Urbanization and Its Effects on Lizards: a Study from a Climate Change Perspective

URBANIZATION

AND ITS EFFECTS ON LIZARDS:

A Study from A Climate Change Perspective

Prepared by:

Dr. Maria Thaker & Madhura Amdekar

Centre for Ecological Sciences, Indian Institute of Science Bangalore, India

FINAL REPORT Submitted to: Environmental Management & Policy Research Institute (EMPRI), Bangalore August 2016 FOREWORD AND ACKNOWLEDGEMENTS

This project was supported by funds from the Environmental Management &

Policy Research Institute (EMPRI), Bangalore, and was conducted in and around the city of Bangalore, Karnataka, India from 12 December 2015 to 30 August

2016. Part of the reported data and some results of this study may be included in the PhD thesis of Miss Madhura Amdekar, Centre for Ecological Sciences, IISc.

EMPRI will be duly acknowledged in the thesis and in any publications that arise from this project. Upon request by the Director General of EMPRI, Ms. Ritu

Kakkar, Appendix A, B and C were included in this report. Appendix A provides the raw data files, Appendix B is the manual and syntax for the software

MAXENT, which was utilized to conduct the niche distribution models, and

Appendix C is the tutorial for running linear models in R used to determine microhabitat preferences.

This project would not have been possible without the help of Mr. Shashank

Balaksrishna, (St Joseph's College, Bangalore), who primarily conducted the thermal biology experiments, and Mr. Abhijit Kumar Nageshkumar for help in taking and analyzing aerial images of the study grids. We would also like to thank Mr. Mihir Joshi, Ms. Anuradha Batabyal and Mr. Arka Pal for help in the collection of field data.

Dr. Maria Thaker

Centre for Ecological Sciences, IISc

TABLE OF CONTENTS

LIST OF TABLES 5

LIST OF FIGURES 6

EXECUTIVE SUMMARY 8

INTRODUCTION AND REVIEW OF KEY LITERATURE 10

Global Climate Change 10

Lizards and environmental temperature 13

Approach to this project 16

Bangalore and the Agamid lizards that live there 16

Peninsula Rock Agama, Psammophilus dorsalis 18

Common Garden Lizard, Calotes versicolor 21

Objectives 24

METHODOLOGY 26

Objective 1: Quantify distribution, abundance, and microhabitat preferences 26

Objective 2: Geographical variation in thermal limit and optimal body temperature 32

Objective 3: Changes in distribution patterns under climate change scenarios 35

RESULTS AND DISCUSSION 38

Objective 1: Quantify distribution, abundance, and microhabitat preferences 38

Objective 2: Geographical variation in thermal limit and optimal body temperature 50

Objective 3: Changes in distribution patterns under climate change scenarios 53

CONCLUSIONS AND RECOMMENDATIONS 59

REFERENCES 62

APPENDIX A 79

Presence, Abundance and Microhabitat Data for 79

Calotes versicolor and Psammophilus dorsalis 79

APPENDIX B 92

A Brief Tutorial on Maxent 92

APPENDIX C 144

Tutorial on running linear models in R 144

LIST OF TABLES

Table 1: Description of plot types……………………………………………………………26

Table 2: Parameters measured in each plot…………………………………………………..30

Table 3: Location coordinates of sampling grids…………………………………………….38

Table 4: Example of the proportion of different substrate types in a plot calculated from image processing……………………………………………………………………………..43

Table 5: Survey of city parks……………………………………………………………...... 45

LIST OF FIGURES

Figure 1: Male (a, b), Female (c) and juvenile (d) of the peninsular rock agama

Psammophilus dorsalis……………………………………………………………………………….19

Figure 2: Global distribution of P. dorsalis (Image Courtesy: IUCN)………………………20

Figure 3: Males of the Common garden lizard Calotes versicolor……………………….…..…22

Figure 4: Global distribution of C. versicolor (Image courtesy: Reptile Database)…………23

Figure 5: Area sampled for presence of P. dorsalis and C. versicolor …………………...…27

Figure 6: Rural landscape commonly seen around the city of Bangalore …………………...28

Figure 7: Representative images of urban areas sampled……………………………………29

Figure 8: Aerial images of the 20 x 20 sampling plots taken with a drone built by Sree Sai

