The authors have declared that no competing interests exist.
Conceived and designed the experiments: RN AD KI. Performed the experiments: RN US AD. Analyzed the data: RN KI. Wrote the paper: RN US AD KI.
Logging and hunting are two key direct threats to the survival of wildlife in the tropics, and also disrupt important ecosystem processes. We investigated the impacts of these two factors on the different stages of the seed dispersal cycle, including abundance of plants and their dispersers and dispersal of seeds and recruitment, in a tropical forest in north-east India. We focused on hornbills, which are important seed dispersers in these forests, and their food tree species. We compared abundances of hornbill food tree species in a site with high logging and hunting pressures (heavily disturbed) with a site that had no logging and relatively low levels of hunting (less disturbed) to understand logging impacts on hornbill food tree abundance. We compared hornbill abundances across these two sites. We, then, compared the scatter-dispersed seed arrival of five large-seeded tree species and the recruitment of four of those species. Abundances of hornbill food trees that are preferentially targeted by logging were two times higher in the less disturbed site as compared to the heavily disturbed site while that of hornbills was 22 times higher. The arrival of scatter-dispersed seeds was seven times higher in the less disturbed site. Abundances of recruits of two tree species were significantly higher in the less disturbed site. For another species, abundances of younger recruits were significantly lower while that of older recruits were higher in the heavily disturbed site. Our findings suggest that logging reduces food plant abundance for an important frugivore-seed disperser group, while hunting diminishes disperser abundances, with an associated reduction in seed arrival and altered recruitment of animal-dispersed tree species in the disturbed site. Based on our results, we present a conceptual model depicting the relationships and pathways between vertebrate-dispersed trees, their dispersers, and the impacts of hunting and logging on these pathways.
Hunting and logging are among the major threats affecting wildlife in tropical forests [
Hunting pressures on vertebrates in the tropics are often deemed unsustainable [
Most of our current understanding of the impacts of hunting and logging on seed dispersal and plant recruitment is from the Neotropics [
In this study we aimed to understand how logging and hunting, through the associated losses of fruiting trees and a key disperser group, affect scatter dispersal and recruitment patterns of large seeded forest trees. We focused on large-seeded tree species as they depend on large-bodied frugivores for dispersal. We selected hornbills, which are the largest avian frugivores and dispersers in Asian tropical forests [
We expected that logging will negatively affect abundance of hornbill food plants, particularly those targeted by logging. This along with hunting, which results in direct removal of individuals from the population, will negatively affect the abundance of hornbills at the heavily disturbed site. We expected that reduced hornbill abundance would result in reduced arrival of scatter-dispersed seeds on forest floor. Given that sites have experienced these threats in the past, we expected to detect a signature of the negative impacts of hunting and logging on recruitment patterns—that is, reduced or altered recruit abundance in the heavily disturbed sites as compared to the less disturbed sites. Based on our results, we present a conceptual model depicting the relationships and feedbacks between vertebrate dispersed trees, their dispersers, and the two main threats to tropical biodiversity, hunting and logging.
The office of the Principal Chief Conservator of Forests (Wildlife and Biodiversity), Government of Arunachal Pradesh, Itanagar approved the study and gave permission (No: CWL/G/13(17)06-07/Pt-III/4219-32) to conduct research in Namdapha Tiger Reserve (Protected Area) and Miao Reserved Forest in Arunachal Pradesh state.
The study was conducted between November 2011 and March 2012 in the Namdapha Tiger Reserve (1985 km2; 200–4,500 m above sea level; 27°23’30”—27°39’40”N and 96°15’2”—96°58’33”E) and the adjoining Miao Reserved Forest (121 km2; 200–1300 m; 27°25’49”—27°30’02”N and 96°8’33”—96°18’59”E) in Arunachal Pradesh, north-east India (
The area shaded gray is the Namdapha Tiger Reserve. The enclosed box shows the intensively sampled area with unbroken lines representing the eight trails in the less disturbed site (Namdapha Tiger Reserve) and broken lines representing the four trails in the heavily disturbed site (Miao Reserved Forest).
