Seed size: a key trait determining species distribution and diversity of dry tropical forest in northern India

Transcription

1 available at journal homepage: Original article Seed size: a key trait determining species distribution and diversity of dry tropical forest in northern India Ekta Khurana, R. Sagar, J.S. Singh * Department of Botany, Banaras Hindu University, Varanasi , India A R T I C L E I N F O A B S T R A C T Article history: Received 16 February 2004 Accepted 14 October 2005 Available online 19 January 2006 Keywords: Seed size Disturbance Seedling Seed weight class Species diversity The study examined the relationships among seed size, plant distribution and abundance in a dry tropical forest of northern India. Results indicated that small-seeded species, which were generally wind-dispersed, were more widely distributed, at this local scale, compared to large-seeded species. However, the proportional abundance and basal cover of seed size categories indicated that the structure of the dry forest was largely determined by the medium- to large-seeded species. There was a considerable amount of redundancy within each seed size group, which added to the species diversity. Variability in seed size and the variable degree of shade-tolerance permit the species to occupy the full range of the gradient of light environments of the forest floor. This study revealed that in little to moderately disturbed locations seedlings of large-seeded species increased in abundance, whereas in extremely perturbed locations seedlings of species with mediumsized to small seeds were more abundant Elsevier SAS. All rights reserved. 1. Introduction * Corresponding author. Tel.: ; fax: address: singh.js1@gmail.com (J.S. Singh) X/$ - see front matter 2006 Elsevier SAS. All rights reserved. doi: /j.actao A variety of factors, such as species life-history traits, competitive interactions within plant groups, habitat heterogeneity and anthropogenic disturbances, determine the spatial pattern of the community composition and relative abundances and coexistence of the inhabiting species. For example, species with greater dispersal ability should be more widely distributed than those of low dispersal capacity (Hanski et al., 1993). Dispersal ability in turn is strongly influenced by the seed size of a species (Rees, 1995). Smaller seeds with greater dispersal ability are readily transported by dispersal agents (Venable and Brown, 1988; Greene and Johnson, 1993) and thus have an advantage in colonization and abundance. Larger and heavier seeds are relatively less abundant but produce seedlings with greater competitive ability than those produced by small seeds, enabling them to establish and survive under various stresses such as competition (McConnaughay and Bazzaz, 1987), moisture (Leishman and Westoby, 1994a), shading (Leishman and Westoby, 1994b), disturbances (Hammond and Brown, 1995), defoliation and herbivory (Armstrong and Westoby, 1993). Colonization-competition trade-offs, interacting with abiotic and biotic environmental factors, may act as a filter to determine the species pool from which local communities assemble. Tropical forests, which cover ~86% of the total forest land in India (Singh and Singh, 1988), vary considerably in terms of microclimate of the forest floor and biotic factors. Tropical dry forests, particularly, are facing large-scale anthropogenic disturbances, such as repeated lopping of trees for fuel wood or leaf fodder, and heavy grazing (Jha and Singh, 1990). This has created a wide range in the gradients of light and resource availability in the dry forest environments. Furthermore, diverse arrays of vertebrate and invertebrate animals disperse and consume seeds and seedlings, thus playing a major role in determining community structure. Life-history traits such as shade tolerance also influence the optimisation of seed size, and thus indirectly influence community

2 197 structure. In view of the above, we assume that the seed size of constituent species plays a major role in determining species distribution and diversity in these dry forests. In this paper we address the following broad questions. What is the pattern of relationship of seed size with species-specific characteristics, viz. dispersal mode and shade tolerance? In what way are plant distribution and abundance in dry tropical forest constrained by seed size? What is the impact of disturbance on species composition with respect to seed size, and how does disturbance affect the temporal pattern of species diversity in relation to seed size? 2. Materials and methods 2.1. Study area The study area is located on the Vindhyan hill tract in the Sonebhadra district of Uttar Pradesh, India, between N and N