The Utility of Protected Areas for Large
Carnivore Conservation
Josephine Arthur September 2014
A thesis submitted for the partial fulfilment of the requirements for the degree
of Master of Science and the Diploma of Imperial College London
Submitted for the MSc in Conservation Science
2
DECLARATION OF OWN WORK
I declare that this thesis, “The utility of protected areas for large carnivore
conservation,” is entirely my own work, and that where material could be construed as
the work of others, it is fully cited and referenced, and/or with appropriate
acknowledgement given.
Signature:
Name of student: Josephine Arthur
Name of Supervisor(s): Dr. Nathalie Pettorelli, ZSL
Miss Clare Duncan, ZSL
3
Table of Contents
ABSTRACT .............................................................................................................................................................................. 5
ACKNOWLEDGEMENTS ..................................................................................................................................................... 6
1. INTRODUCTION ......................................................................................................................................................... 7
1.1. PROBLEM STATEMENT .......................................................................................................................................................... 7 1.2. AIMS AND OBJECTIVES ........................................................................................................................................................ 11 1.3. HYPOTHESES ........................................................................................................................................................................ 11
2. BACKGROUND ......................................................................................................................................................... 12
2.1. LARGE CARNIVORES AND PAS: WHAT DO WE KNOW? ......................................................................................................... 12 2.2. HOW IS CLIMATE CHANGE PREDICTED TO IMPACT PA’S UTILITY FOR CONSERVATION? ................................................. 16 2.3. ASSESSING THE UTILITY OF PAS: POPULATION VIABILITY ANALYSIS (PVA) ................................................................. 17
3. METHODS ................................................................................................................................................................. 19
3.1. LARGE CARNIVORES ............................................................................................................................................................ 19 3.2. PROTECTED AREAS ............................................................................................................................................................. 20 3.3. PRIMARY PRODUCTIVITY DYNAMICS OF PAS ................................................................................................................... 21 3.4. FUTURE CLIMATE SCENARIOS ........................................................................................................................................... 22 3.5. STATISTICAL ANALYSES ...................................................................................................................................................... 23
3.5.1. Future Climate Scenarios ....................................................................................................................................... 23 3.5.2. Predicting PA-specific Carnivore Population Size from Home Range .................................................. 24 3.5.3. Assessing the Viability of Carnivore Populations in PAs ............................................................................ 25 3.5.4. Assessing the factors influencing PA’s utility for carnivore conservation........................................... 26
4. RESULTS .................................................................................................................................................................... 27
4.1. PROTECTED AREAS ............................................................................................................................................................. 27 4.2. CHANGES IN PRIMARY PRODUCTIVITY DYNAMICS .......................................................................................................... 28
4.2.1. Modelling future primary productivity ............................................................................................................. 28 4.2.2. Modelling future seasonality ................................................................................................................................ 29 4.2.3. Predicting changes in productivity dynamics under future climate scenarios ................................. 30
4.3. PREDICTED CHANGES IN HOME RANGE DYNAMICS UNDER FUTURE CLIMATE SCENARIOS ....................................... 33 4.4. PREDICTED CHANGES IN NUMBER OF VIABLE POPULATIONS UNDER FUTURE CLIMATE SCENARIOS ...................... 35 4.5. ASSESSING THE VARIABLES WHICH IMPACT PA’S UTILITY FOR LARGE CARNIVORE CONSERVATION. .................... 36
5. DISCUSSION ............................................................................................................................................................. 37
5.1. PREDICTED CHANGES IN PRIMARY PRODUCTIVITY AND HOME RANGE DYNAMICS .................................................... 37 5.2. WHAT IMPACTS PA’S UTILITY FOR LARGE CARNIVORE CONSERVATION?................................................................... 38 5.3. CONSERVATION IMPLICATIONS ......................................................................................................................................... 40 5.4. LIMITATIONS ........................................................................................................................................................................ 42
5.4.1. Primary productivity dynamics: observations and predictions .............................................................. 42 5.4.2. Home range predictions ......................................................................................................................................... 43 5.4.3. Assessing the viability of populations PAs are able to sustain................................................................. 44
5.5. CONCLUSIONS ............................................................................................................................................................................ 46
REFERENCES ...................................................................................................................................................................... 47
APPENDICES ....................................................................................................................................................................... 54
APPENDIX I. LINEAR MIXED EFFECTS MODEL USED TO PREDICT CARNIVORE HOME RANGE SIZE............................................. 54 APPENDIX II SURVIVAL RATES USED IN PVA .................................................................................................................................... 56
4
LIST OF ACRONYMS
AIC Akaike Information Criterion
CBD Convention on Biological Diversity
GCM Global Climate Model
GIMMS Global Inventory Modelling and Mapping Studies
GHG Greenhouse Gas
GLM Generalised Linear Model
HadGEM2 Hadley Centre Global Environment Model version 2
HR Home Range
iNDVI Integrated Normalised Difference Vegetation Index
IPCC Intergovernmental Panel on Climate Change
IUCN International Union for Conservation of Nature
LME Linear Mixed Model
MCP Minimum Convex Polygon
NDVI Normalised Difference Vegetation Index
PA Protected Area
PVA Population Viability Analysis
RCP Representative Concentration Pathway
UNEP United Nations Environmental Programme
WDPA World Database of Protected Areas
5
ABSTRACT
Conservation strategies often rely on protected areas (PAs) to achieve positive
conservation outcomes. However, PAs are vulnerable to the impacts of climate change
which questions their future utility for conservation. This study assessed the utility of
PAs for large carnivore conservation under current and future climate scenarios.
Utility was determined by whether or not a PA was predicted to be able to sustain viable
populations of its large carnivore species. The population size a PA was predicted to be
able to sustain was estimated using the ecological parameter of home range (HR) size.
HR was predicted using a model, which used PA-specific primary productivity dynamics
and species’ traits, to predict PA-specific HR size for a given carnivore species. PA-
specific primary productivity dynamics under future climate scenarios were modelled
using projected climatic data and then input into this model.
Results showed that the majority of PAs were unable to sustain viable populations of
their large carnivores under current conditions and that this would not change under
future climate scenarios. PA size was found to be the most significant determinant of its
utility, with larger PAs having significantly higher utility. In addition, a latitudinal
gradient of PA utility was identified. PAs in areas of higher latitudes were less likely to
be capable of sustaining viable populations than PAs in lower latitudes.
Therefore, PAs have limited utility for large carnivore conservation as a sole measure to
ensure their persistence under a changing climate, particularly in regions of high
latitude. The future of large carnivore conservation should be in developing novel
approaches to utilising a multiuse landscape which includes PAs and a variety of other
land uses to secure their future.
Word count: 14,887
6
ACKNOWLEDGEMENTS
Thanks to both my supervisors for their hard work helping me complete this project. To
Nathalie Pettorelli for giving me the opportunity to challenge myself and for providing
helpful comments on previous drafts. To Clare Duncan for all her help and patience with
data processing and analysis, providing code for NDVI extraction and smoothing, and
allowing me to use her model of carnivore home range, without all of which, this project
would not have been possible.
Thanks to my fellow ConSciers, especially those who also spent their summer at ZSL or
glued to desks, for all their friendship and support throughout this process. Most of all,
thanks to Ben Preece without whom this year would not have been possible.
7
1. INTRODUCTION
1.1. Problem Statement
Global biodiversity is under threat and there is mounting evidence that biodiversity loss
is altering processes key to the productivity and sustainability of the Earth’s ecosystems
(Hooper et al., 2012). Although there have been periods of significant biodiversity loss in
Earth’s history, recent extinction rates are estimated to be 100 to 1000 times higher
than pre-human levels, and may even increase further in the future (Seddon, Griffiths,
Soorae, & Armstrong, 2014). The key drivers of biodiversity loss are: habitat
degradation and loss, overexploitation, invasive species, pollution and climate change,
all of which are human-induced (Sala, 2000).
Although drivers of biodiversity loss such as habitat degradation and overexploitation
often dominate local changes in biodiversity over the short-term, human-induced
climate change has the capacity to irreversibly alter biodiversity for the long term at a
global scale (Parmesan & Yohe, 2003). However, even though climate change impacts
biodiversity at a global level, the extent of those impacts are often not evenly distributed
(Sala 2000). Some ecosystems or species are more sensitive to the impacts of climate
change than others (Foden et al. 2013).
Voigt et al. (2003) found that sensitivity to climate change significantly increase with
trophic level, with carnivores being the most sensitive to changes in climatic conditions.
Furthermore, McCain and King (2014) found that mammal’s response to current
changes in climatic conditions increase with body size. From these two findings it can be
assumed that large bodied carnivores are highly sensitive to climate change in
comparison to smaller bodied mammalian species.
This may be because large bodied carnivores have relatively high metabolic
requirements. This means that they require high levels of resources (Carbone, Mace,
Roberts, & Macdonald, 1999) and are sensitive to even small changes in these resources.
In addition, their high resource requirement also means that an individual of a given
carnivore species requires large areas which contain sufficient resources to sustain
themselves (Gittleman & Harvey, 1982). This area can be described as an individual’s
home range (HR; Burt 1943). As a result of these large HR requirements, large
carnivores occur at naturally low densities.
8
Moreover, large carnivores have a disproportionately high extinction risk. Declining
carnivore species threatened with extinction are an order of magnitude heavier than
non-threatened species (Cardillo et al., 2005). In addition, due to large carnivores’ large
HR requirements, they often come into conflict with humans as they range widely to find
and secure the prey that they need. Conflict with humans is one of the key causes of
adult mortality and contributes significantly to large carnivores’ high extinction rates
(Woodroffe & Ginsberg, 1998). In summary, large carnivore’s sensitivity to climate
change, coupled with their high extinction risk and vulnerability to human persecution
makes them a high priority for conservation and will therefore be the focus of this study.
The combination of high resource requirement and human-carnivore conflict has led to
large carnivore conservation often being centralised on setting aside large intact areas
of natural habitat with low densities of human settlements (Mills, 1991). These areas
provide large carnivores with the resources and space that they require whilst reducing
the chance of human-carnivore conflict. These areas often taken the form of protected
areas (PAs; Rodrigues et al. 2004).
PAs have been described by The Convention on Biological Diversity (CBD) as the
cornerstone of biodiversity conservation, and have become a key component for
strategies of conserving some large carnivore species (Balmford et al. 2003; Cantú-
Salazar et al. 2013; Gaston et al. 2008). Given conservation’s reliance on PAs, it is
important to understand if they are a suitable tool for conserving a wide range of
ecosystems and species. This study will examine PA’s utility for sustaining viable large
carnivore populations and identify variables which could impact their current and
future utility for large carnivore conservation.
One of the key threats to PA’s utility for large carnivore conservation is climate change
due to the fact that PAs are static entities and cannot shift in response to changing
climatic conditions (Hannah et al., 2007). There is mounting evidence that changes in
climatic conditions, such as alterations in precipitation or temperature patterns, which
lead to alterations in primary productivity dynamics, are impacting the health of the
ecosystems PAs were established to protect, and therefore on their continued utility
(Pettorelli et al., 2012; Singh & Milner-Gulland, 2011).
9
Alterations to primary productivity dynamics, under changing environmental
conditions, produce shifts in the spatiotemporal distribution of prey across landscapes.
Such alterations could impact large carnivores’ HRs, potentially shifting them outside of
PAs resulting in a reduction of the future relevance and potential utility of the current
global PA network to conserve them (Hannah et al., 2007; Parmesan et al., 1999;
Thuiller, 2004).
The high levels of resources which large carnivores require are not static and vary both
spatially and temporally. Varying levels of productivity and seasonality impact the
spatiotemporal predictability of primary productivity and therefore prey distribution
(WallisDeVries, 1996). This variation in the predictability of prey availability has been
shown to contribute significantly to shaping carnivore HR requirements (Duncan et al
submitted).
Areas with high productivity and low seasonality have relatively high prey
predictability. This means that carnivores which live under these conditions have a
more stable and predictable source of prey, so they therefore do not need to range as far
as they would in an area where productivity is low and seasonality is high where prey
would be less abundant and less constant in the landscape. This leads to carnivores in
areas of low productivity and high seasonality requiring larger HRs, than those which
inhabit areas with higher productivity and lower seasonality (Herfindal, Linnell, Odden,
Nilsen, & Andersen, 2005; Mcloughlin, Ferguson, & Messier, 2000; Nilsen, Herfindal, &
Linnell, 2005).
