Abstract
We aim to examine the canopy gap characteristics and evaluate their influence on regeneration, dominance and the early growth of woody species in seasonally dry Shorea robusta forests (Sal forests).
Sixty canopy gaps were surveyed in six randomly located transects belts in seasonally dry subtropical Sal forests of central Nepal. Each transect belt was followed until 10 gap sites were encountered. The equation for the area of an ellipse was used to calculate the size of canopy gap, measuring the longest axis and its perpendicular shorter axis. Number, sizes, ages and causes of tree falls creating canopy gaps along with number and sizes of border trees were identified and recorded. Detailed gap inventories were carried out using square 25-m2 quadrats placed in the middle of each gap. All individuals >2 m in height within the quadrat were identified at the species level and their diameter at breast height was measured. We assigned a nested 4-m2 quadrat to the corner of each 25-m2 quadrat, within which all woody individuals >10 cm tall were identified at the species level, and counted them and their regeneration mechanisms were identified. The height and collar diameter of the tallest individuals were measured. Descriptive statistics was calculated for the variables of interests and Pearson correlation, linear regression, independent-sample t-test and chi-square test were used to relate them and to test for their associations.
The study found mean gap size of 283 m2 and ∼50% gaps of 10–15 years old. Gaps created by natural single-tree falls were significantly more numerous, and their mean size was significantly smaller than those resulting from artificial causes or multiple-tree falls. Gap size correlated with the basal area of felled trees, but it did not correlate with the number of tree falls. While tree fall basal area was significantly positively correlated to the seed-originated seedling to resprout ratio, it was negatively correlated, along with gap area and the basal size of retained trees, to seedling growth. The relative seedling density of Terminalia alata increased with increases in gap areas, while that of S.robusta decreased with increases in tree fall basal area, thereby lowering the plot-level dominance. However, the relative seedling densities of Eugenia operculata and Syzigium cumini were negatively and positively correlated, respectively, with tree fall basal area.
INTRODUCTION
Both natural and artificial disturbances in a forest can cause tree death or injury, which in turn creates openings in the forest cover known as canopy gaps (Yamamoto 2000). These gaps are often filled with other trees, and this replacement phenomenon is termed gap dynamics (Brokaw and Busing 2000). Gap dynamics in tropical and temperate forests have been widely studied (Barton et al. 1989) and are considered to influence plant population dynamics (Lima and Moura 2008; Xi et al. 2008). Regeneration processes in gaps depend on a range of biological factors, such as the life history, physiology and behavior of regenerating species (Lawes et al. 2007); the intensity of the growth of advanced regeneration of predisturbance origin and the colonizing ability of species (Arriaga 2000). However, regeneration also heavily depends on physical gap characteristics. Many authors have proposed gap size to be the most important gap characteristic affecting the recruitment and establishment of various tree species (Brokaw 1985; Li et al. 2005), but it is not the only gap characteristic affecting their establishment (Chandrashekara and Ramakrishnan 1994; Lima and Moura 2008). Other subtle characteristics of canopy gaps influence postgap regeneration (Sapkota et al. 2009), including gap shape; the height and diameter of the surrounding trees (Denslow 1980); gap age (Barik et al. 1992; Chandrashekara and Ramakrishnan 1994; Schnitzer and Carson 2001); the number, causes and sizes of tree fall (Arriaga 2000; Uhl et al. 1988; Xi et al. 2008); gap canopy height (Barik et al. 1992) and the surrounding stand structure (Gagnon et al. 2004). Collectively, these characteristics are commonly termed as ‘gap regimes’ (Gagnon et al. 2004; Yamamoto 2000). The impact of gap regimes on plant population dynamics is of high interest to ecologists (Naaf and Wulf 2007) and must be taken into account when considering population dynamics within forest gaps.
The resources (light, water and nutrients) available to regenerating species are determined by gap characteristics (Van Der Meer and Bongers 1996b). Furthermore, these characteristics often act as sources of within-gap heterogeneity (Heinemann et al. 2000), influencing the availability of suitable microsites in which species can successfully establish and grow (Lima and Moura 2008). For instance, gap size, shape and orientation (Dahir and Lorimer 1996); the characteristics of gap-creating species (Boettcher and Kalisz 1990) and the understory species surviving in gaps (Taylor and Qin 1988) have all been demonstrated to affect the microclimate, and thus species establishment and growth. For these reasons, gap dynamic processes are believed to strongly influence species regeneration (Schnitzer and Carson 2001), and the gap phase regeneration paradigm has widely been accepted in the interpretation of forest organization and ground vegetation management (Lima and Moura 2008).