latitude and E and E longitude. The altitude ranges from 313 to 483 m asl. The climate is dry tropical influenced by monsoon seasonality. The longterm annual rainfall is mm, of which about 86% is received from the southwest monsoon during June August. The year is divisible into three seasons: rainy (monsoon) (mid-june September), winter (November February), and summer (April mid-june). October and March comprise transition periods, respectively, between rainy and winter, and between winter and summer seasons. The soils are Ultisols, sandy loam in texture and reddish to dark grey in colour, and are nutrient-poor. The natural vegetation of the region is tropical dry deciduous forest, locally dominated by Acacia catechu, Anogeissus latifolia, Butea monosperma, Hardwickia binata, Shorea robusta, Tectona grandis, Terminalia tomentosa, etc., and by other species in some cases. (Sagar and Singh, 2003). The region is undergoing rapid changes in vegetation, and is experiencing large-scale human disturbances in the form of mining, thermal power generation through coal-fired power plants, exploitation by the cement industry, etc. In addition to sporadic illegal tree felling, widespread lopping and extraction of non-timber resources are carried out. The forested area is continuously decreasing and the remnant forest cover exists in the form of non-contiguous patches of varying sizes (Singh et al., 1991) Methods Five sites Hathinala, Khatabaran, Majhauli, Bhawani Katariya and Kota were selected on the basis of satellite images and field observations to represent the entire range of conditions in terms of canopy cover and disturbance regimes (Sagar et al., 2003a). The sites were categorized in a gradient of disturbance on the basis of relative impact of composite elements of disturbance (e.g. distance from road, market, or human habitation; intensity of cutting/lopping, grazing/ browsing; impact of wild animals) including rockiness as a natural constraint. Degree of disturbance increased in the following order: Hathinala (HT), Khatabaran (KH), Majhauli (MJ), Bhawani Katariya (BK) and Kota (KT) (Sagar et al., 2003b; Sagar and Singh, 2003). The sites vary in topography, the Kota and Khatabaran sites being relatively flat, with topography gentle at Bhawani Katariya and undulating at Hathinala and Majhauli (Sagar et al., 2003b). At each of the five sites, three contiguous 1-hectare permanent plots were established. Each hectare ( m) was divided into a grid of 100 sub-plots, each m in size. All individuals 30 cm circumference over bark at breast height (CBH) were enumerated by species and the circumference of all the individuals was measured to the nearest millimetre with Freeman s tape (Sagar et al., 2003a, 2003b). The CBH and diameter values were converted to basal area values (Sagar and Singh, 2004, 2005). Within each subplot, a 2 2 m area was marked for enumeration of established seedlings (individuals < 3.2 cm diameter but > 30 cm height) and their diameter was measured at 10 cm above the ground. Seeds of each tree species were collected from five mature individuals in each site where the species occurred. Wherever the site had fewer than five individuals of a species, seeds were collected from similar trees in the adjacent area. Seeds were collected at exposed mid-canopy height. Dry mass of five replicates (each consisting of 50 seeds), from each tree were measured to the nearest milligram, by drying to a constant mass at 80 C. Seeds were grouped into three seed weight classes: (i) g per seed, (ii) g per seed, and (iii) g per seed. Hereafter, these three classes are referred to as SWC-I, -II and -III, respectively. Proportional frequency, per cent of species, proportional abundance and proportional basal area for the set of species in each seed weight class were calculated as follows: Proportional frequency = 100 (total frequency of all species in that seed weight class divided by total frequency of all species in all three seed weight classes). Per cent species = 100 (number of all species in that seed weight class divided by total number of all species in all the three seed weight classes). Proportional abundance = 100 (number of individuals of all species in that seed weight class divided by total number of individuals of all species in all the three seed weight classes). Proportional basal area = 100 (total basal area of all species in that seed weight class divided by total basal area of all species in all the three seed weight classes). The information on dispersal mode and shade tolerance for each species was recorded from Troup (1921) and by direct field observations. The three dispersal modes considered were: wind-dispersed, mammal-dispersed, and other modes (e.g. autochory, adaptation for dispersal by birds). Shade tolerance levels were (Khurana, 2002) as follows: highly shadetolerant (< 5% sunlight), moderately shade-tolerant (20 30% sunlight) and relatively shade-intolerant (80 100% sunlight) Statistical analyses ANOVA was used to identify the impact of dispersal mode and shade tolerance on seed mass. In this analysis, each species was considered as a single data point. The impact of de-