Aerotech (Chennai, India)…………………………………………………………………....31

Figure 9: Thermal gradient setup in Dr. Maria Thaker’s lab to measure optimal body temperature of P. dorsalis……………………………………………………………………….…....34

Figure 10: Schematic work flow of the MaxEnt process …………………………………....36

Figure 11: Representative raw images taken using drone at different sampling locations…..41

Figure12: Raw (left) and processed (right) images for plot types A, B, and C………………42

Figure 13: Sites surveyed so far and sites where P. dorsalis was found present …....………44

Figure 14: Sites surveyed so far (blue dots) and sites where C. versicolor was found present

(green dots)………………………………………………………………………………..….44

Figure 15: Aerial view of Cubbon Park, Bangalore……………………………………….…45

Figure 16: Abundance of P. dorsalis increases with distance from city centre (R2=0.04, p=0.03)...... 46

Figure 17: Sex ratio (female / male) of P. dorsalis in unaffected by distance from city centre

(R2=0.01, p=0.5)……………………………………………………………………………...47

Figure 18: Abundance of P. dorsalis as a function of the proportion of area (400 sq. m) occupied by boulders. R2=0.14, p<0.001………………………………………………..…...48

Figure 19: Abundance of C.versicolor as a function of (a) average height of vegetation and of (b) soil compaction in a 400 sq. m area……………………………………………....49

Figure 20: (a) Critical maximum and (b) critical minimum temperatures of urban and rural males of P. dorsalis …………………...………………………………………………….……51

Figure 21: Body temperature of rural (R) and urban (U) lizards in a thermal gradient…..….52

Figure 22: (a) Current and (b) Future (2070) projected geographical distribution of P. dorsalis…………………………………………………………………………..…….54

Figure 23: (a) Current and (b) Future (2070) projected geographical distribution of C. versicolor………………………………………………………………………………………...55

Figure 24: Projection for maximum temperature of warmest month in 2070 ….…………....56

Figure 25: Annual average temperature in Bangalore ……...... 58

Figure 26: Monthly temperature in Bangalore as a function of distance from city centre ..…58

EXECUTIVE SUMMARY

As ectotherms, lizards are dependent on suitable environmental temperatures for critical physiological processes. Altered and unpredictable patterns of temperature as well as precipitation, caused by climate change, can adversely affect their distribution and survival.

The effects of climate change are compounded by urbanization, which leads to higher temperatures in the city as compared to less developed surrounding areas (i.e. urban heat island effect). In the present study, we extensively surveyed the city of Bangalore and its neighboring areas for the presence, distribution and microhabitat preferences of two agamid lizards,

Psammophilus dorsalis and Calotes versicolor in order to understand the impact of urbanization on these lizards. Using a combination of data on current distribution (measured in the field) and of physiological thermal limits (measured in the laboratory), we determined environmental niche requirements and projected future distribution patterns under prevalent climate models. Microhabitat preferences indicate that the presence of boulders, low vegetation height and suitable soil compaction were important determinants of high density of lizards.

Urban parks that are highly vegetated and lack large boulders and bare soil were not suitable habitats for these species. The temperature tolerance limits of lizards from within the city and outside differed but were still within the predicted environmental temperature range under climate change scenario. Hence at this point, rapid habitat destruction from urbanization is a more pressing concern for the survival of the two lizard species, compared to global warming.

A concerted effort to provide suitable microhabitats that are well connected will significantly improve the survival probability of these important mesocarnivore reptile species in and around the city of Bangalore, India.

Indian Rock Agama, Psammophilus dorsalis (courtesy: flickr)

INTRODUCTION AND REVIEW OF KEY LITERATURE

Global Climate Change

According to the IPCC (2007), earth has warmed by 0.6°C during the last century and continues to warm till date. Diel variation in temperature has declined considerably: not only are daytime temperature increases, so are nighttime temperatures. Seasonal variation has also declined because winter, and not just summer, temperatures are higher. Heat waves have become more frequent in the last decade.

The mechanistic link between accelerating emissions of greenhouse gases and recent patterns of environmental warming (IPCC 2007) is now well established. If emissions continue to accelerate at the current pace, the mean surface temperature of our planet might increase by as much as 6°C by 2100 (IPCC 2007). Such predictions into future decades includes obvious levels of uncertainty. But given the shockingly accurate prediction of the IPCC in 1990, where the predicted rate was 0.15–0.3°C per decade and actual observed rate was 0.2°C per decade, we should pay close attention to our current predictions for the near future.