Several studies have shown the importance of gaps in maintaining the diversity and regeneration of species within old-growth forests in the Indian subtropical (Barik et al. 1992) and southern tropical regions (Chandrashekara and Ramakrishnan 1993, 1994), China (Li et al. 2005; Zang and Wang 2002; Zang et al. 2005) and the temperate (Vetaas 1997) and subtropical forests of Nepal (Sapkota et al. 2009) and North Carolina (Xi et al. 2008). However, the role of gap characteristics in the regeneration and early growth of mature Shorea robusta forests in southeast Asia (hereafter, Sal forests) is still not fully understood (Sapkota et al. 2009). Furthermore, selective logging in Nepalese Sal forests results in frequent gap formation (Sapkota et al. 2009), at rates similar to those observed in India (Chandrashekara and Ramakrishnan 1993, 1994) and Malaysia (Okuda et al. 2003). However, tree regeneration processes in these artificially created gaps have not been studied in Nepalese Sal forests (Sapkota et al. 2009). Therefore, there is a need for more knowledge of the gap dynamics in order to predict future trajectories of species composition and stand structure and to optimize silvicultural strategies for the Nepalese Sal forests, which have been selectively and unsystematically logged for a long time, with too little awareness of the potential consequences.
The aims of this study were to characterize canopy gaps and determine whether gap characteristics influence the mechanisms of seedling recruitment, growth, dominance and relative density of multipurpose tree species. Specifically, we sought to answer the following questions: (i) what are the main attributes of gaps?, (ii) which of them influence the mechanisms of seedling recruitment, their growth and dominance? and (iii) which of them influence the relative density of multipurpose tree species? To meet these aims, we have assessed the main attributes of gaps; the attributes that influence mechanisms of seedling recruitment, growth and dominance and the attributes that influence the relative density of multipurpose tree species in particular.
MATERIALS AND METHODS
Study area
The study was conducted in the state-owned seasonally dry Sal forests of the Nawalparasi district in Nepal (27° 37.5′N, 84° 03′E, ∼190 m above sea level) (Fig. 1). These forests extend continuously from the east to the far west of the country along the East-West National highway and were chosen since they are the most representative Sal forests in Nepal. They cover an area of ∼1 000 ha and are considered to be the largest Sal forests in the region. Mild levels of anthropogenic disturbances (e.g. fuel wood, fodder and litter collection) occur in a uniform pattern across the entire forested region, but no major fire disturbances have been recorded. The forests grow on an alluvial plain that has geological characteristics similar to the Gangetic plain of India in the south, while the underlying sediments originate from tertiary Siwalik material in the north (Anonymous 1994). The Siwalik hills are composed of coarsely bedded stones, crystalline rocks, clays and conglomerates. The climate in the area is tropical to subtropical (Fig. 2), and a typical year can be divided into three main seasons cold, hot and rainy. May and June have the highest mean maximum temperatures (34.6°C), while December and January have the lowest mean minimum temperature (9.9°C). Maximum rainfall occurs during the monsoon season (June to September), with an average monthly rainfall of 531.3 mm, and dew falls from December to February.
location of study area in Sal forests: Chisapani Danda state-managed forests (SMF) in Nawalparasi district of Nepal.
rainfall, relative humidity and temperature data for the study area in Sal forests during January to December (mean of years 2003–2005); average rainfall (vertical bars); average relative humidity (triangles); mean maximum temperature (closed circles); mean minimum temperature (open circles).
Data collection
In the present study we surveyed both naturally and artificially formed canopy gaps resulting from whole-tree fall, where artificial tree fall refers to selective felling carried out by the Department of Forests or illegal loggers. We defined a gap as an area >25 m2 opened by the removal of canopy trees, in which most of the living plants were <5 m tall and <50% of the height of surrounding canopy trees (Lawton and Putz 1988). The upper and lower limit of gap area was determined by assessing whether or not the gaps were formed through tree falls. Irrespective of the size, those gaps were not taken into account, which were formed through the reasons other than tree falls or without tree falls. The East-West National highway's midpoint was taken as a reference point for the gap survey, since the highway divides the forest into two blocks, and we laid out three arbitrary transect belts on both sides of the midpoint (at azimuths of 20°, 50°, 110° and 240°, 250° and 320°), after field reconnaissance and visual interpretation of satellite images. The arbitrary transect belts were each 40 m wide and were laid out by five crew members walking 10 m apart from one another. Each transect belt was simply followed until 10 gap sites were encountered, then each site was identified, labeled with paint and given a serial number. In total, 60 gap sites were surveyed.
The area of each gap was calculated, using the equation for the area of an ellipse, after measuring the longest axis and its perpendicular shorter axis. Trees on the edge of each gap with a diameter at breast height (dbh) >20 cm were identified at the species level and their dbh was measured. In addition, the species, stump diameter, state (standing dead, injured, felled, cut, etc.), mortality type (wind fall, logging, standing dead, trunk broken or uprooted, etc.), and decay class of gap makers (trees with ≥20 cm stump diameter creating a gap) were recorded. The age class of each gap was estimated by visually comparing the decay class of the fallen trees, log remains and stumps from nearby areas subjected to cutting (with the help of office records, persons involved in logging operations and experienced local forest workers), which is an approach that has been successfully used in other studies (Barik et al. 1992; Chandrashekara and Ramakrishnan 1994; Zang and Wang 2002). Where gaps contained multiple fallen trees, we assigned the oldest gap maker as the key determinant of gap age and cause of tree fall, since we considered it to be responsible for creating the gap (Zang and Wang 2002). Spatial autocorrelations and the combined effects of pairs of gaps along the transect belts were avoided by maintaining a minimum distance of 100 m between the border trees of successive sampled gaps.