3 198 gree of disturbance and seed weight class on proportional frequency, % species, proportional abundance and proportional basal area was analysed using multivariate ANOVA (GLM, full factorial). In this analysis, site (representing disturbance intensities) and seed weight class were considered independent variables and proportions of abundance, basal area and frequency in various seed weight classes as dependent variables. Data on proportions from the three plots per site comprised replicates. ANOVA was performed after testing for homogeneity of variance. LSD tests were used to differentiate the means. All statistical analyses were performed using the statistical software package SPSS (S.P.S.S. Inc., 1997). 3. Results 3.1. Impact of seed size on dispersal mode and shade tolerance Table 1 lists the species, seed mass, cumulative total number of individuals of each species at all sites, predominant dispersal mode and shade tolerance. Out of 37 species examined, 46% were wind dispersed, 38% dispersed by mammals (such as jackals, monkeys, bats, rodents and other mammalian fruit/seed predators) and 16% of species showed other modes of seed dispersal. Pterocarpus marsupium produces winged fruits twice a year, February April and May June. The dispersal of fruits maturing from the first fruiting episode occurs due to wind; these fruits contain very light seeds. The fruits formed during the second fruiting episode (May June) contain heavier seeds and these, after ripening, contribute to the seedling recruitment, as their dispersal coincides with the rainy season. Dispersal of these fruits on the ground is mainly due to the activities of rodents, squirrels, etc. Therefore, we have assigned this species to the mammal-dispersed category. Mean seed mass was strongly associated with both dispersal mode and shade tolerance across all sites (Fig. 1). Mean seed mass of the mammal-dispersed species was more than four times greater than that of the wind-dispersed seeds (Table 1). ANOVA indicated significant differences in seed mass among species varying in dispersal mode (F 2, 34 = 3.4, P < 0.05). Small-seeded species (SWC-I) accounted for 60% of the total species, medium-seeded Table 1 Seed weight, abundance, dispersal mode and shade tolerance of dry forest tree species of India. Seeds of some species are known to be dispersed both by birds and mammals. Here we have taken only the predominant mode for each species. The abundance is number of stems in the total 15-ha area (based on Sagar and Singh 2003). Nomenclature follows Verma et al. (1985) Species Seed wt (g per seed) SWC Number of individuals Number of sites occupied Predominant dispersal mode Shade tolerance Adina cordifolia Hook , KH, MJ W IT Ficus benghalensis L , KH O IT Hymenodictyon excelsum Wall , KH W IT Gardenia latifolia Ait , HT, BK M IT Dalbergia sissoo Roxb , KT W IT Anogeissus latifolia Wall All W IT Eriolaena quinquelocularis Wight , HT, MJ M IT Holarrhena antidysenterica DC , KH, HT, MJ, BK, W IT Chloroxylon swietenia DC , KH W IT Cassia siamea Lam , BK W IT Phyllanthus emblica Gaertn , KH, HT, MJ, BK O IT Carissa spinarum DC , KH O IT Ougeinia oogenesis Hochreut , HT W MT Lagerstroemia parviflora Roxb , KH, HT, MJ, BK W IT Bombax ceiba Linn , KH M IT Holoptelea integrifolia Planch , KH W IT Acacia catechu Willd All W IT Albizia odoratissima Benth , HT W IT Nyctanthes arbortris-tris L , KH, HT W HT Butea monosperma Taub , KH, KT W MT Soymida febrifuga A. Juss , HT, MJ, BK W MT Cassia fistula Linn , KH, HT, MJ M MT Boswellia serrata Roxb. ex Coleb , HT, MJ, BK O IT Lannea coromandelica Merr All M HT Bauhinia racemosa Lam , HT, MJ O IT Aegle marmelos Carrea , KH M MT Hardwickia binata Roxb , HT, MJ, BK, KT W IT Tectona grandis Linn , KH M IT Miliusa tomentosa Sincl , HT, MJ, BK M HT P. marsupium Roxb , HT, MJ, BK M HT Buchanania lanzan Spereng All M IT Diospyros melanoxylon Roxb All O MT Semecarpus anacardium Linn.f , KH M IT T. chebula Retz , HT, MJ, BK M IT Zizyphus glaberrima Satap , KH, HT, MJ, BK M HT S. robusta Gaertn , KH, HT, MJ, BK W IT T. tomentosa Wight , KH, HT, MJ, BK M IT SWC = seed weight class, W = wind-dispersed, M = mammal-dispersed, O = other modes, HT = high tolerance, IT = relatively shade-intolerant, and MT = moderate tolerance. HT = Hathinala, KH = Khatabaran, MJ = Majhauli, KT = Kota.

