by milind bunyan
Post on 13-Apr-2022
11 Views
Preview:
TRANSCRIPT
1
EDGE EFFECTS IN A FOREST-GRASSLAND MOSAIC IN SOUTHERN INDIA
By
MILIND BUNYAN
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
2009
2
© 2009 Milind Bunyan
3
To, Preethi- no one deserves this more (Proverbs 31: 29-30)
4
ACKNOWLEDGMENTS
I would like to thank the chair of my Graduate Committee, Dr. Shibu Jose for giving me
the opportunity to pursue a doctoral degree and for his support, encouragement and patience
especially when things did not go as planned. I would like to thank my committee members, Dr.
Ankila Hiremath, Dr. Kaoru Kitajima, Dr. Alan Long, Dr. Michelle Mack and Dr. George
Tanner for their time, guidance and encouragement through my program. I thank Dr. Ankila
Hiremath in particular, for help with the logistics of research in India. I am especially grateful to
Dr. Robert Fletcher for help with my research and methodology and Aditya Singh without who a
large part of this dissertation might not have existed.
I am grateful to the Karnataka and Kerala Forest Departments for permission to conduct
research in BRT Wildlife Sanctuary and Eravikulam National Park and Pampadum shola
National Park respectively. I would like to thank the staff at Ashoka Trust for Research in
Ecology and the Environment (ATREE), Bangalore and the ATREE Field station at BRT for
their assistance with data collection in BRT. My work in Eravikulam National Park was made
possible in no small part by Senthil Kumar, Ancel and Sarvanan and they are gratefully
acknowledged. Dr. Ganesan’s (ATREE) assistance was invaluable in the identification of plant
specimens.
I would like to thank my family in India for their love and support throughout the program.
None of this would have been possible without the endless supply of coffee, food and love from
my wife Preethi and right near the end, smiles and kisses from my one-year old, Rahael.
5
TABLE OF CONTENTS page
ACKNOWLEDGMENTS ...............................................................................................................4
LIST OF TABLES ...........................................................................................................................7
LIST OF FIGURES .........................................................................................................................8
ABSTRACT .....................................................................................................................................9
CHAPTER
1 THE SHOLA-GRASSLAND ECOSYSTEM MOSAIC: A SYNTHESIS OF EXISTING LITERATURE ....................................................................................................11
Introduction .............................................................................................................................11 The Shola-grassland Ecosystem Mosaic ................................................................................12
What is not a Shola? ........................................................................................................13 Flora ........................................................................................................................................13 Fauna .......................................................................................................................................15 Hydrology ...............................................................................................................................16 Soils and Nutrient Cycling ......................................................................................................17 The Shola-grassland Edge ......................................................................................................18 Conclusion ..............................................................................................................................19
2 MICRONENVIRONMENT EDGE EFFECTS IN THE SHOLA GRASSLAND ECOSYSTEM MOSAIC ........................................................................................................25
Introduction .............................................................................................................................25 Methods ..................................................................................................................................29
Study Area .......................................................................................................................29 Sampling Protocol ...........................................................................................................31 Shola Plots .......................................................................................................................31 Grassland Plots ................................................................................................................32
Data Analyses .........................................................................................................................32 Results .....................................................................................................................................34
Comparing Shola and Grassland Plots ............................................................................34 Microenvironment Gradients in Shola Fragments ..........................................................35
Discussion ...............................................................................................................................36 Seasonal and diurnal variation .........................................................................................38 Comparison between One-edge and Multiple Edge Models ...........................................39
Conclusion ..............................................................................................................................40
3 SPECIES DISTRIBUTION PATTERNS IN SHOLA FRAGMENTS ..................................48
Introduction .............................................................................................................................48
6
Methods ..................................................................................................................................51 Study Area .......................................................................................................................51 Sampling Protocol ...........................................................................................................53
Data Analyses .........................................................................................................................54 Vegetation Structure and Composition ...........................................................................54 Edge-interior Gradients in Species Distributions ............................................................55
Results .....................................................................................................................................56 Vegetation Structure and Composition ...........................................................................56 Edge-interior Gradients in Species Distributions ............................................................57
Discussion ...............................................................................................................................58 Conclusion ..............................................................................................................................61
4 EFFECT OF TOPOGRAPHY ON THE DISTRIBUTION OF SHOLA FRAGMENTS: A PREDICTIVE MODELING APPROACH ........................................................................72
Introduction .............................................................................................................................72 Methods ..................................................................................................................................74
Study Area .......................................................................................................................74 Imagery Pre-processing and Data Preparation ................................................................75 Predictive Modeling ........................................................................................................76
Results .....................................................................................................................................77 Combined Model .............................................................................................................77 Data Subsets ....................................................................................................................78
Discussion ...............................................................................................................................78 Comparison of Data Subsets ...........................................................................................80 Implications for the Shola-grassland Edge ......................................................................80
Conclusion ..............................................................................................................................81
5 SUMMARY AND CONCLUSION .......................................................................................91
APPENDIX ....................................................................................................................................95
LIST OF REFERENCES .............................................................................................................111
BIOGRAPHICAL SKETCH .......................................................................................................123
7
LIST OF TABLES
Table page 1-1 Faunal species richness in tropical montane forests ..........................................................22
1-2 Surface soil chemical characteristics in tropical montane forests .....................................23
2-1 Location of tropical montane forest fragments across study sites in the Western Ghats, southern India. ........................................................................................................42
2-2 Plant-essential nutrient concentrations in tropical montane forest and grassland surface soils in the Western Ghats, southern India. ...........................................................42
2-3 Principal components scores for soil nutrient variables in tropical montane forest fragments and grassland soils ............................................................................................43
2-4 Edge-interior gradients in tropical montane forest fragments in the Western Ghats, southern India .....................................................................................................................43
3-1 Location and vegetation characteristics of tropical montane forest fragments studied across the Western Ghats, southern India. .........................................................................63
3-2 Species richness, dominance and diversity estimates in the shola-grassland ecosystem mosaic in the Western Ghats, southern India. ..................................................64
3-3 Similarity and complementarity between tropical montane forest (shola) fragments in the Western Ghats, southern India. ....................................................................................65
3-4 Pearson correlation coefficients for environmental variables with axis scores of NMS ordination. ..........................................................................................................................66
4-1 Topographical variables used to predict presence of shola fragments in the Western Ghats, southern India. ........................................................................................................83
4-2 Topographical parameter coefficients and AuC estimates for predictive models. ............83
A-1 Overstory species presence matrix along an edge-interior gradient across nine fragments in the shola-grassland ecosystem mosaic in the Western Ghats, southern India. ..................................................................................................................................96
A-2 Understory species presence matrix along an edge-interior gradient across nine fragments in the shola-grassland ecosystem mosaic in the Western Ghats, southern India .................................................................................................................................103
8
LIST OF FIGURES
Figure page 1-1 Alternate stable state diagram for the shola-grassland ecosystem mosaic. .......................24
2-1 Location of study sites in the Western Ghats. ....................................................................44
2-2 Schematic representation of estimation of distance to additional edges. ...........................45
2-3 Effect of the shola-grassland edge on microenvironment variables in the shola-grassland ecosystem mosaic ..............................................................................................46
2-4 Edge-interior gradients in microenvironment and soil nutrient parameters in Pampadum shola National Park. Error bars represent one standard error. ........................47
3-1 Species accumulation curves, species richness estimates and rare species (singletons and doubletons) for overstory and understory species.. .....................................................67
3-2 Mean understory density along the edge-interior gradient in tropical montane forest (shola) fragments.. .............................................................................................................68
3-3 Edge-interior gradients in diameter distribution of overstory tropical montane forest (shola) species across nine fragments.. ..............................................................................69
3-4 Non-metric multi-dimensional scaling (NMS) ordination of overstory data.. ...................70
3-5 Non-metric multi-dimensional scaling (NMS) ordination of understory data. ..................71
4-1 Location of study sites in the Western Ghats. ....................................................................84
4-2 Histogram of Intercept coefficients for Eravikulam National Park and Nilgiris data subsets.. ..............................................................................................................................85
4-3 Histogram of Elevation coefficients for Eravikulam National Park and Nilgiris data subsets.. ..............................................................................................................................86
4-4 Histogram of Slope coefficients for Eravikulam National Park and Nilgiris data subsets. ...............................................................................................................................87
4-5 Histogram of Northness coefficients for Eravikulam National Park and Nilgiris data subsets.. ..............................................................................................................................88
4-6 Histogram of Eastness coefficients for Eravikulam National Park and Nilgiris data subsets. ...............................................................................................................................89
4-7 Histogram of Curvature of the Slope coefficients for Eravikulam National Park and Nilgiris data subsets. ..........................................................................................................90
9
Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy
EDGE EFFECTS IN A FOREST-GRASSLAND MOSAIC IN SOUTHERN INDIA
By
Milind Bunyan
August 2009 Chair: Shibu Jose Co-chair: Alan Long Major: Forest Resources and Conservation
Tropical montane forests in the Western Ghats in southern India consist of dense, insular
fragments in a matrix of grasslands separated by an abrupt, natural, edge. I studied edge effects
in the shola-grassland ecosystem mosaic across nine fragments in three study sites in the Western
Ghats. I measured microenvironment and soil variables and overstory species in 10×5 m plots
along an edge-interior gradient at 5 m intervals. Understory density and richness was recorded
from 5×5 m sub plots. Conventional distance to one-edge models indicated edge-interior
gradients in relative humidity (p = 0.018), magnesium (p = 0.027) and potassium (p = 0.008) in
large fragments. In small fragments, gradients in air temperature (p = 0.03), light transmittance
(p = 0.007) and soil moisture (p = 0.0002) were observed as a function of distance to multiple
edges.
We recorded 111 species (77 overstory; 83 understory) across nine fragments but did not
observe any edge-interior trends in overstory density or dominance. Similarly, no edge-interior
gradients were observed in understory density and structure. Non-metric multidimensional
scaling techniques for overstory and understory revealed greater variation among fragments than
could be attributed to edge-related within fragment variation. Overstory composition in mid-
elevation fragments differed significantly from high-elevation fragments while understory
10
vegetation varied as a function of fragment size. Our data indicate that mid-elevation fragments
should not be considered with high elevation fragments in future studies.
We also used a multiple logistic regression model to predict the presence of shola
fragments across two study sites in the Western Ghats. Elevation, slope, aspect (expressed as
eastness and northness), slope curvature and wetness index were used to predict the presence of
shola fragments. We observed that shola fragments were more likely to occur on northern and
western aspects than southern and eastern aspects. Shola fragments were also most likely to
occur on wet, steep slopes. The stability of the shola-grassland edge appears to be driven by fire
rather than frost while exposure to wind might be a driving factor also.
The shola-grassland ecosystem mosaic offers insights into fragmentation related patterns in
small patches and recommendations are made for future investigations.
11
CHAPTER 1 THE SHOLA-GRASSLAND ECOSYSTEM MOSAIC: A SYNTHESIS OF EXISTING
LITERATURE
Introduction
Tropical montane forests (alternatively called tropical montane cloud forests or simply
cloud forests) represent some of the most threatened ecosystems globally. Tropical montane
forests (TMF) are characterized and defined by the presence of persistent cloud cover. A
significant amount of moisture may be captured through the condensation of cloud-borne
moisture on vegetation distinguishing TMF from other forest types. Bruijzneel and Hamilton
(2000) described five kinds of TMF. Four of these, i.e. lower montane forest, lower montane
cloud forest, upper montane cloud forest and subalpine cloud forest, are based on elevation and
tree height whereas the last one an azonal low elevation dwarf cloud forest.
Elevations at which TMF are found, vary with mountain range size and insularity or
proximity to coast. Due to the mass-elevation effect (also known as the Massenerhebung effect),
larger mountain ranges permit the extension of the altitudinal range of plant species. Similarly,
higher humidity levels near coastal mountains enable the formation of clouds at lower altitudes.
On insular or coastal mountain ranges, TMF has been reported from elevations as low as 500m
(Bruijzneel and Hamilton, 2000). As elevation increases, tree height in TMF reduces and leaf
thickness and complexity in tree architecture increases. Other distinctive features of TMF are the
prolific growth of epiphytes and mosses and the lack of vertical stratification. TMF soils are
typically clay-rich, have low pH, abundant organic matter and are often nutritionally poor. TMF
are characterized by high levels of endemism driven by the limited availability of habitat (Bubb
et al. 2004). Located in the headwater catchments of seasonal or perennial streams, TMF
provides often undervalued ecosystem services to downstream communities.
12
Within the Western Ghats-Sri Lanka (WGSL) biodiversity hotspot (Myers et al. 2000),
TMF occurs as a mosaic of forests (locally and hereafter sholas) and grasslands and is commonly
referred to as the shola-grassland ecosystem. With limited exceptions (Werner 1995,
Somanathan and Borges 2000), data from the shola-grassland ecosystem mosaic are rarely
included in biome-wide popular (Bruijzneel and Hamilton 2000) as well as academic (Hamilton
et al. 1995, Bubb et al. 2004) synopses of scientific literature. As such, this document aims to
provide a synthesis of current research and the state of knowledge of the shola-grassland
ecosystem from peer-reviewed literature published on tropical montane forests in the WGSL
biodiversity hotspot. Additionally, a synopsis of research on the sholas of Kerala was also
reviewed (Nair et al. 2001).
The Shola-grassland Ecosystem Mosaic
The Western Ghats located in the WGSL hotspot are a 1600 km long mountain (160,000
km2) chain in southern India. Located above 1700 m, the shola-grassland ecosystem mosaic
consists of rolling grasslands with shola fragments restricted to sheltered folds and valleys in the
mountains separated from the grasslands with a sharp edge. Since, sholas frequently have
persistent cloud cover they can be classified as lower montane cloud forest or upper montane
cloud forest depending on elevation (Bruijzneel and Hamilton, 2000). Ecologists and foresters
have been puzzled over the pattern of the shola-grassland ecosystem mosaic for decades. While
some of the earliest scientific descriptions of the shola-grassland ecosystem described the mosaic
as dual climax (Ranganathan 1938), proponents of the single climax concept (Clements 1936)
argued that the forests represented a biotic (Bor 1938, Noble 1967) or edaphic climax (Jose et al.
1994). A δC13 analysis of peat samples from shola fragments in the Nilgiris indicated that shola
and grasslands have undergone cyclical shifts in dominant vegetative cover. Arid conditions
from 20,000-16000 yr BP led to predominance of C4 vegetation. This was followed by a wetter
13
phase which peaked around 11,000 yr BP leading to a dominance in C3 vegetation. The
weakening of the monsoon around 6000 yr BP led to the expansion of the C4 vegetation again
and the establishment of the current pattern, although a brief warm, wet phase around 600-700yr
BP also occurred (Sukumar et al. 1993).
What is not a Shola?
Arguably, the shola-grassland ecosystem mosaic is among the most distinct ecosystem
types in the WGSL biodiversity hotspot. Although, sholas are typically seen at elevations ≥ 1700
m, sholas at elevations as low as 1050 m have been studied by ecologists (Sudhakara 2001). In
the Anamalais and Nilgiris, the shola-grassland mosaic is characteristically patchy. Often though,
shola fragments are linear strips that may or may not be contiguous with lowland evergreen
forest which contain a different suite of species. While species dominance patterns are distinct
from lowland forest, sholas of different regions exhibit little similarity in species composition.
Yet, physiognomic characteristics of sholas are consistent. Sholas consistof profusely branched,
stunted trees (rarely exceeding 15 m) with prolific epiphytic growth. In order to distinguish shola
from non-shola forest types, despite the varied conditions under which they are found, I propose
that ecologically, a shola be defined as a high elevation (≥1700 m) stunted forest with distinct
physiognomy. Studies on sholas at elevations below 1700 m should be restricted to shola
fragments surrounded by grasslands. Indeed, in plots located at lower elevations, Sudhakara et al.
(2001) recorded families uncommon to sholas but common to lowland forests (Bombaceae,
Clusiaceae, Dichapetalaceae).
Flora
Since Thomas and Palmer (2007) provide a comprehensive review of current research on
grasslands in the shola-grassland ecosystem mosaic, I will restrict this section to reviewing work
on the shola vegetation only. Shola fragments contain species of both tropical and temperate
14
affinities. Also, the grasslands of the Western Ghats show more biogeographic similarity with
Western Himalayan species than TMF in Sri Lanka (Karunakaran et al. 1998).
Phytogeographical analysis of shola genera reveals that genera found on the fringes of shola
fragments and as isolated trees on grasslands are typically temperate (Rubus, Daphiphyllum and
Eurya) or sub-tropical (Rhododendron, Berberis, Mahonia are Himalayan) in origin. Species
within shola fragments on the other hand are IndoMalayan or Indian (rarely Paleotropical) in
origin (Suresh and Sukumar 1999, Nair and Menon 2001). Overstory species in the shola are
dominated by members of Lauraceae, Rubiaceae, Symplocaceae, Myrtaceae, Myrsinaceae and
Oleaceae while dicotyledonous understory species are dominated by Asteraceae, Fabaceae,
Acanthaceae, (Davidar et al. 2007, Swarupanandan et al. 2001). Dominant monocot species in
the understory include members of Poaceae, Orchidaceae & Cyperaceae (Swarupanandan et al.
2001). Along edge-interior gradients in shola fragments, species were found to be significantly
influenced by soil moisture (overstory and understory) and soil nitrogen (understory only) (Jose
et al. 1996). However this study was based on observations from a single shola patch.
Based on our knowledge of species-area curves, we might expect that the limited
availability of suitable habitat for shola species within the shola-grassland ecosystem mosaic
would limit α-diversity. However estimates for α-diversity are highly variable. Estimates for
Shannon-Weiner’s diversity index (H') range from 4.71 (Sudhakara 2001) to 0.87(Murali et al.
1998). Estimates for endemism are also highly variable- from 19.5 % (Suresh and Sukumar
1999) to 83.3 % (Nair and Menon 2001). Historically, the regeneration of arborescent flora in
shola fragments had been expressed as a concern (Vishnu-Mittre and Gupta 1968). A series of
studies now indicate that shola species show adequate regeneration under natural conditions
15
(Jose et al. 1994, Swarupanandan et al. 2001). Additionally, germination rates as high as 95%
have been recorded for shola species in germination trials (Srivastava, 2000).
As with other TMF, shola fragments exhibit prolific epiphytic growth. Studies in TMF
have shown that epiphytic species may constitute up to 25% of all biomass in tropical montane
forests (Foster 2001). They also provide microhabitats for invertebrates and amphibians
(Benzing 1998), store significant amounts of water (Sugden 1981) and influence nutrient cycling
(Benzing 1998). However, given the extent of scientific literature on arborescent flora in the
shola (Jose et al. 1996, Suresh and Sukumar 1999, Davidar et al. 2007, also see Nair et al. 2001),
very limited work exists on epiphytes in the shola-grassland ecosystem mosaic (Abraham 2001).
Similarly, very few studies have quantified productivity in the shola-grassland ecosystem
mosaic. In a comparison of net primary productivity (NPP) patterns of exotic plantations and
native shola forest, NPP and biomass of older exotic plantations (Eucalytpus globulus and Pinus
patula) were significantly higher than that of shola species. However this was at the cost of
lowering of NPP and biomass in the understory in exotic plantations, possibly due to allelopathic
inhibition (Jeeva and Ramakrishnan 1997).
Fauna
The shola-grassland ecosystem mosaic provides habitat for many faunal species of
conservation concern including the tiger (Panthera tigris tigris), dhole (Cuon alpinus), gaur (Bos
gaurus gaurus), Nilgiri langur (Trachypithecus johnii ) and Nilgiri marten (Martes gwatkinsii).
Endemic to the ecosystem-mosaic is the Nilgiri tahr (Niligiritragus hylocrius) which has been
studied meticulously over the years (Davidar 1971, Rice 1984, Rice 1988, Mishra and Johnsingh
1998, Daniels 2006). Although considered a flagship species for the ecosystem, uncertainty over
16
population estimates persists (Daniels et al. 2008) even as the population shows a declining trend
(Alembath and Rice 2008).
Faunal species too have been observed to mirror the shola-grassland ecosystem mosaic
pattern through habitat preferences. Small mammal communities in the Nilgiris (two species of
the nine recorded), showed a high degree of preference for either shola or grassland despite a
lack of resource-driven interspecific competition. However, these patterns were obscured in
exotic plantations (Shanker 2001). Strong habitat selection patterns have also been observed in
avian species in the shola-grassland ecosystem mosaic. Habitat suitability models for the Nilgiri
laughing thrush (Garrulax cachinnans) indicate that habitat use typically restricted to shola cover
might extend to exotic plantations (unsuitable habitat) when located near shola fragments (Zarri
et al. 2008). Other avian species such as the black and orange flycatcher (Ficedula nigrorufa)
have also been known to show a strong preference for shola cover.
Other than those mentioned above, inventories have also been conducted on amphibian,
avian, invertebrate and fish species (Dinesh et al. 2008, also see Nair et al. 2001). However with
the exception of the Nilgiri tahr, the body of scientific literature on faunal species in the shola-
grassland ecosystem mosaic is limited. (See Table 1-1)
Hydrology
Globally, tropical montane forests (alternatively tropical montane cloud forests) have been
shown to significantly influence ecosystem hydrology and biogeochemistry (Bruijnzeel and
Hamilton, 2000). In addition to providing cover and reducing erosion potential, net precipitation
(precipitation reaching the ground) under tropical montane forests is often greater than 100%
(and as high as 180%). This has been attributed to condensation of wind-driven fog on tree
crowns (termed fog drip). In areas of low precipitation such as the Canary Islands, interception
of cloud water can double annual precipitation (Gioda et al., 1995). Protection of tropical
17
montane habitat serves the purpose of hydrological regulation for downstream consumers also.
This is especially significant in the Western Ghats where major rivers originating in the shola-
grassland ecosystem mosaic provide hydrological services to consumers. A study by
Krishnaswamy et al. (2006) demonstrated that individual rainfall events could contribute as
much as 20-30% of the annual sediment load of 239-947 Mg km-2 (Krishnaswamy et al. 2006).
Other studies report significantly lower sediment load estimates (30-97 Mg km-2 year-1) from
other areas (Thomas and Sankar 2001, Sahoo et al. 2006) with as much as 90% of the annual
runoff occurring during the SW Monsoon (Thomas and Sankar 2001).
