffee agroforest and mixed species plantation,diversity ranged from 1.32 to 1.86. Conversely, dominance (D) values were higher in teak and rubber plantations and lowest in evergreen forest.

Table 1 Species diversity and vegetation structure in different land-use types in Kodagu district, India

Fig. 3 Species richness across land-uses; CAF coffee agroforest,EGF evergreen forest, MDF moist deciduous forest, MSP mixedspecies plantation, RP rubber plantation, TMP teak mono-culture plantation. Land-uses carrying same letter are not significantly different at P ＜0.05 (Tukey’s multiple range test)

Vegetation structure

Tree height differed significantly (F5,75= 1.59, P ＜0.05)across land-uses with mean tree height ranging 9.75 ± 0.7 to 20.28 ± 2.1 m (Table 1). Most trees were in the 10-15 m height-class in all land-use types and comparatively taller trees(＞25 m)were recorded in evergreen and coffee agroforests (Fig. 4a). Tree density varied significantly between land-uses (F5,75= 72.7, P ＜0.05) with values ranging from 209 stems ha-1in rubber plantation to 1241 stems ha-1in evergreen forest. A normal reverse-Jshaped diameter distribution curve characterized only evergreen forest (Fig. 4b). We recorded more large trees(＞80 cm dbh)in EGF,MDF and CAF,and fewer in TMP and MSP. Basal area varied significantly (F5,75= 9.35,P ＜0.05) between land uses with higher values recorded in evergreen forest (mean ± SE = 51.2 ± 8.55 m2ha-1)followed by mixed species plantation(mean ± SE = 30.46 ± 7.98 m2ha-1).

Species composition and assemblages

Fig. 4 Size-class distribution of trees with a height and b diameter across land-use types; MDF moist deciduous forest, EGF evergreen forest, CAF coffee agroforest, RP rubber plantation, TP teak monoculture plantation, MSP mixed-species plantation

We recorded differences in species composition between land-use types and five exotic tree species, viz. Acacia mangium, Acrocarpus fraxinifolius, Gliricidia sepium,Grevillea robusta and Hevea brasiliensis. In evergreen forest, 21% of all trees were Chionanthus malabarica while, rest of the species contributed more or less uniformly to species composition. In moist deciduous forest and teak plantation, a substantial proportion of the total number of tree species was contributed by a single species such as Xylia xylocarpa (44%), and Tectona grandis(90%), respectively. The exotic species, G. robusta (silver oak) and H. brasiliensis (rubber) accounted for 68% and 79% of total tree species composition, respectively in coffee agroforests and rubber plantation. Detrended Correspondence Analysis (DCA) was performed on the tree database collected from 96 sample plots. The first DCA axis accounted for 45%and the second axis for 35%of the total variation in species composition. Together the first two axes accounted for over 2/3 of the variation in species composition across land-use types. The contribution of the third and following axes declined rapidly (Table 2) with the third DCA axis accounting for only 12% of the total variation in species composition.Therefore,we considered only first two axes in determining species assemblages and identified three main groups (Fig. 5). There was a greater separation of the groups along Axis-1 than along Axis-2,with Axis-1 having a longer gradient (4.84 SD) compared to Axis 2(2.39 SD).The gradient along Axis-1 was mainly associated with differences in land-use intensity and Axis-2 was associated with the rainfall gradient.

Above-ground biomass

Above-ground biomass (AGB) varied by land-use type(F5,75= 20.34, P ＜0.00) with highly significant differences between evergreen forest and other land-use types(Fig. 6). Across the land-use types, AGB ranged between 59 Mg ha-1in rubber plantation to 327 Mg ha-1in evergreen forest. Mixed species plantation had higher biomass(159 Mg ha-1) than all other land use types. AGB estimates were similar for MDF and CAF, TMP and CAF,TMP and MDF, and TMP and MSP. In most land-use types, higher biomass was contributed by the 20-50 cm dbh class.The exception was evergreen forest where higher biomass was accumulated in 50-80 cm dbh trees(Fig. 7a).Similarly, higher biomass was accumulated in trees with 15-20 m height in all land-use types except evergreen forest where trees of 20-25 m height contributed more to AGB (Fig. 7b).