4 199 were relatively shade-intolerant while 40% were highly shade tolerant Impact of seed size on plant distribution and abundance There was a significant positive relationship between frequency (number of occurrences of the species per 1500 quadrats, i.e. subplots) and number of individuals per 15 ha area (abundance), and between distribution (number of sites occupied by the species) and number of individuals (Fig. 2). However, regression analyses revealed that no significant re- Fig. 1 Mean seed mass of species as a function of (a) their respective dispersal mode and (b) their shade tolerance across five sites in dry tropical forest. Bars represent 1 S.E. Intolerant = relatively shade-intolerant; moderate = moderately shade-tolerant; tolerant = highly shade-tolerant (see text). (SWC-II) 24.5%, and large-seeded (SWC-III) 13.5%. Within SWC-I, 60% of species were wind-dispersed and 11% mammal-dispersed. In SWC-II, 10% of species were wind-dispersed and 60% mammal-dispersed. In SWC-III, 20% of species were wind-dispersed while 80% were mammaldispersed (Table 1). The species were classified into three groups on the basis of shade tolerance: the relatively shade-intolerant species comprising 72% of the total species (mean seed mass 0.25 ± 0.11 g per seed), moderately shade-tolerant species, 14% (mean seed mass 0.43 ± 0.17 g per seed) and highly shadetolerant species, 14% (mean seed mass 0.51 ± 0.18 g per seed). Among the three seed weight classes, 77% of the species belonging to SWC-I were relatively shade-intolerant and only one out of 22 species was highly shade tolerant. In SWC-II, 60% of the species were relatively shade-intolerant while 30% were highly shade tolerant. In SWC-III 40% of the species Fig. 2 (a) Relationships between proportional frequency (F) of species and number of individuals (N i ) in dry tropical forest. Equation for the regression line is N i = 15.82F 0.49, r 2 = 0.97, P = in all sites together. (b) Relationships between number of sites occupied (N s ) and number of individuals (N i ) in all sites together. Equation for the regression line is N i = 69.65N s 75.6, r 2 = 0.43, P =

5 200 sites, Majhauli and Bhawani Katariya, proportional abundance was greatest for species of SWC-III and -II, respectively. The differential contribution of the seed weight classes in terms of proportional abundance and basal area resulted in significant site seed weight class interactions (Table 2) Comparisons between adult and seedling individuals Fig. 3 Triangular relationship between seed mass and number of individuals (all individuals of the species in the total 15 ha area). lationship existed between seed mass and number of sites occupied, or between seed mass and number of individuals. Following Guo et al. (2000), we used triangular envelopes for explaining the relationship between seed mass and number of individuals (Fig. 3). The relationship between seed mass and proportional frequency also showed a similar trend. There were two outliers (S. robusta and T. tomentosa) Impact of disturbance on species composition in relation to seed size classes Dominance of species groups in terms of seed size is reflected by their proportional frequency, proportional abundance and relative contribution to the total basal area of a site. We categorized the species present at each site into the three chosen seed weight classes, and analysed to see whether patterns differed with the intensity of disturbance. At four out of five sites, the proportional frequency was greatest for SWC-I. For Bhawani Katariya, it was highest for SWC-II while for Kota, frequencies in SWC-I and -II were more or less equally distributed (Fig. 4). ANOVA indicated significant differences in proportional frequency of species in different seed weight classes (Table 2). LSD tests indicated that differences