Our current understanding and expectation for patterns of climate change at the global scale is fairly well established. Regional and local changes in temperature, however, are more complex and compounded by specific anthropogenic activities. Removal of vegetation and the building of concrete structures for urbanization increased local temperatures in both terrestrial and aquatic environments (Grimm et al. 2008; Nelson and Palmer 2007; Roth 2002). The elevation in air and surface temperatures in cities is known as the urban heat island effect, and seem to scale logarithmically with the population of a city (Oke 1973). In large cities around the world, urban warming matches or exceeds the rate of global warming. The spatial pattern of urban heat islands depends on a range of factors, from the type of building materials to the

10 location of parks and lakes. Urban warming and global warming are now combining to magnify potential impacts on organisms.

Human induced climate change is projected to be one of the leading causes of biodiversity loss in the near future (Harley 2011). In an extensive recent review, Bellard et al

(2012) examined the effects of climate change on a variety of taxa and found that current models project extinction rates for different animal taxa that could lead to the sixth mass extinction. There are four major trends that describe the effects of climate change on organisms. Organisms have been and are expected to continue to (1) shift in their phenological patterns, (2) shift in their geographic ranges, (3) be disruption in terms of their ecological interactions, and (4) change their primary productivity (see reviews by Hughes 2000; Parmesan

2006; Parmesan and Galbraith 2004; Walther et al. 2002).

To counter the effects of climate change, species can respond to environmental challenges by "shifting their climatic niche along three non-exclusive axes: time (e.g. phenology), space (e.g. range) and self (e.g. physiology).” (Bellard et al. 2012, see figure below). This project was designed to study the effects of climate change at three levels of

Biodiversity (see below): Organisms through Physiology, Populations through Dynamics, and Species through Distributions.

Summary of some of the predicted aspects of climate change and some examples of their likely effects on different levels of biodiversity. (Taken from Bellard et al. 2012. Ecology Letters.)

Lizards and environmental temperature

Environmental temperatures may vary tremendously and thus most organisms must regulate their body temperature through behavior, physiology, and morphology.

Thermoregulation enables an organism to maintain a specific mean or variance of body temperature using neural mechanisms to sense and respond to its environment (Bicego et al.

2007). Endotherms, such as birds and mammals, are capable of maintaining narrow body temperature ranges, but many ectothermic animals can also maintain precise body temperatures despite variation in environmental temperature (Angilletta 2009).

Ectotherms use an external heat source to regulate their body temperature and maintain critical physiological processes (Cowles and Bogert 1944). As ectotherms, lizards exchange heat and thermoregulate by various methods, including through radiation, conduction, convection and evaporation. The thermal attributes of the microhabitat, like air temperature, thermal radiation, wind velocity, body mass, body shape, metabolic heat production and thermal conductance all influence net heat exchange with the environment.

Thus, among the organisms that are expected to be negatively affected by climate change, reptiles are particularly vulnerable. For ectotherms, body temperature is one of the most important ecophysiological variable that can affect their performance. The biochemical and physiological processes, including those underlying behaviour, are directly affected by temperature since temperature affects kinetic energy. Body temperature can also impact different functions like movement, immunity, sensory input, foraging ability, and rate of feeding and growth. The relationship between body temperature and performance is typically in an inverse U-shape. The temperature at which the performance is maximal is called the thermal optimum. The range of temperatures over which performance is possible and is greater than or equal to some fixed level of performance is termed as the performance breadth. Critical

13 thermal limits are the minimum and maximum body temperatures that allow performance.

Although the critical thermal maximum may occur at different temperatures in different species, the behavioural response is the same across a diversity of taxa. Due to these reasons, critical maximum and minimum are a robust indices and standards for evaluating the thermal requirements and physiology of an organism.

Due to their dependence on a narrow range of temperature to carry out physiological activities, reptiles are constrained in space and time (Adolph 1990, Grant and Dunham 1988,

Hertz et al 1982). Thus, when local environmental temperatures exceed beyond the physiological limits of reptiles, biological functions are expected to shut down. The temperature-dependent physiological limit of ectotherms is the primary reason why the distribution patterns of many species worldwide are predicted to shift in most of the climate change projections (Sinervo et al 2010). Recent work has shown that increase in temperature by 2-3°C can limit physiological activities of lizards (Buckley 2008) and can lead to extinction of almost 40% of all lizard species by the year 2080 (Sinervo et al 2010).