Soils and Nutrient Cycling
Soils in the shola-grassland ecosystem mosaic are granitic or metamorphic gneisses in
origin. They are of varying depth, ranging from deep (Ranganathan 1938) to shallow, stony soils
(Gupta and Shankarnarayan 1962). Typically, soils are shallower in the grasslands as compared
to shola soils and are more prone to soil moisture loss. During the dry season, shola soils have
been shown to retain as much as twice the soil moisture in the surrounding grasslands (Thomas
and Sankar 2001). Shola and grassland soils also differ nutritionally. Total N, available P and K
are higher in the sholas as compared to adjoining grasslands. Though this could be attributed to
higher litter decomposition and nutrient recycling rates in the sholas, these differences are rarely
significant. Jose et al. (1994) report organic carbon content in shola surface soils that are
comparable to those recorded in TMF in Ecuador. These values though are much higher than
those recorded by other authors for surface soils under varied types of cover in the shola-
grassland ecosystem mosaic (See Table 1-2). No soil-depth related trends have been reported for
plant essential micronutrients (Cu, Mn, Zn and Fe) in sholas or adjacent grasslands although
differences between sholas and adjacent grasslands have been observed (Nandakumar et al.
2001). Shola soils have higher soil nutrient pools than those under exotic plantations of blue gum
18
(E. globulus) or tea (Camellia sinensis) (Jeeva and Ramakrishnan 1997, Venkatachalam et al.
2007). Nutrient cycling under natural shola vegetation has also been described as steady state
and less likely to suffer losses to leaching since the return through leaf litter is low (Jeeva and
Ramakrishna 1997).
Trends in biogeographical affinities have also been recorded for soil microflora. Though
most soil fungal species recorded in soils in the shola-grassland ecosystem mosaic are
cosmopolitan in distribution, the prevalence of the genus Penicillium is characteristic of
temperate forests (Sankaran and Balasundaram 2001). Soil fungal species diversity is
comparable between shola fragments and grasslands (H'SHOLA=4.18, H'GRASSLAND=4.18) albeit
highly habitat specific (Sankaran and Balasundaram 2001). Shola soils also had significantly
higher soil bacterial and actinomycetes populations than grassland or plantation soils while
fungal populations were highest in grassland soils. Plantation soils under tea, blue gum and black
wattle (Acacia mearnsii) were also consistently observed to have lower soil microbial biomass
than soils under native vegetation (Venkatachalam et al. 2007).
The Shola-grassland Edge
The current dynamic equilibrium between insular shola fragments and grasslands is
indicative of the existence of alternate stable states enforced by environmental parameters (May
1977, Beisner et al. 2003). A change in parameters causes a shift in dominant cover (see Figure
1-1). Applied to the shola-grassland ecosystem mosaic, these parameters might include frost
(Ranganathan 1938, Meher-Homji 1965, 1967), fire (Bor 1938), grazing (Bor 1938, Noble
1967), soil nutrient status (Jose et al. 1994), soil depth (Ganeshaiah per comm.), wind
(Balasubramanian and Kumar 1999) and illegal harvesting (Gupta 1962). The persistence of the
mosaic in areas relatively free of anthropogenic grazing and illegal harvesting in some protected
areas make these two parameters tenuous for explaining the pattern. Although some authors have
19
observed grassland soils to be shallower than shola soils (Jose et al. 1994); others (Nandakumar
et al 2001) did not find a consistent trend. Moreover, shola species have been observed growing
on shallow soils too (Ranganathan 1938). Grassland fires are used as a management tool in the
shola-grassland ecosystem mosaic to reduce fuel loads (Karunakaran et al. 1998) and in some
instances a protective belt is cleared of vegetation around the shola before the grasslands are
fired to preclude fire from the shola (pers observ). These fires could act as an effective deterrent
in the colonization of the grasslands by shola species. Additionally, a study on vegetation fires
during the dry season (February-June) of 2006 revealed that tropical montane forests in the
Indian subcontinent accounted for 8.07% (92 fires) of all fires (Vadrevu et al. 2008). Although
current understanding points to fire as the dominant factor responsible for the maintenance of the
edge, ambiguity remains.
Conclusion
Historically, the shola-grassland ecosystem mosaic has undergone extensive habitat loss.
Plantations of exotic tree species were established in the grasslands aimed at augmenting timber
production as early as 1843 (Palanna 1996) with further introductions in 1870 in the Palni hills
(Srivastava 2001). Plantation programs were expanded under colonial rule to establish extensive
tea plantations in the mosaic. Post independence, tree plantation programs also received national
(federal) budgetary support (Raghupathy and Madhu 2007). Significant populations of invasive
shrubs and herbs (Eupatorium glandulosum, Ulex europaeus and Cytisus scoparius) in the shola-
grassland ecosystem mosaic were reported by early researchers (Bor 1938, Agarwal 1961). This
list continues to expand as new exotic species (e.g. Calceolaria mexicana, Erigeron
mucronatum) have recently been reported from the ecosystem (Seshan 2006). Threats to the
mosaic today include the harvesting of shola species to meet biomass and fuelwood requirements
and cattle grazing (Sekar 2008). In areas adjoining settlements, these threats can be significantly
20
amplified altering patterns in species richness and dominance (Chandrasekhara et al. 2001). The
WGSL biodiversity hotspot is likely to undergo extinctions in plant and vertebrate species due to
the limited availability of habitat (Brooks et al. 2002). Further, globally, TMF experience higher
annual loss in habitat than any other tropical forest biome (FAO 1993). As Sukumar et al. (1995)
suggest, climate change is expected to alter the dynamic equilibrium between the forest and
grassland through a reduction in the incidence of frost coupled with the strengthening of the
monsoon which would select for C3 species. Responding to these threats appropriately requires
the application of current state of knowledge coupled with an identification of gaps in our
knowledge.
Studies detailing the response of vegetation to the edge remain limited. To my knowledge
only one study to date quantifies edge effects in the shola-grassland ecosystem mosaic (Jose et
al. 1996). Unlike the sholas, fragmentation in other tropical montane forests (such as the
neotropics) is often a result of recent anthropogenically induced pressures (e.g. fire, conversion
to pasture). As such, edge effect studies in the shola-grassland ecosystem might be especially
insightful since species in older fragments have had time to equilibrate with fragmentation-
induced pressures (Turner et al. 1996, Harper et al. 2005). Fragmentation studies often observe a
proportional increase in area under edge influence with diminishing fragment size. Small
fragments may then be dominated by edge effect and lack an ‘interior’ or ‘core’, making them
susceptible to complete collapse (Gascon et al. 2000). An edge effect study in the shola-
grassland ecosystem would help us understand patterns in small fragments since shola fragments
in the shola-grassland ecosystem mosaic are often small (~1 ha).
In this study I investigate the effect of the shola-grassland edge on microenvironment
gradients in the ecosystem mosaic. I then assess patterns in species distributions and structure
21
between and within fragments studied. I expect these to provide an understanding of patterns in
smaller, older fragments. In mountainous terrain (where the shola-grassland ecosystem mosaic is
found), topography can significantly determine factors influencing vegetation (such as the
occurrence fire and frost). Topography can thus serve as a surrogate for these variables. I
therefore investigate the topographical variables determining the presence of shola fragments in
the mosaic. Through this, I seek to determine processes driving the shola-grassland edge.
22
Table 1-1. Faunal species richness in tropical montane forests Cover class Site Elevation Taxa Species
richness Species diversity (H')
Percent Endemism Source
Tropical montane forest/ Shola
Mannavan shola
1600-1700 Birds 30 - 20 Nameer (2001) 2000-2100 Birds 40 - 23 Nameer (2001)
Kerala - Fish 24 - - Ghosh (2001) Chembra 1700 Insects 81 4.22 - Mathew et al. (2001) Nilgiris 1800-2500 Small mammals 8 - - Shanker (2001) Nilgiris 2000-2050 Bacteria 93.22 - - Venkatachalam et al. (2007) Nilgiris 2000-2050 Fungi 7.78 - - Venkatachalam et al. (2007)
Grassland Nilgiris 1800-2500 Small mammals 3 - - Shanker (2001) Nilgiris 2000-2050 Bacteria 30.31† - - Venkatachalam et al. (2007) Nilgiris 2000-2050 Fungi 8.89* - - Venkatachalam et al. (2007)
Plantation (Mixed)
Nilgiris 1800-2500 Small mammals 3 - - Shanker (2001) Nilgiris 2000-2050 Bacteria 37.53† - - Venkatachalam et al. (2007) Nilgiris 2000-2050 Fungi 7.66* - - Venkatachalam et al. (2007)
Plantation (Tea)
Nilgiris 1800-2050 Small mammals 4 - - Shanker (2001) Nilgiris 2000-2050 Bacteria 18.54† - - Venkatachalam et al. (2007) Nilgiris 2000-2050 Fungi 5.78* - - Venkatachalam et al. (2007)
†Bacterial biomass (× 107colony forming units -1soil), *Fungal biomass (×105colony forming units-1soil)
23
Table 1-2. Surface soil chemical characteristics in tropical montane forests
Cover class Site Soil nutrients
Authors pH C N P K Ca Mg % kg ha-1
Tropical montane forest/ Shola
Brahmagiri 5.60* 2.80* 0.18*‡ 0.02* 0.14* 0.02* Thomas and Sankar (2001)
Nilgiris 5.44# 1.65# 306.91# 7.27# 138.43# 183.40§a 6.8§a Venkatachalam et al. (2007)
Eravikulam - 22.48# 1.21#‡ 0.02#‡ 0.01#‡ - - Jose et al. (1994) Ecuador (1960 m) 4.60+ 39.00+ 2.10+‡ 0.87+‡ 0.35+‡ 0.36+‡ 0.14+‡ Wilcke et al. (2008)
Ecuador (2090 m) 3.90+ 48.50+ 1.80+‡ 0.57+‡ 0.11+‡ 0.51+‡ 0.06+‡ Wilcke et al. (2008)
Ecuador (2450 m) 4.40+ 35.60+ 1.20+ ‡ 0.34+‡ 0.11+‡ 0.18+‡ 0.03+‡ Wilcke et al. (2008)
Grassland
Brahmagiri 5.00* 2.40* 0.04*‡ 0.01*‡ 0.03*‡ 0.02*‡ Thomas and Sankar (2001)
Nilgiris 4.04# 0.87# 132.92# 1.84# 70.68# - - Venkatachalam et al. (2007)
Eravikulam 18.88# Jose et al. (1994) Plantation (Eucalyptus globulus)
Nilgiris - - 97.50§ 10.60§ 74.50§ 123.60§ 45.70§ Jeeva and Ramakrishnan (1997)
Plantation (Pinus patula)
Nilgiris - - 188.50§ 22.70§ 109.80§ 158.80§ 127.90§ Jeeva and Ramakrishnan (1997)
Plantation (Mixed) Nilgiris 4.45# 1.10# 199.99# 3.67# 88.65# - - Venkatachalam et al. (2007)
Plantation (Tea) Nilgiris 4.06# 0.98# 205.32# 4.14# 100.19# - - Venkatachalam et al. (2007)
§0-10cm, *0-15cm, #0-20cm, +undefined (depth of O horizon), aJeeva and Ramakrishnan 1997, ‡ percent concentration
24
Figure 1-1. Alternate stable state diagram for the shola-grassland ecosystem mosaic. A change
in parameter (e.g. frost, fire) can cause a shift in communities. An increase in fire occurrence can move the dominant community (ball) from shola (A) to grassland (B). A reversal of the parameter can cause the community to return to its original state (along the dashed line).
(B)
(A)
25
CHAPTER 2 MICRONENVIRONMENT EDGE EFFECTS IN THE SHOLA GRASSLAND ECOSYSTEM
MOSAIC
Introduction
The creation of edges in forested habitat is a significant threat to the health and
maintenance of forested ecosystems. The term edge-effect was first used by Aldo Leopold
(1933) who observed higher species diversity near habitat edges. This was followed up by other
reports of increased edge-related species diversity (Johnston 1947). Land managers were advised
to create edges even while research sought to determine the optimum pattern of edge creation
(Yahner 1988). However, studies soon revealed altered ecological processes (Gates and Gysel
1978), species distribution patterns (Anderson et al 1977) and ecosystem structure (Laurance et
al. 1998, Gascon et al. 2000) near edges. At a newly created edge there are immediate changes in
the flow of energy, materials and organisms (termed ecological flows) across the edge as
communities gain access to resources hitherto separated. Species then respond to the altered
levels of resources (resource mapping) by adapting their distributions. Finally, this causes a
change in the interactions between species (Ries et al. 2004). Feedbacks between these
mechanisms (ecological flows, increased access, resource mapping and species interactions)
occur iteratively as primary and secondary responses with increasing edge age.
Numerous studies have quantified the mechanisms and patterns observed at edges as they
relate to fragmentation of forested habitats (for a review see Harper et al. 2005). The
characteristics of an edge are determined to a large extent by the influence of the edge on
overstory structure and composition and its interaction with the abiotic environment. Increased
light transmittance at newly created edges alters microenvironment conditions resulting in
elevated levels of air temperature and reduced relative humidity. Soil characteristics are altered
26
too as elevated soil temperatures, nutrient cycling and litter decomposition rates and diminished
soil moisture content are seen near edges.
As the edge evolves, changes in the abiotic and edaphic variables are reflected by changes
in the biological community. In fragments, this often results in a deepening of area under edge
influence (Gascon et al. 2000, Harper and Macdonald 2002). Interactions with other
anthropogenically induced threats such as fire, logging, introduction of exotic invasives and the
elimination of predators can result in the complete collapse of fragments in threatened
ecosystems (Curran et al. 1999, Gascon et al. 2000). Further, meso-scale climate phenomena
such as the El-Niño Southern Oscillation can amplify susceptibility to threats such as fires
(Cochrane, 2001). Following the creation of an edge, edges may either deepen (as discussed
above) or regenerate as organisms adapt to the physiological constraints of the edge environment
(Laurance et al. 2002). However at habitat edges that are maintained, a dense, vertical layer of
vegetation can develop, sealing off the forest interior from adjacent communities which leads to
the development of a short, sharp edge (Camargo and Kapos 1995, Didham and Lawton 1999).
These edges are more likely to develop when the contrast between adjacent habitats (patch
contrast) is high such as a forest-grassland edge and can affect a fragment’s persistence (Gascon
et al. 2000, Harper et al. 2005). In tropical montane forests too (TMF), the response to habitat
edges may differ based on edge type. For example, López-Barrera et al. (2006) observed light
transmittance and soil moisture changed abruptly across a hard edge in a Mexican TMF.
Studies on microenvironment and soil edge gradients are limited in TMF (Jose et al. 1996,
López-Barrera et al. 2006) as compared to those of lowland forests (Williams-Linera 1990,
Camargo and Kapos 1995, Williams-Linera et al. 1998, Sizer and Tanner 1999, Didham and
Lawton 1999). This is surprising considering the disproportionate endemism associated with
27
TMF (Hamilton et al. 1995). Moreover, microenvironment edge-interior gradients in TMF can
operate at different scales than lowland forests. Microenvironment edge-interior gradients in
lowland forests typically vary between 40-60m from the forest edge (Kapos 1989, Williams-
Linera 1990, Didham and Lawton 1999) with edge-effects being reported as much as 100m away
from the edge (Laurance et al. 2002). In comparison, edges in tropical montane forests and
patterns are complex and even contradictory. Edge-interior gradients in TMF have been observed
to be as short as 15-30m (Jose et al. 1996), while other studies have reported an insignificant
effect of patch contrast on seedling growth and defoliation (López-Barrera et al. 2006). Some
edges may even display ‘reverse’ edge-effects with lower nest predation rates being recorded
near the edge than the forest interior (Carlson and Hartman 2001) unlike patterns observed for
other forest edges.
Conventional edge-effect studies though (irrespective of biome) typically consider the
effect of distance from the nearest edge. This offers the least complex framework to demonstrate
the response of and interaction between abiotic and biotic response variables to (initially) edge
creation and (later) edge habitat. However, it limits the extrapolation of edge effects across
spatial scales (Ries et al. 2004). Although extensive literature exists on edge-effects as a function
of distance from the nearest edge, limited studies exist on multiple edge effects (Malcolm 1994,
Fernández et al. 2002, Zheng and Chen 2003, Fletcher 2005). Interestingly, even large scale
manipulative studies on fragments of various sizes do not consider the effect of additional edges.
This is despite the fact that edge effects can penetrate as much as 100m away from the edge.
(BDFFP, Laurance et al. 2002). The incorporation of multiple edges has been shown to generate
stronger edge effects than models using distance to nearest edge alone. Further these models
become increasingly significant in small fragments that may be dominated by edge habitat
28
(Gascon et al. 2000). Studies have used different techniques to investigate the effect of multiple
edges on response variables. Malcolm (1994) modeled the edge as a collection of points
integrating the effect of each of these points along the edge (point edge effects). However, as
Fletcher (2005) observed, point edge effects offer a robust, yet abstract estimate of edge effects
since point edge effects are hard to measure. Fletcher (2005) also observed that bobolink
(Dolichonyx oryzivorus) distributions were influenced by multiple edges, but the response to
these edges was determined by scale considerations (i.e. multiple edge effects were less
important at a landscape scale). Increasing the number of edges in edge-effect models has
statistical implications too. As the number of parameters increases, the power of the model
reduces (the probability of Type II error increases). Mancke and Gavin (2000) suggested the use
of an integrative index that combines the use of distance to edges in four directions through a
complex ruleset to ensure orthogonal separation in irregular fragments. They suggest though that
simple orthogonal separation can be ensured by estimating distance to 2nd, 3rd and 4th edges in
three cardinal directions with respect to the edge-interior transect.
Within the Western Ghats-Sri Lanka (WGSL) biodiversity hotspot in southern India, TMF
(locally and hereafter termed as sholas) are seen at elevations ≥1700 m and consist of insular
forest fragments in a matrix of grasslands. Shola fragments are typically small (~1 ha) and are
separated from the surrounding grasslands by a sharp edge which is natural in origin. This study
was conducted on the shola-grassland edge in the Western Ghats within the WGSL hotspot. I
quantified effects of the shola-grassland edge on microenvironment and soil variables at a
regional scale in the shola-grassland ecosystem mosaic. I tested for differences in
microenvironment and edaphic variables between shola fragments and grasslands and to
determine if the edge is controlled by these variables. I also tested for gradients in these variables
29
as a function of distance from the edge and by doing so, tested for differences between small and
large fragments. I hypothesize that the small (1-15 ha) fragments will be dominated by edge
habitat (i.e. weak or absent edge-interior gradients) while large fragments (> 50 ha) will exhibit
strong edge-interior gradients. I also tested for edge-effects in grassland plots along a grassland-
edge-shola interior transect and hypothesize that grassland plots too will be affected by distance
from the shola-grassland edge.
Methods
Study Area
The study area is located in the Western Ghats, a region of high conservation importance
(Myers et al. 2000) in southern India. Three study sites were selected in the Western Ghats in
southern India; Biligiri Rangaswamy Temple (BRT) Wildlife Sanctuary in the state of Karnataka
and Eravikulam National Park and Pampadum shola National Park in the state of Kerala ( Fig 2-
1). BRT Wildlife Sanctuary (540km2), is located at the eastern-most edge of the Western Ghats
(11°43’N to 12°08’ and 77° to 77°16’E). Soils in BRT have been classified as Vertisols and
Alfisols and described as shallow to moderately shallow with well drained gravelly clay soils on
the hills and ridges (Krishnaswamy 2004, NATMO 2009). BRT is exposed to both southwestern
and northeastern monsoons with a dry season from November to May. Rainfall ranges between
898 mm to 1750 mm and strong spatial trends have been observed based on topography and
altitude (Krishnasmamy et al. 2004). Vegetation in the sanctuarty is represented by a diversity of
forest types ranging from dry scrub forests to dry and moist deciduous forests, evergreen forests
and the high-altitude TMF-grassland (or shola-grassland) ecosystem mosaic. Within BRT, three
shola fragments were identified as study sites (Figure 2-1, Table 2-1). Accessibility and the
limited availability of insular shola fragments restricted the number of fragments available in
BRT. Although shola vegetation is more extensive in the southern portion of the sanctuary,
30
(particularly near Jodigere shola) most of these areas were considered inappropriate for the
purpose of this study since they are contiguous with lowland forest and do not represent true
forest fragments.
Eravikulam National Park (ENP) is located in the Anaimalai Hills (10°05’ to 10°20’N and
77°0’ to 77°10’E) within the state of Kerala (Fig 2-1). The park (119 km2) consists of a base
plateau at an elevation of 2000m surrounded by peaks with a maximum elevation of 2695 m.
While the mean maximum temperature recorded in lower elevation tea estates outside ENP
varies between 21.9 °C and 22.7 °C, mean maximum temperature within ENP is as low as 16.6
°C. Similarly, mean minimum temperatures within ENP are lower than those recorded in the tea
estates (13.3 °C). The warmest month in ENP is May with a mean maximum temperature of 24.1
°C and January is the coldest month with a mean mimimum temperature 3 °C (Rice 1984). The
soil had been classified as Alfisols and described as Arachean igneous in origin consisting of
granites and gneisses (NATMO 2009). Soils are sandy clay, have moderate depth (30-100 cm)
and are acidic (ph 4.1-5.3). ENP receives rainfall approximately 5200 mm of rainfall annually
from both the southwest and the northeast monsoon with the former contributing as much as 85
% of the annual rainfall. Although contiguous rainforest formations can also be found at lower
elevations, vegetation in the park is predominantly shola-grassland ecosystem mosaic consisting
of rolling grasslands interspersed with dense, insular shola fragments. Five shola fragments were
selected within ENP (Table 2-1).
Pampadum shola National Park (PSNP) located within Idukki district in the state of Kerala
consists of a large shola patch (1.32 km2) contiguous with lowland forest. Although little is
known about this newly notified park (located approximately at 77°15’E and 10°08’N) its
proximity to the larger Mannavan shola (77°09’-77°12’N and 10°10’-10°12’S) provides insights
31
into broad climatic trends. Mean annual temperature in Mannavan shola has been reported to be
approximately 20°C with mean minimum temperature in the coldest months (December and
January) reaching 5°C. Mean annual precipitation in Mannavan shola ranges between 2000-3000
mm. Data from PSNP were collected to compare patterns observed against smaller fragments
chosen in BRT and ENP.