in proportional frequency among seed weight classes were significant for Khatabaran and Bhawani Katariya. Further, except for Hathinala, SWC-I contributed 40 64% to the total number of species on a site. At Hathinala, per cent species was nearly equal in SWC-I and -II. ANOVA indicated significant differences in per cent species in different seed weight classes (Table 2). Proportional abundance, which varied significantly among seed weight classes (Table 2) was greatest for SWC-I at Hathinala, Khatabaran and Kota sites. At the two other Proportional abundances between adults and seedlings of species in different seed weight classes are compared across sites in Table 3. In comparison to the adults, proportional abundances of seedlings of SWC-III were greater, and those of SWC-II were lower, at Hathinala and Khatabaran sites. Further, almost no difference in the proportion of seedlings and adults for the set of SWC-I species was observed at Hathinala. At Majhauli, however, abundance of species belonging to SWC-III was greater than that of other seed weight classes at both adult and seedling stages. At Bhawani Katariya, abundance of seedlings of species in SWC-III was lower by one-half compared to the relative abundance of adults of these species (Table 3). In the Kota site, species in SWC-III were absent at both life stages, and the number of individuals of species of SWC-II at the seedling stage was much lower compared to the abundance of adults of these species. 4. Discussion 4.1. Seed size, dispersal mode and shade tolerance For a neotropical forest of Mabura Hill, Guyana, Hammond and Brown (1995) reported average seed mass of 4.32 ± 0.89 g, compared to 0.32 ± 0.08 g for the dry forest of the present study. The majority of species examined by Hammond and Brown were either mammal-dispersed or gravitydispersed. Our study revealed that variation in seed mass of tropical dry forest tree species was associated with dispersal ability. In our study, the majority of small-seeded species were wind-dispersed, while those of large-seeded species were mammal-dispersed. The former should have greater dispersal ability and colonization and distribution success than the latter group of species. Forty-six per cent of all species present were wind-dispersed. The association of wind dispersal and small seed size represents a general rule for the assembly of plant communities in which small-seeded species with greater dispersal ability exhibit a wider range of abundance and distribution (Guo et al., 2000). In a Mexican rain forest, non-animal dispersal (wind, explosive and gravity) was predominantly associated with smaller seed mass, whereas animal dispersal was predominantly associated with larger seed mass (Chazdon et al., 2003). Small-seeded species generally, were relatively shade-intolerant, whereas a relatively high percentage of the largeseeded species had moderate or high shade tolerance. While surveying seed mass in the British flora, Salisbury (1942) reported that shade-tolerant species had seed mass more than three times greater than that of intolerant species. Foster and Janson (1985) also found that seeds of tropical forest trees tended to be larger in species that could establish un-

6 201 Fig. 4 Proportional frequency (a), percentage of all species (b), and proportional basal area (c) in five sites of dry tropical forest. Narrow bars represent 1 S.E. HT = Hathinala, KH = Khatabaran, MJ = Majhauli, BK = Bhawani Katariya, and KT = Kota. Table 2 Summary of ANOVA on selected parameters in five sites of Vindhayan dry forest of India Site SWC Site seed weight class df F df F df F Proportional frequency NS *** NS Per cent species NS ** *** Proportional abundance NS ** 8 2.2** Proportional basal area NS NS ** *** P < , ** P < 0.05, NS = not significant, SWC = seed weight class.