This is concerning because under future scenarios of climate change, it has been
hypothesised that variance in climatic conditions will increase (IPCC 2007, 2013),
leading to an overall decrease in the spatiotemporal predictability of primary
productivity. This could lead to an overall increase in HR requirements for large
carnivores as the predictability of prey will be reduced in response to the increase in
primary productivity variability. These changes in HR requirements of large carnivores
could threaten the potential future utility of PAs for their conservation.
Under this scenario, PAs which experience the most pronounced changes in their
primary productivity dynamics may be under the most threat, as it can be assumed that
that the large carnivores in these PAs will also experience the most pronounced changes
10
in their HR dynamics. Other factors will, no doubt, also impact PA’s utility for large
carnivore conservation.
Size may play an important role in determining whether not a PA is suitable for large
carnivore conservation. Smaller PAs may have less potential to sustain viable carnivore
populations under current and future climatic conditions than larger ones, as smaller
PAs may struggle to provide sufficient space to accommodate large carnivore’s large HR
requirements (Johnson et al. 2006; Laidlaw 2000; Brashares et al. 2001; Linnell et al.
2001). This situation may be exacerbated for small PAs which are predicted to
experience the most pronounced changes in their primary productivity dynamics, as
these changes may lead to an increase in HR size in large carnivores (Mcloughlin et al.,
2000) and smaller PAs may struggle to provide the area required.
In addition, primary productivity dynamics in the location of the PA, such as
productivity and seasonality, may also impact the probability of whether or not a PA is
useful for large carnivore conservation. PAs in areas with low productivity and high
seasonality will contain large carnivores with relatively larger HRs than PAs with higher
productivity and lower seasonality (Mcloughlin et al., 2000). This may mean that PAs in
these areas are under additional pressure to provide the space that large carnivores
require under these environmental conditions. This would mean that PA’s potential to
sustain viable carnivore populations in these regions, usually found at high latitudes,
could be particularly at risk.
In order to understand how climate change may impact the utility of PAs for large
carnivore conservation, this study will identify the characteristics of PAs which are most
at risk of losing their large carnivores and identify factors which impact PA’s utility to
sustain viable populations of their large carnivores.
By focusing on large carnivores the results of this study will be applicable to wide-
ranging species in general. They will also give a general indication of the utility of PAs to
conserve healthy functioning ecosystems as wide-ranging species, such as large
carnivores, can be used as umbrella species for ecosystem health (Simberloff 1998; Caro
and Doherty 1999).
11
1.2. Aims and Objectives
The aim of this study is to assess the global PA network’s utility for large carnivore
conservation under current and future climate scenarios. The following objectives will
be met to achieve this aim.
Predict primary productivity dynamics of PAs under future climate scenarios.
Predict large carnivore population sizes PAs are able to support under current and
future climate scenarios.
Assess which variables impact PA’s ability to sustain viable large carnivore
populations under current and future climate scenarios.
1.3. Hypotheses
H1 PAs in areas which experience the greatest changes in primary productivity
dynamics will be more likely to lose their utility for large carnivore conservation
PA characteristics hypothesised to impact PA’s utility for large carnivore conservation
(table 1.1).
Table 1.1. The variables hypothesised to impact PA’s utility to conserve carnivores under
current and future climate scenarios
Variable Hypothesis Supporting Information
H2. PA Size
PA size will have a positive impact on the
probability of a PA sustaining viable large
carnivore populations
Baeza and Estades 2010;
Johnson et al. 2006;
Brashares et al. 2001;
Laidlaw 2000
H3. PA location
(latitude)
Latitude will have a negative impact on the
probability of a PA sustaining viable large
carnivore populations
Sala 2000;
Mcloughlin, Ferguson, and
Messier 2000
H4. Productivity
Productivity will have a positive impact on
the probability of a PA sustaining viable
large carnivore populations
Carbone and Gittleman
2002; Nilsen, Herfindal,
and Linnell 2005
H5 Seasonality
Seasonality will have a negative impact on
the probability of a PA sustaining viable
large carnivore populations
Mcloughlin, Ferguson, and
Messier 2000; Herfindal et
al. 2005; Nilsen, Herfindal,
and Linnell 2005;
12
2. BACKGROUND
The following section will (1) provide a brief introduction to the current understanding
of how protected areas (PAs) can contribute to large carnivore conservation and their
limitations and (2) outline the current understanding of how climate change is predicted
to impact utility of PAs for conservation and (3) introduce methods of assessing the
utility of PAs.
2.1. Large carnivores and PAs: what do we know?
Large carnivores, those with a body mass of ≥10kg, are of particular importance and
therefore conservation interest because they play a key role in ecological health and
have experienced substantial population declines, constriction of their geographic
ranges and fragmentation of their habitat in the past two centuries making them
increasingly vulnerable to extinction (Ceballos & Ehrlich, 2002; Morrison, Sechrest,
Dinerstein, Wilcove, & Lamoreux, 2007; Ripple et al., 2014). Even though many species
of large carnivore exist at naturally low densities, they have significant top-down effects
on community composition through direct interactions, such as predation, and indirect
interactions, such as causing an alteration in prey behaviour which consequently
impacts browsing pressure on vegetation (Miller et al., 2001; Ripple et al., 2014). In
addition, the largest bodied carnivores also play a key role in mediating inter-specific
competition amongst carnivores, thus they play a key role in structuring ecosystems and
maintaining their functionality (Estes, 1996).
Large carnivores have large energetic constraints (Carbone et al. 1999), meaning they
require large amounts of resources (prey) to sustain themselves. In order to secure
these resources, they must often roam-widely in search of prey (Cardillo et al. 2004;
2005) leading to large area requirements. The area that an individual requires to sustain
itself and successfully breed is known as its home range (HR; Burt 1943).
For large species of carnivore, such as the tiger, prey abundance has been shown to
largely drive the persistence of viable populations (Karanth, Nichols, Kumar, Link, &
Hines, 2004). Johnson et al. (2006) found that prey in smaller PAs (≤100km2) was
largely absent, and therefore so were the large carnivores. Small PAs do not provide the
13
area large carnivore’s prey, such as ungulates, require (Laidlaw, 2000). A loss of large
prey leads to an increase in competition for smaller prey and a depression or loss of
large carnivore populations (Rabinowitz, 1989). Leading to the conclusion that small
PA’s utility for large carnivore conservation is limited.
This finding is echoed by Laidlaw (2000), who found that the survival of the largest
carnivores was determined by the size of surviving suitable habitat, such as that found
in PAs. Moreover, Laidlaw also found that only the largest patches of suitable habitat
(10,000km2) contained large-bodied wide-ranging species, such as large carnivores.
Brashares et al. (2001) provide further support for the idea that size of PA is an
important factor for the persistence and long-term viability of large carnivore
populations. They showed that PA size is strongly correlated with extinction risk in large
mammal species, meaning that the largest PAs have the highest utility for large
carnivore conservation.
Worryingly, Carroll et al. (2004) outlined a theory that even PAs which currently contain
populations of large carnivores, which may appear to be currently successfully
sustaining them, may in fact be suffering from ‘extinction debt’. Large carnivores are
long-lived. As their habitat becomes more restricted, for example through habitat loss,
large carnivores often retreat into the boundaries of a PA where suitable habitat may
still remain. They may continue to persist within that PA for a number of years, but as
their habitat requirements are, in reality, no longer being provided, their eventual
extinction is inevitable (Doak, 1995).
In addition to PAs providing space and suitable habitat for large carnivores, they can
also provide safety. Conflict with humans has been found to be the most influential
source of adult mortality in large carnivores (Woodroffe & Ginsberg, 1998). PAs can
therefore provide safe havens for large carnivores, and have been key in securing viable
populations of large carnivores, such as tigers, whose decline has been largely due to a
combination of habitat loss and poaching (Johnson et al., 2006). Paviolo et al. (2009)
also outlined the impact of levels of protection in a PA and its utility for large carnivore
conservation. They found a positive relationship between level of protection of a PA and
puma abundance. This relationship has also been observed in other large carnivore
species such as the ocelot (M. S. Di Bitetti, Paviolo, & De Angelo, 2006) and the jaguar
14
(Paviolo, De Angelo, Di Blanco, & Di Bitetti, 2008). From this it can be assumed that PAs
with higher levels of protection have higher utility for certain species of large carnivore
that are vulnerable to anthropogenic threats such as poaching.
It is not just size and level of protection which are important to the utility of PAs for
large carnivore conservation. The management strategy of the PA also impacts its utility.
The IUCN has created seven distinct PA management categories (Table 2.1), ranging
from strict biodiversity protection (category Ia) to areas managed for sustainable
extraction (category VI).
Table 2.1. IUCN PA management categories (Dudley 2008)
Category Management Objectives
Ia Strict nature reserve Strict protection of biodiversity
Ib Wilderness area Protect largely unmodified wilderness
II National park Protect large-scale ecological processes and
provide recreational opportunities
III Natural monument or Feature Protect a specific natural monument
IV Habitat/species management area Protect particular species/habitats with
active management
V Protected landscape/seascape Protect the distinct character of interactions
of people and nature
VI Protected area with sustainable
use of natural resources
Protect ecosystems and sustainably manage
them for natural resource extraction
IUCN PA management categories Ia and Ib are the only categories out of the seven
which provide strict protection and control of human presence and influence within PAs.
Categories V and VI are relatively new designations, developed following the 1992
World Parks Congress (Whitehouse, 2001). It has been argued that these two additional
categories should not be considered protected areas, as their primary objectives are not
to conserve biodiversity but rather to promote sustainable development and therefore
should be reclassified as sustainable development areas (Locke & Dearden, 2005). VI in
particular has drawn criticism as a PA category, because it allows sustainable extraction
of natural resources, which is often seen as the antithesis of biodiversity conservation.
15
However, the objectives of resource extraction and large carnivore conservation do not
necessarily oppose one another. Linnell et al. (2001) propose that the intensive
extractive forestry industries of Scandinavia could provide much needed habitat and
area for populations of Eurasian Lynx. They also go as far to say that the extraction of
timber poses virtually no threat to lynx.
However, it is not always the direct impacts of extraction which threaten large
carnivore’s persistence in an environment. The indirect impacts of having an extractive
industry in an area with large carnivores can also negatively impact the viability of their
populations.
McLellan and Shackleton (1988) found that extractive industries were accompanied by
an increase in roads in an area which have been shown to negatively impact large
carnivore abundance in species such as: Malayan sun bears (Meijaard, 1999);
wolverines (Krebs et al. 2007); badgers (Clarke et al. 1998); tigers (Kerley et al., 2002)
and leopards (Ngoprasert et al. 2007). This list is by means exhaustive but demonstrates
the range of species which all share the same negative relationship with the indirect
impacts of extractive activities such as those which will operate in category VI PAs.
Even PAs with minor non-consumptive anthropogenic impacts, such as recreational use
of a category II PA (National Park), have been shown to negatively impact large
carnivores and therefore their utility for their conservation. Reed and Merenlender
(2008) found that pumas were five times less abundant in PAs which allowed recreation
than PAs with strict protection. This has implications for the utility of any PA which
allows human access or habitation, which is six out of the seven PA types.
In addition to size, level of protection and management strategy of a PA impacting their
utility for large carnivore conservation, the connectivity of a PA also plays a role in its
utility. Carroll et al. (2004) propose that connectivity of a PA is vitally important to
ensure future viability of large carnivore populations. They also put forward the idea
that PAs in northerly regions are often located within a less developed matrix and
therefore often have highly connectivity and therefore higher utility for large carnivore
conservation.
16
Overall, the current understanding of PA utility for large carnivore conservation shows
that it is a complex relationship. PA size, management, levels of protection and
connectivity all shape PA utility for large carnivore conservation.