Sampling Protocol
To investigate edge-interior gradients in shola fragments, transects were laid out
perpendicular to the shola-grassland edge. Plots (10×5 m) were established along these transects
with the first plot being established at the edge itself. The plot was laid out with the longer side
(10 m) running parallel to the edge to capture maximum variability along the edge-interior
gradient. Successive plots along transects were separated by 5 m and extended till the middle of
the patch or to a distance of 35 m from the edge, whichever was shorter. Replicate transects
within a patch were separated by at least 35 m though most often greater than 50 m. The number
of replicate transects (minimum 1, maximum 4) were determined by the size of the fragment.
Shola Plots
Within each 10×5m plot, microenvironment and soil variables were measured. Soil
temperature was measured using a Barnant® Type K thermocouple thermometer (BC Group
International Inc, St Louis, MO) and air temperature and relative humidity were measured using
a digital thermohygrometer (Oakton Instruments, Vernon Hills, IL). Photosynthetically active
radiation (PAR) was measured above and below shola canopy using an AccuPAR LP-80
Ceptometer (Decagon Devices, Pullman, WA). Light transmittance (τ) was then calculated as the
ratio between above and below canopy PAR. Since the shola-grassland ecosystem is
characterized by highly variable light environment, above canopy PAR was measured in the
adjoining grassland prior to measurement of below canopy PAR in the shola plots. From each
32
10×5 m plot, one surface soil (0-30 cm) sample was collected using a soil corer and immediately
double sealed in Ziplock® bags. Soil cores were collected from all but one fragment
(Karadishonebetta, BRT) Gravimetric water content was measured by oven-drying soil samples
for 24 hours at 105°C. Soil samples were analyzed for soil organic carbon (Walkley-Black
method), total nitrogen (micro-Kjeldahl distillation method), and extractable phosphorus (Bray-
Kurtz method). Available potassium and sodium were estimated using the Flame photometric
method (Jackson 1962). Calcium and Magnesium were estimated by EDTA titration (Allen
1974). DTPA extractable iron, manganese, zinc, copper, boron and molybdenum were estimated
using an Atomic Absorption Spectrophotometer (Soultanpour and Schwab 1977).
Grassland Plots
In four of the shola fragments studied, transects were also established in the grassland to
study effects of the shola-grassland edge on the grassland. Since these transects were established
as matching pairs to the shola-transects, they originate at the same point as the shola transect
with replicate transects at least 35 m apart. Microenvironment and soil samples were collected
and soil samples analyzed using protocols described above from two plots at distances of 5 m
and 15 m from the shola-grassland edge to test for gradients along grassland-edge-shola interior
transects.
Data Analyses
In small forest fragments, it was hypothesized that plots would be influenced by more than
one edge. These plots were spatially analyzed using ArcGIS 9.2 (ESRI Labs, Redlands CA) to
determine distance to 2nd, 3rd and 4th nearest edges from each plot (Fig 2-2A). These edges were
determined by measuring the distance to the edge in orthogonal directions, thus achieving a
measure of distance to edge in each cardinal direction. This approach offers an optimum
combination of incorporating information from additional edges while ensuring orthogonal
33
separation (Mancke and Gavin 2004). The spatial analysis of additional edges also revealed
anomalies in the estimation of distance to the nearest edge in small fragments. In some instances,
distances to the edge assumed to be the nearest during data collection would be larger than
distances to the 2nd (on one occasion the 3rd nearest) edge. Although uncommon, these distance
to nearest edge estimates were then revised prior to data analysis (Fig 2-2B). Data obtained from
the large shola fragment (PSNP) were also analyzed separately from other fragments in order to
test for differences between large and small fragments.
An ANOVA was used to test for differences between shola and grassland
microenvironment and soils along grassland-edge-shola interior transects in the fragments (n = 4)
with both shola and grassland plots. Tukey’s HSD was used for multiple comparisons (family-
wise error rate p < 0.05) to test for differences across the six distance classes (two in the
grassland and four in the shola fragments). Additionally, a t-test was used to test for differences
in soil microenvironment conditions between grassland and shola soils. For these analyses, soil
nutrient concentrations were arcsine-square root transformed in order to achieve normality. A
Principal component analysis (PCA) was used to quantify the response of soil nutrient variables
to the shola-grassland edge. PCA is often used to define the underlying structure between
correlated variables and reduce the number of variables to a smaller number of more meaningful
factors and are particularly appropriate when the number of variables is large. Component scores
(or loadings) were used to test for a) differences between shola and grassland soils (n = 4) using
a one-way ANOVA and b) for edge-effects on soil nutrient concentrations (using a simple linear
regression) in shola fragments (n = 8).
A regression analysis was used to study the effect of the forest (shola)-grassland edge on
microenvironment (n = 9) and soil variables (n = 8) in the sholas. A simple linear regression was
34
also used to quantify the response of microenvironment and edaphic variables as a function of
distance from the nearest edge (one-edge model) while a multiple linear regression was used to
quantify the effect of multiple (nearest, 2nd nearest, 3rd nearest and 4th nearest) edges on response
variables (multiple edge model) in small fragments (>1 ha- 13 ha). Response variables were
regressed against distance estimates and log-transformed distance estimates. Log-transformed
distance estimates weight plots near edges higher since I have an a priori expectation that edge-
interior gradients in shola fragments will be short though sharp. All data analysis was performed
using SAS (SAS Institute, Cary NC). Given the small sample size I used a conservative estimate
of p = 0.05 in order to reduce Type I errors.
Results
Comparing Shola and Grassland Plots
Air temperature (p = 0.031) and soil temperature (p < 0.020) declined significantly along
the grassland-edge-shola interior transects (Fig 2-3). Although grassland plots have significantly
lower humidity (p = 0.034) than shola plots no gradients were observed along grassland-edge-
shola interior transects (p = 0.47). Nutritionally, shola and grassland soils differed little (Table 2-
2) and no trends along the grassland-edge-shola interior transect were observed. The use of
principal component analysis (PCA) identified four factors as significant using the latent root
criterion (eigenvalue ≥ 1.0) and explained 73% of the variation (Table 2-3). A varimax rotation
of the selected components was used to maximize orthogonal separation between components
and reduce loading of soil nutrients on more than one principal component (or cross-loading). An
ANOVA of the selected soil nutrient principal components revealed that on the first principal
component (PC1), grassland and shola soils differed marginally (p = 0.054). Organic carbon and
sodium load heavily on PC1 but vary inversely with calcium and magnesium which also load
heavily on PC1 (Table 2-3). None of the other principal components (PC2, PC3 and PC4) varied
35
between habitat type (p ≥ 0.35). Further none of the factors varied along the two-sided grassland-
edge-shola interior transect (p ≥ 0.22).
Microenvironment Gradients in Shola Fragments
In the large shola fragment (PSNP), relative humidity and soil temperature decreased
linearly with increasing distance from edge (Table 2-4, Figure 2-4). While relative humidity
decreased strongly (p = 0.01) with increasing distance from edge and soil temperature decreased
strongly weakly (p = 0.06), no trends were observed in air temperature (p = 0.74) and light
transmittance (p = 0.79). However, light transmittance decreased to 3(±0.3) % of overstory light
conditions within 5 m of the edge. Although most soil nutrients did not vary as a function of
distance from the nearest edge, available potassium increased (p = 0.0089) with increasing
distance from edge (Table 2-3, Fig 2-2). Contrary to other studies that have reported strong non-
linear relationships in soil variables (Jose et al. 1996), I found no evidence of non-linearity.
Adding a quadratic (distance to nearest edge) term did not significantly improve the fit of
relative humidity (p = 0.018) and reduced the significance of estimates of available potassium (p
= 0.030) and other significant variables (p ≥ 0.101). A factor analysis of soil nutrient variables
with quartimax rotation for PSNP soils revealed two factors as significant that explained 72% of
the variation (data not presented). While the first factor loads heavily on organic carbon, total
nitrogen, available phosphorous and calcium, the second factor loads heavily on manganese.
However a linear regression of factors against both factors did not vary with distance from the
nearest edge (p ≥ 0.13).
Small forest fragments though showed no trends in microenvironment (air temperature,
relative humidity, light transmittance) or soil variables (soil temperature, macronutrient and
micronutrient concentrations) as a function of distance of nearest edge (single edge model).
Responses were weaker still when log-distances were used (p ≥ 0.242). The incorporation of
36
information through distance estimates for additional (2nd, 3rd and 4th nearest) edges through the
multiple edge model resulted in a significant improvement of the regression models. With
increasing distance from edge, significant patterns were observed as air temperature (p = 0.0320)
and light transmittance (p = 0.0072) decreased. Weaker trends were observed with relative
humidity increasing and soil temperature decreasing with increasing distance from edge (Table
2-4). Soil macronutrient variables were not correlated with edge distance using either the one-
edge or multiple edge models. Factor analysis of soil nutrient variables produced three factors
(and explained 55%) of the variation). Again, none of these factors were correlated with either
the one-edge or multiple edge models. A canonical correlation was also used to test all soil
nutrient variables against all distances in order to test for correlations between the variables.
However, the analysis failed to find any significant correlations.
Discussion
The microenvironment in grasslands was significantly different from that of shola
fragments. However, no significant differences were observed between grassland and shola
fragment soils. Similar trends in soil response variables have been reported by other authors too
(Jose et al. 1994, Nandakumar et al. 2001). Our estimates for soil nutrient variables for sholas
and grasslands (Table 2-1) are comparable to others from the ecosystem mosaic (Thomas and
Sankar 2001). However, significantly higher estimates for soil organic carbon and soil moisture
have also been reported from the ecosystem mosaic (Jose et al. 1994). Factor analysis of soil
variables indicates that soil organic carbon and calcium load had positive and negative loading
on the first factor (which explained 20% of the variation in the dataset). This is typical of forest
soils that have higher soil organic carbon while basic cations (Ca2+) are lower (as they are lost to
leaching) than grassland soils.
37
Although two-sided edges are less investigated, their proportion vis-à-vis one sided edge
studies has increased considerably (Fonseca and Joner 2007). Their relevance to matrix-based
approaches to understanding edge-effects and restoration ecology is also being increasingly
recognized (Ries and Sisk 2004, Ries et al. 2004, Fonseca and Joner 2007). This could be
significant in the shola-grassland ecosystem mosaic where the matrix surrounding forest
fragments (typically grassland), has been shown to be hostile to animal taxa (Shanker 2001, Zarri
et al. 2008). In our study I found evidence of two-sided edges in microenvironment response
variables. However this pattern was absent in soil response variables with the exception of weak
trends in soil organic carbon. Information on two-sided edges in the shola-grassland ecosystem
mosaic can also assist in ecosystem restoration efforts by providing requisite conditions for re-
establishment of shola fragments in areas converted to exotic tree plantations.
Soils in the shola-grassland ecosystem mosaic had high levels of DTPA extractable iron.
Although much higher than another estimate from the mosaic (Nandakumar et al. 2001) it is
comparable to those observed from TMF-grassland edge in Sri Lanka experiencing dieback
(Ranasinghe et al. 2007). Although Ranasinghe et al. (2007) suggested that the dieback of TMF
could be related to iron toxicity induced diminished nitrogen absorption for soils with iron levels
that are higher (200-400 ppm). I did not observe such high levels of DTPA extractable iron in
our study. However, unlike TMF-grassland soils in Sri Lanka, shola fragment and grassland soils
alike did not record high DTPA extractable manganese or copper levels.
Large shola fragments exhibit different patterns from smaller shola fragments. In
Pampadum shola distance from the nearest edge was sufficient to explain trends in
microenvironment and edaphic variables. However, distance from the nearest edge could not
sufficiently explain the soil variables.
38
Seasonal and diurnal variation
Since individual transects were sampled over the course of a single day, a diurnal variation
in relative humidity and air temperature can be expected. As the day progresses, ambient air
temperatures would increase leading to a lowering of relative humidity. An inverse relationship
between relative humidity and air temperature, it can be argued, could be an artifact of either
distance from the forest fragment (shola)-grassland edge or diurnal variation. However, in our
study, air temperature and relative humidity were not inversely related. In smaller fragments, we
did indeed observe such trends (air temperature decreased with increasing distance while relative
humidity increased). However in the large fragment (PSNP), relative humidity decreased with
increasing distance from edge (and as the day progressed) while air temperature showed no edge-
interior trends. Such changes in relative humidity trends as a function of distance from the edge
could also be an artifact of seasonality. Sampling of the large fragment (PSNP) was done in the
wet season and often under constant cloud cover leading to higher relative humidity near the
shola-grassland edge and lower humidity inside (Figure 2-4). Since the sampling of small
fragments was carried out in both wet and dry seasons, trends are more intuitive (increasing
humidity with increasing distance from edge) albeit weakened by wet season data.
Our estimate of light transmittance (3% of overstory light conditions) is significantly lower
than light transmittance at the edge (12%) reported by Jose et al. (1996) although similar to
estimates from old growth rainforest-pasture edges in Mexico (4.7%, Williams-Linera et al.
1998). Other studies too have reported an abrupt change in light conditions across a high contrast
edge (López-Barrera et al. 2006). Shola fragments thus represent deeply shaded habitats and low
depth of influence (DEI) with regard to light transmittance. Analysis of the large shola fragment
(PSNP) revealed trends (although weak) widely reported from other edge-effect studies (e.g. soil
39
temperature) both in tropical montane forests and lowland tropical forests (Jose et al. 1996,
Laurance et al. 2002).
The present study investigates edge-interior gradients in microenvironment and soil
nutrient availability across nine tropical montane (shola) fragments. To my knowledge only other
study in the shola-grassland ecosystem investigates edge-interior gradients (Jose et al. 1996).
Although the study reported distinct edge-interior gradients for both microenvironment and
edaphic variables, this dataset is based on data from a single fragment.
Comparison between One-edge and Multiple Edge Models
Studies on edge-effects across biomes have demonstrated the effect of distance from edge
on microenvironment and edaphic variables. In tropical forests, edge effects have been
particularly worrisome as microenvironment gradients extend as much as 100m away from the
edge of forested habitat (Laurance et al. 2002). As a corollary to this, smaller fragments have
been hypothesized to be dominated by edge habitat (Gascon et al. 2000). Most studies in tropical
fragments though measure the response variables as a function of distance from nearest edge and
do not incorporate the effect of distance from additional edges (however, see Malcolm 1994). In
the shola-grassland ecosystem mosaic, large shola fragments exhibited different patterns from
smaller shola fragments. In Pampadum shola distance from the nearest edge was sufficient to
explain trends in microenvironment and edaphic variables. However in smaller forest fragments,
microenvironment and edaphic variables did not show a response with distance from the nearest
edge suggesting that fragments are dominated by edge habitat. However, the incorporation of
additional edges as predictor variables in the model significantly improved model fit. Moreover
multiple edge-models were more parsimonious (lower AIC values) than one-edge models (Table
2-4). To my knowledge, no other study from the shola-grassland ecosystem and indeed other
other tropical montane forest fragments test for multiple edge effects. The incorporation of
40
multiple edge-effect models might better explain the absence of edge interior gradients in other
edge effect studies also (Williams-Linera et al. 1998).
Delineating edges and determining edges using spatial analysis techniques may not always
be feasible though. In our study, insular forest fragments are isolated from other forest types by a
high contrast grassland edge. This made determination of distance to additional edges relatively
easy. When edges separate similar habitat types (such as a native forest-tree plantation edge)
distinguishing habitat types might require the use of other techniques (landcover/landuse
classification).
Conclusion
The present study describes the response of microenvironment and edaphic variables to
distance from a tropical montane forest (shola)-grassland edge. Nutritionally, grassland and shola
soils differed little. Since such results have been consistently observed across multiple study sites
in the ecosystem mosaic, an edaphically controlled shola-grassland edge appears unlikely. The
influence of other driving variables such as fire or frost needs further investigation. I observed
that conventional one-edge models sufficiently explained variation trends in microenvironment
variables along the edge-interior gradient in large fragments. As with other studies on small
fragments though, I observed no edge effects with the use of a conventional one-edge model.
However, the inclusion of multiple edges in small fragments significantly improved model fit. I
can conclude that small fragments observed to be dominated by edge habitat may in fact
resemble larger fragments with the inclusion of multiple edges. Our models did not evaluate non-
linear effects which often better explain patterns in edge-interior gradients (Malcolm 1994). The
incorporation of such non-linear models in the system might further improve model fit. Finally,
further research is required in investigating the effect of multiple edge models in predicting edge
41
effects across fragment sizes, edge types and biomes in order to improve our understanding of
edge effects.
42
Table 2-1. Location of tropical montane forest (shola) fragments across study sites in the Western Ghats, southern India.
Fragment (shola name)
Fragment code
Study site Sampling season
Longitude (°E)
Longitude (°N)
Patch area (ha)
Karadishonebetta KR BRT Dry - - - Gummanebetta GU BRT Dry 77° 10' 38" 12° 1' 35" 1.73 Jodigere JG BRT Dry 77° 11' 12" 11° 47' 6" 0.20 Suicide point SP ENP Wet 77° 1' 33" 10° 8' 23" 1.24 Chinnanaimudi CA ENP Wet 77° 3' 24" 10° 9 '4" 1.66 Kolathan KS ENP Wet 77° 4' 31" 10° 12' 57" 1.58 Pusinambara PS ENP Wet 77° 4' 3" 10° 12' 56" 1.75 Mukkal mile MM ENP Wet 77° 5' 7" 10° 13' 51" 13.42 Pampadum PP PSNP Wet 77° 15' 5" 10° 7' 28" 132.00 Table 2-2. Plant-essential nutrient concentrations in tropical montane forest (shola) and
grassland surface soils in the Western Ghats, southern India. Values are means (±1 SE) expressed as g kg-1 unless noted otherwise.
Soil nutrient parameter Shola Grassland Macronutrients Soil organic carbon 11.28* (0.293) 10.52* (0.625) Nitrogen 0.53 (0.0090) 0.55 (0.0152) Phosphorous 0.144 (0.0036) 0.143 (0.0060) Potassium 0.135 (0.00033) 0.133 (0.00699) Calcium 0.139 (0.0033) 0.145 (0.0052) Magnesium 0.142 (0.0033) 0.134 (0.0069) Micronutrients Iron 0.121(0.0016) 0.124 (0.0040) Boron 0.0057 (0.00013) 0.0058 (0.00029) Manganese 0.00013 (0.0000046) 0.00014 (0.0000117) Zinc 0.048 (0.0011) 0.050 (0.0017) Copper 0.000056 (0.0000014) 0.000061 (0.0000029) Molybdenum 0.000138 (0.0000045) 0.000134 (0.0000111) * P < 0.05
43
Table 2-3. Principal components scores for soil nutrient variables in tropical montane forest
(shola) fragments and grassland soils (n = 4) in the Western Ghats, southern India (eigenvalue ≥ 1). Bold values indicate the component on which the variable loads after a varimax rotation was applied.
Variable PC1 PC2 PC3 PC4 Eigenvalue 2.6339 1.7456 1.5472 1.1782 Proportion variance explained 0.2634 0.1746 0.1547 0.5927 Cumulative variance explained - 0.4380 0.5927 0.7105 OGC 0.8230 -0.2062 0.2420 -0.1481 NIT -0.2197 0.8292 -0.0645 -0.0431 PHO -0.0569 0.7863 0.1720 0.3743 POT -0.1527 0.0845 -0.0660 0.8147 CAL -0.6289 0.0689 0.5139 0.2867 MAG -0.7399 -0.0993 0.3650 -0.311 SOD 0.6805 -0.1299 0.1163 -0.1697 MAN 0.0357 0.6671 0.1054 -0.5675 MOL 0.0254 0.1001 0.8919 0.0920 COP -0.0454 0.0006 -0.6009 0.2412 OGC: Organic carbon, NIT: Total Nitrogen, PHO: Available phosphorous, POT: Potassium, CAL: Calcium, MAG: Magnesium, SOD: Sodium, MAN: Manganese, MOL: Molybdenum, COP: Copper Table 2-4. Edge-interior gradients in tropical montane forest (shola) fragments in the Western
Ghats, southern India. Small fragments (n = 7) varied between 0.2-13 ha whereas the large fragment (n = 1) was 132 ha.
Variable
Fragment size Large Small Predictor variables
r2 p Predictor variables
r2 p
Air temperature (°C) 0.017 0.7492 logdist, logdist2
0.174 0.0320
Relative humidity (%) logdist 0.443 0.0181 0.142 0.0625 Light transmittance (%) 0.006 0.7978 logdist,
logdist2 0.268 0.0072
Soil temperature (°C) 0.302 0.0637 0.180 0.0695 Total nitrogen 0.010 0.7568
0.0637 0.1068
Magnesium logdist 0.3990 0.0276 0.002 0.7418 Potassium (%) logdist 0.511 0.0089 0.007 0.5864 Avail. Phosphorous (%) 0.316 0.0570 0.114 0.1979 Calcium 0.067 0.4149 0.114 0.3305 Significant relationships are in bold (p < 0.05), number of predictor variables used in the model was chosen based on Akaike information criterion (AIC). Predictor variables are only displayed for significant models.
44
Figure 2-1. Location of study sites in the Western Ghats. Insular forest fragments were selected from three sites- Biligiri Rangaswamy Temple Wildlife Sanctuary (BRT), Eravikulam National Park (ENP) and Pampadum shola National Park (PSNP). Three fragments were studied in BRT (A), Gummanebetta (GU), Jodigere (JG) and Karadishonebetta (KR). Five fragments were studied in Eravikulam National Park (B), Chinnanaimudi (CA), Kolathan (KS), Mukkal Mile (MM), Pusinambara (PS) and Suicide point (SP). See Table 2-1 for details on.fragments. For inset maps, 1cm equals 4 kilometers
45
Figure 2-2. Schematic representation of estimation of distance to additional edges. (A) d1, d2, d3 and d4 represent distance estimates to 1st, 2nd 3rd and 4th nearest edge. (B) In small fragments, estimated distances to additional edges (gray letters) differed from actual distances (black letters). These distance estimates were revised prior to data analysis. Black dots represent the plot center and d1-d4 lines represent distance to nearest, 2nd, 3rd, and 4th nearest edges respectively.