7 202 Table 3 Proportional abundances of adults and seedlings, belonging to three seed weight classes, of species in five sites of Vindhayan dry forest of India. Values are mean ± 1 S.E. The abundance of seedlings is based on Sagar and Singh (2004). SWC = seed weight class SWC-I SWC-II SWC-III Hathinala Adults ± ± ± 8.07 Seedlings ± ± ± 9.03 Khatabaran Adults ± ± ± 4.46 Seedlings ± ± ± 6.90 Majhauli Adults ± ± ± 4.76 Seedlings ± ± ± 0.89 Bhawani Katariya Adults ± ± ± 4.06 Seedlings ± ± ± 5.67 Kota Adults ± ± ± 0.0 Seedlings ± ± ± 0.0 der closed canopies or in small gaps than in species that require large gaps. Large-seeded species under closed canopies had an advantage both by initially reaching into a better light environment, and by surviving longer in the thick litter layers because they are able to support respiration longer while experiencing a net carbon deficit (Westoby et al., 1992). The majority of the species studied, in this study, belonged to the relatively shade-intolerant category. This result could be explained by the relatively open nature of dry forest canopy Species distribution and abundance in relation to seed size The observed triangular relationship between seed mass and percent frequency and abundance indicated that individuals of small-seeded species were distributed more widely than those of large-seeded species. Similar results were obtained by Rees (1995); Guo et al. (2000). Several researchers (Leishman et al., 1995; Parciak, 2002) have reported that small seeds are produced in greater numbers than the large seeds. Large crops of small seeds may be an adaptation to facilitate dispersal because of the greater mobility of small seeds and the larger numbers of offspring that can potentially disperse (Baker, 1972; Howe and Richter, 1982; Sallabanks 1992). Smaller seeds also enter into the soil more easily than large seeds, and thus have a greater probability of being found in persistent soil seed banks (Khurana and Singh, 2001). Further, large seeds, which comprise generally the first crop of germinants in the early rainy season, may die off due to the synchronization of insect outbreaks with the early rainy season (Garwood, 1983). Small seeds provide a back-up crop of seedlings and may then dominate the site. Seeds in this dry forest mature and disperse in the period preceding the rainy season (Singh and Singh, 1992). Further, a high level of redundancy in SWC-I (69% of species examined had small seeds) indicates that this trait, due to better dispersal ability, adds to the species diversity of relatively open-canopied dry forest. We suggest that the mobile dispersal phase plays a key role in the determination of species abundance and distribution in the dry tropical forest. However, the importance of large seeds should not be under-emphasized, in view of their competitive ability and establishment success, particularly under various stresses (see, Khurana and Singh, 2000, 2004a, 2004b). For example, species with medium-sized to large seeds comprised a significant proportion, even higher than that of small-seeded species, at two sites, i.e. Majhauli and Bhawani Katariya, where species with medium-sized and those with large seeds, together, comprised about 60 70% of individuals. In the extremely disturbed Kota site, a complete absence of large-seeded species was observed, which could be due to selective removal. Thus, this study suggests that while the preponderance of smallseeded species could be a general trend for the dry tropical forest due to their larger seed crops and dispersal ability, the distinctive composition of the community is ultimately determined by species with medium-sized or large seeds. The ability of large-seeded species to compete might allow them to grow efficiently in moderately disturbed areas and in the interior of forest patches Impact of seed size on species recruitment in relation to environmental heterogeneity and disturbance The occurrences of small- and large-seeded species reflected the availability of niches with variable micro-environmental conditions at the floor of the dry forest. For example, the canopy of dry forest is not very dense and there occurs a continuum of regeneration niches, with most areas receiving intermediate levels of sunlight, and thus showing intermediate levels of temperature and soil moisture. The patchy canopies of dry forest communities create forest floor microenvironments with variable resource availability. Seedlings from large-seeded species may be able to better cope with temporary carbon deficit resulting from such moderate disturbances as grazing and browsing (see also Armstrong and Westoby, 1993). According to Hammond and Brown (1995), largeseeded species, with their greater amounts of reserves that are amenable to mobilisation, would