2.2. How is climate change predicted to impact PA’s utility for conservation?
The Intergovernmental Panel on Climate Change (IPCC)’s latest report predicts that
global earth temperatures will increase by between 1.5°C and 4.5°C by the end of the
21st century, depending on levels of green house gas (GHG) emissions (IPCC 2013). It is
most likely that it will be the unprecedented rate of change rather than the magnitude of
change that will impact biodiversity and therefore the stability of Earth’s ecosystems
(Bradshaw & Holzapfel, 2006).
These climate change induced changes in temperature and precipitation patterns are
predicted to change species’ ranges (Root et al. 2003; Parmesan and Yohe 2003). Species
which currently occur within PA boundaries may shift their ranges in response to
changing environmental conditions and move ‘out’ of PAs (Hannah et al., 2007;
Parmesan et al., 1999). PA’s utility for conservation under a changing climate is
therefore called into question (Peters & Myers 1991).
Hannah et al. (2007) found that the current PA network required additional PAs to
accommodate future range shifts under moderate climate change. They also predicted
that increasing the current PA extent to accommodate future range shifts was the most
efficient solution to solving the current PA’s questionable effectiveness in representing
all species ranges (Rodrigues et al., 2004) whilst also ensuring its future utility for
biodiversity conservation.
Even if species to do not shift their ranges out of PAs, their area requirements may
change under future climatic conditions. Environmental variability has been shown to
have a positive impact on the amount of area an individual requires to survive and
breed; its home range (Mcloughlin et al., 2000). With climate change predicted to
increase the variability of environments in many regions of the world (IPCC 2013) this
could lead to PAs struggling to provide the space for increasing home range
requirements.
17
Therefore is can be concluded that PAs are potentially an extremely useful tool to
conserve biodiversity under climate change but only if their utility under future climate
change is maximised by expanding the current PA network according to predicted future
range shifts of species and predicted increases in area requirements under climate
change.
2.3. Assessing the utility of PAs: population viability analysis (PVA)
A key way to measure whether or not a PA is useful for conservation of a given species is
to determine whether or not it is capable of sustaining viable populations of said species.
One of the most commonly used methods to assess the viability of populations is
population viability analysis (PVA; Shaffer 1981; Boyce 1992).
There is no agreed upon single process of exactly what constitutes a PVA, despite its
common usage as a tool in endangered species management (Coulson et al. 2001; Reed
et al. 2002). However, there are some components which they all share. They all include
a measure of extinction risk (quasi-extinction) or projected population growth. This is
achieved by using life-history data such as, survival and fecundity rates or population
growth rates, to predict population size from one year to the next over a set period of
time. The complexity of a PVA can vary dramatically.
The simplest PVA should include stochasticity in life-history to represent variation in
environmental conditions which impact survival, fecundity and population growth rates
(Boyce, 1992). Another level of complexity would be to incorporate density dependence
and a carrying capacity of maximum population size. Environmental disasters can also
be incorporated, whereby the population growth rate or juvenile survival will be zero
for a random year due to an ‘environmental disaster’. PVAs can also be spatially explicit,
whereby the life-history traits used to project the population vary spatially within a
landscape depending on where the population you are assessing occur (Carroll et al.,
2004). Most PVAs do not include immigration but this could also be incorporated.
PVAs are useful and extremely flexible tools in comparing the effects of different future
scenarios on population viability, such as future climate scenarios or management plans.
In this instance, PVAs can be used to measure the relative impact of two or more
scenarios on population viability and therefore do not need to be explicitly precise
(Beissinger & Westphal 1998; Possingham et al. 1993).
18
A number of issues have been highlighted in PVA’s use and application to conservation.
These are primarily related to the estimation of parameters used to project populations
using minimal data and some PVA’s simple construction (Coulson et al., 2001). In
particular, the validity of future population projections has been questioned when a PVA
assumes that variation in vital rates will remain stationary in the future or attempts to
predict future changes in vital rates (Coulson et al., 2001).
However, PVAs still provide vital insights into how future scenarios could impact the
viability of populations, perhaps not to explicitly predict a population’s trajectory, but to
assess the relative impact of different possible future scenarios in comparison to current
conditions. In addition, PVA is currently the best available tool for predicting extinction
risk. Many of the alternatives are more subjective and are likely to provide less accurate
predictions (Zeckhauser & Viscusi 1990). Furthermore, PVAs have been shown to
project populations accurately when compared to observed data (Brook et al., 2000)
demonstrating their utility for conservation. It is, however, important to interpret PVA
results with caution and to be aware of how parameters have been determined and
exactly how a PVA has been constructed to ensure that all assumptions of the model are
clearly communicated and understood.
19
3. METHODS
This study assessed protected area’s (PAs) utility to sustain viable populations of large
carnivores and how this may vary under current climate change scenarios outlined by
the IPCC’s 2013 report. This was achieved by predicting (1) changes in primary
productivity dynamics under predicted changes in climate, (2) how these changes will
impact carnivore home range requirements, and (3) how those changes in home range
requirements could impact population sizes PAs were potentially able to sustain. Figure
3.1 shows the general outline of the methodological process of this project.
Fig. 3.1. General outline of project methodological framework
3.1. Large Carnivores
Only large bodied carnivores, defined here as those with an average adult body mass of
≥10kg were considered for the analysis. Data on average adult body size was taken from
the PanTHERIA database (Jones et al., 2009) and species’ diets were classified into two
categories: omnivores or obligate carnivores according to Nowak (1999). Only solitary
species were included in the analysis. From this subset, only species which had all data
required for analysis available from the PanTHERIA database (Jones et al., 2009) were
Predict which
factors impact
PA’s utility for
large carnivore
conservation
Assess primary
productivity
dynamics in a
random sample
of PAs
Predict changes in
primary productivity
dynamics under future
climate scenarios in
sampled PAs
Identify which
PAs occur within
large carnivore’s
range
Use primary productivity
dynamics and carnivore
ecology to predict changes
in population sizes
sampled PAs can support
under current and future
climate scenarios
Predict what kind
of PAs may lose
their carnivore
species under
future climate
scenarios
20
considered, this resulted in a sample of 10 species of solitary large carnivore (table 3.1).
Conducting an extensive literature search to increase the sample size by adding data to
the PanTHERIA database was beyond the scope of this study. Geographic distribution
maps of the large carnivores being considered were downloaded from the IUCN Red List
(2014).
Table 3.1. Species included in analysis
Species Common name Average adult body
mass (kg) (Jones et al., 2009)
Diet
(Nowak, 2005)
Ursus arctos Brown Bear 196.29 Omnivore
Panthera tigris Tiger 161.91 Carnivore
Ursus americanus Black Bear 110.50 Omnivore
Ursus thibetanus Asian Black Bear 99.71 Omnivore
Puma concolor Puma 53.95 Carnivore
Panthera pardus Leopard 52.40 Carnivore
Acinonyx jubatus Cheetah 50.58 Carnivore
Lynx lynx Lynx 19.30 Carnivore
Gulo gulo Wolverine 12.79 Carnivore
Leopardus pardalis Ocelot 11.88 Carnivore
3.2. Protected Areas
Terrestrial PA information was taken from the World Database of Protected Areas
(WDPA; IUCN & UNEP 2014). Only PAs located within the global range of the large
carnivores being assessed were considered.
All PAs in the WDPA database have been classified into one of seven IUCN Protected
Area Management Categories (Dudley 2008). IUCN Category III PAs, whose primary
objective is to “protect specific outstanding natural features and their associated
biodiversity and habitats” were excluded from the analyses because these PAs are often
very small in area and only encompass a single natural or human influenced feature
such as a prehistoric human cave dwelling or a single waterfall, making carnivore
conservation beyond their objectives. PAs for which the validity of productivity indices
21
can be reduced, i.e. areas in extreme high latitudes or large wetland areas were also
omitted. PAs not nationally gazetted or not legally or formally declared, were also
excluded from the analyses.
For each of the five continents where the large carnivores selected occur, a randomly
selected sample, according to the probability distribution of continent-specific PA size,
of 125 PAs was identified, giving a total sample size of 625 PAs.
3.3. Primary productivity dynamics of PAs
The Normalised Difference Vegetation Index (NDVI) was used to index primary
productivity dynamics (Pettorelli 2013). NDVI is derived from the ratio of red:near-
infrared (R:NIR) reflectance, generating values between -1 and +1 using the following
calculation: NDVI = (NIR – R) / (NIR + R). Pixels encompassed in the set of PAs
identified were extracted from the Global Inventory Modelling and Mapping Studies
dataset (GIMMS; Tucker et al. 2005). The GIMMS dataset provides a bimonthly, 8km
resolution, 16-day composite global measure of NDVI; the period considered for the
analysis was 1982 – 2011. The GIMMS data represents the longest NDVI time-series
dataset available and is therefore the most appropriate source of data to determine
long-term changes in primary productivity.
This NDVI time-series data was ‘smoothed’ to correct for environmental noise (Pettorelli
et al., 2005). NDVI values can become distorted by atmospheric variation contaminating
the radiation reaching the satellite sensor and therefore misrepresenting changes in
primary productivity on the ground (Achard and Estreguil 1995; Tanre, Holben, and
Kaufman 1992). These distortions can be detected by identifying rapid changes in NDVI
values (of 0.25 or more from one composite to the next) followed by a rapid return to
original values (Garonna et al. 2009). Once these values are identified, they can be
replaced with the average of the previous and following composites, this will ‘smooth’
the annual NDVI curve for that pixel (Pettorelli et al., 2012). When there was a rapid
change in NDVI values which did not return to original values for two or more
composites, that pixel was removed. PAs which had more than 50% of their pixels
removed were not considered in further analysis.
22
To index PA-specific estimates of productivity, the average annual Integrated NDVI
(iNDVI; the sum of all NDVI composites within a given year; Pettorelli 2013) was taken
from all pixels and across all years for each PA. In order to index the annual level of
temporal variability in primary productivity, the level of seasonality (or ‘contingency’) in
NDVI was calculated across the time series data for each PA sensu Colwell (1974).
Contingency varies from 0 to 1, with higher values indicating greater levels of
seasonality (Colwell 1974).
In order to calculate this index of seasonality in NDVI, the time series data were
classified into 10 classes (NDVI = 0–0.1, 0.1–0.2, 0.2–0.3, 0.3–0.4, 0.4–0.5, 0.5–0.6, 0.6–
0.7, 0.7–0.8, 0.8–0.9, and 0.9–1.0; English et al. 2012). It is not expected that data
discretisation will result in a loss of information that would impact the results because
these analyses were performed at a global-scale.
3.4. Future Climate Scenarios
Data on the current climatic conditions and future climate scenarios were extracted
from the Hadley Centre Global Environmental Model: version 2 (HadGEM2; Collins et al.,
2011) global climate model (GCM) downloaded from WorldClim (Worldclim, 2005;
Hijmans et al. 2005). Current climate data were available for the time period 1950 –
2000 and future projections were available for the time periods 2041 – 2060 and 2061 –
2080. For each of the three time periods being considered, data were available as one-
year monthly averages of the climatic conditions taken from across the time-series.
To examine different possible future climate trajectories, each of the Intergovernmental
Panel on Climate Change’s (IPCC) four Representative Concentration Pathways (RCPs;
IPCC 2013) were considered. Each RCP represents a possible climate future, each with
varying levels of greenhouse gas (GHG) concentrations due to different emissions levels
(table 3.2). The numbers of the RCPs represent the radiative forcing, which is the
difference between how much sunlight is absorbed by the earth and how much is
radiated back to space (W m-2). Radiative forcing is calculated for the year 2100 relative
to pre-industrial levels in 1750 (IPCC, 2013).
23
Table 3.2. IPCC’s four RCP climate scenarios considered in the analyses (IPCC 2013)
Climate
Scenario Description
Predicted range of global mean surface temperature change in 2100 relative to 1986 – 2005
RCP 2.6 Global GHG emissions peak in 2010-2020
then significantly decline +0.3 – 1.7°C
RCP 4.5 Global GHG emissions peak in 2040 then
significantly decline +1.1 – 2.6°C
RCP 6.0 Global GHG emissions peak in 2080 then
significantly decline +1.4 – 3.1°C
RCP 8.5 Global GHG emissions continue to rise
throughout the 21st century +2.6 – 4.8°C
3.5. Statistical Analyses
3.5.1. Future Climate Scenarios
Primary productivity dynamics were predicted for each of the sampled PAs using linear
modelling under each of the four RCPs, for the two future time periods examined.