A B
46
Figure 2-3. Effect of the shola-grassland edge (dashed line) on microenvironment variables in the shola-grassland ecosystem mosaic. (A) Soil temperature. (B) Air temperature. (C) relative humidity. Different letters indicate significant differences as determined by Tukey’s HSD comparison (p < 0.05). Error bars represent one standard error.
47
2.5 12.5 22.5 32.5
Mag
nesi
um (%
)
0.010
0.012
0.014
0.016
0.018
Rel
ativ
e hu
mid
ity (%
)
50
60
70
80
90
100
Pota
ssiu
m (%
)
0.008
0.010
0.012
0.014
0.016
0.018
0.020
A
B
C
Distance to edge (m)
Figure 2-4. Edge-interior gradients in microenvironment and soil nutrient parameters in Pampadum shola National Park (~132 ha). Error bars represent one standard error.
48
CHAPTER 3 SPECIES DISTRIBUTION PATTERNS IN SHOLA FRAGMENTS
Introduction
The creation of an edge in forested habitat alters both abiotic and biotic components of
ecosystems. Immediate changes to microenvironment conditions at the edge— higher air and soil
temperatures, lowered relative humidity and soil moisture— are a result of elevated insolation
and turbulence caused by an alteration of the boundary interface. Changes are manifested in
altered ecosystem processes: evapotranspiration, litter decomposition, nutrient cycling and
dispersal. These ecosystem processes interact iteratively through feedback loops with the biotic
component of ecosystems causing a shift in species composition through changes in recruitment,
growth, reproduction and mortality. Although edges affect the biotic and abiotic components, the
influence of abiotic processes diminishes as species adapt to altered conditions (Laurance et al.
2002). Biological responses to edge creation include elevated tree mortality (Williams-Linera
1990) and reduced canopy and sub-canopy cover (Laurance 1991). This opening of the canopy
manifests itself in increased seed rain from generalist species at the edge and an increase in
density of liana and understory species (Janzen 1983, Laurance 1991). These shifts in biota last
longer than microenvironment response to edge creation and may even be permanent. Spatially
too, the biotic response to edge creation is of a much larger magnitude than those of
microenvironment edge-interior gradients (i.e., high DEI, Harper et al. 2005). Long-term, large
scale manipulative experiments on Amazonian fragments reveal that while microenvironment
gradients extend up to 100m from the forest edge, edge effects on tree mortality can penetrate as
much as 300m from the edge (Laurance et al. 2002). Larger estimates of species response to
forest disturbance (~10 km) have also been attributed to edge creation (Curran et al. 1999).
49
However these might be an artifact of the collapse of predator populations in fragments rather
than edge effects per se (Ickes and Williamson 2000, Laurance 2000).
In tropical montane forests (TMF), species richness and vegetation structure can change
rapidly over short distances (Nadkarni and Wheelwright 2000). β-diversity was observed to
increase with an increase in elevation in TMF fragments in Mexico indicating that fragment
complementarities increased with increasing elevation (Williams-Linera et al. 2002). Species
within TMF fragments are often dispersal limited, such that late successional species probably
regenerate from seeds originating within the patch itself (del Castillo and Rios 2008). As a result
TMF fragments maybe characterized by high endemism and the number of unique species
assemblages is negatively correlated with fragment size (Young and Leon 1995, Ganeshaiah et
al. 1997). Smaller fragments are thus more likely to form unique habitats and, from a
conservation standpoint, are equally as important as larger fragments. Studies on species
response to a forest-grassland (or forest-pasture) edge in fragmented TMF ecosystems reveal
contrasting trends. For example, Jose et al. (1996) found trends in species distributions as a
function of distance from the edge in TMF fragments in southern India with sub-tropical and
temperate species near the edge while species with tropical affinities were located in interior
plots. On the other hand, studies in other TMF fragments report weak or complex distance to
edge trends in species diversity (Oosterhoorn and Kappelle 2000) and stem density (Young and
Keating 2001).Often though, TMF fragments may be a greater source of variation than distance
to edge per se (Williams-Linera 2003). Reduced canopy heights have also been recorded from
edge habitats in Costa-Rican TMF fragments. Combined with increased sub-canopy heights this
resulted in a simplification of vertical structure of edge habitat as compared to fragment interiors
(Oosterhoorn and Kappelle 2000).
50
The theory of island biogeography (IBT, Macarthur and Wilson 1963, 1967) offers an
elegant and robust explanation of the process of colonization and speciation on islands.
Numerous studies since have sought to apply the theory to fragmented and insular terrestrial
habitats (for a review see Laurance 2008). However, as Laurance (2008) indicates, system
properties associated with terrestrial fragments (e.g. non-random habitat conversion, edge effects
and matrix hostility) render IBT inadequate in explaining the effects of habitat fragmentation. In
terrestrial ecosystems, unlike true oceanic islands, the surrounding matrix has a significant
influence on the magnitude and direction of species responses to edge effects. In a study of
Amazonian forest fragments, fragments bordering cattle pastures (a high-contrast edge) had
higher tree mortality rates than low-contrast forest-secondary regrowth edges (Mesquita et al.
1999). In the Western Ghats, TMF (hereafter shola) occurs as insular forest fragments in a matrix
of grasslands separated by a stable, natural, high contrast edge. Since the edge is natural in
origin, we can expect equilibrium in species distribution patterns as a function of edge-related
processes. Although the hostility of the matrix, as observed in the response of species to the edge
(Shanker 2001) is often suggestive of true oceanic islands, it is unlikely that IBT alone will
sufficiently explain patterns observed in the ecosystem.
The study of created (or anthropogenic) edges in forested habitats often has direct
management implications. However, natural edges (such as those observed in the shola-grassland
ecosystem mosaic) allow us to observe patterns and processes operating at edges while isolating
human influence. Edge effects in naturally patchy ecosystems though may be suppressed or even
absent (Pavlacky and Anderson 2007). In a study of nest predation on TMF in Tanzania, contrary
to expectation, higher nest predation rates were reported for nests away from forest edges
(Carlson and Hartman 2001). Natural fragments may also maintain a high diversity of tree
51
species while vertebrate (especially mammal) diversity may not be comparable (Bowman and
Woinarski, 1994). Finally, edge-effect related processes are also dependent on patch size. As
fragment size reduces the proportional area under edge influence increases. This effectively
reduces the core-area of a fragment that land managers and conservationists are advised to
maximize by selecting for larger fragments (Gascon et al. 2000). However, smaller fragments
may serve as long-term refugia for threatened species (Turner and Corlett 1996), differ
compositionally from original forests (Tabarelli et al. 1999) and even contain unique species
assemblages not associated with larger fragments (Ganeshaiah et al. 1997).
This study was conducted in the shola-grassland ecosystem mosaic in southern India. I
quantified species distribution patterns as a function of distance from the edge. I hypothesize that
there will be edge-interior gradients in species distributions and expect abundance and
dominance patterns to be correlated with microenvironment and edaphic variables. Since edges
offer increased light transmittance to understory vegetation I expect edges to have greater
understory structural complexity and higher overstory basal area than interior plots. Further,
since the edge is a high contrast edge, I expect higher densities near the edge. Finally, since data
is available at a regional scale, I anticipate compositional differences between overstory species
in sub-montane and montane forests.
Methods
Study Area
Three study sites were selected in the Western Ghats in southern India within the Western
Ghats-Sri Lanka (WGSL) biodiversity hotspot; Biligiri Rangaswamy Temple (BRT) Wildlife
Sanctuary in the state of Karnataka and Eravikulam National Park and Pampadum shola National
Park in the state of Kerala ( Fig 2-1). BRT Wildlife Sanctuary (540km2), located at the eastern-
most edge of the Western Ghats (11°43’N to 12°08’ and 77° to 77°16’E), occurs at the
52
confluence of the Eastern Ghats and the Western Ghats. Soils in BRT have been classified as
Vertisols and Alfisols and described as shallow to moderately shallow with well drained gravelly
clay soils on the hills and ridges (Krishnaswamy 2004, NATMO 2009). Rainfall in BRT—
which is exposed to both southwestern and northeastern monsoons —ranges between 898 mm to
1750 mm and strong spatial trends have been observed based on topography and altitude
(Krishnasmamy et al. 2004). The dry season extends from November to May with February
recorded as the driest month. Vegetation in the sanctuary is represented by a diversity of forest
types ranging from dry scrub forests, dry and moist deciduous forests, evergreen forests and the
high-altitude TMF-grassland (or shola-grassland) ecosystem mosaic. Within BRT, three shola
fragments were identified as study sites (Figure 2-1, Table 3-1). Accessibility and the limited
availability of insular shola fragments restricted the number of fragments available in BRT.
Although shola vegetation is more extensive in the southern portion of the sanctuary,
(particularly near Jodigere shola) most of these areas were considered inappropriate for the
purpose of this study since they are contiguous with lowland forest and do not represent true
forest fragments.
Eravikulam National Park (ENP) is located in the Anaimalai Hills (10°05’ to 10°20’N and
77°0’ to 77°10’E) within the state of Kerala (Fig 2-1). The park (119 km2) consists of a base
plateau at an elevation of 2000m surrounded by peaks with a maximum elevation of 2695 m. The
mean maximum temperature recorded within ENP is 16.6 °C while the mean minimum is 6°C.
The warmest month in ENP is May with a mean maximum temperature of 24.1 °C and January
is the coldest month with a mean mimimum temperature 3 °C (Rice 1984). The soil had been
classified as Alfisols and described as Arachean igneous in origin consisting of granites and
gneisses (NATMO 2009). Soils are sandy clay, have moderate depth (30-100 cm) and are acidic
53
(ph 4.1-5.3). ENP receives rainfall approximately 5200 mm of rainfall annually from both the
southwest and the northeast monsoon with the former contributing as much as 85 % of the
annual rainfall. Although contiguous rainforest formations can also be found at lower elevations,
vegetation in the park is predominantly shola-grassland ecosystem mosaic consisting of rolling
grasslands interspersed with dense, insular shola fragments. Five shola fragments were selected
within ENP (Table 3-1).
Pampadum shola National Park located within Idukki district (approximately at 77°15’E
and 10°08’N) in the state of Kerala consists of a large shola patch (1.32 km2) contiguous with
lowland forest. Mean annual temperature in the adjacent Mannavan shola has been reported to be
approximately 20°C with mean minimum temperature in the coldest months (December and
January) reaching 5°C. Mean annual precipitation in Mannavan shola ranges between 2000-3000
mm. Data from PSNP were collected to compare patterns observed against smaller fragments
chosen in BRT and ENP.
Sampling Protocol
To investigate edge-interior gradients in species distributions in shola fragments, transects
were laid out perpendicular to the shola-grassland edge. Overstory structure was characterized in
terms of basal area and density of tree species in 10×5 m plots along each transect with the first
plot being established at the edge itself. Plots were laid out with the longer side (10 m) running
parallel to the edge to capture maximum variability along the edge-interior gradient. In each plot,
trees ≥ 5 cm dbh were identified to species and measured. Overstory species were assigned to
one of 5 diameter classes based on dbh (5-10 cm, 11-20 cm, 21-40 cm and > 40 cm). Nested 5×5
m sub-plots were established to identify and measure understory vegetation. For the purpose of
this study all vegetation ≥ 50cm in height was defined as understory vegetation. Understory
individuals were assigned to one of five height classes (50-99 cm, 100-149 cm, 150-199 cm,
54
200-249 cm and > 250 cm) based on maximum height (Hmax). Specimens were collected for
overstory and understory individuals that could not be identified confidently and were later
identified with the help of taxonomists at Ashoka Trust for Ecology and the Environment,
Bangalore. I also collected data on microenvironment (soil and air temperature, relative
humidity, light transmittance, soil moisture) and soil macronutrient (organic carbon, total
nitrogen, available phosphorous, potassium) variables from 10×5 m plots. Details on techniques
used to record microenvironment data and collect soil nutrient samples are presented earlier in
Chapter 2.
Successive plots along transects were separated by 5 m and extended till the middle of the
patch or to a distance of 35 m from the edge, whichever was shorter. Edge effects in shola
fragments are negligible at this distance (Jose et al. 1996). Thus, data are available from four
classes- 2.5 m, 12.5 m, 22.5 m and 32.5 m (as measured from the shola-grassland edge to the
mid-point of the plot)- along the edge-interior gradient. Replicate transects within a patch were
separated by at least 35 m though most often greater than 50 m.
Data Analyses
Vegetation Structure and Composition
I assessed plant species richness between fragments and completeness of species
inventories using a sample-based rarefaction (species accumulation curves) approach as
calculated by EstimateS (Colwell 2006). Species accumulation curves were calculated using the
Mao-Tau function (Colwell et al. 2004, Mao et al. 2005). Although the Mao-Tau function does
not require resampling runs to calculate species accumulation curves (Sobs), the sampling order
was randomized over 100 runs to compute singletons (species occurring in one sample) and
doubletons (species occurring in two samples). Species diversity was assessed using Fisher’s
alpha index and richness was estimated using the nonparametric Bootstrap estimator (Smith and
55
van Belle 1984) since it is well suited to quadrat data (Chao 2005). Moreover, it provided the
most stable and least biased estimate of species richness for overstory and understory species.
Species diversity was assessed across all fragments using Shannon-Weiner index (H') (Magurran
1988). Beta-diversity was computed using the Chao Abundance-based Sørensen similarity index.
These indices correct for the undersampling bias associated with traditional similarity indices
(e.g., Sørensen similarity index) and are especially suitable for use when rare species are
numerous (Chao et al. 2005). Fragment complementarity (C) between pairs of fragments was
assessed by calculating the Marczewski-Steinhaus distance (Colwell and Coddington 1994).
Complementarity varies from zero (when all species are common to both fragments) to unity
(when no species are common to both fragments). Regression analysis was used to test for trends
in overstory dominance and vertical structure of overstory (dbh distribution) and understory
(Hmax) as a function of distance to the shola-grassland edge. Soil samples were analyzed for
macronutrient concentration using standard techniques (see Chapter 2). A stepwise regression
was used to determine the microenvironment and edaphic variables determining overstory
dominance.
Edge-interior Gradients in Species Distributions
I quantified the variation in species compositional patterns as a function of distance to the
shola-grassland edge across fragments using non-metric multidimensional scaling (NMS) as
implemented in PC-ORD (McCune and Mefford 2006). NMS was run in the ‘medium’ mode
(200 iterations, 0.0001 instability criterion, 50 runs and four starting axes) of the autopilot
procedure using the Sørensen distance measure to calculate dissimilarity coefficients. A Monte-
Carlo randomization test (50 runs) was used to test the best solution against a randomized
dataset. A three dimensional solution yielding a stress significantly lower than obtained by
chance was selected for each of the overstory (p = 0.019) and understory datasets (p = 0.019). A
56
joint plot was constructed to explore the relationship between species and environmental
variables (soil temperature, air temperature, relative humidity, soil moisture, soil organic carbon,
total nitrogen, available phosphorous and potassium). Axis scores were correlated with
environmental measures to assess the ecological significance of axis in ordination space. For the
understory dataset light transmittance was included as an environmental variable.
Results
Vegetation Structure and Composition
I recorded 2379 individuals representing 111 species from 49 families in the nine
fragments studied. Fragment size varied from 0.2 ha to 132 ha (with most fragments ~1 ha). The
most abundant families are Lauraceae (19 species), Rubiaceae (18 species), Myrsinaceae (7
species), Myrtaceae and Oleaceae (5 species each). Species accumulation curves failed to reach
an asymptote for overstory (Figure 3-1a) and understory (Figure 3-1b) species. Curves for
singletons (species occurring once) did not decline for overstory species but declined marginally
for understory species indicating adequate sampling for rare species in understory vegetation.
Shannon-Weiner diversity indices (H') were 3.96 and 3.57 for overstory and understory species
respectively (Table 3-2). The Bootstrap estimator indicated that 12 additional overstory species
and 13 additional understory species are estimated as not being represented in the dataset
assuming that the selected fragments were representative of the shola ecosystem (Figure 3-1a, 3-
1b). On a landscape scale, shola fragments had high variation between fragments. This was
especially true for overstory species with complete separation often observed between mid
elevation and high elevation fragments (Chao-Sørsensen index: 0.0). For understory species
mean complementarity was lower (C = 0.85) than overstory (C = 0.90) and only one pair of
fragments (JG-MM) showed complete separation.
57
Edge-interior gradients in vegetation were largely absent from shola fragments. Although
edge plots (2.5m) on average had higher understory density than interior plots (Fig 3-2), no
statistically significant trends were observed along the edge-interior gradient (p = 0.190).
Structurally, there were no edge-interior trends across overstory dbh classes (p ≥ 0.202) and
trends were inconsistent. With increasing distance from edge, the total number of individuals
across diameter classes decreased in some fragments while other fragments saw an increase in
total number of individuals (Figure 3-3). Others still appeared to have a bimodal distribution
(MM, GU). Only in the largest fragment (PP) were the largest trees (> 40cm dbh) present across
all distances. In the understory, while the density of individuals in the smallest height class (50-
99cm) declined weakly (p = 0.079) along the edge-interior gradient, such trends were absent
across other height classes (p ≥ 0.581). Mean plot basal area for overstory species across
fragments varied between 15.5m2ha-1 and 78.5m2ha-1 (Table 3-1). However, no trends were
observed in overstory basal area along the edge-interior gradient (p = 0.897). Stepwise regression
revealed that soil temperature was significantly related to the variation in overstory dominance (p
= 0.039). However, this variation may be an effect of rather than a cause of the variation in
overstory dominance. No other microenvironment or soil variables sufficiently explained the
variation in overstory dominance.
Edge-interior Gradients in Species Distributions
For the overstory dataset NMS ordination produced three significant axes which accounted
for 57% of the variation between species in the ordination space and the original space. A stable
solution (instability = 0.00005) was reached in 200 iterations. NMS separated the plots
reasonably well in ordination space (Figure 3-4). The alignment of overstory plots was better in
terms of fragments than distance to edge classes with lower elevation shola fragments occupying
the ‘south eastern’ portion of the ordination cloud. Axis 2 showed a strong correlation with
58
relative humidity (r = 0.536) and soil moisture (r= 0.560) and slightly weaker correlations with
soil organic carbon (r = 0.458), total nitrogen (r = 0.467), available phosphorous (r = 0.350) and
potassium (r = 0.414). Axis 2 thus appears to represent soil nutrient availability well. Axis 3 was
weakly correlated with soil temperature (r = 0.284) and air temperature (r = 0.245) alone while
Axis 1 was not correlated with any of the environmental variables (Table 3-4).
NMS ordination indicated three axes as significant which accounted for 59.3% of the
variation between understory species. A stable solution was reached in 169 iterations (instability
< 0.00001). For understory and overstory datasets, final stress (which is a measure of the
dissimilarity between plots in ordination space and original space) was relatively high (18.13 for
overstory; 16.79 for understory). Similar to overstory, understory plots aligned better as
fragments than distance classes. However the separation between shola fragments based on
elevation was not seen. A separation between plots based on fragment size was observed with
larger shola fragments (PP and MM) occupying south-eastern portion of the ordination cloud
(Figure 3-5). Axis 1 was strongly correlated with the environmental variables relative humidity
(0.447) and soil moisture (r = 0.452) and weakly with soil organic carbon (r = 0.328). Axis 2 was
weakly correlated with air temperature (r = 348), soil temperature (r = 0.320) total nitrogen (r =
0.374) and available phosphorous (r = 0.383) while correlations with Axis 1 were weak.
Discussion
Patterns in small fragments may contrast from those observed in larger fragments and
small fragments may be dominated by edge habitat (Tabarelli et al. 1999, William-Linera et al.
1998). In this study, I observed an absence of edge-interior gradients in overstory dominance and
understory density in tropical montane forest (shola) fragments. Although this is typical of a high
contrast edge (low DEI, Harper et al. 2005), shola fragments are not ‘sealed’ by a wall of
vegetation across strata as seen at other high contrast edges (Strayer et al. 2003, Laurance et al
59
2002). I also observed that vertical complexity of the understory did not vary along the edge-
interior gradient. In contrast, a study on Costa Rican tropical montane forest fragments observed
a decrease in density and simplification in understory structure with increasing distance from
edge (Oosterhoorn and Kappelle 2000). Departures in my dataset from patterns observed in
neotropical montane fragments could be due to differences in the origin of tropical montane
fragments studied. While neotropical fragments (such as those in Costa Rica) are often
anthropogenic in origin, tropical montane forest fragments in the Western Ghats are natural in
origin (Sukumar et al. 1993). Other naturally fragmented tropical montane forests too have been
observed to be characterized by atypical edge effects (Carlson and Hartman 2001). With
increasing fragment age, species equilibrate with the microenvironment leading to a softening of
edge effects (Harper et al. 2005). Structurally, mean plot basal area ranged between 15.5 m2ha-1
to 78.5 m2ha-1, with high elevation shola fragments generally supporting a much higher basal
area (>50 m2ha-1) than mid-elevation fragments. The basal area values reported are within the
range of those reported for other TMF fragments (Williams-Linera 2002) but higher than those
reported from other shola fragments (Jose et al. 1994). The variation in overstory basal area
could be significantly, albeit weakly, explained only by soil temperature. Soil temperature has
been shown to be a good indicator of the penetration of edge effect on microclimatic variables
since it is coupled with changes in insulating overstory vegetation (Laurance and Yensen 1991).
I found 111 species from 49 families across the nine fragments studied which is higher
than other studies from shola fragments (Jose et al. 1994, Davidar et al. 2007) possibly because
data were collected from mid-elevation as well as high elevation shola fragments. Bootstrap
species richness estimators also indicated that 12 overstory species and 13 understory species
were not represented in our dataset. My estimates for species diversity (H') are within the range
60
of reported values for shola fragments (Jose et al. 1994, Menon 2001) and other tropical montane
forest fragments (Oosterhoorn and Kappelle 2000).