Although it is commonly assumed that the effects of temperature rises will be negligible in tropical areas, this is a fallacy as climate change is projected to alter many other aspects of weather conditions, including precipitation patterns in tropical areas as well. The reproduction of many terrestrial reptiles in India is restricted to narrow windows of time (typically summer and pre-monsoon), when temperature and precipitation conditions are suitable for critical natural history activities such as foraging and mating. Altered and erratic weather conditions, which are projected for India, could lead to alteration of habitats and change in patterns of temperature and precipitation. These may cause adverse effects on animals such as habitat fragmentation, alteration of spatial distribution and species ranges, changes in community dynamics, altered sex ratio and thermal limits. All of this could result in the disruption of activity patterns and reproduction in reptiles.

To compound the effects of climate change, urbanization is impacting reptiles worldwide, especially when areas deemed suitable for urban development coincide with areas that support high native species richness. The urban heat island effect is expected to continually increase as cities grow, and is expected to negatively affect the survival of the most temperature sensitive species (Parris and Hazell 2005).

Previous studies in our lab have been aimed at examining the effects of local density and disturbance on lizard physiology and behaviour. In light of climate change and the urban heat island effect, these lizard species are at risk of local extirpation if we fail to understand their microhabitat requirements and physiological tolerances. Many lizards function at their physiological optimum at temperatures just below their critical tolerance limit, thus, even a small increment in habitat temperature may prove lethal (Ackley et al 2015). If lizards remain active at higher temperatures, it leads to an increase in their energy expenditure in order to maintain higher metabolic rates (Angilletta 2009). The need to maintain physiological processes within tolerance limits are compounded by the fact that these agama species feed on insects which also require suitable microhabitat conditions to sustain themselves. Therefore, if temperatures are too higher, lizards will require to feed more but may be limited by the availability of insects (Ackley 2015).

Approach to this project

We used an integrative ecological approach to study the distribution, abundance, habitat preferences, and temperature limits of two lizard species across the urban-rural gradient in

Bangalore. This involved extensively surveying the city of Bangalore and the surrounding rocky habitats to determine the distribution pattern of both species of lizard. In doing so, we determined the current microhabitat conditions that favour the presence of these species.

Quantifying thermal limits, which is the maximum and minimum temperature at which the species shows physiological and behavioural stress, as well as the preferred body temperature of lizards across the urban-rural gradient, provided us with critical information about whether and in what way lizards have adjusted to consistently different environmental temperatures.

The urban heat island effect predicts a gradient of temperature from high (on average, and in maximum and minimum) in Bangalore city to low in the surrounding rocky outcrops and rural areas. By studying the physiological limits of individuals across the current spatial gradient of environmental temperature, we can further predict the range of responses for these species under projected climate change scenarios. This space-for-time substitution can be a powerful approach to study and predict the effects of climate change on vulnerable species.

Bangalore and the Agamid lizards that live there

The city of Bangalore and the surrounding rocky habitats are home to multiple reptile species, including the peninsula rock agama, Psammophilus dorsalis, and the common garden lizard, Calotes versicolor. These species have a widespread distribution, occupying both human inhabited and un-inhabited areas. They are the largest lizard species in the area and thus are important mesopredators in the environment.

Agamid lizards are an Old World family of diurnal lizards that usually possess a crest along the centre of their back (Daniel 1983). These lizards are found in Africa, Europe, Asia and Australia. Agamids are phylogenetically related to the iguanas of the New World. Different species of this family are characterised by the type and arrangement of scales on their body.

They are found in a variety of habitats like forests, rocky outcrops and sandy areas. Most agamid lizards are insectivorous, but are also known to eat small mammals and birds whenever possible. There are many species in the family of Agamid lizards that live in India, and they occupy diverse biotopes (Daniel 1983).