Primary productivity dynamics considered were productivity (mean iNDVI) and
seasonality (NDVI contingency). Relationships were identified between climatic
variables, which have been shown to impact productivity and seasonality (Grosso et al.,
2008; Potter & Brooks, 1998), and mean iNDVI and NDVI contingency of the PAs
sampled. These relationships were then used to predict future PA-specific productivity
and seasonality. The incorporation of site-specific non-climatic variables such as habitat
type was beyond the scope of this study. The climatic variables considered are outlined
in table 3.3. Model averaging was used to select the best fitting model using R package
MuMIn (Barton 2014).
24
Table 3.3. Climatic variables considered to predict iNDVI and seasonality
Productivity (mean iNDVI) Seasonality (NDVI contingency)
Annual mean temperature Standard deviation in annual mean monthly
temperature
Annual mean precipitation Standard deviation in annual mean monthly
precipitation
Standard deviation in annual mean monthly
temperature
Coefficient of variation of annual mean
monthly temperature
Standard deviation in annual mean monthly
precipitation
Coefficient of variation of annual mean
monthly precipitation
The resolution of the NDVI data extracted from the GIMMS time-series and the
HadGEM2 data were of different spatial resolutions (8km and 4.5km respectively). To
account for this, an average of all PA-specific pixels of average monthly temperature and
average monthly precipitation were taken to calculate the predictor variables (table
3.5.1.1) for each PA from the HadGEM2 to predict future primary productivity dynamics
using the linear models.
These results can be used to make some qualitative predictions about how future
climate scenarios may impact primary productivity dynamics of PAs.
3.5.2. Predicting PA-specific Carnivore Population Size from Home Range
A linear mixed effects model (lme) of carnivore home range (HR) size, developed by
Duncan et al. (submitted), was used to predict PA-specific population-level HR size for a
given carnivore species. This model used PA-specific primary productivity dynamics
(iNDVI and NDVI contingency), and species-specific body mass and dietary category to
predict population-level HR size. Full details of this model are in Appendix I.
The number of females for each species which could be accommodated within the PA
was then calculated by dividing the area of the PA by the predicted individual HR size.
This assumed that female HRs do not overlap, as is often seen in large carnivores
(Rabinowitz and Notthingham 1986) and that male’s HRs overlapped completely at a 1:1
ratio to female’s HRs. A further assumption is that the entire PA contains suitable habitat
and an equal distribution of resources which also allows females to defend exclusive
HRs (Mcloughlin et al., 2000). Defining a level of overlap between females would have
25
increased the number of assumptions in comparison to assuming no overlap, so was
concluded to be the most appropriate method for calculating population size from HR
size.
3.5.3. Assessing the Viability of Carnivore Populations in PAs
A female-based single population model (assumes a 1:1 ratio of females to males) which
included environmental stochasticity and density dependence was used to calculate
extinction risk of each species’ predicted population for each PA. The model used adult
and juvenile survival rates and birth rates per female per year to predict population
trajectory (figure 3.2).
Fig 3.2. Population model used to assess viability of populations. Nt = size of population (N) at
time (t); Nt+1 = size of population at Nt plus one year; β = no. of female offspring
per female per year; Sα = adult survival; Sj = juvenile survival
The model includes density dependence which was applied to the juvenile survival rate.
The carrying capacity for each population was determined by multiplying the PA area by
average population density of the species being considered. Population density data
were taken from the PanTHERIA database (Jones et al., 2009).
A literature search was conducted to determine adult and juvenile survival rates.
Juvenile survival was defined here as the survival probability from birth to one year.
Only species with at least three annual survival rates were included. For each year the
population was projected, the adult and juvenile survival rates were randomly chosen
from the normal distribution of survival rates, using the mean and standard deviation of
these rates, bounded between 0 and 1 to represent environmental variation. Details of
survival rates for each species can be found in Appendix II.
Number of females born per female per year was calculated from the PanTHERIA
database (Jones et al., 2009) by dividing litter size by birth interval and dividing this in
two, assuming an equal ratio of male to female offspring. The data from PanTHERIA are
already averaged across the literature therefore variation in fecundity was unavailable.
26
The starting population (Nt) was the number of female HRs a PA was predicted to be
able contain within its boundaries. The model assumed no immigration. Each population
was projected for 100 years for 1000 iterations. A population was considered extinct
(quasi-extinct) when it contained less than five individuals. This number was chosen as
a population of this size would become extremely vulnerable to demographic stochastic
effects (Caughley & Sinclair 1994). Extinction risk was calculated by dividing the
number of iterations which resulted in quasi extinction by the total number of iterations.
For a population to be considered viable, the extinction risk must be ≤5% (Wielgus,
2002).
The results of these analyses were examined using Pearson’s Chi-squared test to
determine whether or not a PA’s ability to sustain viable large carnivore populations,
and therefore a PA’s utility, changed significantly under the future climate scenarios.
3.5.4. Assessing the factors influencing PA’s utility for carnivore conservation
The results of the population viability analyses were examined using generalised linear
models (GLMs). Each predictor variable was modelled independently with the response
variable as significant relationships were expected between the explanatory variables.
The response variable was the probability of a PA sustaining viable populations of its
carnivore species. The predictor variables considered were PA area, PA location
(latitude), productivity (iNDVI) and seasonality (NDVI contingency).
If a significant difference is found in the utility of PAs to sustain viable large carnivore
populations under future climate scenarios, this will be repeated under each climate
scenario considered to assess whether factors impacting a PA’s utility for carnivore
conservation is predicted to change under future climate scenarios. All NDVI data
processing and statistical analyses were carried out in R version 3.1.1 (R development
core team 2014).
27
4. RESULTS
4.1. Protected Areas
388 PAs were included in the analysis after the removal of PAs through ‘smoothing’ the
NDVI data (fig 4.1). On average each PA contained 1.6 species of large carnivore.
Figure 4.1 Final PA sample included in analyses
North America n=66; South America n = 118; Europe n=19; Africa n=112; Asia n=73
The size distribution of sampled PAs differed significantly between the five continents
(F(4, 383) = 50.02, P<2e–16). Europe and North America having on average the smallest
PAs, and Africa and South America having the largest (fig 4.2).
Figure. 4.2. Size distribution of sampled PAs across the five continents assessed. Bars which do
not share letters are significantly different from one another (p<0.001).
28
The size distribution of sample PAs also significantly differed between the IUCN
Management categories (F(5, 382) = 9.802, p = 7.92e –09) (fig. 4.3).
Figure 4.3. Sampled PA size distribution by IUCN management category. Bars which do not
share letters are significantly different from one another (p<0.001).
4.2. Changes in primary productivity dynamics
4.2.1. Modelling future primary productivity
The top five performing linear models (lm) which explained the most variation in
productivity (mean iNDVI) are outlined in table 4.1. Model averaging was used to select
the best fitting model as four of the models were within 2 AIC values of each other. Table
4.2 shows the estimated impact of each variable included in the best fitting model.
Table 4.1. Best performing linear models considered to predict productivity
Model AIC ΔAIC Akaike
Weight R2
Annual mean precipitation + precipitation SD + annual
mean temperature
1870.52 0 0.37 0.564
Annual mean precipitation + precipitation SD 1871.11 0.37 0.31 0.562
Annual mean precipitation + temperature SD +
precipitation SD
1872.16 1.42 0.18 0.562
Annual mean precipitation + annual mean temperature +
temperature SD + precipitation SD 1872.71 1.97 0.144 0.563
Annual mean precipitation 1891.94 21.19 8.79E-06 0.54
SD = Standard deviation of the annual monthly mean precipitation and temperature
29
Table 4.2. Estimates of impact for each variable included in best fitting model used to predict
productivity
Variable Estimate Standard Error t value p value
Annual mean precipitation 0.145 0.0095 15.27 <2e-16
Precipitation SD -0.0285 0.00057 -5.03 0.0000008
Annual mean temperature 0.145 0.019 1.553 0.121
4.2.2. Modelling future seasonality
The top five performing linear models which explained the most variation in seasonality
(NDVI contingency) are outlined in table 4.3. Model averaging was used to select the
best fitting model as three of the models were within 2 AIC values of each other. Table
4.4 shows the estimated impact of each variable included in the best fitting model.
Table 4.3. Best performing linear models considered to predict seasonality.
Model AIC ΔAIC Akaike
Weight R2
Precipitation SD + temperature SD -588.55 0 0.45 0.41
Precipitation SD + temperature SD + precipitation COV -587.62 0.94 0.28 0.41
Precipitation SD + temperature SD + temperature COV -586.55 2.00 0.17 0.41
Precipitation SD + temperature SD + temperature COV +
precipitation COV -585.57 2.98 0.10 0.41
Temperature SD + precipitation COV -563.84 24.71 1.90e-06 0.37
SD = Standard deviation of the annual monthly mean precipitation and temperature
COV = Coefficient of variation in annual monthly mean precipitation and temperature
Table 4.4. Estimates of impact for each variable included in best fitting model used to predict
seasonality
Variable Estimate Standard Error t value p value
Precipitation SD 0.0299 0.00187 15.967 <2e-16
Temperature SD 0.0011 0.000182 6.071 3.05e-09
30
4.2.3. Predicting changes in productivity dynamics under future climate scenarios
The majority of PAs were predicted to experience changes in productivity and
seasonality in comparison to current conditions (figures 4.4 and 4.5). However, on a
global scale neither productivity nor seasonality was predicted to significantly change
overall when comparing values from all PAs (productivity: F(8, 3483) = 0.111 p = 0.999;
Seasonality: F(8, 3483, df = 8, p = 0.306)).
Figures 4.4 and 4.5 show predicted changes in productivity (mean iNDVI) and
Seasonality (NDVI contingency) in each PA under each RCP for the time period 2061 –
2080 in comparison to current primary productivity dynamics (1950 – 2000). Changes
in PA primary productivity dynamics were site specific and varied from PA to PA.
However, some overall regional trends in changes of productivity and seasonality across
the future climate scenarios considered were identified. The most pronounced of these
were in arid regions, where productivity and seasonality were both predicted to
increase to the greatest extent.
There was an overall trend of decreasing productivity and seasonality in Southern
Africa. An opposite increasing trend was predicted for productivity of PAs in West
Africa, but seasonality was still predicted to decrease under future climate scenarios.
PAs in Europe were predicted to experience increases in seasonality and a decrease in
productivity. The majority of sampled PAs in South America were also predicted to
experience an increase in seasonality.
31
Fig. 4.4. Predicted proportional changes in productivity (iNDVI) under each RCP for the time period 2061 – 2080 in comparison to
current conditions (1950 – 2000)
RCP 2.6
RCP 4.5
RCP 6.0
RCP 8.5
-51 – -75%
-26 – -50%
-1 – -25%
0%
1 – 25%
26 – 50%
51 – 75%
76 – 100%
+100%
% Change
32
Fig. 4.5. Predicted proportional changes in seasonality (NDVI contingency) under each RCP for the time period 2061 – 2080 in
comparison to current conditions (1950 – 2000)
RCP 2.6
RCP 4.5
RCP 6.0
RCP 8.5
-51 – -75%
-26 – -50%
-1 – -25%
0%
1 – 25%
26 – 50%
51 – 75%
76 – 100%
+100%
% Change
33
4.3. Predicted changes in home range dynamics under future climate
scenarios
Table 4.5 shows the minimum, maximum and mean HR size predicted for each species
considered. The model predicted a much smaller range of HR sizes than is observed in
data taken from the scientific literature on observed home range. The mean observed
and predicted HRs varies to the greatest extent in the brown bear (Ursus arctos), the
wolverine (Gulo gulo) and the lynx (Lynx lynx).
Table 4.5. Predicted and observed carnivore home range sizes (Duncan et al. Submitted).