be able to rebalance their carbon budget after being cut and lopped, by deploying new leaf area until photosynthetic income could again support requirements for respiration. Larger seeds are particularly rich in secondary compounds, which could be translocated from seeds to seedlings, thus reducing the load of herbivory (Foster, 1986). Under extreme disturbances, species with small and medium-sized seeds would benefit by their greater dispersal ability. According to Baker (1972); Foster (1986); Hammond and Brown (1995), fluctuations in environmental conditions such as temperature, light intensity, rainfall, and predation and herbivory may simultaneously favour intra-population variation in seed size, and maintain species with different seed sizes. Thus, both biotic and abiotic processes combine to determine the characteristics of species pools from which local communities assemble. We compared abundances of adult and seedling individuals of the examined species at all sites. In less disturbed sites, viz., Hathinala and Khatabaran, we observed disproportionately high recruitment of large-seeded species and low recruitment of small-seeded species. In contrast, in two

8 203 intensely disturbed sites, seedlings of the large-seeded species were fewer (in Bhawani Katariya) and the abundance of seedlings of small-seeded species was greater (in Kota). Dominance of large-seeded species at both adult and seedling stages at the Majhauli site indicated that this site is probably below a threshold limit of disturbance intensity, beyond which the recruitment pattern drastically changes. While working on 10 wet tropical forests of Costa Rica, Chazdon et al. (2003) found that in second-growth forests (frequently disturbed) relative abundance of species with explosive and wind-dispersed seeds, or those with insect pollination, was higher, and relative abundance of species with animal dispersal, or those with mammal pollination, was lower. In the Hathinala and Khatabaran sites, the greater number of mature trees may act as dispersal barriers for incoming wind-dispersed small seeds, leading to a decline in the abundance of small-seeded species within the community. This may result in increased abundance of large-seeded, mammal dispersed, non-pioneer, and shade-tolerant species. The dominance (~90%) of seedlings of small-seeded species in Kota, where seven species occurred both as adults and seedlings and four species occurred only as seedlings, indicated the absence of dispersal barriers. The above pattern in regeneration is also indicative of a general successional process. Hammond and Brown (1995) found that extreme disturbance increased the likelihood of local extinction of species with slow-growing seedlings and very large seeds that are dispersed by gravity and terrestrial mammals. Sagar and Singh (2004) reported that the percentage of species with declining population, as inferred from a comparison of the proportions of seedlings and mature individuals at a single point in time, was maximal at the Kota site. Seedlings of large-seeded species such as Terminalia tomentosa and Diospyros melanoxylon were abundant at Hathinala, Khatabaran and Majhauli sites, and were totally absent or at very low densities at Bhawani Katariya and Kota sites. In addition, the preponderance of large-seeded fleshy-fruited species, which are attractive to animals (frugivores), indicates that other trophic levels (i.e. consumers) would also be amply supported in the Hathinala and Khatabaran sites, leading to further enhancement in biodiversity. Further, both small-seeded and relatively larger-seeded species possess a variety of phenotypic traits reported for the dry tropical forest vegetation (Sagar and Singh, 2003). In conclusion, the study indicates colonization competition trade offs between species, and we hypothesize that the interplay of biotic interactions, disturbance intensity and dispersal events jointly control local community composition and species diversity. The different seed sizes permit the occupation of the variety of niches available in the dry forest with respect to gradients of light, moisture and temperature, contributing to the maintenance of overall species diversity. Acknowledgements Funding support from Ministry of Environment and Forests, New Delhi, Government of India, is gratefully acknowledged. One of us (J.S.S.) is supported by the CSIR Emeritus Scientist Scheme. REFERENCES Armstrong, D.P., Westoby, M., Seedlings from large seeds tolerate defoliation better: a test using phylogenetically independent contrasts. Ecology 74, Baker, H.G., Seed weight in relation to environmental conditions in California. Ecology 53, Chazdon, R.L., Carega, S., Webb, C., Vargas, O., Community and phylogenetic structures of reproductive traits of woody species in wet tropical forests. Ecol. Monogr. 