Detailed gap inventories were carried out between November 2006 and March 2007 using square 25-m2 quadrats placed in the middle of each gap where the two measured axes of the gap intersected. All individuals >2 m in height within the main quadrat were identified at the species level and their dbh was measured. We also assigned a nested 4-m2 quadrat to the corner of each 25-m2 quadrat, within which all woody individuals >10 cm tall were identified at the species level and counted. We grouped these trees into seedlings (individuals 10 cm to 2 m tall) and saplings (individuals >2 m tall) based on their measured heights. The collar diameter and height of the tallest seedlings were also measured in each plot. Natural regeneration mechanisms (seed vs. resprouting) at the seedling layer were identified for all species, mostly through morphological appearance, although belowground analysis of roots was also performed in some cases. Species were identified in situ when possible and by comparison with voucher specimens from the National Herbarium.
Data analysis
To analyze the relationships between the variables of interest, the following were assigned as dependent variables: the density of seed-originated seedlings and resprouts (hereafter collectively referred as seedlings); the height and collar diameter of the tallest seedlings; the average sapling dbh; the plot-level sapling basal area; the ratio of height to collar diameter of the tallest seedling (HD ratio); the ratio of seed-originated seedling density to resprout density; and the ratio of individual species seedling density to total seedling density (relative seedling density). Gap plot-level basal areas of fallen trees and border trees (trees retained around the gap) were calculated and used together with the gap areas as independent variables. In addition, percentage values were calculated for each gap size class, number of tree falls, gap age, and tree fall causes (i.e. natural and artificial).
Descriptive statistics were calculated for each dependent and independent variable used in the analyses. Pearson correlations were then used to relate the measured gap parameters (i.e. gap areas, fallen tree basal areas and number of tree falls) and to see the extent of correlation between each independent and dependent variable. Particularly, correlation analyses between independent variables and the HD ratio and the seed-originated seedling to resprout ratio were performed in order to see the seedlings’ growth performance type (height vs. radial) and recruitment mechanisms (seed vs. resprout) in the forest community. The mean gap sizes resulting from natural and artificial tree falls were compared using an independent-sample t-test. A chi-square test was used to test for significant differences in gap frequencies between gaps created by single natural tree falls and those created by multiple-tree falls and/or artificial causes. In addition, each independent variable was grouped into five to six classes, and mid values of each class was regressed against the plot-level mean values of Simpson's index of dominance (Magurran 2004). Regression analyses were made to see the rate of change in plot-level dominance index with per-unit change in gap area, basal area of felled tree and basal area of retained tree. Prior to analysis, the data sets were normalized by log transformation as deemed necessary. All analyses were performed using SPSS version 14 (SPSS for Windows; SPSS Inc., Chicago, IL).
RESULTS
Gap characteristics
The total number of dead trees creating a gap was 97; of which 56%, 30% and 14% of dead trees were of dead standing, cut and wind fall, respectively. Gap size varied between 67 and 1 418 m2 (Table 1), with an average of 287 ± 28 m2 (mean ± SE). The median gap size was 216. Of the total number of gaps, almost 50% were of medium size (200–400 m2). The next most frequent class was small (60–200 m2), while large gaps (>600 m2) accounted for <10% (Fig. 3a). The number of gaps resulting from a single-tree fall was more than double the number formed by two or more tree falls (Fig. 3b). Furthermore, almost 50% of the gaps were 10–15 years old, the next most frequent were 5–10 years old, while only 5% were 15–20 years old (Fig. 3c). In addition, gaps resulting from natural causes were far more frequent (75%) than those formed by artificial tree falls (Fig. 3d).