Table 2 Eigen values and percentage of variance explained by different axes

Species composition and AGB

In moist deciduous forest,Xylia xylocarpa,with 128 stems ha-1accounted for 6.3 m2ha-1basal area and contributed 33 Mg ha-1to total AGB (Table 3). Tectona grandis was the second most dominant tree species (78 stems ha-1with 6.2 m2ha-1basal area) and contributed most to AGB in moist deciduous forest. Although the stem density of Olea dioica(25 stems ha-1)and Mallotus philippensis(13 stems ha-1) was higher in moist deciduous forest, these species contributed less to AGB. In evergreen forest, Veteria indica and Kingiodendron pinnatum with fewer stem numbers (60 and 45 stems ha-1, respectively) contributed more to AGB when compared to Chionanthus malabarica,Cinnamomum spp and Atlantia spp. (193,148 and 140 stems ha-1, respectively). In coffee agroforest, Ficus bengalensis, Grevillea robusta, Pterocarpus marsupium,Aporusa lindliana and Terminelia peniculata contributed most to total biomass. In contrast, Terminalia crenulata,Dalbergia latifolia, Terminalia bellerica, Lagerstroemia lanceolata and Stereospermum spp. contributed more to total biomass under mixed species plantation.In rubber and teak plantations higher density,basal area and biomass was accumulated by the plantation species H. brasiliensis and T. grandis, T. crenuleta and D. latifolia with density of 33 and 35 stems ha-1and basal area of 7.2 and 4.93 m2ha-1contributed 42.5 and 26.8 Mg ha-1of AGB, respectively.Though the density of T. grandis and L. lanceolata was higher but contributed less to AGB.

Discussion

Variation in diversity, structure and composition

Our results revealed high plant species diversity and richness in evergreen forest compared to other land-uses.Moist deciduous forest, coffee agroforest and mixed species plantations were on par with each other, and supported more tree species compared to rubber and Teak plantations.Variation in species diversity and composition may be as a result of different land-uses. Land-use in evergreen forests is protection oriented where unsustainable anthropogenic activities are checked unlike moist deciduous forests which are subject to disturbances due to livestock grazing,illegal felling and collection of non-timber forest products.Chima and Uwaegbulem(2012)evaluated tree species populations under different land-use types in Port Harcourt region of Nigeria and reported that tree species richness was higher in biodiversity conservation area than unprotected secondary regrowth and arable farmland.Murthy et al.(2016)reported on the Western Ghats of India where the more disturbed evergreen and deciduous forests had low species diversity compared to less disturbed forests.