Species
Home range km2
Predicted from lme Observed
Minimum Maximum Mean Minimum Maximum Mean
Ursus arctos 30.20 178.70 98.16 3.31 8171 683.7
Panthera
tigris 78.66 402.17 268.17 10.46 1160 254
Ursus
americanus 12.91 124.42 51.84 4.6 1721 37.55
Ursus
thibetanus 10.95 103.95 44.32 - - -
Puma
concolor 23.10 246.66 79.90 39 826 255.43
Panthera
pardus 30.32 292.56 100.13 8.7 1137 151
Acinonyx
jubatus 59.89 216.37 118.40 - - -
Lynx lynx 21.86 130.99 69.57 106 1515 439
Gulo gulo 4.82 16.57 10.93 90 2876 553
Leopardus
pardalis 6.79 36.78 15.75 1.17 29.58 12.5
Changes in home range (HR) size under each climate scenario considered, varied site
by site for each species. However, there were some overall global trends which
appeared to be closely correlated with changes in seasonality (fig 4.6). The strongest
increase in seasonality in arid regions was also accompanied by increases in HR sizes
of large carnivores which occur in the region. The decrease in seasonality in Sub-
Saharan Africa was coupled with a general decrease in HR size in the region. The
overall trend of an increase in seasonality across South America was accompanied by
increases in HR sizes of most large carnivore species.
34
Figure 4.6. Predicted proportional changes in home range size (km2) under each RCP for the time period 2061 – 2080 in comparison to
current conditions (1950 – 2000)
RCP 2.6
RCP 4.5
RCP 6.0
RCP 8.5
-51 – -75%
-26 – -50%
-1 – -25%
0%
1 – 25%
26 – 50%
51 – 75%
76 – 100%
+100%
% Change
35
Climate Scenario
% p
op
ula
tions v
iable
0
10
20
30
40
50
60
70
Current RCP 2.6 RCP 4.5 RCP 6.0 RCP 8.5 RCP 2.6 RCP 4.5 RCP 6.0 RCP 8.5
4.4. Predicted changes in number of viable populations under future climate
scenarios
There was no significant difference in any of the species assessed between the number
of viable populations under current and future climate scenarios (asian black bear χ2 =
0.06, df = 8, p = 1; puma χ2 = 1.33, df = 8, p = 0.50; leopard χ2 = 0.04, df = 8, p = 1;
cheetah: χ2 = 0.04, df = 8, p = 1; lynx: χ2 = 0.47, df = 8, p = 0.99; ocelot χ2 = 0.04, df = 8, p =
1). The number of viable brown and black bear populations in PAs assessed did not
change under future climate scenarios compared to current conditions. No PA assessed
was predicted to have a viable tiger or wolverine population under current climatic
conditions or any of the future climate scenarios considered. Of the large carnivore
populations predicted to occur in European PAs, not one was viable and only one PA in
North America was predicted to be able to sustain viable populations of its large
carnivore species. All other continents contained PAs which were able to sustain some
viable large carnivore populations.
Of the 633 populations assessed, only 19.9% were predicted to be viable under current
conditions (fig 4.7) and this did not change significantly under future climate scenarios
(χ2 = 0.1697, df = 8, p= 1).
Figure 4.7. Proportion of populations in sampled PAs viable under current and future climate
scenarios n=633. Years: 1950 – 2000; 2041 - 2060; 2061 - 2080
36
Even in PAs where seasonality and productivity were predicted to drastically change
(Figs 4.4 and 4.5), this was not related to a change in a PA’s ability to sustain viable
populations of its large carnivore species. Therefore H1 can be rejected. Out of the 633
populations assessed, only 15 were predicted to either lose or gain their viability under
future climate scenarios in comparison to current conditions. Each population was in a
separate PA, and each PA had varying levels of change in productivity and seasonality in
both directions. There was no clear pattern predicted for what kind of changes in
primary productivity dynamics lead to a loss or gain of a viable large carnivore
population.
4.5. Assessing the variables which impact PA’s utility for large carnivore
conservation.
As expected there were significant relationships between the explanatory variables.
Linear models showed that seasonality and latitude had a significant positive
relationship (R2 = 0.31, t = 13.09, p<0.001). Latitude and area had a significant negative
relationship (R2 = 0.28, t = -15.61, p<0.001). Seasonality and area had a significant
negative relationship (R2 = 0.27, t = 3.46, p = 0.000612) and seasonality and productivity
had a significant negative relationship (R2 = 0.039, t = -4.071, p<0.001). Consequently,
each variable was modelled individually to assess its impact on the probability of a PA
being able to sustain viable populations of its large carnivores.
PA size had a significant positive impact on the probability of a PA sustaining viable
populations of its large carnivore populations (H2; estimate = 1.108 ± 0.105 (1 s.e.), df =
387, p<2e-16). There was a significant negative effect of latitude on the probability of a
PA sustaining viable carnivore populations (H3; estimate = -0.08 ± 0.009 (1 s.e.), df =
387, p<2e-16). Productivity did not impact the probability of a PA being able to sustain
viable population of its large carnivore species (H4; estimate = -0.006 ± 0.024 (1 s.e.), df
= 387, p = 0.799). Seasonality had a significant negative effect on the probability of a PA
sustaining viable populations its carnivore species (H5; estimate = -1.51 ± 0.68 (1 s.e.), df
= 387, p = 0.0273). This was not repeated under future climate scenarios as there was no
difference between the number of PAs able to sustain viable populations under future
climate scenarios in comparison to current conditions.
37
5. DISCUSSION
The aim of this study was to assess the utility of the global PA network for large
carnivore conservation under current and future climate scenarios. The main finding of
this study was that, globally, the vast majority of PAs are not able to sustain viable
populations of their large carnivores under current conditions, and that this will not
change under future climate scenarios. Therefore, the current global PA network’s utility
for large carnivore conservation under current and future climate scenarios is
questionable.
5.1. Predicted changes in primary productivity and home range dynamics
Some of the most extreme changes in primary productivity in this study were predicted
for PAs which occur in the biome of desert and xeric shrublands (Olson et al., 2001). PAs
located in this biome were all predicted to experience the same trend of a substantial
increase in productivity and seasonality. The changes were particularly strong for PAs in
this biome located in North Africa.
This trend of increasing productivity in PAs in North Africa echoes Hoerling et al. (2006)
who also predicted an increase in precipitation and an increase in productivity in the
Sahel and Southern Saharan regions of Africa in this biome. However, there is not a
consensus in the literature with this finding as other studies have predicted an opposite
‘drying’ trend leading to a decrease in productivity (Hulme, Doherty, Ngara, New, &
Lister, 2001; Jenkins, Adamou, & Fongang, 2002). Regardless of which direction this
change takes, there is a general consensus that the desert and xeric shrubland biome is
highly sensitive to changes in precipitation patterns caused by climate change (Smith,
Monson, and Anderson 1998) and therefore the impacts of climate change.
Consequently, PAs located within these biomes could potentially experience the most
pronounced changes in their environmental conditions, and also in their utility for large
carnivore conservation (Weltzin et al. 2003).
Another particularly strong trend in primary productivity dynamics predicted was a
pronounced decrease in seasonality across most of Sub-Saharan Africa. For East and
West Africa, this could be a continuation of the same trend identified in the regions by
38
Pettorelli et al. (2012) for the time period 1982 – 2008. In addition, Southern Africa was
predicted to experience a reduction in productivity which is probably due to a decrease
in winter precipitation in the region (IPCC 2013). This ‘drying’ could also be causing the
decrease in seasonality, as the difference in precipitation between the winter and
summer could be reduced under these conditions.
Changes in home range (HR) size seem to be strongly correlated with changes in levels
of seasonality. A predicted increase in seasonality in South America led to a predicted
increase in HR size in the region, and a predicted decrease in seasonality in Africa led to
a predicted decrease in HR sizes in Africa. These trends show support for the theory that
HR requirements increase with increasing environmental variability (Mcloughlin et al.,
2000). However, increases in HR size were not correlated with a reduction in the
viability of a large carnivore population in a given PA.
This study predicted that changes in primary productivity dynamics driven by climate
change, even if drastic and causing changes in HR size, they would not impact a PA’s
ability to sustain viable populations of its large carnivores. These results mean that
climate change will not impact PA’s utility for large carnivore conservation. It also
suggests that factors, other than environmental conditions driving HR size, are the main
factors determining whether or not a PA is useful for large carnivore conservation.
However, it is important to note that the apparent lack of impact of climate change on
the utility of PAs may also be due to limitations in the methodology which will be
discussed in section 5.4.1.
5.2. What impacts PA’s utility for large carnivore conservation?
This study identified several important variables which impact the utility of PAs for
large carnivore conservation. The most influential of these was size of a PA. Larger PAs
are significantly more likely to be capable of sustaining their large carnivore populations
and therefore have a higher utility for large carnivore conservation. This is concerning
because over half of all PAs which occur within the large carnivore species’ ranges
considered in this study are under 1km2, with only 7.5% more than 100km2 (WDPA
2014). Considering that the largest bodied carnivores, such as brown bears, can require
home ranges as large as 8000km2 (McLoughlin et al. 1999) this is clearly a limiting
39
factor in PA’s utility to contribute to large carnivore conservation under current and
future climatic conditions.
The fact that PAs often fail to provide adequate area to support viable populations of
wide-ranging species, such as large carnivores, is not a new finding and has been
documented extensively in the scientific literature (Baeza & Estades 2010; Johnson et al.
2006; Mulongoy and Chape 2004; Simonetti & Mella, 1997; Newmark 1985, 1996; Soule,
Wilcox, & Holtby, 1979). In addition, Woodroffe and Ginsberg (1998) found that PA size
and extinction risk in carnivores is highly correlated, further demonstrating the
significance of PA size to determining their utility for carnivore conservation.
Another interesting finding from this study is an apparent latitudinal gradient of the
utility of PAs. PAs in lower latitudes are significantly more likely to sustain their large
carnivore species than those in higher latitudes. As latitude and seasonality are strongly
correlated, one hypothesis could be that PAs in higher latitudes are not as capable of
supporting their large carnivore populations because their seasonality, and hence their
environmental variability, is higher than that of PAs in lower latitudes.
Environmental variability could decrease the utility of PAs for large carnivore
conservation because it has been shown to have a positive relationship with home range
size in carnivores (Mcloughlin, Ferguson, and Messier 2000). There is an overall trend
that the more variable an environment, the larger home range an individual requires to
secure resources. Therefore, one hypothesis to explain this latitudinal gradient in PA
utility could be that the larger home range requirements of populations living in PAs
with higher seasonality are not being provided by PAs in those regions, leading to a
reduction in their utility for large carnivore conservation. In addition, Gompper and
Gittleman (1991) found that latitude and home range size in carnivores had a significant
positive relationship, providing further support for this hypothesis.
Another explanation of the latitudinal gradient of PA utility for large carnivore
conservation is that PAs in areas of higher latitude are significantly smaller than those at
lower latitudes. Particularly in Europe and North America, where PAs are significantly
smaller than PAs located in regions of lower latitude such as Sub-Saharan Africa and the
tropics of South America. Therefore, regardless of the impact of environmental
40
conditions on home range requirements, PAs in higher latitudes may be simply too small
to provide adequate space to sustain their large carnivore populations.
In reality, it is likely that a combination of these variables has lead to a latitudinal
gradient in PA utility for large carnivore conservation. PAs in areas of high latitude tend
to have higher levels of seasonality. Higher seasonality leads to larger home range
requirements. PAs in high latitudes tend to be smaller than PAs in lower latitudes. Due
to the increase in area requirements of large carnivores and the decrease in the area of
PAs at high latitudes compared to lower latitudes, they are less likely to be able to
sustain viable large carnivore populations, and therefore their utility is reduced.
5.3. Conservation Implications
According to the results of this study, the majority of PAs are not capable of sustaining
viable large carnivore populations and this is unlikely to change under future climate
scenarios. Therefore it can be assumed that they have little utility for large carnivore
conservation as a sole measure for their conservation. This means that PAs may not be
the most appropriate management strategy for large carnivore conservation, especially
in a changing climate. This is particularly true at higher latitudes in Europe and North
America where PAs are smallest and seasonality is relatively high.
Of the European PAs assessed in this study, none were predicted to be capable of
sustaining a large carnivore population. This was because, of the 19 PAs assessed, 18
were less than 1km2, and therefore not capable of supporting a single individual’s home
range requirements, let alone a viable population. This finding echoes Linnell et al.
