Satellite images indicate vegetation degradation due to invasive herbivores

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Recent studies have documented changes in vegetation due to invasive herbivores in the Andaman Islands. In this study we demonstrate that the change is large enough and rapid enough to be detected by remotely sensed data.

Using freely available Normalised Difference Vegetation Index (NDVI) imagery, we examined changes in vegetation cover due to presence invasive herbivores in the Andaman Islands. Two time periods were analysed using different types of imageries. Changes between 1985-95 and also between 2001 and 2005 across four sites with different combinations of introduced chital deer and elephants were examined. Results indicate that area with deer have faster rates of degradation. The maximum rate of degradation occurred at sites with both elephants and deer, and the minimum where neither of the two animal species occurred. There is a need to eliminate invasive herbivores in the island.

KEYWORDS : invasives, remote sensing, degradation, change detection

A recent publication pointed out that introduced herbivores in the Andaman Islands were causing vegetational changes (Ali, 2004). These included a reduction in the number of tree species as well as a reduction in basal area in areas where Chital or Spotted Deer(Axis axis) and Elephant (Elephas maximus) are found. . In this paper we establish that the change caused by these herbivores is large enough and occurs over a time frame that is short enough to be detected by freely available satellite imagery. We also discuss policy issues that need to be addressed urgently to tackle the problems caused by animals that have become invasive.

The earlier study had identified Interview Island as an uninhabited site which has both deer and elephant. Little Andaman has neither, but has a recently settled human population which has caused forest changes, major in some parts, in the last two decades. The West side of Little Andaman is relatively free of logging, encroachment and other human pressures, and has a small population of Onge tribals, This part has been chosen for the analysis. The Mahatma Gandhi Marine National Park (MGMNP) has deer on its islands which are uninhabited, except for small settlements on Rutland island which have been eliminated from the analysis. Jarawa Reserve has a small population of Jarawa tribals, about 240 spread out over 700 sq km(UNESCO, 2010). It also has deer and their densities are lower here than on Interview Island because of hunting by non-tribals on its periphery. The differences in these areas are summed up in Table 1.

Vegetationally, these areas were all wet evergreen forest, and are free of significant human presence. However, large numbers of deciduous species, especially Lagerstromia hypoleuca, have started increasing in dominance in areas where chital are found, and this appears to be due to the fact that this species is not browsed by chital (Ali, 2004).

Numerous other studies have also shown degradation by invasives. Examples of these include degradation of vegetation by feral goats on Aldabra Atoll (Rainbolt & Coblenz, 1999), the spread of exotic plant species by exotic mammalian herbivores (Davis et al., 2009)

To assess the change in vegetation cover across the four sites a comparison of Normalised Difference Vegetation Index (NDVI) trends over time would indicate the difference between the trends in vegetation cover. Areas near habitations and near the coastline have been eliminated from the analysis.

NDVI values are based on the principle that chlorophyll very strongly absorbs visible light but reflects near-Infrared light. If there is more near-IR reflectance then the vegetation in that pixel is likely to be dense.

NDVI analysis has been used for a number of different kinds of analyses, including monitoring land degradation (Tanser and Palmer, 1999), calculate biomass and grazing intensity (Kawamura et al., 2003) estimate species richness ( Katti et al., 2002; Bawa et al., 2002), and to assess malaria risk (Dhiman, 2000).

Satellite imagery has also been increasingly used in the detection and management of invasive species. Bonneau et al. (1999) used imagery to identify healthy stands of hemlock affected by insect pests. Kimothi et al. (2010) used imagery to map Lantana camara. Rouget et al. (2003) used satellite imagery to map longterm landuse changes partly caused by invasives. Bergstrom et al. (2009) demonstrated that cat removal resulted in an increase in rabbit numbers causing environmental degradation at a World Heritage site. However, satellite imagery as not been used as much as its potential indicates: Mack's (2000) statement that success with it still remains elusive, still appears to be valid. This is possibly the first instance it has been used to monitor invasive herbivore impacts on vegetation.


In this analysis we use NDVI as a proxy for vegetation cover or vigour.

For 1986-1995, we used the calibrated vegetation index (CVI) data compiled and mosaic by the National Institute of Environmental Studies, Japan. The data were radiometrically and geometrically corrected prior to the calculation of the normalized vegetation index. So this is no identical to the NDVI, in that it differs slightly in the processing. The CVI was derived from the Advanced Very High Resolution Radiometer (AVHRR) satellite platform bands 1 and 2.