Fig. 5 Detrended Correspondence Analysis (DCA) of species distribution overlaid with six land-uses (1 = moist deciduous forest,2 = evergreen forest, 3 = coffee agroforest, 4 = rubber plantation,5 = teak mono-culture plantation, 6 = mixed-species plantation. The species drawn in this diagram are abbreviated into eight characters i.e.first four letters of genus and species name as Acro.fraxi = Acrocarpus fraxinifolius, Alst.scho = Alstonia scholaris, Apor.lind = Aporosa lindleyana, Arto.hirs = Artocarpus hirsutus,Arto.lako = Artocapus lakoocha, Atlan.spp = Atlantia spp, Bacc.-cour = Baccaurea courtallensis, Bomb.ceib = Bombax ceiba, Cass.-fist = Cassia fistula, Calo.aust = Calophyllum austroindicum,Chio.mala = Chionanthus malabarica, Cinn.spp = Cinnamomum spp, Dalb.lati = Dalbergia latifolia, Dill.pent = Dillenia pentagyna,Eryt.indi = Erythrina indica, Ficu.beng = Ficus benghalensis, Flac.-mont = Flacourtia montana, Glyri.sepi = Gliricidia sepium,Grev.robu = Grevillea robusta, Grew.tili = Grewia tiliaefolia,Heve.bras = Hevea brasiliensis, Holi.arno = Holigarna arnottiana,Hope.parvi = Hopea parviflora, King.pin = Kingiodendron pinnatum, Knem.atte = Knema attenuata, Lage.lanc = Lagerstroemia lanceolata, Lage.parv = Lagerstroemia parviflora, Lann.cora = Lannea coromandelica, Maca.pelt = Macaranga peltata,Mall.phil = Mallotus philippensis, Myri.spp = Myristica spp, Nari.cren = Naringi crenulata, Olea.dioc = Olea dioica, Pter.mars = Pterocarpus marsupium, Pter.acer = Pterospermum acerifolium,Schl.oleo = Schleichera oleosa, Ster.chil = Stereospermum chelonoides, Syzi.cumi = Syzygium cumini,Syzi.lanc = Syzygium lanceolatum, Tect.gran = Tectona grandis, Term.bell = Terminalia bellerica, Term.cheb = Terminalia chebula, Term.cren = Terminalia crenulata, Term.pani = Terminalia paniculata, Tetr.nudi = Tetrameles nudiflora, Trem.orie = Trema orientalis, Vate.indi = Vateria indica, Vite.alti = Vitex altissima, Xyli.xylo = Xylia xylocarpa

Fig. 6 Above-ground biomass distribution in different land-uses;CAF coffee agroforest, EGF evergreen forest, MDF moist deciduous forest, MSP mixed-species plantation, RP rubber plantation, TMP teak mono-culture plantation. Land-uses carrying same letter are not significantly different at P ＜0.05 (Tukey’s multiple range test)

In production land-use types such as coffee agroforests,rubber, teak, and mixed species plantations, species diversity was lower than in evergreen forest, possibly due to management regime. It is well established that increasing management intensities reduces the diversity of different taxa, such as, trees, birds, amphibians in tropical agroforestry systems (Philpott et al. 2008; Clough et al.2010; Wanger et al. 2010). For example, the traditional coffee agroforestry in this region, which comprised large number of native tree species in the past,are being replaced by ‘‘modernized’’ monocultures aimed at increasing the coffee productivity. In this study we found that the exotic Grevillea robusta was dominant in coffee based agroforestry system which accounted to 67.5% of the total tree species composition. Pinard et al. (2014) reported that planted exotic tree species were dominant on Kenyan coffee farms, whereas most native species occurred at low abundances.Similarly,Ruf et al.(2006)have reported that the introduction of new hybrid cocoa varieties has led to the gradual shift towards the elimination of native shade trees in cocoa based agro-ecosystems in Ghana. Further,farmers have found it necessary to eliminate forest tree species to effect high performance of hybrid varieties.

Fig. 7 Biomass distribution in different a diameter and b height classes across the land-uses; MDF moist deciduous forest, EGF evergreen forest, CAF coffee agroforest, RP rubber plantation, TMP teak mono-culture plantation, MSP mixed-species plantation

Stem density per hectare (trees with dbh ＞10 cm) in tropical landscape varies from 245 (which is considered low) to intermediate values of 420-617 and high value of more than 639 stems ha-1(Suratman 2012). Based on this categorization, the stem density of the plots of our study could be classified as low to intermediate. Except evergreen forest, no other land-use types showed normal reverse J shaped size class distribution indicating the abundance of trees in few diameter/height classes. We found lesser number of trees in lower diameter classes(＜10 cm and 10-20 cm DBH)and higher number of trees in intermediate class (20-50 cm DBH) in moist deciduous forest,rubber,teak and mixed species plantations probably due to the fact that these land-uses face continuous pressure in the form of wood removal primarily, fuel wood and small timber for domestic use. Similar effects of disturbance on stem density and forest stand structure have been reported on the Western Ghats of India (Murthy et al.2016).