descriptive statistics of all variables used for regression and correlation analyses in the gap study of seasonally dry Sal forests
| Variables | n | Minimum | Maximum | Mean ± SE |
| Gap area (m2) | 60 | 67 | 1 417 | 287.4 ± 28.5 |
| Tree fall basal area (cm2) | 60 | 804 | 12 096 | 3 566 ± 293 |
| Border tree basal area (cm2) | 60 | 1 624 | 17 110 | 7 241 ± 589 |
| Number of tree falls | 60 | 1 | 7 | 1.6 ± 0.13 |
| Number of seed-originated seedlings | 60 | 5 | 30 | 15 ± 1 |
| Number of resprouts | 60 | 5 | 45 | 19 ± 1 |
| Seed-originated seedling to resprout ratio | 60 | 0.16 | 3.13 | 0.9 ± 0.07 |
| Tallest seedling height (cm) | 60 | 50 | 200 | 132 ± 4.4 |
| Collar diameter of tallest seedlings (mm) | 60 | 4.6 | 29.5 | 14.3 ± 0.71 |
| HD ratio | 60 | 40.4 | 206.5 | 102.3 ± 4.6 |
| Average sapling dbh (cm) | 54 | 1.1 | 8.8 | 3.2 ± 0.23 |
| Plot-level sapling basal area (cm2) | 54 | 1.6 | 280 | 76.2 ± 10.2 |
| Simpson's index of dominance | 60 | 0.26 | 0.87 | 0.52 ± 0.02 |
| Relative seedling density | ||||
| Shorea robusta | 60 | 0.27 | 0.93 | 0.67 ± 0.018 |
| Lagerstroemia parviflora | 22 | 0.03 | 0.20 | 0.07 ± 0.009 |
| Terminalia alata | 16 | 0.01 | 0.14 | 0.06 ± 0.009 |
| Eugenia operculata | 15 | 0.02 | 0.17 | 0.06 ± 0.017 |
| Terminalia bellirica | 5 | 0.02 | 0.06 | 0.04 ± 0.006 |
| Syzigium cumini | 5 | 0.05 | 0.14 | 0.09 ± 0.020 |
| Variables | n | Minimum | Maximum | Mean ± SE |
| Gap area (m2) | 60 | 67 | 1 417 | 287.4 ± 28.5 |
| Tree fall basal area (cm2) | 60 | 804 | 12 096 | 3 566 ± 293 |
| Border tree basal area (cm2) | 60 | 1 624 | 17 110 | 7 241 ± 589 |
| Number of tree falls | 60 | 1 | 7 | 1.6 ± 0.13 |
| Number of seed-originated seedlings | 60 | 5 | 30 | 15 ± 1 |
| Number of resprouts | 60 | 5 | 45 | 19 ± 1 |
| Seed-originated seedling to resprout ratio | 60 | 0.16 | 3.13 | 0.9 ± 0.07 |
| Tallest seedling height (cm) | 60 | 50 | 200 | 132 ± 4.4 |
| Collar diameter of tallest seedlings (mm) | 60 | 4.6 | 29.5 | 14.3 ± 0.71 |
| HD ratio | 60 | 40.4 | 206.5 | 102.3 ± 4.6 |
| Average sapling dbh (cm) | 54 | 1.1 | 8.8 | 3.2 ± 0.23 |
| Plot-level sapling basal area (cm2) | 54 | 1.6 | 280 | 76.2 ± 10.2 |
| Simpson's index of dominance | 60 | 0.26 | 0.87 | 0.52 ± 0.02 |
| Relative seedling density | ||||
| Shorea robusta | 60 | 0.27 | 0.93 | 0.67 ± 0.018 |
| Lagerstroemia parviflora | 22 | 0.03 | 0.20 | 0.07 ± 0.009 |
| Terminalia alata | 16 | 0.01 | 0.14 | 0.06 ± 0.009 |
| Eugenia operculata | 15 | 0.02 | 0.17 | 0.06 ± 0.017 |
| Terminalia bellirica | 5 | 0.02 | 0.06 | 0.04 ± 0.006 |
| Syzigium cumini | 5 | 0.05 | 0.14 | 0.09 ± 0.020 |
descriptive statistics of all variables used for regression and correlation analyses in the gap study of seasonally dry Sal forests
| Variables | n | Minimum | Maximum | Mean ± SE |
| Gap area (m2) | 60 | 67 | 1 417 | 287.4 ± 28.5 |
| Tree fall basal area (cm2) | 60 | 804 | 12 096 | 3 566 ± 293 |
| Border tree basal area (cm2) | 60 | 1 624 | 17 110 | 7 241 ± 589 |
| Number of tree falls | 60 | 1 | 7 | 1.6 ± 0.13 |
| Number of seed-originated seedlings | 60 | 5 | 30 | 15 ± 1 |
| Number of resprouts | 60 | 5 | 45 | 19 ± 1 |
| Seed-originated seedling to resprout ratio | 60 | 0.16 | 3.13 | 0.9 ± 0.07 |
| Tallest seedling height (cm) | 60 | 50 | 200 | 132 ± 4.4 |
| Collar diameter of tallest seedlings (mm) | 60 | 4.6 | 29.5 | 14.3 ± 0.71 |
| HD ratio | 60 | 40.4 | 206.5 | 102.3 ± 4.6 |
| Average sapling dbh (cm) | 54 | 1.1 | 8.8 | 3.2 ± 0.23 |
| Plot-level sapling basal area (cm2) | 54 | 1.6 | 280 | 76.2 ± 10.2 |
| Simpson's index of dominance | 60 | 0.26 | 0.87 | 0.52 ± 0.02 |
| Relative seedling density | ||||
| Shorea robusta | 60 | 0.27 | 0.93 | 0.67 ± 0.018 |
| Lagerstroemia parviflora | 22 | 0.03 | 0.20 | 0.07 ± 0.009 |
| Terminalia alata | 16 | 0.01 | 0.14 | 0.06 ± 0.009 |
| Eugenia operculata | 15 | 0.02 | 0.17 | 0.06 ± 0.017 |
| Terminalia bellirica | 5 | 0.02 | 0.06 | 0.04 ± 0.006 |
| Syzigium cumini | 5 | 0.05 | 0.14 | 0.09 ± 0.020 |
| Variables | n | Minimum | Maximum | Mean ± SE |
| Gap area (m2) | 60 | 67 | 1 417 | 287.4 ± 28.5 |
| Tree fall basal area (cm2) | 60 | 804 | 12 096 | 3 566 ± 293 |