Ordination of species data indicates that no edge-interior gradients were apparent in shola-
(TMF) fragments for understory and overstory data. However, a separation of plots based on
elevation was noticed among overstory plots. Plots at elevations lower than 1500 m (BRT) were
distinct from those at elevations greater than 1700 m in species ordination space. This is also
supported by patterns observed in the diameter distribution of overstory species along the edge-
interior gradient (Figure 3-3). Data from mid-elevation shola fragments (GU, JG, KR) reveal
that, the smallest trees (5-10cm dbh) were often absent across all distances leading to a distinct
separation between overstory and understory strata. Such differentiation is typically not
associated with tropical montane forest (Shola) fragments in the Western Ghats and is indeed
distinct from patterns observed in high elevation fragments in this study. Although this is not
surprising, numerous studies have characterized lower elevation forest fragments as shola or
tropical montane forests both within BRT (Ganeshaiah et al. 1997, Murali et al. 1998) and in
other medium elevation ecosystems within the Western Ghats (Sudhakara, 2001). While medium
elevation forests in BRT have previously been recognized as distinct from lowland evergreen
forests (Murali et al. 1998), this study clearly demonstrated their separation from ‘true’ high
elevation montane forest in the Western Ghats.
Soil nutrient availability was well represented in the overstory ordination cloud and
assumed greater significance for high elevation fragments (Figure 3-4). Globally, tropical
montane forests are found on nutritionally poor soils (Hamilton et al. 1995). In my study too, soil
nutrient availability became significant only in high elevation shola fragments. The separation of
fragments in general and mid-elevation fragments from high elevation fragments in particular is
61
also expressed in fragment complementarity scores and the Chao Abundance-based Sørensen
similarity indices for our overstory dataset. While across the dataset fragment complementarity is
high, separation between mid-elevation fragments and high elevation fragments is often
complete (C = 1.00).
For understory vegetation, fragment size was a greater source of variation than elevational
differences. While Axis 1 was correlated with relative humidity, soil moisture and organic
carbon, Axis 2 was weakly correlated with air temperature and negatively with soil temperature
and available phosphorous. Thus, in larger fragments microenvironment and soil nutrient
availability determined the distribution of species whereas in smaller fragments only air
temperature appeared to determine variation in species distributions. Light transmittance was
very weakly correlated with Axis 1. Since shola fragments are deeply shaded, this clearly
demonstrates that understory species are well adapted to low light conditions as seen in other
tropical montane forest fragments (Williams-Linera 2003)
Conclusion
This study, to my knowledge, represents the first attempt to investigate edge-interior
gradients in species richness and distributions across multiple tropical montane forest (shola)
fragments in the Western Ghats. Previous studies have investigated similar gradients in a single
fragment (Jose et al. 1996). Patterns observed in species density and structure are indicative of a
stable, mature shola-grassland edge. Across fragments, differences in overstory species
distributions were seen based on elevation while differences in understory species distributions
were based on fragment size. In both instances species distributions were not determined by edge
effects. Such distinct elevation based separation of fragments highlights the need to sub-classify
tropical montane forests as seen in other tropical montane forest ecosystems (Bruijzneel and
Hamilton 2000). From a management standpoint, the high degree of complementarity between
62
fragments exhibited by the ecosystem emphasizes the need to conserve the present extent of the
landscape and restore disturbed fragments where possible. My designed preponderance to
smaller fragments limits extensive characterization of fragments using fragment metrics (e.g.
core area, edge/area ratio, edge length). Although such an analysis would add invaluable
knowledge to the existing study, investigative efforts are probably best directed to filling the
current void of process based studies in the ecosystem. As stable, natural fragments in a hostile
grassland matrix, the study of shola fragments offers tremendous opportunities to study
fragmentation patterns and processes.
63
Table 3-1. Location and vegetation characteristics of tropical montane forest (shola) fragments studied across the Western Ghats, southern India.
Fragment (shola name)
Fragment code
Study site
Fragment area (ha)
Elevation (m)
Species richness Species diversity (H')
Overstory basal area (m2 ha-1)
Understory density (indiv ha-1) OS* US† OS* US†
Karadishonebetta KR BRT 1311 9 15 2.09 2.40 15.47 2120 Gummanebetta GU BRT 1.73 1433 11 21 2.00 2.52 44.47 2288 Jodigere JG BRT 0.20 1550 3 9 0.62 1.32 27.06 4450 Suicide point SP ENP 1.24 1810 17 20 2.62 2.83 38.59 1900 Chinnanaimudi CA ENP 1.66 2000 26 24 2.96 1.60 78.51 6956 Kolathan KS ENP 1.58 2205 16 21 2.41 2.37 69.06 5070 Pusinambara PS ENP 1.75 2222 17 14 2.64 1.99 59.43 7900 Mukkal mile MM ENP 13.42 2066 15 22 2.40 2.64 67.96 7574 Pampadum PP PNSP 132.00 2005 23 22 2.70 2.12 70.10 17300 *OS: overstory species (≥ 5 cm dbh) †US: understory species (≥ 50 cm in height but ≤ 5 cm dbh)
64
Table 3-2. Species richness, dominance and diversity estimates in the shola-grassland ecosystem mosaic in the Western Ghats, southern India.
Area
Elevation range (m)
Basal Area (m2ha-1)
Density (indiv ha-1)
Species richness
Species diversity (H')
Reference
Overstory ENP+BRT+PSNP 52.29 1711 77 3.96 This study Eravikulam 48.00‡ 1884 53 4.86* Jose et al. 1994 Mannavan shola 1550-1750 372.44 16710 55 4.71 Sudhakara 2001 Pampadum 1750-1950 114.94 21894 44 4.53 Sudhakara 2001 Eravikulam 2100 59.37 2.50 Swarupanandan et al. 2001 Silent Valley 1950-2300 - - - 2.73 Nair and Baburaj 2001 Eravikulam 2050-2200 - - - 2.71 Nair and Baburaj 2001 Kukkal RF 58.47 451 45 12.1† Davidar et al. 2007 Understory BRT+ENP+PSNP 1500-2200 - 6173 83 3.57 This study Eravikulam - - 65 4.86* Jose et al. 1994 †Fisher’s Alpha, *H' calculated for cumulatively for overstory and understory strata, ‡Overstory ≥ 10cm gbh, Overstory ≥ 5cm dbh for our study, ≥10 cm for other studies, understory ≥ 50cm for our study. BRT, ENP, PSNP represent BRT Wildlife Sanctuary, Eravikulam National Park and Pampadum shola National Park respectively.
65
Table 3-3. Similarity and complementarity between tropical montane forest (shola) fragments in the Western Ghats, southern India. The Chao- Sørsensen abundance based index varies from 0 (completely dissimilar) to 1 (completely similar) whereas Complementarity varies from 0 (all species shared) to 1 (no shared species).
Fragments Chao-Sørsensen index Complementarity GU JG SP CA KS PS MM PP† GU JG SP CA KS PS MM PP† Overstory KR 0.78 0.11 0 0.06 0 0.05 0 0.10 0.66 0.90 1.00 0.97 1.00 0.95 1.00 0.96 GU 0.59 0 0.05 0.06 0 0.03 0.19 0.72 1.00 0.97 0.96 1.00 0.96 0.93 JG 0 0 0.08 0 0 0.04 1.00 1.00 0.94 1.00 1.00 0.96 SP 0.82 0.23 0.28 0.22 0.29 0.61 0.90 0.86 0.85 0.88 CA 0.08 0.44 0.20 0.41 0.97 0.86 0.92 0.80 KS 0.60 0.56 0.20 0.73 0.81 0.91 PS 0.42 0.19 0.76 0.91 MM 0.13 0.94 Understory KR 0.76 0.63 0.31 0.05 0.54 0.03 0.03 0.03 0.61 0.66 0.90 0.94 0.79 0.97 0.97 0.94 GU 0.95 0.10 0.05 0.25 0.15 0.07 0.02 0.69 0.94 0.92 0.87 0.95 0.97 0.95 JG 0.05 0.06 0.20 0.12 0 0.18 0.96 0.90 0.85 0.92 1.00 0.89 SP 0.76 0.12 0.11 0.41 0.50 0.70 0.86 0.92 0.80 0.80 CA 0.10 0.11 0.06 0.18 0.88 0.84 0.85 0.78 KS 0.55 0.19 0.17 0.75 0.87 0.90 PS 0.56 0.23 0.65 0.83 MM 0.49 0.84 Karadishonebetta (KR), Gummanebetta (GU) and Jodigere (JG) sholas were located with BRT Wildlife Sanctuary (BRT). Suicide point (SP), Chinnanaimudi (CA), Kolathan (KS), Pusinambara (PS) and Mukkal mile sholas were located within Eravikulam National Park (ENP). Pampadum shola (PP) is located within Pampadum shola national park (PSNP). †PP represents the single large fragment (~132 ha) as most other fragments were smaller (0.2-13 ha).
66
Table 3-4. Pearson correlation coefficients (r2) for environmental variables with axis scores of
NMS ordination. Correlation coefficients accounting for more than 30 % of the variation in individual factors are highlighted in bold.
Environmental variable Axis1 Axis 2 Axis 3 Overstory Soil temperature -0.077 0.150 0.284 Air temperature -0.124 0.231 0.245 Relative humidity 0.158 0.536 0.295 Soil moisture P 0.020 0.560 0.232 Organic carbon -0.019 0.458 0.341 Total nitrogen -0.013 0.467 0.317 Available phosphorous -0.001 0.350 0.194 Available potassium 0.207 0.414 0.160 Understory Light transmittance -0.268 0.138 -0.124 Soil temperature -0.082 -0.320 -0.211 Air temperature -0.005 0.348 -0.246 Relative humidity 0.447 -0.219 0.137 Soil moisture 0.452 -0.252 0.001 Organic carbon 0.328 -0.285 -0.164 Total nitrogen 0.270 -0.374 -0.010 Available phosphorous 0.155 -0.383 -0.086 Available potassium 0.244 -0.291 0.009
67
Ric
hnes
s
0
20
40
60
80
100
120Sobs (Mao Tau) BootstrapSingletonsDoubletons
Cumulative number of plots
0 10 20 30 40
Ric
hnes
s
0
20
40
60
80
100
Figure 3-1. Species accumulation curves (Sobs), species richness estimates (Bootstrap) and rare species (singletons and doubletons) for (a) overstory and (b) understory species from tropical montane forest (shola) fragments in the Western Ghats, southern India. Species data were pooled across 4 distances within the 9 fragments (n=35).
a
b
68
Distance from edge (m)
2.5 12.5 22.5 32.5
No.
of i
ndiv
idua
ls
0
10
20
30
40
5050-99100-149 150-199 200-249 > 250
Figure 3-2. Mean understory density (number of individuals/plot) along the edge-interior gradient in tropical montane forest (shola) fragments (n = 9). Classes represent maximum height (Hmax) of understory individuals recorded in centimeters.
69
No.
of i
ndiv
idua
ls (n
)0
2
4
6
8
Distance to edge (m)2.5 12.5 22.5
No.
of
indi
vidu
als (
n)
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
No.
of i
ndiv
idua
ls (n
)
0
4
8
12
Distance to edge (m)2.5 12.5 22.5 32.5
No.
of i
ndiv
idua
ls (n
)
0
4
8
12
16
No.
of i
ndiv
idua
ls (n
)
0
2
4
6
No.
of i
ndiv
idua
ls (n
)
0
4
8
12
16
5-10 cm10-20 cm20-40 cm>40 cm
No.
of i
ndiv
idua
ls (n
)
0
6
12
18
No.
of i
ndiv
idua
ls (n
)
0
6
12
18
Distance to edge (m)2.5 12.5 22.5 32.5
No.
of i
ndiv
idua
ls (n
)
0
4
8
12
GU
JG
SP
CA PP
MM
KS
PS
KR
Figure 3-3. Edge-interior gradients in diameter distribution of overstory tropical montane forest (shola) species across fragments. Individuals are for each plot averaged across transects within a patch. Karadishonebetta (KR), Gummanebetta (GU) and Jodigere (JG) sholas were located within BRT Wildlife Sanctuary. Suicide point shola (SP), Chinnanaimudi (CA), Pusinambara (PS), Kolathan (KS) and Mukkal mile sholas are located within Eravikulam National Park while Pampadum shola (PP) was located within Pampadum shola National Park
70
Figure 3-4. Non-metric multi-dimensional scaling (NMS) ordination of overstory data. KR, GU
and JG represent mid elevation shola (TMF) fragments while the other fragments represent high elevation shola fragments. Points represent plots showing their distribution across fragments. Arrows indicate the strength and direction of correlations with environmental variables (moist: soil moisture, nit: total nitrogen, orgc: organic carbon, relhum: relative humidity) on the two axes. Letters denote fragment names as in Table 3-1 and numbers indicate distance from edge classes.
71
Figure 3-5. Non-metric multi-dimensional scaling (NMS) ordination of understory data. Points
represent plots showing their distribution across fragments. Arrows indicate the strength and direction of correlations with environmental variables (moist: soil moisture, relhum: relative humidity)on the two axes. Letters denote fragment names as in Table 3-1 and numbers indicate distance from edge classes.
72
CHAPTER 4 EFFECT OF TOPOGRAPHY ON THE DISTRIBUTION OF SHOLA FRAGMENTS: A
PREDICTIVE MODELING APPROACH
Introduction
The influence of topography on vegetation across biomes is well known. For example,
increasing elevation brings with it a change in plant communities. For a given latitude, a change
in elevation implies a shift ‘upward’ in species composition and assemblages progressively to
alpine or boreal communities. Although the influence of other topographical variables on
processes has been known for a while (e.g. aspect, Merz 1953), research continues to be directed
at it. Recently, topographical variables have been used to determine species richness (Hofer et al.
2008), regional biodiversity patterns (Coblentz and Ritters 2004), forest canopy health (Stone et
al. 2008), species distributions (Warren 2008) and gradients in exotic species (Qian 2008). As
Coblentz and Riitters (2004) identify, this is driven in part by the global availability of high
resolution elevation data (e.g., GLCF, www.landcover.org) coupled with the availability of high
speed computing platforms. Within the Western Ghats, the potential of geographic information
systems (GIS) and remotely sensed data in tropical ecosystems (such as the Western Ghats
hotspot) from the perspective of a landscape was recognized early (Menon and Bawa 1997).
Studies have since utilized these tools to highlight gaps in the network of conservation areas
(Ramesh et al. 1997), quantify threats (Barve et al. 2005) and prioritize areas for conservation
(Das et al. 2006)
Topographical based analyses are especially insightful in mountainous regions,
presumably due to the extent of topography related heterogeneity. In a study on tropical montane
forests (TMF) in Ethiopia, topography related variation in soil physicochemical characteristics
and proximity to water accounted for differences in forest communities (Aerts et al. 2006).
Similarly, a study in an Indonesian TMF revealed that topography induced variation in relative
73
humidity and wind velocity influenced species presence (Sri-Ngernyuang et al. 2003). While it is
well known that the abiotic environment (e.g., temperature, soil moisture) of a plant is an
important determinant in it’s ability to establish, grow and reproduce (Gurevitch et al. 2002), an
extensive determination of the abiotic environmental variables at the scale of a landscape (or
larger) may not be feasible. However, the relative ease of determination of topographical
variables using extensive, large scale elevation data makes these variables a suitable surrogate
for abiotic environmental conditions.
In tropical montane forests (TMF) topography can result in the creation and maintenance
of abrupt boundaries or edges. For instance, the Cordilleras in the Dominican Republic exhibit a
discrete TMF-pine forest ecotone maintained by a combination of fire and frost (Martin et al.
2007). Similarly, abrupt boundaries have been observed for tree islands in the Ecuadorian Andes
(Coblentz and Keating 2008). In the Western Ghats, tropical montane forests (or sholas) consist
of insular forest fragments in a matrix of grasslands. Carbon isotope analysis of peat samples
from the ecosystem mosaic reveal a cyclical shift in dominant vegetation type (forest or
grassland) corresponding to glacial cycles (Sukumar et al. 1993). The edge is thus assumed to be
natural in origin although its persistence may be due to a combination of natural (fire, frost,
edaphic) and anthropogenic (fire) factors. Shola fragments are characterized by high species
diversity (α-diversity) and endemism (Jose et al. 1994, Nair and Menon 2001). In addition,
fragments have been shown to have a high degree of complementarity (high diversity among
fragments, Chapter 3). Since patterns in plant distributions are non-random (Grieg-Smith 1979),
predicting their occurrence is dependent on quantifying ecological variables controlling
incidence (Franklin 1995). In this study, the influence of topographical variables on determining
the presence of shola fragments was tested. A multiple logistic regression approach was used to
74
predict presence or absence of insular fragments in the matrix of grasslands in two disjunct study
sites in southern India using elevation data.
Since the southwest monsoon contributes the bulk of the rainfall in both our study sites, I
hypothesize that the presence of fragments will be negatively correlated with eastern aspects
(hereafter Eastness). Further, I expect weak or no correlations with northern aspects (hereafter
Northness), since N-S aspects are not expected to be significant at the low latitudes of our study
sites. Preliminary field observations indicate that shola fragments are more likely to be found on
steeper, wetter slopes. Accordingly, I hypothesize that presence of shola fragments will be
positively correlated with wetness index (Moore et al. 1993, Gessler et al. 1995) and curvature of
the slope (McNab 1989)
At a regional scale, I expect differences between study sites at a macro scale (as defined by
Delcourt and Delcourt 1998). Although topographical variables are assumed to be acting
similarly on the same ecosystem mosaic, differences in site history may result in differences in
the response to topographical variables (Bader and Ruijten 2008).
Methods
Study Area
Two distinct areas were selected for the purpose of our study. Eravikulam National Park
(ENP) is located in the Anaimalai Hills within the state of Kerala (Fig 4-1). The park consists of
a base plateau at an elevation of 2000m surrounded by peaks with a maximum elevation of
2695m at Anaimudi. The soil had been classified as Vertisols and described as Arachaen igneous
in origin and consisting of granites and gneisses. Soils are sandy clay, have moderate depth (30-
100cm) and are acidic (ph 4.1-5.3). The mean maximum temperature recorded within ENP is
16.6 °C while the mean minimum temperature is 6°C. ENP soils had been classified as Alfisols
consist of granites and gneisses (NATMO 2009) are sandy clay, have moderate depth (30-100
75
cm) and are acidic (ph 4.1-5.3). ENP receives rainfall approximately 5200 mm of rainfall
annually from both the southwest and the northeast monsoon with the former contributing as
much as 85 % of the annual rainfall with a brief dry period (January to April). Although
contiguous rainforest formations can also be found at lower elevations, vegetation in the park is
predominantly shola-grassland ecosystem mosaic consisting of rolling grasslands interspersed
with dense, insular shola fragments. Additionally, exposed rock formations and cliffs and lower
elevation forests occupy 5.9% and 8.5% of the total area of ENP (Menon 2001).
The second study area is the Nilgiris (NIL) which is located in the state of Tamilnadu and
includes the Mukurthi National Park in the Nilgiri Biosphere Reserve. Details on the second
study site can be found in Zarri et al. (2008). Together our two study sites represent the two
largest contiguous extents of the high elevation shola-grassland ecosystem mosaic.
Imagery Pre-processing and Data Preparation
Elevation was derived by digitizing topographical maps at a scale of 1:50,000 scale (58
F/4, 58 F/3, Survey of India) at 20 m intervals for Eravikulam National Park (ArcGIS 9.2) to
generate a digital elevation model (DEM). Remotely sensed ETM+ imagery (acquisition date:
January 14, 2001) was used to identify and digitize shola fragments in Eravikulam National Park.
Images were pan-sharpened and a PCA low-pass filter was used to enable clearer delineation of
shola fragments from grasslands (ERDAS Imagine 9.2). Similarly a DEM (at 30 m contour
intervals) and a landscape map were generated for the Nilgiri Hills (see Zarri et al. 2008 for
details). In both these areas, these contour intervals represent the finest scale of elevation
information available. In addition to elevation, all other topographic covariate layers were
developed using DEMs (Table 4-1). A grid of random points was generated for Eravikulam
(1341 points) and the Nilgiris (2188 points). Shola points were identified by overlaying the
points with digitized shola fragments and declaring all other points as non-shola. In order to
76
further distinguish shola fragments from non-shola forest types, all points below 1700m were
masked and excluded from further data analysis. Estimates of topographic covariates were
extracted for these points (ENP: 401 shola, 940 non-shola; NIL: 666 shola, 1522 non-shola).
The DEM(s) were used to derive the following topographical variables: slope, aspect
(eastness and northness), curvature of the slope and wetness index (Table 4-1). The wetness
index (also termed Compound Topographical Index) combines topographical variables with
resource variables (moisture) while the other variables represent direct topographical variables.
Curvature of the slope was calculated using ArcGIS 9.2. A positive curvature indicates that the
surface is downwardly convex at that cell while a negative curvature indicates that the surface is
upwardly concave at that cell. A value of zero indicates that the surface is flat.
Predictive Modeling
In order to quantify topography dependent variables that influence the distribution of
sholas in the ecosystem mosaic 3 models were generated. For the first model, the data for the two
areas were combined and a dummy variable was introduced to test for differences between sites.
Since these two areas represent disjunct ecosystems (distance between ENP-NIL ~ 120km), the
dummy variable would test for regional differences. A multiple logistic regression model was
used for deriving estimates of the coefficients for topographical parameters and area under the
receiver operating characteristic curve (AuC). The AuC value is the proportion of randomly
drawn pairs of shola and non-shola points that are correctly classified using the multiple logistic
regression model. A value of 0.50 therefore indicates a random process (i.e. there is an equal
probability of the point being shola or non-shola) whereas a value ≥ 0.50 indicates a non-random
process. Seventy-five percent of the data were randomly selected for training the model and 25%
were retained as a cross-validation set. This process was repeated to 1000 iterations and a dataset
containing parameter coefficients and AuC estimates for each iteration was created. Two-tailed t-
77
tests were used to test for a) AuC estimates to be significantly different from random (AuC >
0.50) and b) significance of parameter coefficients (βi > 0). We used R programming language
throughout (R, 2008).
Two additional models were developed using our data subsets (ENP and NIL) to cross
validate each other. This was done to obtain conservative estimates of our parameter coefficients
and increase the robustness of our model. For the second model, the ENP data subset was used to
train the model and this was used to predict the presence of shola fragments in NIL. Similarly,
for the third model, the NIL data subset was used to train the model and this was used to predict
the presence of shola fragments in ENP. As with Model 1 (the combined model) a multiple
logistic regression model was used and repeated to 1000 iterations on the second and third model
to obtain topographical coefficients and AuC estimates for NIL and ENP respectively. AuC
values from Models 2 and 3 were then compared using a t-test (p < 0.05). When the two models
are compared, the data subset for the model with a significantly larger AuC was declared as a
better predictor of the presence of shola fragments based on topographical variables. Two-tailed
t-tests were used to compare the strength and direction of topographical parameter coefficients
across Models 2 and 3.