(2001) who also found that no single PA in Scandinavia would be capable of exclusively
supporting a viable lynx population due to their large home range requirements and
lack of large PAs in suitable habitat.
The results of this study found a similar situation in North America. Only one PA was
predicted to be capable of sustaining viable populations of its large carnivore species,
and that PA was 905km2. The average PA size across all PAs which occur within the
range of the ten large carnivore species considered in this study for Europe is 14km2 and
for North America is 76km2. Considering that the average home range size predicted, of
41
an individual, from across all ten species considered in this study, was 86km2, it is
therefore unsurprising that the utility of PAs in these two continents is so limited.
Additonally, the largest PAs and therefore the PAs with the highest utility were in IUCN
management categories II and VI. Both these categories have high levels of human
activity. Category II PAs are National Parks (NP) and category VI allow resource
extraction. This means that although they provide the area large carnivores require they
may not actually be the most suitable areas for large carnivore conservation due to
potential negative impacts of human-carnivore conflict (Woodroffe & Ginsberg, 1998)
and the secondary impacts of the extractive industry (McLellan & Shackleton, 1988).
Although the current global PA network’s utility for large carnivore conservation is
limited, it could still contribute to the overall conservation effort for large carnivores.
Hannah et al. (2007) propose that the shortcomings in the extent of the global PA
network and their future utility for conservation under climate change could be secured
by using and expanding the current PA network. However, this does not seem like a
viable option in a world with an ever expanding human population with competition for
space and resources increasing every year. In addition, to secure viable populations of
large carnivores which would be resilient to the future impacts of climate change would
require a huge proportion of land to be set aside as PAs.
Another approach would be to focus on conserving large carnivores, not only in PAs, but
outside them too, in a multiuse landscape. Even in Africa and South America where PAs
are significantly larger than Europe and North America, and have comparatively higher
utility, they still only represent a tiny proportion of the available land which could be
used for large carnivore conservation.
The current global PA network could be useful for large carnivore in a number of ways.
PAs can provide area for a ‘core’ population with additional individuals living outside of
PAs (Linnell et al., 2001). Additionally, although the PAs in Europe and North America
are significantly smaller than those in Africa, Asia and South America, they are more
numerous (WDPA 2014). Therefore, in these areas a network of smaller PAs within a
landscape that is suitable for large carnivores, such as some forms of agricultural land,
may also contribute to large carnivore conservation.
42
Fundamentally, the current global PA network is not sufficient to exclusively sustain
viable populations of large carnivores, so the area outside of PAs must be the target for
large carnivore conservation. However, this involves many challenges, principally
concerned with habitat suitability and conflict with land-uses outside of PAs.
Baeza and Estades (2010) argue that the current utility of PAs for large carnivore
conservation can be improved, not by expanding the PA network as Hannah et al. (2007)
suggest, but by improving the matrix (the area outside of PAs) and targeting large
carnivore conservation in this multiuse landscape. This strategy may be particularly
applicable to areas of high latitude with high seasonality, such as Europe, because Baeza
and Estades also found that even small enhancements to the matrix under scenarios of
high environmental variability lead to significant improvements in the probability of a
species’ persistence inside a PA.
However, there are significant challenges to overcome to increase the utility of areas
outside of PAs for large carnivore conservation. Historically, people have not been
tolerant of large carnivores (Mech 1995; 1996) and human-induced mortality still
remains one of the key causes of adult mortality in large carnivores, even within PAs
(Woodroffe & Ginsberg, 1998).
The impact of human-carnivore conflict on the conservation of large carnivores cannot
be underestimated, as efforts for large carnivore conservation often rely directly on
public acceptance of large carnivores (Kaltenborn, Bjerke, and Vittersoslash 1999).
Therefore, it would be necessary to reduce, as much as possible, the opportunity for
human-carnivore conflict by focusing conservation efforts in areas outside PAs where
land use is not in direct conflict with large carnivore conservation.
5.4. Limitations
5.4.1. Primary productivity dynamics: observations and predictions
A potential limitation to this study is the use of the GIMMS NDVI time-series data used to
index primary productivity dynamics in PAs, as they are of a coarse resolution (8km2).
Many of the PAs assessed were smaller than the pixel size of these data and therefore
this questions the validity of their representativeness of PA-specific primary
43
productivity dynamics. This issue could be resolved by using finer-scale NDVI data (up
to 250m resolution) available from the Moderate Resolution Imaging Spectroradiometer
(MODIS) sensors (http://glcf.umd.edu/data/ndvi/). However, although MODIS data
would provide a more accurate measurement of primary productivity dynamics at
smaller scales, it is only available for a limited time period (2001 – 2006). Therefore, the
large time scale available from the GIMMS NDVI data (1981 – 2011) makes it a more
suitable choice for this study.
The linear models of productivity (mean iNDVI) and seasonality (NDVI contingency)
were able to explain a significant amount of variation in these two indices of
productivity, R2 = 0.564 and 0.41 respectively. The remaining unexplained variability
may be due to other non-climatic variables such as latitude or habitat type which were
not included in the modelling process. The potential limitation of the productivity
models is that both appear to predict inflated values for areas of very low productivity
and low seasonality, such as deserts. Therefore this shortcoming in the models could be
responsible for the trend predicted for PAs in these regions of a pronounced increase in
productivity and seasonality in these regions.
The models both used values of predicted precipitation and temperature from the
HadGEM2 climate model. However, these values were only available from WorldClim as
one-year monthly means for 20 year time periods. This led to a loss of important
information on variation in climate patterns, particularly important for predicting
seasonality. Region-specific climate models may provide more detailed, and more
accurate, predictions of variation in temperature and precipitation under climate change
scenarios but were not appropriate to use for a global level study.
A major limitation to this study is also the assumption that primary productivity equals
suitable habitat and resource availability for large carnivores. Spatial shifts in habitat or
prey species were not included when modelling future primary productivity dynamics
as this was beyond the scope of this study. However, this would be key in predicting PA’s
future utility for large carnivore conservation.
5.4.2. Home range predictions
44
The model used to predict population-level home range (HR) size developed by Duncan
et al (submitted) explained a large amount of the variation in HR size (R2 = 0.53).
However, it did predict a much narrower range of HR sizes for species compared to
observed data it was based on. This could be due to differences between study sites. The
PAs sampled for this study may not cover as large a variety of environmental conditions
as the observed data on home range size the model used, potentially causing the
reduction in HR size predictions.
On the other hand the source of variation in observed HR values and predicted may be
due to species-specific ecology which is not accounted for in the model. For example, the
predicted HR size of wolverines was significantly smaller than those observed in the
scientific literature. This is most likely because the wolverine has an unusually large HR
size in comparison to its body mass (Dawson, Magoun, Bowman, & Ray, 2010) and the
model uses body size linearly to predict HR size.
In addition, interactions between species can also have a significant impact on HR size
and were also not included in the model. For example, cheetah movements have been
shown to be particularly sensitive to lion activity (Durant, 2000). This avoidance
behaviour of other large carnivores is not unique to cheetah and can also be seen in
pumas and ocelots, which have been shown to adjust their movement to jaguars (Mario
S. Di Bitetti, De Angelo, Di Blanco, & Paviolo, 2010). Without the inclusion of these
interactions, HR size may have been underestimated.
The final potential limitation to this method of predicting PA-specific population sizes
through HR predictions is that the assumption that a HR only changes in size and not
location. In reality, changes in primary productivity dynamics in PAs may not impact the
size of HR but it may impact its location or shape, potentially shifting large carnivores
outside of PAs.
5.4.3. Assessing the viability of populations PAs are able to sustain
As with any population viability analysis (PVA) this PVA contains several potentially
subjective parameters which can limit its applicability to reality. These include the
threshold at which a population was considered extinct, length of simulation, starting
population size and carrying capacity. Starting population size was determined by
45
predicting how many female HRs a PA could contain assuming no overlap, therefore this
parameter was objectively set. The carrying capacity was based on average species-
specific population densities according to the scientific literature, and was therefore also
not subjective. The threshold at which the population was considered extinct, its quasi-
extinction, was chosen based on an assumption that a population of less than five
individuals would be highly vulnerable to demographic stochasticity. In addition, the
length of simulation (100 years) was chosen based on what previous studies have done
when carrying out PVAs (Boyce, 1992; Shaffer, 1981) and so could be argued is
somewhat arbitrarily chosen. Furthermore, the extinction threshold of ≤5% was chosen
based on what previous studies had set (Wielgus, 2002) and that it is also commonly
accepted as the scientific standard (p<0.05) for accepting significance.
This PVA included variation in survival rates due to environmental variation but not
fecundity. It assumed that all females bred every year and birthed the average number
of offspring each time. In reality, each year the number of females breeding and the
number of female offspring would vary. Another potential limitation of this study is that
the PVA assumed the same rate of sexual maturation of females for all species, which
was one year. For most species of large carnivore, females do not become sexually
mature until two or three years old and therefore this PVA has over-estimated
reproductive potential of populations.
This PVA also assumes that variation in vital rates due to environmental variability will
remain the same as currently observed variation when predicting viability under future
climate scenarios. However, under future climate scenarios most regions are predicted
to increase in variability so the variation in vital rates will also likely increase. The PVA
also did not include any kind of ‘environmental disaster’ such as a drought or flood
where vital rates would significantly drop in reality.
However, this does not mean that the results of the PVA are not valid or do not provide
useful insights into PA’s utility for large carnivore conservation under current and
future climate scenarios. It simply means that the extinction risk is conservative due to
over estimating the reproductive potential of populations. The results of this study
therefore paint a ‘best case scenario’ of whether or not a PA could sustain a viable
population of a given carnivore species if its reproductive output was at its maximum
capacity. Given that even under these circumstances, the PVA still predicted that the
46
majority of PAs were not capable of sustaining viable populations of their large
carnivore species under current or future climate scenarios provides further evidence of
their limited utility for large carnivore conservation.
5.5. Conclusions
These results provide clear evidence that at a global scale, the current PA network has
limited utility for large carnivore conservation as a sole measure to secure their
continued persistence under future climate change. The future direction of large
carnivore conservation should focus on developing novel approaches to using a multiuse
landscape, and locating areas where the opportunity for human-carnivore conflict can
be mitigated and the vast area that large carnivores require can be secured.
This research could be improved by repeating the analysis but assess only PAs which are
currently capable of sustaining viable populations and predicting whether or not this
will change under future climate scenarios. The vast majority of PAs are simply too small
to accommodate viable populations of large carnivores, by excluding these, the impact of
climate change on PA’s utility for large carnivore conservation may become significant.
Future areas of research could use similar techniques used in this study to predict ‘hot
spots’ of human-carnivore conflict under climate change, by predicting where known
large carnivore HRs may increase and come into contact with human settlements. This
may be particularly useful knowledge for the management of PAs with high densities of
human settlement on the edges of PAs as it would allow them to mitigate future threat of
human-carnivore conflict.
47
REFERENCES
Achard, F., & Estreguil, C. (1995). Forest Classification of Southeast Asia Using NOAA AVHRR Data, 208(June), 198–208.
Baeza, A., & Estades, C. F. (2010). Effect of the landscape context on the density and persistence of a predator population in a protected area subject to environmental variability. Biological Conservation, 143(1), 94–101.
Barton, K (2014) MuMIn: Multi-model inference. R package version 1.10.3.http://CRAN.R-project.org/package=MuMIN
Beissinger, S.R., Westphal, M.I., 1998. On the use of demographic models of population viability in
endangered species management. Journal of Wildlife Management 62, 821–841
Boyce, M. S. (1992). Population viability. Annual Review of Ecological Systems, 23, 481–506.
Bradshaw, W. E., & Holzapfel, C. M. (2006). Climate change. Evolutionary response to rapid climate change. Science (New York, N.Y.), 312(5779), 1477–8.
Brashares, B. J. S., Abrahms, B., Fiorella, K. J., Christopher, D., Hojnowski, C. E., Marsh, R. A., … Withey, L. (2010). underlying causes.
Brashares, J. S., Arcese, P., & Sam, M. K. (2001). Human demography and reserve size predict wildlife extinction in West Africa. Proceedings. Biological Sciences / The Royal Society, 268(1484), 2473–8.