| Border tree basal area (cm2) | 60 | 1 624 | 17 110 | 7 241 ± 589 |
| Number of tree falls | 60 | 1 | 7 | 1.6 ± 0.13 |
| Number of seed-originated seedlings | 60 | 5 | 30 | 15 ± 1 |
| Number of resprouts | 60 | 5 | 45 | 19 ± 1 |
| Seed-originated seedling to resprout ratio | 60 | 0.16 | 3.13 | 0.9 ± 0.07 |
| Tallest seedling height (cm) | 60 | 50 | 200 | 132 ± 4.4 |
| Collar diameter of tallest seedlings (mm) | 60 | 4.6 | 29.5 | 14.3 ± 0.71 |
| HD ratio | 60 | 40.4 | 206.5 | 102.3 ± 4.6 |
| Average sapling dbh (cm) | 54 | 1.1 | 8.8 | 3.2 ± 0.23 |
| Plot-level sapling basal area (cm2) | 54 | 1.6 | 280 | 76.2 ± 10.2 |
| Simpson's index of dominance | 60 | 0.26 | 0.87 | 0.52 ± 0.02 |
| Relative seedling density | ||||
| Shorea robusta | 60 | 0.27 | 0.93 | 0.67 ± 0.018 |
| Lagerstroemia parviflora | 22 | 0.03 | 0.20 | 0.07 ± 0.009 |
| Terminalia alata | 16 | 0.01 | 0.14 | 0.06 ± 0.009 |
| Eugenia operculata | 15 | 0.02 | 0.17 | 0.06 ± 0.017 |
| Terminalia bellirica | 5 | 0.02 | 0.06 | 0.04 ± 0.006 |
| Syzigium cumini | 5 | 0.05 | 0.14 | 0.09 ± 0.020 |
gap plot characteristics and their distribution in seasonally dry Sal forests: the percentages of gaps based on gap size classes (a), percentages of gaps based on number of tree falls (b), percentages of gaps based on gap age (c) and percentages of gaps based on causes of tree fall (d).
Significantly more gaps were created by a single natural tree fall than by multiple-tree falls or artificial causes (; P < 0.0001). The gap sizes formed by natural tree falls were also significantly smaller (log value ± SE = 2.3 ± 0.04; original value ± SE = 260 ± 31) than those formed by artificial tree falls (2.5 ± 0.08; 375 ± 66) (t[58] = 3.8; P = 0.05). Furthermore, there was no correlation between gap size and the number of fallen trees (r = −0.049; n = 60; P = 0.707). However, significant correlation was found between the number of tree falls and the basal area of the fallen trees (r = 0.420; n = 60; P = 0.001). The correlation between gap size and the basal area of the fallen trees was also significant (r = 269; n = 60; P = 0.038).
Correlations of regeneration and growth attributes with gap characteristics
There was a significant positive correlation between the basal area of fallen tree stumps and the seed-originated seedling to resprout ratio (Table 2), and all of the studied gap characteristics were significantly negatively correlated with at least one seedling/sapling growth attribute (Table 2). In addition, gap area was negatively correlated with the average sapling dbh in the plots, while border basal area was negatively correlated with both the average sapling basal area and dbh. Furthermore, the HD ratio of the tallest seedlings decreased as tree fall basal area increased (Table 2). However, none of the regeneration attributes were explained by gap area or border tree basal area.
correlation (r) of the recruitment and growth attributes with the gap plot characteristics (n = 60) in seasonally dry Sal forests
| Recruitment and growth attributes | Gap plot characteristics | ||
| Gap area (log10) | Border BA | Tree fall BA | |
| Seed-originated seedling density | −0.025 | −0.130 | −0.027 |
| Resprout density | −0.192 | −0.132 | −0.200 |
| Seed-originated seedling to resprout ratio | 0.112 | 0.038 | 0.298* |
| Seedling height | 0.074 | 0.006 | −0.114 |
| Seedling collar diameter | 0.035 | 0.042 | 0.141 |
| HD ratio | 0.025 | −0.057 | −0.272* |
| Average sapling dbh | −0.251* | −0.386** | 0.105 |
| Plot-level sapling basal area | −0.244 | −0.356** | 0.059 |
| Recruitment and growth attributes | Gap plot characteristics | ||
| Gap area (log10) | Border BA | Tree fall BA | |
| Seed-originated seedling density | −0.025 | −0.130 | −0.027 |
| Resprout density | −0.192 | −0.132 | −0.200 |
| Seed-originated seedling to resprout ratio | 0.112 | 0.038 | 0.298* |
| Seedling height | 0.074 | 0.006 | −0.114 |
| Seedling collar diameter | 0.035 | 0.042 | 0.141 |
| HD ratio | 0.025 | −0.057 | −0.272* |
| Average sapling dbh | −0.251* | −0.386** | 0.105 |
| Plot-level sapling basal area | −0.244 | −0.356** | 0.059 |
BA refers to basal area.