Results
Combined Model
For the combined model, topographical variables significantly predict whether the
vegetation type is shola or non-shola. The distribution of shola fragments was significantly
different from random as indicated by the area under the receiver operating curve (AuC = 0.714,
SE 0.0005, p < 0.0001). All topographical variables were highly significant (p < 0.0001, Table 4-
2). Coefficients for northness were positive while those of eastness were negative as related to
the presence of shola fragments (p < 0.0001). Coefficients for wetness index, slope and elevation
78
were positive and negative for curvature of the slope (for all p < 0.0001). Significant differences
were also seen between the study sites (p < 0.0001) with the Nilgiris (NIL) data subset predicting
the presence of shola fragments in Eravikulam (ENP) better than the use of the ENP dataset to
predict shola fragments in the Nilgiris (p < 0.0001).
Data Subsets
The Nilgiris (NIL) data subset was able to distinguish shola fragments from non-shola in
Eravikulam better (AuCNIL = 0.624) than vice versa (AuCERV= 0.709). While all parameter
coefficients from both models differed in magnitude, they did not differ in direction (Table 4-2,
Figure 4-2 to Figure 4-7). For both models, aspect (as quantified in eastness and northness)
strongly determined the presence of shola fragments and to a lesser extent wetness index (βCTI)
and curvature of the slope (βCUR).
Discussion
Results from our multiple logistic regression models clearly indicate the influence of
topography induced variation on the location of tropical montane forest (shola) fragments
relative to the surrounding non-shola matrix. AuC estimates provide statistical validation for the
strong non-random pattern of shola fragments in the mosaic. The effect of elevation on
determining the presence of shola fragments was unexpected in all our models a priori. This
could be due to our exclusion of points below 1700 m. However, at elevations below 1700 m,
separation of shola vegetation from evergreen forest may require a more intensive landcover
classification model which was beyond the scope of this investigation. Shola fragments below
1700 m (mid-elevation sholas) have also been shown to differ compositionally and structurally
from high elevation shola fragments and might warrant a formal partitioning into forest sub-
types (Chapter 3). Digitized shola fragments in ENP and field observations indicate the presence
of fragments (as large as 2 ha) at 2500 m and absence only from the highest peaks, possibly due
79
to the extremely steep terrain on those peaks (e.g. Anaimudi). The topographical limitation on
elevation in the Western Ghats prohibits the formation of true tree line as seen in other tropical
montane forest ecosystems (Troll 1973, Bader and Ruijten 2008). This might explain the lack of
strength associated with elevation in our models.
Shola fragments were also likely to be found on northern and western aspects. The
proportion of rainfall attributed to the southwestern monsoon in Eravikulam (ENP) and the
Nilgiris (NIL) might explain the preponderance of shola fragments on western slopes which
support my hypothesis. The dependence of shola fragments on the availability of soil moisture is
also reflected in the direction of the response to the wetness index (Compound Topographical
Index, CTI). The response to wetness index is significant as topographical variables that combine
a measure of topography with resource variable (such as CTI) have been shown to predict
patterns in vegetation better than those that measure topography alone (Franklin et al. 2000). The
significant bias of shola fragments to north-facing slopes was unexpected and contrary to our
hypothesis. Aspect induced variations in solar exposure along a north-south gradient are unlikely
to be significant at the latitude associated with the study sites (10°05’-11°32’). During the
months corresponding to the southwestern monsoon (May-October) significant differences in
wind exposure are likely as southern aspects (exposed to monsoonal winds) are likely to have
higher exposure to wind than northern aspects. This might limit the establishment of tree cover
(shola fragments). However evidence for wind-exposure related restriction of shola fragments in
this study is tenuous at best. Bader & Ruijten (2008) suggest that in addition to wind, diurnal
variation in radiation received might cause aspect-related differences in the tree line in Andean
tropical montane forest They surmise that more radiation is received earlier in the day (i.e.
eastern aspects) which ‘causes photoinhibition in maladapted plants’. Lack of data on diurnal
80
variation in intensity of radiation from our study however precludes me from making such
conclusions. However, this might be worth investigating in the system.
Curvature of the slope (or Terrain shape index) is a measure of the shape of the landform.
For shola fragments, presence of shola fragments was significantly linked to negative
coefficients which are indicative of concave slopes (see Equation 4-1). This is in keeping with
field observations of sholas and provides proof for my hypothesis. Similarly, the coefficient for
slopes was positive, i.e. sholas were present on steeper, more inaccessible terrain.
Comparison of Data Subsets
The use of the Nilgiris (NIL) dataset to predict the presence of shola fragments in
Eravikulam (ENP) produced a better model than vice versa. However, the coefficient of
topographical parameters was stronger for the NIL model. This can be explained by differences
in the history of the study sites as Bader and Ruitjen (2007) suggested from their study in
Andean tropical montane forests. Among sites in this study, ENP represents the better protected
of the two sites having received formal protection from colonial times and was used as a game
reserve by colonial settlers. Further, unlike tropical montane forests in other parts of the Western
Ghats (including the Nilgiris) colonial management of the grasslands (and indeed even the
sholas) in ENP did not involve extensive afforestation with exotic plantations. While the
influence of topographical variables on shola presence/absence in ENP (and consequently using
that data subset to predict presence/absence of fragments) is more indicative of processes with
limited human influence, both sites are not completely free of human influence.
Implications for the Shola-grassland Edge
Historically, theories on the persistence of the shola-grassland edge in the ecosystem
mosaic include frost, fire and edaphic limitation. While shola and grassland soils have been
shown to exhibit little variation in soil physicochemical properties and soil depth (Table 1-1,
81
Chapter 2), topographical variables such as wetness index can also provide insights into patterns
of fire and frost. Areas with a high wetness index also indicate wetter areas that are likely to be
precluded by fire, protecting tropical montane fragments. Studies have shown that patterns in
water accumulation and frost occurrence are often similar (Blennow 1998). Areas with a high
wetness index (depressions etc.) are likely to have higher occurrence of frosts reducing the
potential for shola fragment persistence. This is in contrast to the shola-grassland ecosystem
mosaic where sholas are located within depressions. Thus, the positive correlation of wetness
index with the presence of shola fragments indicates that, in addition to the strong linkage to
hydrological regulation, among fire and frost, fire rather than frost might be responsible for the
maintenance of the edge. Patterns in exposure to wind too might be responsible for the edge.
Conclusion
A common problem associated with predictive modeling is the lack of discrete boundaries
between habitat patches and high variation within patches (Hofer et al. 2008). The sharp edge
between shola fragments and grasslands in the shola-grassland ecosystem mosaic can therefore
provide significant insights without the limitations associated with other systems. Tropical
montane forest (shola) fragments in the Western Ghats have a high variation among fragments.
Further edge-interior studies reveal that fragments show little variation along edge-interior
gradients (or within fragments, see Chapter 3). This study provides unambiguous support for the
maintenance of the edge by fire rather than frost in addition to strong linkages to hydrological
regulation. Predictive models are useful tools for restoration efforts and at larger scales can
provide inputs into existing restoration efforts. The use of topographical variables to predict the
presence of shola fragments represents a subset of static, environmental sorting variables
influencing the distribution of shola fragments. The inclusion of dynamic variables such as fire
82
history and regime and biotic interactions (e.g. topography induced dispersal limitation) might
provide further insights and enhance our understanding of the shola-grassland ecosystem mosaic.
83
Table 4-1. Topographical variables used to predict presence of shola fragments in the Western Ghats, southern India.
Model parameter Code Description Elevation ELE Slope SLO Northness(NOR) NOR sin (Aspect) Eastness (EAS) EAS cos (Aspect) Curvature of the slope CUR Wetness index CTI ln (As / (tan(β)) †‡ Location Code LOC Boolean (ENP = 1, NIL = 0) As: Flow accumulation grid, β: Slope (radians), † Moore et al. 1993., ‡Gessler et al. 1995 Table 4-2. Topographical parameter coefficients and AuC estimates for predictive models.
Coefficients represent mean values from 1000 runs. Training datasets for the models were: Pooled data from the Eravikulam (ENP) and Nilgiris (NIL) data subsets for the Combined model; ENP for the NIL model and NIL for the ENP model. All parameter coefficients were significant at p < 0.0001.
Model output Model type (number)
Combined NIL ENP (1) (2) (3)
AuC 0.71 0.62 0.70 Parameter coefficients Intercept (INT) -8.2730 -9.6763 -6.7185 Elevation (ELE) 0.0032 0.0038 0.0024 Slope (SLO) 0.0234 0.0459 -0.0023 Northness (NOR) 0.3895 0.5703 0.3327 Eastness (EAS) -0.1324 -0.2673 -0.1080 Curvature of the slope (CUR) -0.0416 -0.0634 -0.0365 Wetness index (CTI) 0.0582 0.0215 0.0914 Location (LOC) -0.2786 - -
84
Figure 4-1. Location of study sites in the Western Ghats. Two study sites were selected for the
study are (A), Nilgiris (NIL) and (B), Eravikulam National Park (ENP), For inset maps, 1cm equals 5.5 kilometers
85
Figure 4-2. Histogram of Intercept coefficients (INTC) for Eravikulam National Park (ERV) and Nilgiris (NIL) data subsets.
Superimposed curves represent normal curves.
86
Figure 4-3. Histogram of Elevation coefficients (DEM) for Eravikulam National Park (ERV) and Nilgiris (NIL) data subsets.
Superimposed curves represent normal curves.
87
Figure 4-4. Histogram of Slope coefficients for Eravikulam National Park (ERV) and Nilgiris (NIL) data subsets. Superimposed curves represent normal curves.
88
Figure 4-5. Histogram of Northness coefficients for Eravikulam National Park (ERV) and Nilgiris (NIL) data subsets. Superimposed curves represent normal curves.
89
Figure 4-6. Histogram of Eastness coefficients for Eravikulam National Park (ERV) and Nilgiris (NIL) data subsets. Superimposed curves represent normal curves.
90
Figure 4-7. Histogram of Curvature of the Slope coefficients (CURV) for Eravikulam National Park (ERV) and Nilgiris (NIL) data subsets. Superimposed curves represent normal curves.
91
CHAPTER 5 SUMMARY AND CONCLUSION
Tropical montane forest (sholas) in the Western Ghats in southern India consist of insular
forest fragments in a matrix of grasslands. The boundary between the sholas and the grasslands is
thought to be natural in origin and is relatively stable. As with other tropical montane forest
(TMF) systems, the sholas consist of dense stands of stunted trees (~15m). Physiognomically,
branch architecture is twisted or gnarled and branches are often covered in epiphytic growth.
Given the absence of anthropogenically induced fragmentation in the mosaic, the primary aim of
this study was to investigate edge effects in shola fragments. Nine fragments across three study
sites were selected in the Western Ghats to represent a range of elevations (1400m-2200m) and
fragment sizes (0.2ha-132ha). Data on microenvironment (soil and air temperature, relative
humidity, light transmittance. soil moisture), soil macronutrient (organic carbon, total nitrogen,
available phosphorous, potassium calcium and magnesium) and micronutrient (boron, copper,
molybdenum, manganese, iron and zinc) variables were collected. Data were also collected on
overstory (richness and dominance) and understory (richness and vertical structure) vegetation.
Data were collected along grassland exterior-edge-fragment interior gradients for
microenvironment variables and edge-interior gradients for species variables.
Edge effect studies typically consider the effect of the response of a variable (for e.g., soil
moisture) as a function of distance to the nearest edge. This is driven in part by the nature of
conventional edge effect studies which quantify the effect of the creation of an edge between
original (often forested) habitat and converted habitat (e.g., agriculture, road). However, studies
have shown that the response to an edge can be better explained when distance to multiple edges
is considered. The inclusion of distance to multiple edges in edge models becomes increasingly
important in highly fragmented ecosystems and/or small fragments. Since the study was
92
designed to study edge effects in small fragments distance-to-edge in three additional directions
(2nd nearest, 3rd nearest and 4th nearest edge) mutually orthogonal to our original distance were
obtained. The use of the distance to nearest edge alone found no evidence of edge-interior
gradients similar to patterns observed in other small fragments. However the inclusion of
distance to multiple edges significantly improved the predictability of our models in small
fragments. Highly significant patterns for light transmittance (p = 0.0072) and soil moisture (p =
0.0002) and weaker patterns for air temperature (p = 0.032) were observed as a function of
distance to multiple edges. For larger fragments though distance to the nearest edge was
sufficient to observe gradients in relative humidity and potassium. With the exception of soil
temperature no exterior-edge-shola interior gradients in microenvironment and soil nutrient
variables were observed. Further, nutritional differences between grassland and shola fragment
soils did not differ statistically as has been observed in other studies.
No gradients in overstory dominance and vertical complexity in understory as a function of
distance to edge were observed. Nonmetric multidimensional scaling (NMS) ordination
techniques revealed a higher variation in species distributions between fragments than within
fragments along the edge-interior gradient. Compositionally, significant differences were
observed between mid-elevation fragments and high elevation fragments for overstory species.
Since species found in high elevation (≥ 1700 m) fragments are more representative of tropical
montane forest habitat the exclusion of fragments ≤ 1700 m from future tropical montane forest
studies is recommended possibly designating these areas as premontane habitat. However, this
requires a formal exercise to clearly delineate the two habitat subtypes.
The apparent non-random pattern of shola fragments in the ecosystem mosaic also
prompted a study on the effect of topographical variables in determining the presence (or
93
absence) of shola species. For the purpose of this study the two largest high-altitude extents of
the shola-grassland ecosystem mosaic- Eravikulam National Park (ENP) and the Nilgiris (NIL)
were used. Elevation, slope, curvature of the slope (or Terrain Shape index), aspect (expressed as
Eastness and Northness) and wetness index were selected as the topographical variables used to
predict shola presence. Three models were used- a combined model, and two models using ENP
and NIL individually as the training datasets-to test hypotheses. Although all topographical
variables quantified were highly significant (p < 0.0001), the response to aspect was unexpected.
While the preference of shola fragments for western aspects captures monsoonal related trends
(and hence soil moisture) the preference for northern aspects was contrary to expectation. Due to
the limitations of the dataset I am unable to make conclusive arguments for the existence of this
pattern. Soil moisture related variability is also captured in the response to wetness index with
fragments preferring wetter sites.
The persistence and relative stability of the shola-grassland edge continues to puzzle
ecologists and foresters. Our predictive model provides support for fire and hydrological
regulation as driving forces maintaining the edge as shola fragments preferred wetter habitats
(which are less likely to burn). Since research has shown that the occurrence of frosts mirrors
patterns in the flow of water, wetter areas are more likely to have ground frosts. This would
inhibit the establishment of shola, which is in contrast to our observations. We can conclude that
it is unlikely that frost maintains the edge. Maintenance of the edge due to edaphic differences is
unlikely since nutritionally, shola and grassland habitat do not differ significantly.
The shola-grassland ecosystem mosaic offers insights into fragmentation-related processes
in the absence of human influence. The persistence of small fragments (~ 1ha) in the mosaic also
needs to be highlighted. However, more needs to be learnt about processes driving the observed
94
patterns. In addition to the need for basic hydrological studies, data are also lacking about the
role of the shola epiphytic community in rainfall interception and nutrient cycling observed from
other tropical montane forest ecosystems (e.g., Nadkarni et al. 2000). Physiological data for
shola species are conspicuous by their absence as are experimental studies on seedling
establishment essential for ecosystem restoration efforts (however, see Sekar 2008). Information
on genetic diversity of plant species within the mosaic is scarce (Jain et al. 2000, Deshpande et
al. 2001) and studies need to address this given the patchy nature of sholas both at a mosaic level
and at a regional level.
I conclude by highlighting the dearth of process oriented studies in the ecosystem and
suggest that further research needs to be directed to it.
95
APPENDIX
96
Table A-1. Overstory species (≥ 5 cm dbh) presence matrix along an edge-interior gradient across nine fragments in the shola-grassland ecosystem mosaic in the Western Ghats, southern India. Karadishonebetta (KR), Gummanebetta (GU) and Jodigere (JG) sholas were located within BRT Wildlife Sanctuary (BRT). Suicide point shola (SP), Chinnanaimudi (CA), Pusinambara (PS), Kolathan (KS) and Mukkal mile (MM) sholas were located within Eravikulam National Park (ENP). Pampadum shola (PSNP) is represented by a single fragment. Presence across 4 distances (2.5 m, 12.5 m, 22.5 m and 32.5 m marked 1, 2, 3 and 4 respectively) is indicated by gray squares.
BRT ENP Sps. KR GU JG SP CA PS KS MM 1 2 3 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4ACSI ACSP ALCO ALSP APLA ARBL BESP CATR CETI CHMA CHRO CIMA CISP CISU DESP DEVE
ACSI: Acacia sinuata; ACSP: Actinodaphne spp.; ALCO: Allophylus cobbe; ALSP: Albizzia spp.; APLA: Apodytes laerithanniana; ARBL: Ardisia blatteri; BESP: Beilschmiedea spp.; CATR: Canthium travancorium; CETI: Celtis timorensis; CHMA: Chionanthus malabarica; CHRO: Chrysophyllum roxburghii; CIMA: Cinnamomum malabarica; CISP: Cinnamomum spp.; CISU: Cinnamomum sulphuratum; DESP: Derris spp.; DEVE: Debregeasia velutina.
97
Table A-1. Continued BRT ENP Sps. KR GU JG SP CA PS KS MM 1 2 3 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4DIOP ELSE EMBA EMRI EMSP EUSP EXRO FASP GASP GOCO HYAL ILSP ILTH ISSP IXNO IXSP LABL LASP LGSP LIDE LISP
DIOP: Dioscorea oppositifolia; ELSE: Elaeocarpus serratus; EMBA: Embelia basaal; EMRI:Embelia ribes; EMSP: Embelia spp.; EUSP: Eugenia spp.; EXRO: Excoecaria robusta; FASP: Fagraea spp.; GASP: Garcinia spp.; GOCO: Gomphandra coriaceae; HYAL: Hydnocarpus alpina; ILSP: Ilex spp.; ILTH: Ilex thwaitesii; ISSP: Isonandra spp.; IXNO: Ixora notoniana; IXSP: Ixora spp.; LABL: Lasianthus blumeanus; LASP: Lasianthus spp.; LGSP: Ligustrum spp.; LIDE: Litsea deccanensis; LISP: Litsea spp.
98
Table A-1. Continued BRT ENP Sps. KR GU JG SP CA PS KS MM 1 2 3 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4LIWI MAPE MALE MAPH MAAR MASP MESI NECA NEFI NEFO NESC NESP NEZE OLGL PEMA PHWI PISP PINE PSCO
LIWI: Litsea wightiana; MAPE: Maesa perrottetiana; MALE: Mahonia leschenaultii; MAPH: Mallotus philippensis; MAAR: Mastixia arborea; MASP: Mastixia spp.; MESI: Meliosma simplicifolia; NECA: Neolitsea cassia; NEFI: Neolitsea fischeri; NEFO: Neolitsea foliosa; NESC: Neolitsea scrobiculata; NESP: Neolitsea spp.; NEZE: Neolitsea zeylanica; OLGL: Olea glandulifera; PEMA: Persea macrantha; PHWI: Phoebe wightii; PISP: Piper spp.; PINE: Pittosporum neelegherrense; PSCO: Psychotria congesta.
99
Table A-1. Continued BRT ENP Sps. KR GU JG SP CA PS KS MM 1 2 3 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4PSNI PSSP RHNI RUCO SAFO SCRO SCWA SCCR STAL STIS SYMO SYCA SYDE SYSP TAAS TESP TRCO TUNE VALE VANE VASP
PSNI: Psychotria niligiriensis; PSSP: Psychotria spp.; RHNI: Rhododendron nilagaricum; RUCO: Rubia cordifolia; SAFO: Saprosoma foetens ssp. ceylanicum; SCRO: Schefflera rostrata; SCWA: Schefflera wallichiana; SCCR: Scolopia crenata; STAL: Strobilanthes integrifolius; STIS: Strobilanthes isophyllus; SYMO: Symplocos monatha; SYCA: Syzygium carophyllatum; SYDE: Syzygium densiflorum; SYSP: Syzygium spp.; TAAS: Tarenna asiatica; TESP: Ternstroemia spp.; TRCO: Tricalysia apiocarpa; TUNE: Turpinia nepalensis; VALE: Vaccinium leschenaultii; VANE: Vaccinium neilgherrense; VASP: Vaccinium spp.
100
Table A-1. Continued Sps. PSNP 1 2 3 4 ACSI ACSP ALCO ALSP APLA ARBL BESP CATR CETI CHMA CHRO CIMA CISP CISU DESP DEVE DIOP ELSE EMBA EMRI EMSP EUSP EXRO FASP GASP GOC HYAL ILSP ILTH
ACSI: Acacia sinuata; ACSP: Actinodaphne spp.; ALCO: Allophylus cobbe; ALSP: Albizzia spp.; APLA: Apodytes laerithanniana; ARBL: Ardisia blatteri; BESP: Beilschmiedea spp.; CATR: Canthium travancorium; CETI: Celtis timorensis; CHMA: Chionanthus malabarica; CHRO: Chrysophyllum roxburghii; CIMA:Cinnamomum malabarica; CISP: Cinnamomum spp.; CISU: Cinnamomum sulphuratum; DESP: Derris spp.; DEVE: Debregeasia velutina; DIOP: Dioscorea oppositifolia; ELSE: Elaeocarpus serratus; EMBA: Embelia basaal; EMRI: Embelia ribes; EMSP: Embelia spp.; EUSP: Eugenia spp.; EXRO: Excoecaria robusta; FASP: Fagraea spp.; GASP: Garcinia spp.; GOCO: Gomphandra coriaceae; HYAL: Hydnocarpus alpina; ILSP: Ilex spp.; ILTH: Ilex thwaitesii
101
Table A-1. Continued Sps. PSNP 1 2 3 4 ISSP IXNO IXSP LABL LASP LGSP LIDE LISP LIWI MAPE MALE MAPH MAAR MASP MESI NECA NEFI NEFO NESC NESP NEZE OLGL PEMA PHWI PISP PINE PSCO PSNI
ISSP: Isonandra spp.; IXNO: Ixora notoniana; IXSP: Ixora spp.; LABL: Lasianthus blumeanus; LASP: Lasianthus spp.; LGSP: Ligustrum spp.; LIDE: Litsea deccanensis; LISP: Litsea spp.; LIWI: Litsea wightiana; MAPE: Maesa perrottetiana; MALE: Mahonia leschenaultii; MAPH: Mallotus philippensis; MAAR: Mastixia arborea; MASP: Mastixia spp.; MESI: Meliosma simplicifolia; NECA: Neolitsea cassia; NEFI: Neolitsea fischeri; NEFO: Neolitsea foliosa; NESC: Neolitsea scrobiculata;NESP: Neolitsea spp.; NEZE: Neolitsea zeylanica; OLGL: Olea glandulifera; PEMA: Persea macrantha; PHWI: Phoebe wightii; PISP: Piper spp.; PINE: Pittosporum neelegherrense; PSCO: Psychotria congesta; PSNI: Psychotria niligiriensis.