One kilometre maximum NDVI Composited Data Set for the period 2000 - 2005 was obtained via a Goddard Distributed Active Archive Centre (DAAC) Internet site. The product is derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) produced by selecting cloud-free values of NDVI from successive daily acquisitions of data and retaining the maximum value, which is assumed to be least affected by clouds, for each pixel location. The compositing period generally is limited to a 10-day or semi-monthly time interval to minimize temporal variations in the resultant data product. Although ten-day NDVI images are available, we downloaded and analysed annual maximum value composites. Change in vegetation over time was determined by fitting a regressional slope trend line (Pelkey et al., 2002). Positive slopes indicate increase in vegetation while negative slopes indicate decrease in vegetation vigour. The number of sample grids selected in each site is given in Table 1.

Pairwise comparisons were then done using box plots and linear functions. The linear function represents the a test of the linear hypothesis the mean A- Mean B = 0. The confidence interval represent Tukey's adjustment for multiple comparisons (Chambers & Hastie, 1992; Hothorn, Bretz & Westfall, 2008) . This test was performed using Hothorn, Bretz & Westfall (2009).


Fig 2a and 2b show the change in the NDVI at the sampled sites for the two data sets. It is apparent that Little Andaman, with no herbivores, has suffered the least degradation; in contrast, Interview Island, with two herbivores species, has suffered the maximum degradation. The other two, with only chital, are in between.

This trend has been repeated in all the succeeding analysis. The Box plots shown in Fig. 3a and 3b show same pattern.

Two statistical tests were used to test this. The linear functions in Figs 4a and 4b reflect these trends for the two data sets. The results using the MODIS data are more clearcut than the results using the AVHRR data, since the within site variation is much lower. Interview Island had a significantly higher rate of degradation than both Jarawa Reserve and Little Andaman.

Finally, Tables 5a and 5b give the results of a Kruskal-Wallis test for the two data sets. Little Andaman has a significantly lower rate of degradation than both Interview island and Jarawa Reserve for the 1985-1995 data; In the 2001-2005 data set, Interview Island degrades faster than the Jarawa-MGMNP cluster, which in turn degrades faster than Little Andaman. All but one of these is significant.


The vegetation changes in Little Andaman and Interview Island have already been documented (Ali, 2004). This study further strengthens the findings of the previous field surveys. In this study, the presence of indigenous tribal populations is noted but is unlikely to have an effect because both groups of tribals, being foragers, have very little impact on the forest. The Jarawa also have never been recorded as hunting deer, though poaching from the settlements outside the reserve may have ironically slowed down forest degradation.

Interview Island has a high deer population, as well as a feral elephant population (Ali, 2005). The two deer-only places (We are including Jarawa Reserve among these since the Jarawa tribals, being foragers, do not hunt deer, or cut down trees) have a lower rate of degradation since tree damage by elephants does not occur.

There are numerous sources of error that are possible. Areas near habitation have been avoided to eliminate errors due to logging and encroachments. There are also differences in satellite sensor alignment between different years. We have demonstrated that this does not make a difference, and the differences are great enough to be picked up through remote sensing techniques. We avoided grids along the coast to avoid the influence of changes in coastline due to the sensor alignment problem. The only source of difference that appears to remain is the presence or absence of invasive herbivores such as the chital and the elephant.

Plant transects on Interview Island suggest the causes of this degradation. The regeneration in logged areas here consists almost entirely of Lagerstromia hypoleucos, one of two common species not browsed by chital (the other being Pongamia pinnata, a coastal species on the island).

The answer seems to be the combination of chital and elephants. Elephants have damaged the vegetation badly in recent years, and have created a situation where forest regeneration rates are higher than normal. The chital prevent regeneration by browsing on the seedlings.

The degradation observed over relatively short time spans is worrying, and indicates that the problem of invasives here needs to be tackled on an urgent footing. Other invasives in the islands that have not been studied but need attention include feral dogs and cats. Unfortunately, no other studies on invasive herbivores have been done in this part of the world, and there is no baseline for comparing relative degradation rates.

The internationally accepted scientific principle for dealing with invasives is extremely clear: in brief, if it is introduced and causes economic and environmental damage, remove it. One does not need census figures to validate the decision to eliminate invasives. It is enough to show that degradation is happening. We have shown a quick method for establishing this.


We thank Srinivas Vaidyanathan for his comments and inputs.