P values <0.05.
correlation (r) of the recruitment and growth attributes with the gap plot characteristics (n = 60) in seasonally dry Sal forests
| Recruitment and growth attributes | Gap plot characteristics | ||
| Gap area (log10) | Border BA | Tree fall BA | |
| Seed-originated seedling density | −0.025 | −0.130 | −0.027 |
| Resprout density | −0.192 | −0.132 | −0.200 |
| Seed-originated seedling to resprout ratio | 0.112 | 0.038 | 0.298* |
| Seedling height | 0.074 | 0.006 | −0.114 |
| Seedling collar diameter | 0.035 | 0.042 | 0.141 |
| HD ratio | 0.025 | −0.057 | −0.272* |
| Average sapling dbh | −0.251* | −0.386** | 0.105 |
| Plot-level sapling basal area | −0.244 | −0.356** | 0.059 |
| Recruitment and growth attributes | Gap plot characteristics | ||
| Gap area (log10) | Border BA | Tree fall BA | |
| Seed-originated seedling density | −0.025 | −0.130 | −0.027 |
| Resprout density | −0.192 | −0.132 | −0.200 |
| Seed-originated seedling to resprout ratio | 0.112 | 0.038 | 0.298* |
| Seedling height | 0.074 | 0.006 | −0.114 |
| Seedling collar diameter | 0.035 | 0.042 | 0.141 |
| HD ratio | 0.025 | −0.057 | −0.272* |
| Average sapling dbh | −0.251* | −0.386** | 0.105 |
| Plot-level sapling basal area | −0.244 | −0.356** | 0.059 |
BA refers to basal area.
P values <0.05.
Relationships of dominance and relative seedling density with gap characteristics
The Simpson's index of dominance showed a significant linear but negative relationship with the basal area of fallen trees at the plot level (Fig. 4b), but it was not significantly related with either gap area or border tree basal area. The relative seedling density of Terminalia alata increased with gap area (Table 3). Furthermore, tree fall basal areas were negatively correlated with relative seedling densities of S.robusta and Eugenia operculata and positively correlated with those of Syzigium cumini. However, there was no significant relationship between the border tree basal area and the relative seedling density of examined tree species at the plot level (Table 3).
correlation (r) of the relative seedling densities of six multipurpose tree species with gap plot characteristics in seasonally dry Sal forests
| Gap plot characteristics | ||||
| Species | Main uses | Gap area | Border BA | Tree fall BA |
| Shorea robusta | Fo, Fu, T, M | −0.153 | −0.029 | −0.358* |
| Lagerstroemia parviflora | Fu, T, M | 0.203 | 0.032 | 0.243 |
| Terminalia alata | Fo, Fu, T, M | 0.557* | 0.380 | 0.068 |
| Eugenia operculata | Fo, Fu, T, M | −0.047 | −0.148 | −0.557* |
| Terminalia bellirica | Fo, Fu, M | 0.644 | 0.705 | −0.437 |
| Syzigium cumini | Fo, Fu, T, M | 0.324 | −0.016 | 0.882* |
| Gap plot characteristics | ||||
| Species | Main uses | Gap area | Border BA | Tree fall BA |
| Shorea robusta | Fo, Fu, T, M | −0.153 | −0.029 | −0.358* |
| Lagerstroemia parviflora | Fu, T, M | 0.203 | 0.032 | 0.243 |
| Terminalia alata | Fo, Fu, T, M | 0.557* | 0.380 | 0.068 |
| Eugenia operculata | Fo, Fu, T, M | −0.047 | −0.148 | −0.557* |
| Terminalia bellirica | Fo, Fu, M | 0.644 | 0.705 | −0.437 |
| Syzigium cumini | Fo, Fu, T, M | 0.324 | −0.016 | 0.882* |
BA refers to basal area. Fo, Fu, T and M refer to fodder, fuel wood, timber and medicine, respectively.