102
Table A-1. Continued Sps. PSNP 1 2 3 4 PSSP RHNI RUCO SAFO SCRO SCWA SCCR STAL STIS SYMO SYCA SYDE SYSP TAAS TESP TRCO TUNE VALE VANE VASP
PSSP: Psychotria spp.; RHNI: Rhododendron nilagaricum; RUCO: Rubia cordifolia; SAFO: Saprosoma foetens ssp. ceylanicum; SCRO: Schefflera rostrata; SCCR: Scolopia crenata; SCWA: Schefflera wallichiana; STAL: Strobilanthes integrifolius; STIS: Strobilanthes isophyllus; SYMO: Symplocos monatha; SYCA: Syzygium carophyllatum; SYDE: Syzygium densiflorum; SYSP: Syzygium spp.; TAAS: Tarenna asiatica; TESP: Ternstroemia spp.; TRCO: Tricalysia apiocarpa; TUNE: Turpinia nepalensis; VALE: Vaccinium leschenaultii; VANE: Vaccinium neilgherrense; VASP: Vaccinium spp.
103
Table A-2. Understory species presence matrix along an edge-interior gradient across nine fragments in the shola-grassland
ecosystem mosaic in the Western Ghats, southern India. All vegetation greater than 50 cm in height but less than 5 cm dbh was defined as understory. Karadishonebetta (KR), Gummanebetta (GU) and Jodigere (JG) sholas were located within BRT Wildlife Sanctuary (BRT). Suicide point (SP), Chinnanaimudi (CA), Pusinambara (PS), Kolathan (KS) and Mukkal mile (MM) sholas were located within Eravikulam National Park (ENP). Pampadum shola (PSNP) is represented by a single fragment. Presence across 4 distances (2.5 m, 12.5 m, 22.5 m and 32.5 m marked 1, 2, 3 and 4 respectively) is indicated by gray squares.
Sps. BRT ENP KR GU JG SP CA PS KS MM 1 2 3 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4AGAD ALVE ARBL ARSP ASSP ATMO BESP CASP CATR CAPA CETI CHMA CHOD CIMA CISP CISU
AGAD: Ageratina adenophora; ALVE: Alstonia venenata; ARBL: Ardisia blatteri; ARSP: Ardisia spp.; ASSP: Asparagus spp.; ATMO: Atalantia monophylla; BESP: Beilschmiedea spp.; CASP: Calamus spp.; CATR: Canthium travancorium; CAPA: Cassine paniculata; CETI: Celtis timorensis; CHMA: Chionanthus malabarica; CHOD: Chromolaena odorata; CIMA: Cinnamomum malabarica; CISP: Cinnamomum spp.; CISU: Cinnamomum sulphuratum
104
Table A-2. Continued Sps. BRT ENP KR GU JG SP CA PS KS MM 1 2 3 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4CIIN COAR COSP DESP DIOP ELKO ELSE EMBA EMRI EUSP EXRO FASP GAGU GASP GLSP HEDY ISSP IXSP LACA LABL LASP LIOL
CIIN: Cipadessa indica; COAR: Coffea arabica; COSP: Coleus spp.; DESP: Derris spp.; DIOP: Dioscorea oppositifolia; ELKO: Elaegnus kologa; ELSE: Elaeocarpus serratus; EMBA: Embelia basaal; EMRI: Embelia ribes; EUSP: Eugenia spp.; EXRO: Excoecaria robusta; FASP: Fagraea spp.; GAGU: Garcinia gummi-gutta; GASP: Garcinia spp.; GLSP: Glochidion spp.; HEDY: Hedyotis spp.; ISSP: Isonandra spp.; IXSP: Ixora spp.; LACA: Lantana camara; LABL: Lasianthus blumeanus; LASP: Lasianthus spp.; LIOL: Ligustrum oleaceae
105
Table A-2. Continued Sps. BRT ENP KR GU JG SP CA PS KS MM 1 2 3 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4LGSP LIDE LISP LIWI MAIN MAPE MALE MAPH MAAR MESI MEUM NEFI NESC NEZE NONI PEMA PHWI PISP POSP POAC PSCO PSNI
LGSP: Ligustrum spp.; LIDE: Litsea deccanensis; LISP: Litsea spp.; LIWI: Litsea wightiana; MAIN: Maesa indica; MAPE: Maesa perrottetiana; MALE: Mahonia leschenaultii; MAPH: Mallotus philippensis; MAAR: Mastixia arborea; MESI: Meliosma simplicifolia; MEUM: Memecylon umbellatum; NEFI: Neolitsea fischeri; NESC: Neolitsea scrobiculata; NEZE: Neolitsea zeylanica; NONI: Nothapodytes nimmoniana; PEMA: Persea macrantha; PHWI: Phoebe wightii; PISP: Piper spp.; POSP: Polygonum spp.; POAC: Polyscias acuminata; PSCO: Psychotria congesta; PSNI: Psychotria niligiriensis
106
Table A-2. Continued Sps. BRT ENP KR GU JG SP CA PS KS MM 1 2 3 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4PSSP RUSP SCCA SCCR SORO STGU STAL STIS STSP SYCO SYMO SYCU SYSP TASP TESP TOAS TRAP unBA unFE unSH VASP VIPU WETH
PSSP: Psychotria spp.; RUSP: Rubus spp.; SCCA: Schefflera capitata; SCCR: Scolopia crenata; SORO: Solanum robustum; STGU: Sterculia guttata; STAL: Strobilanthes integrifolius; STIS: Strobilanthes isophyllus; STSP: Strobilanthes spp.; SYCO: Symplocos cochinchinensis; SYMO: Symplocos monatha; SYCU: Syzygium cuminii; SYSP: Syzygium spp.; TASP: Tarenna spp.
107
Table A-2. Continued Sps. BRT ENP KR GU JG SP CA PS KS MM 1 2 3 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4TESP TOAS TRAP unBA unFE unSH VASP VIPU WETH
TESP: Ternstroemia spp.; TOAS: Toddalia asiatica; TRAP: Tricalysia apiocarpa; unBA: Unidentified bamboo; unFE: Unidentified Fern; unSH: Unidentified shrub; VASP: Vaccinium spp.; VIPU: Viburnum punctatum; WETH: Wendlandia thyrsoidea
108
Table A-2. Continued Sps PSNP 1 2 3 4 AGAD ALVE ARBL ARSP ASSP ATMO BESP CASP CATR CAPA CETI CHMA CHOD CIMA CISP CISU CIIN COAR COSP DESP DIOP ELKO ELSE EMBA EMRI EUSP EXRO FASP GAGU GASP GLSP
AGAD: Ageratina adenophora; ALVE: Alstonia venenata; ARBL: Ardisia blatteri; ARSP: Ardisia spp.; ASSP: Asparagus spp.; ATMO: Atalantia monophylla; BESP: Beilschmiedea spp.; CASP: Calamus spp.; CATR: Canthium travancorium; CAPA: Cassine paniculata; CETI: Celtis timorensis; CHMA: Chionanthus malabarica; CHOD: Chromolaena odorata; CIMA: Cinnamomum malabarica; CISP: Cinnamomum spp.; CISU: Cinnamomum sulphuratum; CIIN: Cipadessa indica; COAR: Coffea arabica; COSP: Coleus spp.; DESP: Derris spp.; DIOP: Dioscorea oppositifolia; ELKO: Elaegnus kologa; ELSE: Elaeocarpus serratus; EMBA: Embelia basaal; EMRI: Embelia ribes; EUSP: Eugenia spp.; EXRO: Excoecaria robusta; FASP: Fagraea spp.; GAGU: Garcinia gummi-gutta; GASP: Garcinia spp.; GLSP: Glochidion spp.
109
Table A-2. Continued Sps PSNP 1 2 3 4 HEDY ISSP IXSP LACA LABL LASP LIOL LGSP LIDE LISP LIWI MAIN MAPE MALE MAPH MAAR MESI MEUM NEFI NESC NEZE NONI PEMA PHWI PISP POSP POAC PSCO PSNI
HEDY: Hedyotis spp.; ISSP: Isonandra spp.; IXSP: Ixora spp.; LACA: Lantana camara; LABL: Lasianthus blumeanus; LASP: Lasianthus spp.; LIOL: Ligustrum oleaceae; LGSP: Ligustrum spp.; LIDE: Litsea deccanensis; LISP: Litsea spp.; LIWI: Litsea wightiana; MAIN: Maesa indica; MAPE: Maesa perrottetiana; MALE: Mahonia leschenaultii; MAPH: Mallotus philippensis; MAAR: Mastixia arborea; MESI: Meliosma simplicifolia; MEUM: Memecylon umbellatum; NEFI: Neolitsea fischeri; NESC: Neolitsea scrobiculata; NEZE: Neolitsea zeylanica; NONI: Nothapodytes nimmoniana; PEMA: Persea macrantha; PHWI: Phoebe wightii; PISP: Piper spp.; POSP: Polygonum spp.; POAC: Polyscias acuminata; PSCO: Psychotria congesta; PSNI: Psychotria niligiriensis
110
Table A-2. Continued Sps PSNP 1 2 3 4 PSSP RUSP SCCA SCCR SORO STGU STAL STIS STSP SYCO SYMO SYCU SYSP TASP TESP TOAS TRAP unBA unFE unSH VASP VIPU WETH
PSSP: Psychotria spp.; RUSP: Rubus spp.; SCCA: Schefflera capitata; SCCR: Scolopia crenata; SORO: Solanum robustum; STGU: Sterculia guttata; STAL: Strobilanthes integrifolius.; STIS: Strobilanthes isophyllus; STSP: Strobilanthes spp.; SYCO: Symplocos cochinchinensis; SYMO: Symplocos monatha; SYCU: Syzygium cuminii; SYSP: Syzygium spp.; TASP: Tarenna spp.; TESP: Ternstroemia spp.; TOAS: Toddalia asiatica; TRAP: Tricalysia apiocarpa; unBA: Unidentified bamboo; unFE: Unidentified Fern; unSH: Unidentified shrub; VASP: Vaccinium spp.; VIPU: Viburnum punctatum; WETH: Wendlandia thyrsoidea
111
LIST OF REFERENCES
Abraham, T.K. 2001. Meliolaceous and soil fungi and lichen flora of the sholas of Munnar and Waynad. Pages 118-135 in Nair, K.K.N., Khanduri, S.K., Balasubramanayam, K., editors. Shola forests of Kerala: Environment and Biodiversity. Kerala Forest Research Institute, Peechi, India.
Aerts, R., Overtveld, K.V., Haile, M., Hermy, M., Deckers, J., Muys, B. 2006. Species composition and diversity of small Afromontane forest fragments in northern Ethiopia. Plant Ecology 187: 127-142.
Agrawal, S.C., Madan, U.S., Chinnamani, S., Rege, N.D. 1961. Ecological studies in the Nilgiris. Indian Forester 87: 376-389.
Alembath, M., Rice, C. 2008. Nilgiritragus hylocrius. In: IUCN 2008. 2008 IUCN Red List of Threatened Species. www.iucnredlist.org. Downloaded on 30 January 2009.
Allen S. E. 1974. Chemical analysis of ecological materials. Blackwell Scientific, Oxford.
Anderson, S.H.,Mann, K. Shugart,H.H. 1977. The effect of transmission-lines corridors on bird populations. American Midland Naturalist 97: 216-221.
Bader, M.Y., Ruijten, J.J.A. 2008. A topography-based model of forest cover at the alpine tree line in the tropical Andes. Journal of Biogeography 35: 711-723.
Balasubramanian, K., Kumar, K.K. 1999. The riddle of the shola. Evergreen 42: 1-5.
Barve, N., Kiran, M.C., Vanaraj, G., Aravind, N.A., Rao, D., Shaanker, R.U., Ganeshaiah, K.N., Poulsen, J.G. 2005. Measuring and mapping threats to a wildlife sanctuary in southern India. Conservation Biology 19: 122-130.
Beisner, B.E., Haydon, D.T., Cuddington, K. 2003. Alternative stable states in ecology. Frontiers in Ecology and the Environment 1: 376-382.
Benzing, D.H. 1998. Vulnerabilities of tropical forests to climate change: the significance of resident epiphytes. Climatic Change 39: 519-540.
Blennow, K. 1998. Modelling minimum air temperature in partially and clear felled forests. Agricultural and Forest Meteorology 91: 223-235.
Bor, N.L. 1938. The vegetation of the Nilgiris Indian Forester 64: 600-609.
Brooks, T.M., Mittermeier, R.A., Mittermeier, C.G., da Fonseca, G.A.B., Rylands, A.B., Konstant, W.R., Flick, P., Pilgrim, J., Oldfield, S., Magin, G., Hilton-Taylor, C. 2002. Habitat loss and extinction in the hotspots of biodiversity. Conservation Biology 16: 909-923.
112
Bruijnzeel, L.A., Hamilton, L.S. 2000. Decision time for cloud forests: IHP Humid tropics programme series No. 13. UNESCO, Paris, France.
Bowman, D.M.J.S., Woinarski, J.C.Z. 1994. Biogeography of Australian monsoon rainforest mammals: Implications for the conservation of rainforest mammals. Pacific Conservation Biology 1: 98-106.
Bubb, P. May, I. Miles, L., Sayer, J. 2004. Cloud Forest Agenda. UNEP-WCMC, Cambridge UL. Online at http://www.unep-wcmc.org/resources/publications/UNEP_WCMC_bio_series/20.htm
Camargo, J.L.C., Kapos, V. 1995. Complex edge effects on soil moisture and microclimate in central Amazonian forest. Journal of Tropical Ecology 11: 205-221.
Carlson, A., Hartman, G. 2001. Tropical forest fragmentation and nest predation- an experimental study in an Eastern Arc montane forest, Tanzania. Biodiversity and Conservation 10: 1077-1085.
Clements, F.E. 1936. Nature and structure of climax. Journal of Ecology 24: 252-284.
Chandrasekhara, U.M. Muraleedharan, P.K., Sibichan, V.2001. Disturbed sholas of Kerala and strategies for its conservation and management. Pages 395-437 in Nair, K.K.N., Khanduri, S.K., Balasubramanayam, K., editors. Shola forests of Kerala: Environment and Biodiversity. Kerala Forest Research Institute, Peechi, India.
Chandrasekharan, C. 1962. Forest types of Kerala. Indian Forester 88: 837-847.
Chao, A. 2005. Species richness estimation. Pages 7909-7916 in Balakrishnan, N., Read, C. B., Vidakovic, B., editors. Encyclopedia of Statistical Sciences. Wiley, New York, USA.
Chao, A. Chazdon, R.L., Colwell, R.K., Shen, T.J. 2005. A new statistical approach for assessing similarity of species composition with incidence and abundance data. Ecology Letters 8: 148-159.
Coblentz, D., Keating, P.L 2008. Topographic controls on the distribution of tree islands in the high Andes of south-western Ecuador. Journal of Biogeography 35: 2026-2038.
Coblentz, D.D., Riitters, K.H. 2004. Topographic controls on the regional scale biodiversity of the south-western USA. Journal of Biogeography. 31: 1125-1138.
Cochrane, M.A. 2001. Synergistic interactions between habitat fragmentation and fire in evergreen tropical forests. Conservation Biology 15: 1515-1521.
Colwell, R.K. 2006. EstimateS: Statistical estimation of species richness and shared species from samples. Version 8. Persistent URL <purl.oclc.org/estimates>.
Colwell, R. K., Coddington, J. A. 1994. Estimating terrestrial biodiversity through extrapolation. Philosophical Transactions of the Royal Society (Series B) 345: 101-118.
113
Colwell, R.K., Mao, C.X., Chang, J. 2004. Interpolating, extrapolating and comparing incidence-based species accumulation curves. Ecology 85: 2717-2727.
Curran, L.M. Caniago, I., Paoli, G.D., Astianti, D., Kusneti, M., Leighton, M., Nirarita, C.E., Haeruman, H. 1999. Impact of El-Niño and logging on canopy recruitment in Borneo. Science 286: 2148-2188.
Daniels 2006. The Nilgiri Tahr: An endemic south Indian mountain goat. Macmillan India, New Delhi, India.
Daniels, R.J.R.; Easa, P.S.; Alembath, M. 2008. Distribution and status of the endangered Nilgiri tahr. Current Science 94: 10-11.
Das, A., Krishnaswamy, J., Bawa, K.S., Kiran, M.C., Srinivas, V., Kumar, N.S., Karanth, K.U. 2006. Prioritization of conservation areas in the Western Ghats, India. Biological Conservation 133: 16-31.
Davidar, E.R.C. 1971. A note on the status of the Nilgiri Tahr in the grass hills in the Anamalais. Journal of the Bombay Natural History Society 68: 347-354.
Davidar, P., Mohandass, D., Vijayan, S.L. 2007. Floristic inventory of woody plants in a tropical montane forest in the Palni hills of the Western Ghats, India. Tropical Ecology 48: 15-25.
del Castillo, R.F., Rios, M.A.P. 2008. Changes in seed rain during secondary succession in a tropical montane forest region in Oaxaca, Mexico. Journal of Tropical Ecology 28: 433-444.
Delcourt, H.R. Delcourt, P.A. 1988. Quaternary landscape ecology: Relevance scales in space and time. Landscape Ecology 2: 32-44.
Deshpande, A.U., Apte, G.S., Bahulikar, R.A., Lagu, M.D., Kulkarni, B.G., Suresh, H.S., Singh, N.P., Rao, M.K.V., Gupta, V.S., Pant, A., Ranjekar, P.K. 2001. Genetic diversity across plant populations of three montane plant species from the Western Ghats, India revealed by intersimple sequence repeats. Molecular Ecology 10: 2397-2408.
Didham, R.K., Lawton, J.H. 1999. Edge structure determines the magnitude of changes in microclimate and vegetation structure in tropical forest fragments. Biotropica 31: 17-30.
Dinesh, K.P., Radhakrishnan, C., Bhatta, G. 2008. A new species of Nyctibatrachus Boulenger (Amphibia: Anura: Nyctibatrachidae) from the surroundings of Bhadra Wildlife Sacntuary, Western Ghats, India. Zootaxa 1914: 45-56.
FAO (Food and Agricultural Organization). 1993. Forest resources assessment 1990: tropical countries. FAO Forestry Paper No. 112. Rome
Fernández, C., Acosta, F.J., Abellá, G., López, F., Díaz, M. 2002. Complex edge effect fields as additive processes in patches of ecological systems. Ecological Modelling 149: 273-273.
114
Fletcher, R.J. 2005. Multiple edge effects and their implications in fragmented landscapes. Journal of Animal Ecology 74: 342-352.
Fonseca, C.R., Joner,F.Two sided edge effect studies and the restoration of endangered ecosystems. Restoration Ecology 15: 613-619.
Foster, P. 2001. The potential negative impacts of global climate change on tropical montane cloud forests. Earth Science Reviews 55: 73-106.
Franklin, J. 1995. Predictive vegetation mapping, geographic modelling of biospatial patterns in relation to environmental gradients. Progress in Physical Geography 19:474–499.
Franklin, J., McCullough, P., Gray, C. 2000. Terrain variables used for predictive mapping of vegetation communities in southern California. Pages 331-353 in: Wilson, J., Gallant, J., editors. Terrain analyses: principles and applications, John Wiley and sons, New York, USA.
Ganeshaiah, K.N., Umashaanker, R.U., Bawa, K.S. 1997. Diversity of species assemblages of islands: Predictions and their test using tree species composition of shola fragments. Current Science 73: 188-194.
Gardner, F.P., Pearce, R.B., Mitchell, R.L. 1984. Physiology of crop plants. Iowa State University Press, Ames, IA, USA.
Gascon, C.G., Williamson, B., da Fonseca, G.A.B. 2000. Receding forest edges and vanishing reserves. Science 288: 1356-1358.
Gates, J.E., Gysel, L.W. 1978. Avian nest dispersion and fledgling success in field-forest ecotones. Ecology 59: 871-883.
George, M., Mohandas, K., Brijesh, C.M. 2001. Insect fauna of the sholas of Idukki and Wayanad districts. Pages 317-340 in Nair, K.K.N., Khanduri, S.K., Balasubramanayam, K., editors. Shola forests of Kerala: Environment and Biodiversity. Kerala Forest Research Institute, Peechi, India.
Gessler, P.E., Moore, I.D. McKenzie, N.J., Ryan, P.J. 1995. Soil-landscape modeling and spatial prediction of soil attributes. International Journal of GIS. 9: 421-432.
Ghosh, D.S. 2001. Fish diversity and associated species in the sholas of Munnar, Idukki District. Pages 353-364 in Nair, K.K.N., Khanduri, S.K., Balasubramanayam, K., editors. Shola forests of Kerala: Environment and Biodiversity. Kerala Forest Research Institute, Peechi, India.
Gioda, A. Maley, J., Guasp, R.E., Baladón, A.A. 1995. Some low elevation fog forests of dry environments: application to African palaeoenvironments. Pages 156-164 in Hamilton, L.S., Juvik, J.O., Scatena, F.N., editors. Ecological studies 110: Tropical Montane Cloud Forests. Springer-Verlag, New York, USA.
115
Greig-Smith, P. 1979. Pattern in vegetation. Journal of Ecology 67:755–779.
Gupta, R.K. 1962. Studies in some shola forests on the Palni hills near Kodaikanal. Indian Forester 88: 848-853.