More than 490 bird species have been recorded from Namdapha and adjoining areas [
Our study site, inside Namdapha Tiger Reserve (less disturbed site), was on a plateau (14 km2; 500–700 m above sea level) near the western border of the reserve. The other site in Miao Reserved Forest (heavily disturbed site) was c. 10 km2 with elevation ranging from 400 to 700 m and 20 km straight line distance from our study site in Namdapha.
In spite of a ban on hunting under the Wildlife (Protection) Act, 1972, [
Miao Reserved Forest is located within one kilometer of Miao town, which is inhabited by several communities (Singpho, Tangsa, Wancho, Nyishi, Nocte, Adi, Lisu, Chakma and Nepali) with some of these tribes using hornbill body parts. Members of some of these tribes hunt ungulates, arboreal mammals, hornbills and small birds with muzzle loader guns (locally fabricated firearms), rifles, air guns or catapults (R. Naniwadekar, pers. obs.). In the past four years (2008–12), we recorded four instances of hunting of hornbills (three instances of Great Hornbills and one of Wreathed hornbills). Miao was notified as a Reserved Forest in 1962 and the earliest record of systematic logging operations is from 1978 (Working Plan, Jairampur Forest Division). A ban on logging through a Supreme Court order was in effect from 1996 till 2008 [
The two sites were similar to each other in geology, rainfall and elevation and forest/vegetation type. The straight-line distance between the furthest sampling sites was approximately 28 km (
We identified hornbill food plant species at both sites based on prior information [
To model the effect of logging on hornbill food plants, we used Generalized Linear Models (GLM) on the count data of hornbill food plants with sites (less disturbed and heavily disturbed), type of hornbill food plant (logged food plants, not logged food plants and strangler figs), and an interaction between these two factors as predictor variables. We initially ran a GLM with Poisson errors but the results indicated over-dispersion in the data. Therefore, we used a negative binomial GLM with a log link.
We carried out trail walks at both sites between December 2011 and February 2012. In the less disturbed site, we walked trails in the mornings (0600–0900 hr) and late afternoons (1300–1500 hr), as hornbill activity is high during these time periods. We had eight trails, which were walked 12 times each. We largely avoided monitoring the same trail twice on the same day, except on seven occasions. In the heavily disturbed site, trails were walked only in the early mornings (0600–0900 hr) as human activity during the day could affect hornbill presence/detection in the late afternoons. One or two observers walked these trails and recorded hornbill species and number of individuals. The species identity was noted on hearing calls. However, we used only the visual detections for the analysis. We monitored each trail four times in a month. The total effort in less disturbed and heavily disturbed sites was 144 km and 72 km, respectively. Since we recorded only seven individuals in three detections in the heavily disturbed site in 72 km of effort, we were unable to estimate detection probability for the two sites separately. However, hornbills are large, striking and canopy-dwelling birds and are unlikely to be missed.
We used Generalized Linear Mixed-effect Models (GLMM) with Poisson errors to compare counts of hornbills sighted between less disturbed and heavily disturbed sites. There were differences in counts of hornbills across the three months (December = 173 individuals, January = 32 individuals, February = 41 individuals; effort = 48 km in each month in the less disturbed site), therefore, we incorporated ‘month’ effects as a random effect in the model. In addition, we walked each trail 12 times during the study duration, there was variation in the total number of hornbill sightings across the different trails (range: 2–105 individuals; effort: 18 km per trail). Therefore, we also used ‘trail’ as a random effect. In the heavily disturbed site, the total number of hornbill individuals seen was very low throughout the sampling period. The GLMM with trail and month effects as random and site effects (less disturbed and heavily disturbed) as fixed indicated over-dispersion in the data. We, therefore, incorporated individual observations as a random effect in the model following Elston et al. [