Brook, B. W., O’Grady, J. J., Chapman, a P., Burgman, M. a, Akçakaya, H. R., & Frankham, R. (2000). Predictive accuracy of population viability analysis in conservation biology. Nature, 404(6776), 385–7.
Burt, W. H. (1943) Territoriality and home-range concepts as applied to mammals. Journal of Mammalogy, 24, 346-352.
Cantú-Salazar, L., Orme, C. D. L., Rasmussen, P. C., Blackburn, T. M., & Gaston, K. J. (2013). The performance of the global protected area system in capturing vertebrate geographic ranges. Biodiversity and Conservation, 22(4), 1033–1047.
Carbone, C., & Gittleman, J. L. (2002). A common rule for the scaling of carnivore density. Science (New York, N.Y.), 295(5563), 2273–6. doi:10.1126/science.1067994
Carbone, C., Mace, G. M., Roberts, S. C., & Macdonald, D. W. (1999). Energetic constraints on the diet of terrestrial carnivores Energetic constraints on the diet of terrestrial carnivores. Nature, 402, 1997–2000.
Cardillo, M., Mace, G. M., Jones, K. E., Bielby, J., Bininda-Emonds, O. R. P., Sechrest, W., … Purvis, A. (2005). Multiple causes of high extinction risk in large mammal species. Science (New York, N.Y.), 309(5738), 1239–41.
Cardillo, M., Purvis, A., Sechrest, W., Gittleman, J. L., Bielby, J., & Mace, G. M. (2004). Human population density and extinction risk in the world’s carnivores. PLoS Biology, 2(7), E197. Caro, T. M., & Doherty, G. O. (1999). On the Use of Surrogate Species in Conservation Biology. Conservation Biology, 13(4), 805–814.
Carroll, C., Noss, R. F., Paquet, P. C., & Schumaker, N. H. (2004). Extinction Debt of Protected Areas in Developing Landscapes. Conservation Biology, 18(4), 1110–1120.
48
Caughley, G., Sinclair, A.R.E., 1994. Wildlife Ecology and Management. Blackwell, Scientific Publications,
Boston
Ceballos, G., & Ehrlich, P. R. (2002). Mammal population losses and the extinction crisis. Science (New York,
N.Y.), 296(5569), 904–7.
Clarke, G. P., White, P. C. L., & Harris, S. (1998). Effects of roads on badger Meles meles populations in south-west England. Biological Conservation, 86, 117–124.
Collins, W. J., Bellouin, N., Doutriaux-Boucher, M., Gedney, N., Halloran, P., Hinton, T., … Woodward, S. (2011). Development and evaluation of an Earth-system model – HadGEM2. Geoscientific Model Development Discussions, 4(2), 997–1062.
Colwell, R. K. (1974) Predictability, constancy and contingency of periodic phenomena. Ecology, 55, 1148-1153.
Coulson, T., Mace, G. M., Hudson, E., & Possingham, H. (2001). The use and abuse of population viability analysis. Trends in Ecology & Evolution, 16(5), 219–221. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11301139
Dawson, F., Magoun, A., Bowman, J., & Ray, J. (2010). Wolverine , Gulo gulo , Home Range Size and Denning Habitat in Lowland Boreal Forest in Ontario. The Canadian Field Naturalist, 124, 139–145.
Di Bitetti, M. S., De Angelo, C. D., Di Blanco, Y. E., & Paviolo, A. (2010). Niche partitioning and species coexistence in a Neotropical felid assemblage. Acta Oecologica, 36(4), 403–412.
Di Bitetti, M. S., Paviolo, a., & De Angelo, C. (2006). Density, habitat use and activity patterns of ocelots (Leopardus pardalis) in the Atlantic Forest of Misiones, Argentina. Journal of Zoology, 270,
Doak, D. F. (1995). Source-Sink Models and the Problem of Habitat Degradation: General Models and Applications to the Yellowstone Grizzly. Conservation Biology, 9(6), 1370–1379.
Duncan, C., Nilsen, E., Linnel, J. & Pettorelli, N. (submitted) Resource dynamics and life history attributes determine home-range sizes in Carnivora. American Naturalist.
Durant, S. M. (2000). Living with the enemy: avoidance of hyenas and lions by cheetahs in the Serengeti. Behavioral Ecology, 11(6), 624–632.
Eisenberg, J.F., 1980. The density and biomass of tropical mammals. In: Soule, M., Wilcox, B.A. (Eds.),
Conservation Biology: An Evolutionary–Ecological Perspective Sinauer Associates, Sunderland,
Massachusetts.
English, A. K., Chauvenet, A. L. M., Safi, K., & Pettorelli, N. (2012). Reassessing the determinants of breeding
synchrony in ungulates. PloS One, 7(7), e41444.
Estes, J. A. (1996). Carnivores and ecosystem management. Wildlife Society Bulletin, 24, 390–396.
Foden, W. B., Butchart, S. H. M., Stuart, S. N., Vié, J.-C., Akçakaya, H. R., Angulo, A., … Mace, G. M. (2013). Identifying the world’s most climate change vulnerable species: a systematic trait-based assessment of all birds, amphibians and corals. PloS One, 8(6), e65427.
Garonna, I., Fazey, I., Brown, M. E., & Pettorelli, N. (2009). Rapid primary productivity changes in one of the last coastal rainforests: the case of Kahua, Solomon Islands. Environmental Conservation, 36(03), 253.
49
Gaston, K. J., Jackson, S. F., Cantú-Salazar, L., & Cruz-Piñón, G. (2008). The Ecological Performance of Protected Areas. Annual Review of Ecology, Evolution, and Systematics, 39(1), 93–113.
Gittleman, J. L., & Harvey, P. H. (1982). Carnivore home-range size, metabolic needs and ecology. Behavioral Ecology and Sociobiology, 10(1), 57–63.
Gompper, M. E., & Gittleman, J. L. (1991). Home range scaling: intraspecific and comparative trends. Oecologia, 87, 343–348.
Grosso, S., Parton, W., Stohlgren, T., Zheng, D., Bachelet, D., Prince, S., … Olson, R. (2008). Global potential net primary production predicted from vegetation class, precipitation, and temperature: comment. Ecology, 89(8), 2117–2126.
Hannah, L., Midgley, G., Andelman, S., Araújo, M., Hughes, G., Martinez-Meyer, E., … Williams, P. (2007). Protected area needs in a changing climate. Frontiers in Ecology and the Environment, 5(3), 131–138.
Herfindal, I., Linnell, J. D. C., Odden, J., Nilsen, E. B., & Andersen, R. (2005). Prey density, environmental productivity and home-range size in the Eurasian lynx (Lynx lynx). Journal of Zoology, 265(1), 63–71.
Hijmans, R., Cameron, S., Parra, J., Jones, P., & Jarvis, A. (2005). Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology, 25, 1965 – 1978.
Hoerling, M., Hurrell, J., Eischeid, J., & Phillips, A. (2006). Detection and Attribution of Twentieth-Century Northern and Southern African Rainfall Change. Journal of Climate, 19(16), 3989–4008.
Hooper, D., Adair, E. C., Cardinale, B. J., Byrnes, J. E. K., Hungate, B. a, Matulich, K. L ., … O’Connor, M. I. (2012). A global synthesis reveals biodiversity loss as a major driver of ecosystem change. Nature, 486(7401), 105–8.
Hulme, M., Doherty, R., Ngara, T., New, M., & Lister, D. (2001). African climate change : 1900 – 2100. Climate Research, 17, 145–168.
Intergovernmental Panel on Climate Change (IPCC) (2007) Climate change 2007: The physical science basis. Cambridge University Press, Cambridge, UK.
Intergovernmental Panel on Climate Change (IPCC) (2013) The Physical Science Basis. Contribution of
Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate
Change. Cambridge University Press, Cambridge, United Kingdom
International Union for Conservation of Nature (IUCN) Red List Spatial Data Available
www.iucnredlist.org Accessed [13th May 2014]
IUCN and UNEP (2014) World Database on Protected Areas (WDPA) UNEP-WCMC Cambridge UK
Available: www.protectedplanet.net Accessed: [8th June 2014]
Jenkins, G. S., Adamou, G., & Fongang, S. (2002). THE CHALLENGES OF MODELING CLIMATE VARIABILITY
AND CHANGE IN WEST AFRICA, 263–286.
Johnson, a., Vongkhamheng, C., Hedemark, M., & Saithongdam, T. (2006). Effects of human?carnivore conflict on tiger (Panthera tigris) and prey populations in Lao PDR. Animal Conservation, 9(4), 421–430.
Jones, K., Bielby, J., Cardillo, M., Fritz, S., Dell, J., Orme, D., & et al. (2009). PanTHERIA : a species-level database of life history, ecology, and geography of extant and recently extinct mammals. Ecology, 90(9), 2648.
50
Kaltenborn, B. P., Bjerke, T., & Vitters;oslash;, J. (1999). Attitudes toward large carnivores among sheep farmers, wildlife managers, and research biologists in Norway. Human Dimensions of Wildlife, 4(3), 57–73.
Karanth, K. U., Nichols, J. D., Kumar, N. S., Link, W. a, & Hines, J. E. (2004). Tigers and their prey: Predicting carnivore densities from prey abundance. Proceedings of the National Academy of Sciences of the United States of America, 101(14), 4854–8.
Kerley, L. L., Goodrich, J. M., Miquelle, D. G., Smirnov, E. N., Quigley, B., Hornocker, M. G., & Miqueitle, D. G. (2002). Effects of Roads and Human Disturbance on Amur Tigers. Conservation Biology, 16(1), 97–108.
Krebs, J., Lofroth, E. C., & Parfitt, I. (2007). Multiscale Habitat Use by Wolverines in British Columbia, Canada. Journal of Wildlife Management, 71(7), 2180.
Laidlaw, R. K. (2000). Effects of Habitat Disturbance and Protected Areas on Mammals of Peninsular Malaysia Effects Mammals of Habitat Disturbance Peninsular. Conservation Biology, 14(6), 1639–1648.
Linnell, J. D. C., Andersen, R., Kvam, T., Andren, H., Liberg, O., Odden, J., & Moa, P. F. (2001). Home Range Size and Choice of Management Strategy for Lynx in Scandinavia. Environmental Management, 27(6), 869–879.
Locke, H., & Dearden, P. (2005). Rethinking protected area categories and the new paradigm. Environmental Conservation, 32(1), 1–10.
McCain, C. M., & King, S. R. B. (2014). Body size and activity times mediate mammalian responses to climate change. Global Change Biology, 20(6), 1760–9.
McLellan, B., & Shackleton, D. M. (1988). Grizzly Bears and Resource-Extraction Industries: Effects of Roads on Behaviour, Habitat Use and Demography. Journal of Applied Ecology, 25(2), 451–460.
Mcloughlin, P. D., Ferguson, S. H., & Messier, Ë. O. I. S. (2000). Intraspecic variation in home range overlap with habitat quality : a comparison among brown bear populations. Evolutionary Ecology, 14, 39–60.
Mech, L. D. (1995). The Challenge and Opportunity of Recovering Wolf Populations. Conservation Biology, 9(2), 270–278. doi:10.1046/j.1523-1739.1995.9020270.x
Mech, L. D. (1996). Conservation strategies A new era for carnivore conservation, 24(3), 397–401.
Meijaard, E. (1999). Human-imposed threats to sun bears in Borneo. Ursus, 11, 185–192.
Miller, B., Dugelby, B., Foreman, D., Martinez, C., Noss, R., Phillips, M., … Willcox, L. (2001). The Importance of Large Carnivores to Healthy Ecosystems. Endangered Species UPDATE, 18(5), 202–210.
Mills, M. (1991). Conservation management of large carnivores in Africa. Koedoe, 34(1), 81–90.
Morrison, J. C., Sechrest, W., Dinerstein, E., Wilcove, D. S., & Lamoreux, J. F. (2007). Persistence of Large Mammal Faunas as Indicators of Global Human Impacts PERSISTENCE OF LARGE MAMMAL FAUNAS AS INDICATORS OF GLOBAL HUMAN IMPACTS, 88(6), 1363–1380.