P values <0.05.
correlation (r) of the relative seedling densities of six multipurpose tree species with gap plot characteristics in seasonally dry Sal forests
| Gap plot characteristics | ||||
| Species | Main uses | Gap area | Border BA | Tree fall BA |
| Shorea robusta | Fo, Fu, T, M | −0.153 | −0.029 | −0.358* |
| Lagerstroemia parviflora | Fu, T, M | 0.203 | 0.032 | 0.243 |
| Terminalia alata | Fo, Fu, T, M | 0.557* | 0.380 | 0.068 |
| Eugenia operculata | Fo, Fu, T, M | −0.047 | −0.148 | −0.557* |
| Terminalia bellirica | Fo, Fu, M | 0.644 | 0.705 | −0.437 |
| Syzigium cumini | Fo, Fu, T, M | 0.324 | −0.016 | 0.882* |
| Gap plot characteristics | ||||
| Species | Main uses | Gap area | Border BA | Tree fall BA |
| Shorea robusta | Fo, Fu, T, M | −0.153 | −0.029 | −0.358* |
| Lagerstroemia parviflora | Fu, T, M | 0.203 | 0.032 | 0.243 |
| Terminalia alata | Fo, Fu, T, M | 0.557* | 0.380 | 0.068 |
| Eugenia operculata | Fo, Fu, T, M | −0.047 | −0.148 | −0.557* |
| Terminalia bellirica | Fo, Fu, M | 0.644 | 0.705 | −0.437 |
| Syzigium cumini | Fo, Fu, T, M | 0.324 | −0.016 | 0.882* |
BA refers to basal area. Fo, Fu, T and M refer to fodder, fuel wood, timber and medicine, respectively.
P values <0.05.
linear relationships between Simpson's index and gap plot characteristics in seasonally dry Sal forests: Simpson's index vs. gap size (a), Simpson's index vs. tree fall basal area (b), and Simpson's index vs. border tree basal area.
DISCUSSION
The mean gap areas observed in this study were similar to those recorded in studies carried out in north-eastern India (Arunachalam and Arunachalam 2000; Barik et al. 1992) and central Japan (Yamamoto 1995). However, other studies performed in north-eastern Mexico (Arriaga 2000), the Atlantic Montane Rain Forest (Lima and Moura 2008) and Western Ghat India (Chandrashekara and Ramakrishnan 1994) reported smaller mean gap areas than those we observed. The suggestions of these differences are mainly of 2-folds. First, the gap size is often affected by forest developmental stage (Yamamoto 2000), since mean gap size is generally larger in old-growth stands than in younger stands (Spies et al. 1990). The investigated forest was an old-growth forest (Sapkota et al. 2009), and ∼50% of the observed gaps were of a medium size (200–400 m2), which in turn raised the measured mean gap size. Second, the method used for gap measurement also greatly influences the mean gap size (Lima 2005). For instance, the variation in using the method for gap measurement between our study and the study made by Lima and Moura (2008) has resulted in the variation in mean gap size. Gap size is influenced by various factors, such as the number, causes and sizes of tree falls (Lima and Moura 2008; Yamamoto 2000). Although many previous studies have found gap size to be correlated with the number of tree falls (Lima and Moura 2008), our results do not support such a relationship. Instead, our data support the conclusion of a study from north-eastern Mexico that the number of tree falls does not necessarily explain gap size (Arriaga 2000).
The positive correlations observed between the basal area of the fallen trees and both their number and gap size were consistent with observations from studies performed in Atlantic Montane Rain forests (Lima and Moura 2008) and Nouragues (Van Der Meer and Bongers 1996a). These relationships may be linked to the cause, size, and number of tree falls during a gap formation event. The areas of gaps formed by artificially felled trees are generally bigger, since large trees are normally logged to maximize timber production (Gagnon et al. 2004). In addition, logging operations may knock down additional trees during the felling process (Chandrashekara and Ramakrishnan 1994; Lima and Moura 2008). Therefore, logging operations may have caused the large gaps observed in our study, in which most of the trees that had been felled were large, single or multiple trees. The significantly higher occurrence of gaps created by natural, single-tree falls (e.g. standing dead and uprooted trees due to high wind velocities) seen in the present study, and their association with small gaps, supports this hypothesis.
Considering the positive correlation between the basal area of fallen trees and gap size, the positive association between the seed-originated seedling to resprout ratio and basal area of fallen trees can be attributed to sizeable canopy openings caused by large, single- or multiple-tree falls and their impact on the mechanisms of seedling origin. The surface of the soil in a large tree fall opening generally receives higher light intensities, and thus the soil temperature is generally higher than that in small tree fall openings or areas with a closed canopy (Denslow et al. 1998). Furthermore, Bullock (2000) clearly demonstrated that the competition for light, nutrients and water is weaker in large gaps than in intact vegetation and/or small gaps. Hence, the ability of a species to establish from the seed is likely to be higher in large tree fall gaps than in small tree fall gaps and/or in intact vegetation (Bullock 2000). However, the observed negative relationship between the HD ratio and basal area of fallen trees may indicate that gaps caused by multiple falls of large trees favor the radial growth of seedlings over their height growth. These findings support the results of a study performed in Indonesia (Tuomela et al. 1996), where a negative relationship was observed between seedling height and gap size. This relationship may be attributable to gap size-related differences in the seedlings’ access to and/or competition for light (Dekker et al. 2007). Seedlings in larger gaps often have continuous access to light, leading to relatively weak competition among individuals for light and space. Therefore, a higher investment in radial stem growth at the cost of height gain may occur (Sterck and Bonger 1998), since equal access to light is often ensured for all individuals. Conversely, the lower amounts of light in small gaps lead to seedlings competing for this resource (Bullock 2000), and individuals with rapid height growth compete for light more successfully, in accordance with the ‘competitive exclusion’ among individuals principle (Dekker et al. 2007). Furthermore, similar relationships have been observed in Ghana (Doland Nichols et al. 1998) and the Netherlands (Dekker et al. 2007).