In both sites along each trail, we established 200 (1 m × 1m) plots on the forest floor to record the arrival of scatter-dispersed seeds (1600 m2 in less disturbed site; 800 m2 in heavily disturbed site). We monitored these plots every 15 days (except once when the monitoring interval was 29–31 days) from December 2011–February 2012 (five occasions). We monitored the arrival of dispersed seeds of five large-seeded hornbill food plants—
Tree species | Fruit length | Fruit width | Seed length | Seed width |
---|---|---|---|---|
36.5 (0.9) | 28.8 (1.3) | 27.8 (3.6) | 16.9 (0.96) | |
36.7 (0.4) | 27.4 (0.1) | 34.3 (2.8) | 22.9 (1.2) | |
30.4 (0.5) | 18.3 (0.6) | 28.3 (0.95) | 17.2 (0.5) | |
40.6 (0.8) | 24.7 (0.6) | 33.8 (0.96) | 15.1 (1.1) | |
43.3 (0.7) | 21.8 (0.4) | 35.6 (2.2) | 17.4 (1.0) |
Although seed arrival rates at the less disturbed site were similar between the two years, they differed in species composition [
Along each trail at both sites, we established belt transects measuring 1500 m × 3 m. For two trails in the less disturbed site, we sampled 750 m × 3 m. We recorded recruits of four of the five large-seeded hornbill food plants,
To compare the recruits in four size classes between the less disturbed and the heavily disturbed site, we used GLM on the count data of recruits with sites (less disturbed and heavily disturbed), size of the recruits (10–30 cm, 30–50 cm, 50–100 cm, 100–150 cm) and an interaction between these two factors as predictor variables. We used the natural logarithm of area sampled as an offset to control for varying sampling effort between trails. We detected only 6 and 10 recruits of
We carried out all the analysis using R Language, version 2.15.1 [
In the heavily disturbed site, the density of cut logs/stumps (GBH≥100 cm) was 11±1 logs/stumps per ha (mean ± SD). There was no logging in the less disturbed site. Despite a lower sampling effort in the heavily disturbed site, we recorded hunters (men with guns) on six occasions, while we heard human presence only once in the less disturbed site. Additionally, we saw feathers of Wreathed Hornbills on the forest floor next to a temporary camp in the heavily disturbed site, outside of the trail sampling.
We identified six species of figs and 15 species of non-fig food plants (
The overall abundance of hornbill food plants was two times higher in the less disturbed site as compared to the heavily disturbed site (negative binomial GLM,
(a) logged, food plants (b) not logged and (c) strangler figs in the less disturbed site (Namdapha) and the heavily disturbed site (Miao).
Tree species | Tree density in less disturbed | Tree density in heavily disturbed | Seed arrival in less disturbed | Seed arrival in heavily disturbed |
---|---|---|---|---|
0.375 (0.292) | 0.167 (0.167) | 863 (525) | 0 (0) | |
2.458 (0.804) | 0.250 (0.160) | 256 (71) | 0 (0) | |
1.500 (0.508) | 0.000 (0.000) | 50 (25) | 0 (0) | |
0.042 (0.042) | 0.25 (0.25) | 38 (16) | 13 (13) | |
0.125 (0.061) | 0.083 (0.083) | 306 (92) | 150 (74) |
During trail walks, we detected four species of hornbills (Great, Wreathed, Rufous-necked and Brown Hornbill) in the less disturbed site and two species (Wreathed and Rufous-necked Hornbill) in the heavily disturbed site. The total number of hornbills seen across the trails varied from 2–105 individuals at the less disturbed site and 0–6 individuals at the heavily disturbed site. The total number of individuals detected varied across months in the less disturbed site (December: 173, January: 41 and February 32). Overall encounter rates of hornbills was 22 times higher in the less disturbed site as compared to the heavily disturbed site (
We detected seeds of all the five species (
Bootstrap mean and 95% confidence intervals of number of seeds per day per 200 m2 (200 1 m2 plots per trail) in the less disturbed site (Namdapha) and the heavily disturbed site (Miao).