Predominance of large sized trees (＞50 cm dbh) in coffee agroforests could be attributed to the management practices. Coffee farmers of this region retain large sized trees, through intermittent crown pruning, for the purpose of shade to under-storey coffee and as a standard for pepper wines. Luke et al. (2011) equally attributed the predominance of large sized trees to the management practices in cocoa agroforests of Ghana, which begins with the selective thinning of the original forest stand to leave a few large desirable tree species as shade for the developing cocoa. Subsequent regeneration of tree species is considered as weed growth and they are therefore removed during weeding, or under natural circumstances they may not persist due to the heavy shading from the cocoa canopy.DCA ordination analyses uncovered the significant effect of land-use intensities on species composition. We found that certain species were unique to particular land-use systems and not in others which calls for conservation priorities. For example, Ficus bengalensis, which is considered as the key stone species, was recorded only in coffee agroforest and not in other land-uses. Dawson et al.(2013) emphasized the conservation of flagship species in agriculture dominated landscapes to enhance the conservation value of production systems.

Changes in above-ground biomass

Above-ground biomass (AGB) differed significantly between land-use types. Biomass estimated in evergreen forest in our study (327.33 Mg ha-1) is slightly lower when compared to the biomass value of 439 Mg ha-1recorded in tropical wet evergreen forests of Uttar Kannada district of northern Western Ghats in India (Bhat et al.2000). Similarly, the biomass values of coffee agroforests(100.82 Mg ha-1) and teak monoculture plantations(138.88 Mg ha-1)were also slightly lower when compared with the biomass estimates reported by Devakumar et al.(2007) for coffee agroforests (138 Mg ha-1) and Subramanian et al. (2009) for teak plantations (179 Mg ha-1).The amount of biomass in evergreen forests is around 2-2.5 times higher than in other land-use types. These differences probably arise due to the variations in stand structure. Among the structural parameters, stem density and basal area are the important determinants of biomass production (Chave et al. 2003). However, biomass is not always reflected by the density as in the present study.Though moist deciduous and coffee agroforest had higher stem density (314 and 281 stems ha-1, respectively)however, contained lesser biomass compared to mixedspecies plantation which had lesser number of individuals(248 stems ha-1). Conversely, basal area is a strong indicator of growing stock and AGB as evidenced in the present study where land-use types with higher tree basal area contributed for higher AGB.Large trees play a vital role in influencing the AGB storage across a range of tropical ecosystems (Letcher and Chazdon 2009; Silva-Costa et al.2012; Chaturvedi and Raghubanshi 2015) since trees with large diameters (＞10 cm DBH) often make the largest contribution to AGB (Kirby and Potvin 2007; Djuikouo et al. 2010). The majority of AGB in our study was found within taller trees and trees with a larger diameter. Their removal substantially alters the biomass dynamics in this region.The removal of large trees by intensive logging hadsignificant impacts on species richness, diversity and composition of forests (Lee et al. 2005; Suratman 2012).

Table 3 Contribution of top ten tree species (based on IVI) to density, basal area and biomass across land-use types

Table 3 continued

We conclude that species diversity and above-ground biomass varied between land-use types with highest diversity and biomass in evergreen forests and lowest in rubber plantations. Next to natural forests, coffee agroforests contain high diversity and AGB.Our results showed that coffee agroforests and mixed species plantations are more desirable compared to monocultures for conserving biodiversity and to store more biomass.We alsofound that certain species were unique to particular land-use systems and not in others.For example,Ficus bengalensis,which is considered as the key stone species, was recorded only in coffee agroforest and not in other land-uses.Therefore,the promotion of native trees on the farms, with specific attention to rare species and species with low population densities should have priority.