Gupta, R.K., Shankarnarayan, K.A. 1962. Ecological status of the grasslands in south India. Tropical Ecology 3: 75-78.
Gurevitch, J., Scheiner, S.M., Fox G.A. 2002. The Ecology of plants Sinauer Associates, Sunderland, Massachusetts, USA.
Hamilton, L.S., Juvik, J.O., Scatena, F.N. 1995. Ecological studies 110: Tropical Montane Cloud Forests. Springer-Verlag, New York, USA.
Harper, K.A., MacDonald, S.E. 2002. Structure and composition of edges next to regenerating clear-cuts in mixed-wood boreal forest. Journal of Vegetation Science 13: 535-546.
Harper, K.A., Macdonald, S.E., Burton, P.J., Chen, J.Q., Brosofske, K.D., Saunders, S.C., Euskirchen, E.S., Roberts, D., Jaiteh, M.S., Esseen, P.A.. 2005. Edge influence on forest structure and composition in forest fragments. Conservation Biology 19: 768-782.
Hofer, G., Wagner, H.H., Herzog, F., Edwards, P.J. 2008. Effects of topographic variability on the scaling of plant species richness in gradient dominant landscapes. Ecography 31: 131-139.
Ickes, K., Williamson, B.G. 2000. Edge effects and ecological processes: are they on the same spatial scale? Trends in Ecology and Evolution 15: 373.
Jain, A., Pandit, M.K., Elahi, S., Jain, A., Bhaskar, A., Kumar, V. 2000. Reproductive behaviour and genetic variability in geographically isolated populations of Rhododendron arboreum (Ericaceae). Current Science 79: 1377-1381.
Jackson, M.L. 1962. Soil chemical analysis. Prentice-Hall, Englewood Cliffs, NJ, USA.
Janzen, D.H. 1983. No park is an island- Increase in interference from outside as park size decreases. Oikos 41: 402-410.
Jeeva, V., Ramakrishnan, P.S. 1997. Studies on impact of plantation forestry in Nilgiri Hills of the Ghats on soil quality and nutrient cycling. Tropical Ecology 38: 215-235.
Johnston, V.R. 1947. Breeding birds of the forest edge in Illinois. Condor 2: 45-53.
Jose, S. Sreepathy, A. Mohan Kumar, B., Venugopal, V.K. 1994. Structural, floristic and edaphic attributes of the shola-grassland forests of Eravikulam in peninsular India. Forest Ecology and Management 65: 279-291.
116
Jose, S. Gillespie, A.R., Goerge, S.J., Kumar, M.K. 1996. Vegetation response along edge-interior gradients in a high altitude tropical forest in peninsular India. Forest Ecology and Management 87: 51-62.
Kapos, V. 1989. Effects of isolation on the water status of forest patches in the Brazilian Amazon. Journal of Tropical Ecology 5: 173-185.
Karunakaran, P.V., Rawat, G.S., Uniyal, V.K. 1998. Ecology and conservation of the grasslands of Eravikulam National Park and the Western Ghats. Wildlife Institute of India, Dehradun, India.
Kruskal, J.B. 1964. Nonmetric multi-dimensional scaling: a numerical method. Psychometrika 29: 115-129.
Krishnaswamy, J., Kiran, M.C., Ganeshaiah, K.N. 2005. Tree model based eco-climatic vegetation classification and fuzzy mapping in diverse tropical deciduous ecosystems using multi-date NDVI. International Journal of Remote Sensing 25: 1185-1205.
López-Barrera, F. Newton, A., Manson, R. 2005.Edge effects in a tropical montane forest mosaic: experimental tests of post-dispersal acorn removal. Ecological Research. 20: 31-40.
Laurance, W.F. 1991. Edge effects in tropical forest fragments: application of a model for the design of nature reserves. Biological Conservation 57: 205-219.
Laurance, W.F. 2000. Reply from W.F. Laurance. Trends in Ecology and Evolution 15: 373.
Laurance, W.F. 2008. Theory meets reality: How habitat fragmentation research has transcended island biogeographic theory. Biological Conservation 141: 1731-1744.
Laurance,W.F., Ferreira,L.V. Rankin-De Merona, J.M., Laurance, S.G., Hutchings, R.W. Lovejoy, T.E. 1998. Effects of forest fragmentation on recruitment patterns in Amazonian tree communities. Conservation Biology 12: 460-464.
Laurance, W.F., Lovejoy, T.E., Vasconcelos, H.L., Bruna, E.M., Didham, R.D., Stouffer, P.C., Gascon, C., Bierregaard, R.O., Laurance, S.G., Sampaio, E.G. 2002. Ecosystem decay of Amazonian forest fragments: a 22-Year investigation. Conservation Biology 16: 605-618.
Laurance, W.F., Yensen, 1991. Predicting the impacts of edge effects in fragmented habitats. Biological Conservation 69: 23-32.
Leopold,A. 1993.Game management. Scribner Sons, New York, USA.
López-Barrera, F., Manson, R.H. González-Espinosa, M., Newton, A.C. 2006. Effects of the type of montane forest edge on oak seedling establishment along forest-edge-exterior gradients. Forest Ecology and Management 225: 234-244.
117
López-Barerra, F. Newton, A., Manson, R. 2005.Edge effects in a tropical montane forest mosaic: experimental tests of post-dispersal acorn removal. Ecological Research. 20: 31-40.
Macarthur, R.H., Wilson, E.O. 1963. An equilibrium theory of insular zoogeography. Evolution 17: 373-387.
Macarthur, R.H., Wilson, E.O. 1967. The Theory of Island Biogeography. Princeton University Press, Princeton, New Jersey, USA.
Magurran, A.E. 1988. Ecological diversity and its measurement. Croom Helm, London, UK.
Malcolm, J.R. 1994. Edge effects in central Amazonian forest fragments. Ecology 75: 2438-2445.
Mancke, R.G.,Gavin, T.A.2000. Breeding bird density in woodlots: Effects of depth and buildings at the edges. Ecological Applications 10: 598-611.
Mao, C.X., Colwell, R.K., Chang, J, 2005. Estimating species accumulation curves using mixtures. Biometrics 61: 433-441.
Martin, P.H., Sherman, R.E., Fahey, T.J. 2007. Tropical montane forest ecotones: climate gradients, natural disturbance, and vegetation zonation in the Cordillera Central, Dominican Republic. Journal of Biogeography 34: 1792-1806.
Mather, P.M. 1976. Computational methods of multivariate analysis in physical geography. J Wiley and sons, London, UK.
May, R.M. 1977. Thresholds and breakpoints in ecosystems with multiplicity of states. Nature 269: 471-477.
McCune, B., Mefford, M.J. 2006. PC-ORD. Multivariate analysis of ecological data. Version 5. MjM Software, Glenden Beach, Oregon, USA.
McNab, W.H. 1989. Terrain Shape Index: Quantifying effect of minor landforms on tree height. Forest Sciance 35: 91-104.
Meher-Homji, V.M. 1965. Ecological status of the montane grasslands of the south Indian hills: A phytogeographic reassessment. Indian Forester 91: 210-215.
Meher-Homji, V.M. 1967. Phytogeography of the south Indian hill stations. Bulletin of the Torrey Botanical Club 94: 230-242.
Menon, A.R.R. 2001.Mapping and analysis of the shola-grassland vegetation of Eravikulam, Idukki district. Pages 95-115. in Nair, K.K.N., Khanduri, S.K., Balasubramanayam, K., editors. Shola forests of Kerala: Environment and Biodiversity. Kerala Forest Research Institute, Peechi, India.
118
Menon, S., Bawa, K.S. 1997. Applications of geographic information systems, remote-sensing and a ladscape ecology approach to biodiversity conservation in the Western Ghats. Current Science 73: 134-145.
Merz, R.W. 1953. Site-index estimates made easy. Journal of Forestry 51: 749-750.
Mesquita, R.C.G., Delamônica, P., Laurance, W.F. 1999. Effect of surrounding vegetation on edge-related mortality in Amazonian forest fragments. Biological Conservation 91: 129-134.
Mishra, C., Johnsingh, A.J.T. 1998. Population and conservation status of the Nilgiri tahr Hemitragus hylocrius in Anamalai hill, south India. Biological Conservation 86: 199-206.
Moore, I.D., Gessler, P.E., Nielsen, G.A., and Petersen, G.A. 1993. Terrain attributes: estimation methods and scale effects. Pages 189-214 in: Jakeman, A.J., Beck, M.B. McAleer, M., editors. Modeling Change in Environmental Systems. Wiley, London, UK.
Murali, K.S., Shetty, R.S., Ganeshaiah, K.N., Shaanker, R.U. 1998. Does forest type classification reflect spatial dynamics of vegetation? An analysis using GIS techniques. Current Science 75: 220-227.
Myers, N., Mittermeier, R.A.,Mittermeier, C.G., da Fonseca, G.A.B., Kent, J. Biodiversity hotpots for conservation priorities. Nature 403: 853-858.
Nadkarni, N.M. Wheelwright, N.T. 2000. Monteverde: Ecology and Conservation of a Tropical Montane Cloud Forest. Oxford University Press, New York, USA.
Nadkarni, N.M., Lawton, R.O., Clark, K.L., Matelson, T.J., Schaefer, D. 2000. Ecosystem Ecology and Forest Dynamics. Pages 303-350 in Nadkarni, N.M. Wheelwright, N.T., editors. Monteverde: Ecology and Conservation of a Tropical Montane Cloud Forest. Oxford University Press, New York, USA.
Nair, K.K.N., Menon, A.R.R. 2001. Endemic arborescent flora of the sholas of Kerala and its population and regeneration status. Pages 209-236 in Nair, K.K.N., Khanduri, S.K., Balasubramanayam, K., editors. Shola forests of Kerala: Environment and Biodiversity. Kerala Forest Research Institute, Peechi, India.
Nair, K.K.N., Khanduri, S.K., Balasubramanayam, K. 2001. Shola forests of Kerala: Environment and Biodiversity. Kerala Forest Research Institute, Peechi, India.
Nameer, P.O. 2001. Richness of avifauna in the sholas of Munnar, Idukki District. Pages 365-389 in Nair, K.K.N., Khanduri, S.K., Balasubramanayam, K., editors. Shola forests of Kerala: Environment and Biodiversity. Kerala Forest Research Institute, Peechi, India.
Nandakumar, V., Rajendran, P., Narendra Babu, K. 2001. Characterization of soils in the sholas of Idukki and Wayanad districts Pages 25-70 in Nair, K.K.N., Khanduri, S.K., Balasubramanayam, K., editors. Shola forests of Kerala: Environment and Biodiversity. Kerala Forest Research Institute, Peechi, India.
119
NATMO.2009. Soil map of India (National Atlas and Thematic Mapping Organization). Available online at www.indiabiodiversity.org. Last accessed on May 04th 2009.
Noble, W.A. 1967. The shifting balance of grasslands, shola forests and planted trees on the upper Nilgiris, southern India. Indian Forester 93: 691-693.
Oosterhoorn, M. Kappelle, M. 2000. Vegetation structure and composition along an edge-interior gradient in a Costa Rican montane cloud forest. Forest Ecology and Management 126: 291-307.
Palanna, R.M. 1996. Eucalyptus in India. In Kashio, M. White, K. (editors) Reports submitted to the regional expert consultation on Eucalyptus. Vol II. Available online at http://www.fao.org/docrep/005/ac772e/ac772e06.htm#bm06
Pavlacky, D.C., Anderson, S.H. 2007. Does avian species richness in natural patch mosaics follow the forest fragmentation paradigm? Animal Conservation 10: 57-68.
Qian, H. 2008. A latitudinal gradient of beta diversity for exotic vascular plant species in North America. Diversity and Distributions 14: 556-560.
R Development Core Team (2008). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org.
Ramesh, B.R., Menon, S., Bawa, K.S. 1997. A vegetation based approach to biodiversity gap analysis in the Agastyamalai region, Western Ghats, India. Ambio 26: 529-536.
Ranasinghe, P.N., Dissanayake, C.B., Samarasinghe, D.V.N., Galappatti, R. The relationship between soil geochemistry and die back of montane forests in Sri Lanka: a case study. Environmental Geology 51: 1077-1088.
Ranganathan, C.R. 1938.Studies in the ecology of the shola grassland vegetation of the Nilgiri Plateau. Indian Forester 64: 523-541.
Raghupathy, R., Madhu, M. 2007. The natural grasslands of the Nilgiris: Past and present scenario. Range Management and Agroforestry 28: 271-273.
Rice, C. 1984. The behavior and ecology of the Nilgiri Tahr (Hemitragus hylocrius Ogilby, 1838). Dissertation. Texas A & M University, College Station, Texas, USA.
Rice, C. 1988. Habitat, population dynamics and conservation of the Nilgiri Tahr, Hemitragus hylocrius. Biological Conservation 44: 137-156.
Ries, L. Fletcher, R.J., Battin, J. 2004. Ecological responses to habitat edges: Mechanisms, models and variability explained. Annual Review of Ecology Evolution and Systematics 35: 491-522.
Ries, L., Sisk, T.D. 2004. A predictive model of edge effects. Ecology 85: 2917-2926.
120
Sahoo, D.C., Sharda, V.N., Jayakumar, M., Tripathi, K.P., Padmanabhan, M.V., Raghunath, B., Chandran, B. 2006. Hydrology of small watersheds in high hills of Nilgiris. Indian Journal of Soil Conservation 34: 97-101.
Sankaran, K.V., Balasundaram, M. 2001. Soil microflora of the sholas of Eravikulam National Park, Idukki district. Pages 151-178 in Nair, K.K.N., Khanduri, S.K., Balasubramanayam, K., editors. Shola forests of Kerala: Environment and Biodiversity. Kerala Forest Research Institute, Peechi, India.
Schaller, G.B. 1970. Mountain monarchs: Wild sheep and goats of the Himalayas. The University of Chicago Press, Chicago, USA.
Sekar, T. 2008. Observations on survival and growth of different Shola species under a Shola afforestation programme in Nilgiris District, Tamil Nadu. Indian Forester 134: 451-457.
Seshan, S. 2006. Notes from the edge. Available online at : http://www.gbsanctuary.org/articles/Notes_from_the_Edge.pdf
Shanker, K. 2001. The role of competition and habitat in structuring small mammal communities in a tropical montane forest ecosystem in southern India. Journal of Zoology 253: 15-24.
Sizer, N., Tanner, E.V.J. 1999. Responses of woody plant seedlings to edge formation in forest fragments in central Amazonia. Biological Conservation 91: 135-142.
Smith, E.P., van Belle, G. 1984. Nonparametric estimation of species richness. Biometrics 40: 119-129.
Somanathan, H., Borges, R.M. 2000. Influence of exploitation on population structure, spatial distribution and reproductive success if dioecious species in fragmented cloud forest in India. Biological Conservation 80: 9-15.
Soltanpour, P.M., Schwab, A.P. 1977. A new soil test for simultaneous extraction of macro and micro nutrient in alkaline soils. Communications in Soil Science Plant Analysis 8: 195-207.
Sugden, A.M. 1981. Aspects of ecology of vascular epiphytes in two Colombian cloud forests: II. Habitat preferences of Bromeliaceae in the Serrania de Macuira. Selbyana 5: 264-273.
Suresh, H.S., Sukumar, R. 1999. Phytogeographical affinities of Flora of Nilgiri Biospehere Reserve. Rheedea 9: 1-21.
Sukumar, R. Suresh, H.S., Ramesh, R. 1995. Climate change and its impact on tropical montane ecosystems in southern India. Journal of Biogeography 22: 533-536.
Sri-Ngernyuang, K., Kanzaki, M., Mizuno, T., Noguchi, H., Teejuntuk, S., Sungpalee, C., Hara, M., Yamakura, T., Sahunalu, .P, Dhanmanonda, P., Bunyavejchewin, S. 2003. Habitat fragmentation of Lauraceae species in a tropical lower montane forest in northern Thailand. Ecological Reseach 18:1—14.
121
Srivastava, R.K. 2000. Seed germination tests of pioneer shola species in Nilgiris. Annals of Forestry 8: 293-294.
Srivastava, R.K. 2001. Biotic pressure and entry of exotics in shola grassland ecosystem of upper Palnis. Indian Journal of Forestry 24: 324-327.
Still, C.J., Foster, P.N. Schneider, S.H. 1998. Simulating the effects of climate change on tropical montane forests. Nature 398: 608-610.
Stone, C., Kathuria, A., Carney, C., Hunter, J. 2008. Forest canopy health and stand structure associated with bell miners (Manorina melanophrys) on the central cpast of New South Wales. Australian Forestry 71: 294-302.
Strayer, D.L., Power, M.E., Fagan, W.F., Pickett, S.T.A., Belnap, J. 2003. A classification of ecologi al boundaries Bioscience 53: 723-729.
Sudhakara, K. 2001. Inventory and computerized herbarium of higher plants in the sholas of Munnar, Idukki district. Pages 179-208 in Nair, K.K.N., Khanduri, S.K., Balasubramanayam, K., editors. Shola forests of Kerala: Environment and Biodiversity. Kerala Forest Research Institute, Peechi, India.
Svenning, J-C. 2001. Environmental heterogeneity, recruitment limitation and the mesoscale distribution of palms in a tropical montane forest (Maquipucuna, Ecuador). Journal of Tropical Ecology 17: 97-113.
Swarupanandan, K., Sashidharan, N., Chacko, K.C., Basha, S.C. 2001. Pages 237-258 in Nair, K.K.N., Khanduri, S.K., Balasubramanayam, K., editors. Shola forests of Kerala: Environment and Biodiversity. Kerala Forest Research Institute, Peechi, India.
Thomas, S.M. Palmer, M.W. 2007. The montane grasslands of the Western Ghats, India: Community ecology and conservation. Community Ecology 8: 67-73.
Thomas, T.P., Sankar, S. 2001. The role of sholas in maintaining watercourses in the High Ranges of Kerala. Pages 71-115 in Nair, K.K.N., Khanduri, S.K., Balasubramanayam, K., editors. Shola forests of Kerala: Environment and Biodiversity. Kerala Forest Research Institute, Peechi, India.
Swarupanandan, K., Sasidharan, N., Chacko, K.C., Basha, S.C. 2001. Floristic and ecological studies on the sholas of Idukki district. Pages 259-286 in Nair, K.K.N., Khanduri, S.K., Balasubramanayam, K., editors. Shola forests of Kerala: Environment and Biodiversity. Kerala Forest Research Institute, Peechi, India.
Tabarelli, M., Mantovani, W., Peres, C.A. 1999. Effects of habitat fragmentation on plant guild structure in the montane Atlantic forest of southeastern Brazil. Biological Conservation 91: 119-127.
Troll, C. 1973. The upper timberlines in different climatic zones. Arctic and Alpine Research 5: A3-A18.
122
Turner, I.M., Corlett, R.T. 1996. The conservation value of small, isolated fragments of lowland tropical rain forest. Trends in Ecology and Evolution 11: 330-333.
Warren, R.J. 2008. Mechanisms driving understory herb distributions across slope aspects: as derived from landscape position. Plant Ecology 198: 297-308.
Williams-Linera, G. 1990. Vegetation structure and environmental conditions of forest edges in Panama. Journal of Ecology 78: 356-373.
Williams-Linera, G., Domínguez, Garcia-Zurita, M.E. 1998. Microenvironment and floristics of different edges in a fragmented tropical rainforest. Conservation Biology 12: 1092-1102.
Williams-Linera, G. 2002. Tree species richness, complementarity, disturbance and fragmentation in a Mexican tropical montane cloud forest. Biodiversity and Conservation 11: 1825-1843.
Williams-Linera, G. 2003. Temporal and spatial phonological variation of understory shrubs in a tropical montane forest. Biotropica 35: 28-36.
Venkatachalam, S., Kalaiselvi, T., Neelakantan and Gunasekaran, S. 2007. A comparative study on soil microflora and nutrient status of sholas and adjoining vegetation. Indian Journal of Forestry 30: 135-140.
Vishnu-Mittre, Gupta, H.P. 1968. A living fossil community in south Indian hills. Current Science 37: 671-672.
Werner, W.L. 1995. Biogeography and ecology of the upper montane rain forest of Sri Lanka (Ceylon). Pages 343-352 in Hamilton, L.S., Juvik, J.O., Scatena, F.N., editors. Ecological studies 110: Tropical Montane Cloud Forests. Springer-Verlag, New York, USA.
Yahner, R.H.1988. Changes in communities near edges. Conservation Biology 2: 333-339.
Young, K.R., Keating, P.L. 2001. Remnant forests of Volcán Cotacachi, Northern Ecuador. Arctic, Antarctic and Alpine Research 33: 165-172.
Young, K.R., Leon, B. 1995. Distribution and conservation of Peru’s montane forests: Interactions between the biota and human society. Pages 362-276 in Hamilton, L.S., Juvik, J.O., Scatena, F.N., editors. Tropical Montane Cloud Forests. Springer-Verlag, New York, USA.
Zarri, A. A., Rahmani, A.R., Singh, A., Khushwaha, S.P.S. 2008. Habitat suitability assessment for the endangered Nilgiri laughing thrush: A multiple logistic regression approach. Current Science 94: 1487-1494.
Zheng,D.,Chen, J. 2000. Edge effects in fragmented landscapes: a generic model for delineating area of edge influences (D-AEI). Ecological Modelling 132: 175-190.
123
BIOGRAPHICAL SKETCH
Milind Christopher Bunyan was born in Bangalore, India in 1979. The pursuit of an
undergraduate degree in Environmental Science at St. Joseph’s College of Arts and Science,
Bangalore, which he received in 1999, was more by chance than by design. Numerous hikes later
though, he discovered an interest in the natural world. He went on to receive a Master’s in Forest
Ecology and Management from the Forest Research Institute, Dehradun, India in 2001. After a
brief stint as a Research Assistant with Wildlife Conservation Society, India, he worked as a
Research Associate in the Ashoka Trust for Research in Ecology and the Environment (ATREE)
between 2002 and 2004. It was at ATREE that his interest in Ecology encouraged him to pursue
a doctoral degree. Under the supervision of Shibu Jose, he earned his PhD from the School of
Forest Resources and Conservation, at the University of Florida in Gainesville, Florida studying
edge effects in tropical montane forest fragments in southern India.
top related