Mulongoy, K. J., & Chape, S. (2004). Protected areas and biodiversity: an overview of key issues.
Newmark, W. D. (1985). Legal and biotic boundaries of western North American national parks: A problem of congruence. Biological Conservation, 33(3), 197–208.
51
Newmark, W. D. (1996). Parks and the Local Insularization of Tanzanian Extinction of Large Mammals. Conservation Biology, 10(6), 1549–1556.
Ngoprasert, D., Lynam, A. J., & Gale, G. a. (2007). Human disturbance affects habitat use and behaviour of Asiatic leopard Panthera pardus in Kaeng Krachan National Park, Thailand. Oryx, 41(03), 343–351.
Nilsen, E. B., Herfindal, I., & Linnell, J. D. C. (2005). Can intra-specific variation in carnivore home-range size be explained using remote-sensing estimates of environmental productivity? Ecoscience, 12(1), 68–75.
Nowak, R. (2005). Walker’s Carnivores of the World. The John Hopkins University Press.
Nowell, K., and P. Jackson. 1996. Wild cats: status survey and action plan. IUCN, Gland, Switzerland.
Olson, D. M., Dinerstein, E., Wikramanayake, E. D., Burgess, N. D., Powell, G. V. N., Underwood, E. C., … Kassem, K. R. (2001). Terrestrial Ecoregions of the World: A New Map of Life on Earth. BioScience, 51(11), 933.
Parmesan, C., Ryrholm, N., Stefanescu, C., Hill, J. K., Thomas, C. D., Descimon, H., … Warren, M. (1999). Poleward shifts in geographical ranges of butterfly species associated with regional warming. Nature, 399, 579–583.
Parmesan, C., & Yohe, G. (2003). A globally coherent fingerprint of climate change impacts across natural systems. Nature, 421(6918), 37–42.
Paviolo, A., Blanco, Y. E. Di, Angelo, C. D. De, & Bitetti, M. S. Di. (2009). Protection Affects the Abundance and Activity Patterns of Pumas in the Atlantic Forest PROTECTION AFFECTS THE ABUNDANCE AND ACTIVITY. Journal of Mammalogy, 90(4), 926–934.
Paviolo, A., De Angelo, C. D., Di Blanco, Y. E., & Di Bitetti, M. S. (2008). Jaguar Panthera onca population decline in the Upper Paraná Atlantic Forest of Argentina and Brazil. Oryx, 42(04), 554.
Peters RL and Myers JP. 1991. Preserving biodiversity in a changing climate. Issues Science and Technology
8: 66–72.
Pettorelli, N., Chauvenet, A. L. M., Duffy, J. P., Cornforth, W. a., Meillere, A., & Baillie, J. E. M. (2012). Tracking the effect of climate change on ecosystem functioning using protected areas: Africa as a case study. Ecological Indicators, 20, 269–276.
Pettorelli, N., Vik, J. O., Mysterud, A., Gaillard, J.-M., Tucker, C. J., & Stenseth, N. C. (2005). Using the satellite-derived NDVI to assess ecological responses to environmental change. Trends in Ecology & Evolution, 20(9), 503–10.
Possingham, H. P., Lindenmayer, D., & Norton, T. (1993). A framework for the improvement of threatened species based on population viability analysis. Pacific Conservation Biology, 1, 39–45.
Potter, C. S., & Brooks, V. (1998). Global analysis of empirical relations between annual climate and seasonality of NDVI. International Journal of Remote Sensing, 19(15), 2921–2948.
R Core Team (2014) R: A language and environment for statistical computing R foundation for statistical computing. http://R-project.org/
Rabinowitz, A. (1989). Density and behavior of large cats in dry tropical forest. Natural History Bulletin, 37(2), 235–251.
52
Rabinowitz, A., & Notthingham, B. (1986). Ecology and behaviour of the Jaguar (Panthera onca) in Belize, Central America. Journal of Zoology, 210, 149–159.
Reed, J. M., Mills, L. S., Jr, J. B. D., Menges, E. S., Kelvey, K. S. M. C., Frye, R., … Miller, P. (2002). Emerging Issues in Population Viability Analysis. Conservation Biology, 16(1), 7–19.
Reed, S. E., & Merenlender, A. M. (2008). Quiet, Nonconsumptive Recreation Reduces Protected Area Effectiveness. Conservation Letters, 1(3), 146–154.
Ripple, W. J., Estes, J. a, Beschta, R. L., Wilmers, C. C., Ritchie, E. G., Hebblewhite, M., … Wirsing, A. J. (2014). Status and ecological effects of the world’s largest carnivores. Science (New York, N.Y.), 343(6167), 1241484.
Rodrigues, A. S. L., Andelman, S. J., Bakarr, M. I., Boitani, L., Brooks, T. M., Cowling, R. M., … Animale, B. (2004). Effectiveness of the global protected area network in representing species diversity. Nature, 428(April), 9–12.
Root, T. L., Price, J. T., Hall, K. R., Schneider, S. H., Rosenzweig, C., & Pounds, J. A. (2003). Fingerprints of global warming on wild animals and plants. Nature, 421(6918), 57–60.
Sala, O. E. (2000). Global Biodiversity Scenarios for the Year 2100 . Science, 287(5459), 1770–1774.
Seddon, P. J., Griffiths, C., Soorae, P., & Armstrong, D. (2014). Reversing defaunation: Restoring species in a changing world. Science, 406(345), 406–412.
Shaffer, M. L. (1981). Minimum Population Sizes for Species Conservation. BioScience, 31(2), 131–134.
Simberloff, D. (1998). FLAGSHIPS , UMBRELLAS , AND KEYSTONES : IS SINGLE-SPECIES MANAGEMENT PASSI IN THE LANDSCAPE ERA ? Biodiversity and Conservation, 83(3), 247–257.
Simonetti, J. A., & Mella, J. E. (1997). Park size and the conservation of Chilean mammals. Revista Chilena de Historia Natural, (70), 213–220.
Singh, N. J., & Milner-Gulland, E. J. (2011). Conserving a moving target: planning protection for a migratory species as its distribution changes. Journal of Applied Ecology, 48(1), 35–46.
Smith, S., Monson, R., & Anderson, J. (1998). Physiological Ecology of North American Desert Plants. New Phytologist, 746–747.
Soule, M. E., Wilcox, B. A., & Holtby, C. (1979). BENIGN NEGLECT : A MODEL OF FAUNAL COLLAPSE IN THE GAME RESERVES OF EAST AFRICA. Biological Conservation, (15), 259–272.
Tanre, D., Holben, B. N., & Kaufman, Y. J. (1992). Atmospheric Correction Algorithm for NOM-AVHRR Products : Theory and Application, 30(2).
Thuiller, W. (2004). Patterns and uncertainties of species’ range shifts under climate change. Global Change Biology, 10(12), 2020–2027.
Tucker, C., Pinzon, J., Brown, M., Slayback, D., Pak, E., Mahoney, R., … El Saleous, N. (2005). An extended AVHRR 8-km NDVI dataset compatible with MODIS and SPOT vegetation NDVI data. International Journal of Remote Sensing, 26(20), 4485–4498.
Voigt, W., Perner, J., Davis, A., Eggers, T., Schumacher, J., Bahrmann, R., … Sander, F. (2003). Trophic levels are differentially sensitive to climate ¨. Ecology, 84(9), 2444–2453.
53
WallisDeVries, M. F. (1996). Effects of resource distribution patterns on ungulate foraging behaviour : a modelling approach. Forest Ecology and Management, 88, 167–177.
Weltzin, J. F., Loik, M. E., Schwinning, S., Williams, D. G., Fay, P. a., Haddad, B. M., … Zak, J. C. (2003). Assessing the Response of Terrestrial Ecosystems to Potential Changes in Precipitation. BioScience, 53(10), 941.
Whitehouse, J. F. (2001). IVth World Congress on National Parks and Protected Areas Caracas , Venezuela , February 1992. Australian Zoologist, 28(February 1992), 39–47.
Wielgus, R. B. (2002). Minimum viable population and reserve sizes for naturally regulated grizzly bears in British Columbia. Biological Conservation, 106(3), 381–388.
Woodroffe, R., & Ginsberg, J. R. (1998). Edge Effects and the Extinction of Populations Inside Protected Areas. American Association for the Advancement of Science, 280(5372), 2126–2128.
WorldClim (2014) HadGEM2-AO Available: www.worldclim.org [Accessed: 2nd July 2014]
54
APPENDICES
Appendix I. Linear mixed effects model used to predict carnivore home range size
Duncan et al (submitted) developed a linear mixed effects model (LME) of population
level home range (HR) in carnivores based on data collected on observed home range
size from the scientific literature over 35 years from 21 species (table 1).
Table 1. Species used to create carnivore HR LME.
Species Body Mass (kg) Diet
African wild dog 22.00 Carnivore
American marten 0.87 Carnivore
Black bear 110.50 Omnivore
Bobcat 6.37 Carnivore
Brown bear 196.29 Omnivore
Canadian lynx 9.68 Carnivore
Cougar 53.85 Carnivore
Coyote 11.99 Carnivore
Eurasian badger 11.88 Omnivore
Eurasian lynx 19.30 Carnivore
Fisher 3.75 Carnivore
Grey fox 3.83 Omnivore
Grey wolf 31.76 Carnivore
Leopard 52.40 Carnivore
Lion 158.62 Carnivore
Ocelot 11.88 Carnivore
Racoon 6.37 Omnivore
Red fox 4.82 Omnivore
Spotted hyena 63.37 Carnivore
Tiger 161.91 Carnivore
Wolverine 12.79 Carnivore
Home range (km2) was modelled as a function of: Sex * Group + Seasonality (NDVI
contingency) + Productivity (iNDVI) + log(BM) + Diet. * denotes an interaction between
55
two variables in the model. The model controlled for the random effect of study site and
contours as the observed data included minimum convex polygon (MCP) contours of
both 100 and 95%. Table 2 outlines the estimated impact of each variable included in
the model.
Table 2. Estimates for variables which influence population-level variation in home
range size in carnivores (log km2)
Parameter Estimate SE t-value p
Intercept 2.21 0.32 6.79 <0.01
Sex (M) 0.00 0.06 0.00 1.00
Group (Solitary) -0.47 0.11 -4.37 <0.01
Contingency 2.73 0.45 6.01 <0.01
iNDVI -0.09 0.02 -5.22 <0.01
log(Body Mass (kg)) 0.81 0.04 19.06 <0.01
Diet (Omnivore) -1.59 0.11 -13.94 <0.01
Sex (M) : Group (Solitary) 0.90 0.09 10.44 <0.01
56
Appendix II Survival rates used in PVA
Table 3. Survival rates taken from the literature used in PVA.
Species Adult Juvenile (0 – 1 year)
References Mean SD Mean SD
Ursus arctos 0.85 0.082 0.622 0.126
Fagan 2004; Harris et al. 2007; Hebblewhite et al 2003;
Kovach et al. 2006; Mace & Waller 1998; Sterling et al 2003
Panthera
tigris 0.74 0.095 0.61 0.150 Goodrich et al 2008; Karanth et al 2010;
Ursus
americanus 0.83 0.171 0.75 0.068 Clark & Eastridge 2006; Freedman et al 2006;
Ursus
thibetanus 0.85 0.055 0.82 0.053 Horino & Miura 2000;
Puma
concolor 0.72 0.095 0.57 0.065
Cunningham et al 2001; Spencer et al 2001; Logan 2001;
Jansen & Jenks 2012; Lambert et al 2006
Panthera
pardus 0.76 0.109 0.578 0.139 Bailey 2005; Balme et al 2009, 2010, 2012;
Acinonyx
jubatus 0.78 0.068 0.30 0.213
Durant 2004; Kelly & Durant 2000; Laurenson 2009;
Mills & Mills 2013; Oliver et al. 2011
Lynx lynx 0.82 0.178 0.41 0.095 Andren et al 2006; Boutros et al. 2007; Breitenmoser-Wursten 2007;
Gulo gulo 0.82 0.143 0.71 0.042 Krebs et al 2004; Persson et al 2003; Persson 2003
Leopardus
pardalis 0.76 0.168 0.71 0.057 Haines 2005, 2006; Laack et al 2006;