73, Foster, S.A., Janson, C.H., The relationship between seed size and establishment conditions in tropical woody plants. Ecology 66, Foster, S.A., On the adaptive value of large seeds for tropical moist forest trees: a review and synthesis. Bot. Rev. 52, Garwood, N.C., Seed germination in a seasonal tropical forest in Panama: a community study. Ecol. Monogr. 53, Greene, D.F., Johnson, E.A., Seed mass and dispersal capacity in wind-dispersed diaspores. Oikos 67, Guo, Q., Brown, J.H., Valone, T.J., Kachman, S.D., Constraints of seed size on plant distribution and disturbance. Ecology 81, Hammond, D.S., Brown, V.K., Seed size of woody plants in relation to disturbance, dispersal, soil type in wet neotropical forests. Ecology 76, Hanski, I., Kouki, J., Halkka, A., Three explanations of the positive relationship between distribution and abundance of species. In: Ricklefs, R.E., Schluter, D. (Eds.), Species Diversity in Ecological Communities. University of Chicago Press, Illinois, USA, pp Howe, H.F., Richter, W.M., Effects of seed size on seedling size in Virola surinamensis: a within and between tree analysis. Oecologia 53, Jha, C.S., Singh, J.S., Composition and dynamics of dry tropical forest in relation to soil texture. J. Veg. Sci. 1, Khurana, E., Germination and seedling growth of selected dry tropical forest species. Ph.D. thesis. Banaras Hindu University, Varanasi, India. Khurana, E., Singh, J.S., Influence of seed size on seedling growth of Albizia procera under different soil water levels. Ann. Bot. 86, Khurana, E., Singh, J.S., Ecology of seed and seedling growth for conservation and restoration of tropical dry forest: a review. Environ. Conserv. 28, Khurana, E., Singh, J.S., 2004a. Impact of elevated N inputs on seedling growth of five dry tropical tree species as affected by life history traits. Can. J. Bot. 82, Khurana, E., Singh, J.S., 2004b. Germination and seedling growth of five tree species from tropical dry forest in relation to water stress: impact of seed size. J.Trop.Ecol. 20, Leishman, M.R., Westoby, M., 1994a. The role of seed size in seedling establishment in dry soil conditions experimental evidence from semiarid species. Journal of Ecology 82, Leishman, M.R., Westoby, M., 1994b. The role of large seed size in shaded conditions: experimental evidence. Func. Ecol. 8, Leishman, M.R., Westoby, M., Jurado, E., Correlates of seed size variation: a comparison among five temperate floras. J. Ecol 83,

9 204 McConnaughay, K.D.M., Bazzaz, F.A., The relationship between gap size and performance of several colonizing annuals. Ecology 68, Parciak, W., Seed size, number and habitat of a fleshyfruited plant: consequences for seedling establishment. Ecology 83, Rees, M., Community structure in sand dune annuals: is seed size weight a key quantity? J. Ecol. 83, Sagar, R., Raghubanshi, A.S., Singh, J.S., 2003a. Asymptotic models of species area curve for measuring diversity of dry tropical forest tree species. Curr. Sci. 84, Sagar, R., Raghubanshi, A.S., Singh, J.S., 2003b. Tree species composition, dispersion and diversity along a disturbance gradient in a dry tropical forest region of India. For. Ecol. Manag. 186, Sagar, R., Singh, J.S., Predominant phenotypic traits of disturbed tropical dry deciduous forest vegetation in northern India. Comm. Ecol. 4, Sagar, R., Singh, J.S., Local plant species depletion in tropical dry deciduous forest of northern India. Environ. Conserv. 31, 1 8. Sagar, R., Singh, J.S., Structure, diversity and regeneration of tropical dry deciduous forest of northern India. Biod and Conser. 14, Salisbury, E.J., The reproductive capacity of plants. Bell, London, England. Sallabanks, R.S., Fruit fate, fugivory, and fruit characteristics: a study of the hawthorn, Crataegus monogyna (Rosaceae). Oecologia 91, Singh, J.S., Singh, K.P., Agrawal, M., Environmental degradation of the Obra-Renukoot-Singrauli area, India and its impact on natural and derived ecosystems. The Environmentalist 11, Singh, K.P., Singh, J.S., Certain structural and functional aspects of dry tropical forests and savanna. Inter. J. Ecol. Environ. Sci. 14, Singh, J.S., Singh, V.K., Phenology of seasonally dry tropical forest. Curr. Sci. 63, S.P.S.S. Inc., SPSS/PC for the IBM PC/XT/AT. Chicago, IL.. Troup, R.S., The Silviculture of Indian Trees, vols. I and II. Oxford at the Clarendon Press, UK. Venable, D.F., Brown, J.S., The selective interactions of dispersal, dormancy and seed size as adaptations for reducing risk in variable environments. Amer. Nat. 131, Verma, D.M., Pant, P.C., Hanfi, M.I., Flora of Raipur, Durg and Rajnandgaon. Botanical Survey of India, Howrah. Westoby, M., Jurado, E., Leishman, M., Comparative evolutionary ecology of seed size. Trends Ecol. Evol. 7,