However, the negative relationship between gap area and average sapling dbh in our study corroborated an earlier study in Indonesia, in which a negative effect of larger gaps on seedling growth was found (Tuomela et al. 1996). Furthermore, large openings may increase plant resources during early developmental stages, while slowly diminishing them later (Fraver et al. 1998). In addition, the higher light availability in large gaps often increases plant resources during a certain period after gap creation, and the resultant plant growth increases resources for herbivores and, in turn, predators. Consequently, populations of herbivores and predators that feed on plants and young leaves (Richards and Coley 2007) may increase, ultimately reducing the growth and survival of saplings in larger gaps (Brokaw 1985). Our results also agree with a study performed in Panama (Fraver et al. 1998), in which rapid growth rates resulting from the creation of large gaps diminished after 42 months of regrowth and followed the rates similar to those found in the understory.
Similarly, the negative relationship between the border tree basal area of gaps and the saplings’ dbh and plot-level basal area can be attributed to pests (Richards and Coley 2007) and/or root competition between the border trees and saplings (Brokaw 1985). For instance, Bylund (1997) and Sapkota et al. (2009) demonstrated that old-growth forests containing large-sized trees (Bullock 2000) often serve as host trees for various herbivores (e.g. ants, termites, grasshoppers; Indra Sapkota), which feed on the young leaves of seedlings. Hence, these herbivores might have affected the seedling population, resulting in the current sapling populations’ lower growth and size. Root competition for nutrients and water between the seedlings and bordering trees might also contribute to reductions in seedling growth and survival, thereby negatively affecting the plot-level sapling basal area.
The positive correlation observed between gap area and the relative seedling density of T.alata may be due to its light requirements. This species is a strong light demander, and its germination is successful in bare ground if soil moisture is available (Sen et al. 2008). Furthermore, its seedlings are tolerant to continuous light and/or dry conditions due to their deep root systems (Sen et al. 2008). In contrast, the Simpson's index of dominance and relative seedling density of S.robusta decreased with increases in the basal area of fallen trees, suggesting that the dominance of this species is reduced in large tree fall gaps. These reductions may be attributed to its typical ephemeral seedling stock, as observed in other members of the Dipterocarpaceae family (Rosario 1982). Dipterocarps may be heavily adversely affected by shocks caused by sudden exposure to sunlight, which are often coupled with wind factors (Rosario 1982). Furthermore, dipterocarps require partial shade between the germination and pole stages, while they require increasing amounts of light as they mature (Mauricio 1985). Therefore, the natural regeneration of dipterocarps is likely to be most efficient in relatively small gap environments, which provide partial light and wind shelter (Tuomela et al. 1996).
Relatively high soil moisture-based emergence and survival strategy of larger seeded as well as shade-tolerant species such as E.operculata (Jackson 1994) may explain their lower relative seedling density in large tree fall gaps. Openings, due to multiple falls of large trees, may allow continuous light availability, thereby increasing water stress and hence mortality of young seedlings (Bullock 2000). However, a positive correlation was observed between the basal area of fallen trees and the relative seedling density of S.cumini, which was probably related to its seed dispersal processes. S.cumini seeds are widely dispersed by birds and bats (Morton 1987), and preferential dispersal may occur in openings created by multiple-tree falls since bird and bat activity mostly occurs in such gaps (Bullock 2000).
The results of our study suggest that large gaps do not seem suitable for overall seedling growth and regeneration for most of the examined species. Furthermore, multiple-tree falls, resulting in high fallen tree basal areas, open up more space in the canopy, which does not seem beneficial to the early growth of woody species either. In addition, large trees around the gap lower the seedling growth and regeneration performance. Therefore, we recommend the single tree-based selective logging, while ensuring the creation of small gaps, since this may assist natural regeneration and the early stimulation of seedling growth. Removing large trees in order to reduce the basal area of retained trees may also be beneficial. However, longer term studies are needed to validate our findings. The longer term studies are also needed to elucidate the species-specific survival and growth strategies related to gap openings.
FUNDING
Sida.
We thank Bijaya Raj Paudel, Padam Prasad Nepal, Lok Raj Nepal and Bhairab Prasad Ghimire for their help with logistics. We are grateful to Shesh Kanta Bhandari, Bishnu Bahadur Thapa, Shyam Sundar Bhandari and Tek Bahadur Rayamajhi for their constant support during the forest inventory. Meena Kunwar and Poorneshwor Subedi provided the satellite images. We thank two anonymous reviewers for valuable suggestions. We also thank Chaudhary, Dr. Sushim Ranjan Baral and Puran Prasad Kurmi for their help with species identification in the field and the Herbarium. Dr. John Blackwell edited our manuscript for English language.