The overall abundance of recruits across all size classes was 20 times higher for
Bootstrap mean and 95% confidence intervals of number of recruits per ha in the less disturbed site (Namdapha) and the heavily disturbed site (Miao) for a)
We found that the heavily disturbed site had reduced abundances of fruit plants, hornbills and scatter-dispersed seeds, and showed altered recruitment patterns. These findings suggest that logging can result in lowered abundance of hornbill food tree species indirectly affecting hornbill abundance, scatter-dispersal of seeds and their recruitment. Hunting can result in direct reduction of dispersers like hornbills, indirectly affecting scatter-dispersal of seeds and their recruitment. Our inferences are based on comparing two similar tropical forest sites close to each other (~ 20 km) that primarily differed in the extent of hunting and logging. Due to the absence of multiple comparable sites representing the less disturbed scenario, we did not have more replicate sites. However, we found that several different stages of the seed dispersal cycle consistently showed differences between our two sites as expected by the hypothesized effects of hunting and logging. This suggests that our findings were robust and that logging and hunting underlie these differences. Based on the findings of this study and other expected impacts, we propose a conceptual model that outlines the direct and indirect impacts of logging and hunting on the seed dispersal cycle (
The relationship (black solid arrow) across different stages of seed dispersal (shown in open rectangular boxes) and direct (color coded oval box without outline and broken arrow) and indirect (color coded rectangular box without outline) impacts of logging (dark grey rectangular box with outline) and hunting (light grey rectangular box with outline) on the different stages of the seed dispersal. Dashed and dotted line shows additional likely impacts of logging (not explored in this study) on plant recruitment.
In this study, we compared four different stages of the seed dispersal cycle—abundance of food plants, abundance of frugivores, net seed arrival and recruitment patterns of selected large-seeded biotically dispersed plants (
In the short-term, hunting will have no impacts on food plant abundance. Hunting of frugivores will result in their reduced abundances in the ecosystem (
Given that logging likely affects the abundance of food plants of frugivores and hunting impacts the frugivores themselves, negative impacts on the further stages of seed dispersal cycle of frugivore-dispersed trees, including seed dispersal and recruitment, can be expected. Scatter dispersal of seeds as has been documented in this study is mediated by frugivores through fruit removal. While studies have documented reduced frugivore visitation, especially large-bodied frugivores in sites experiencing anthropogenic disturbances [
Reduced recruitment in the disturbed site of certain high value timber species that are animal-dispersed has consequences for long-term persistence of these species and affect the sustainability of timber harvests in logged forests in the long-term. Given that these logged forests also experience hunting of important seed dispersers like hornbills, the negative impacts on the timber species dependent on large vertebrates for seed dispersal are exacerbated. Given that even the less destructive practices like reduced-impact logging [
Hornbill food plants were categorized into strangler figs, hornbill food plants, which are logged, and hornbill food plants that are not logged. Class A-E (as per the Working Plan, Jairampur Forest Division) represents decreasing order of preference for timber value.
(DOCX)
Results from the GLM with negative binomial errors comparing hornbill food plant abundance across three categories (food plants which are logged, food plants which are not logged and strangler figs) between the two sites (Namdapha—with no logging and low hunting pressures and Miao—with logging and high hunting pressures). Parameter estimates (intercept and contrasts), standard errors (SE) and hypothesis tests for parameters are shown.
(DOCX)
Results from the GLMM with Poisson errors comparing hornbill abundance between Namdapha (with no logging and low hunting pressures) and Miao (with logging and high hunting pressures). Parameter estimates (intercept and contrast), standard errors (SE) and hypothesis tests for parameters are shown.
(DOCX)
Results from the GLMM with Poisson errors comparing abundance of scatter-dispersed seeds between Namdapha (with no logging and low hunting pressures) and Miao (with logging and high hunting pressures). Parameter estimates (intercept and contrast), standard errors (SE) and hypothesis tests for parameters are shown.
(DOCX)
Results from the GLM with negative binomial errors, for
(DOCX)
We thank the Arunachal Pradesh Forest Department and Field Director, Namdapha Tiger Reserve for granting us permission to conduct the study. We are grateful to Mr. Japang Pansa, Ashwin Vishwanathan, Akhi Nathany, and Phupla Singpho for logistical help during fieldwork. We thank our field staff for their assistance during fieldwork. We thank Navendu Page and Dr. K. Haridasan for help with plant identification. RSN thanks Jahnavi Joshi, Charudutt Mishra, M.D. Madhusudan, Soumya Prasad, Raman Kumar, Ameya Naniwadekar, Geetha Ramaswami, Anand MO and Kulbhushansingh Suryawanshi for useful discussions and help during fieldwork. We thank two anonymous reviewers for their comments that helped improve the manuscript.