Peninsula Rock Agama, Psammophilus dorsalis

The peninsular rock agama, Psammophilus dorsalis, is widely distributed in the semi- arid regions of Southern India, where they are found on rocky boulders and sheet rocks interspersed with scrub vegetation (Figure 2). They are usually found basking on rocks or wall ledges as perches for thermoregulation. These lizards have a characteristic large head, a flattened body, scales around the middle of the body and a fold in the skin or the throat. The males of this species are larger (snout to vent length range 107-145 mm) than the females (snout to vent length range 74-108 mm, Balakrishna and Thaker 2016). The males also develop bright colours during the breeding season (April-September) unlike the females (Figure 1). The females and the juveniles of this species are olive brown in colour, with dark brown and white coloured spots on the neck. Males of P. dorsalis show rapid physiological colour changes similar to chameleons, but have two distinct bands on their body, as opposed to irregular patches, that change during social interactions (Batabyal and Thaker, in review). These colour changes are specific to the social context. During encounters with females, the dorsal band of males physiologically changes from dull yellow to bright orange/red and the lateral band shifts from patchy brown to dark black. During encounters with other males, the dorsal band becomes bright yellow and the lateral band shifts to orange (Batabyal and Thaker, in review). Their diet mostly consists of ants (Balakrishna and Thaker 2016).

Research from our lab has already shown that P. dorsalis from urban and rural populations differ in multiple ways. In particular, lizards from rural areas consume more food and have more diverse prey types in their diet than lizards from urban areas (Balakrishna and

Thaker 2016). The colour changing ability (hue, chroma and brightness, as well as range of colour), and the social and antipredator behaviours of lizards from urban and rural areas are also markedly different (Batabyal and Thaker, in review).

a b

Figure 1: Male (a, b), Female (c) and juvenile (d) of the peninsular rock agama

Psammophilus dorsalis (Images courtesy: Anuradha Batabyal)

Figure 2: Global distribution of Psammophilus dorsalis (Image Courtesy: IUCN)

Common Garden Lizard, Calotes versicolor

Despite the name, the common garden lizard, Calotes versicolor, is not common in all gardens. Calotes versicolor is a medium sized lizard species (snout to vent length 105 mm, Ji et al 2002a) in which the males of the species are bigger than the females (Ji et al 2002b). These lizards can be identified by their medium size, laterally compressed body and two spines behind the tympanum on each side of their head. They also possess large, equal sized scales in the centre of their back and a crest of scales on their nape (Daniel 1983). They are brown in colour and have spots and bars on their dorsal side. The juveniles and females also possess two lateral stripes interspersed with patterns (Daniel 1983). Calotes versicolor is known to be present in a wide range of habitats including forests, plains and at an altitude of 2000m (Daniel 1983). They are found in most of the Indian subcontinent and South East Asia. These lizards mostly feed on insects, with ants forming a major component of their diet. Sometimes the diet may also include small birds, frogs and other small animals (Daniel 1983).

During the breeding season (April- September), males become conspicuously coloured while courting females. Certain body parts of the males, especially the head, shoulder and forelegs turn bright red or crimson. The males also defend territories from where they display head bobbing and push up behaviour (Daniel 1983).

Figure 3: Males of the Common garden lizard, Calotes versicolor

(Image (Image courtesy: Reptile Database) versicolor

alotes C istribution of igure igure 4: Global d F

Objectives

1) Quantify distribution, abundance, and microhabitat preferences of

Psammophilus dorsalis and Calotes versicolor in and around the city of

Bangalore, Karnataka

2) Measure geographical variation in thermal limits and optimal body temperature in Psammophilus dorsalis

3) Predict changes in species distribution patterns under prevailing climate change scenarios

Common Garden lizard, Calotes versicolor (Image Courtesy: wishwas anarchic)

METHODOLOGY

Objective 1: Quantify distribution, abundance, and microhabitat preferences

Between January-June 2016, we used a rectangular grid sampling method to quantify the microhabitat and survey the distribution and abundance of agamids. This survey was carried out over an area 60 km in radius from the centre of the city of Bangalore such that a gradient of urban (human inhabited and developed) to rural (undeveloped) areas was covered (Figure

5). Specific sampling locations were picked systematically such that each location was at least

15 km apart. At each of the grids, we selected 20 X 20m plots for opportunistic sampling. To ensure that we capture the variation in microhabitat types within each grid, we sampled three types of plots that are classified as follows (Table 1):

Table 1: Description of plot types

Plot type Description

A Maximum built up area

B Equal built up area and vegetated/barren area

C Maximum area vegetated/barren

This sampling method enabled us to cover several environmental gradients: 1) at broad scale gradient from urban to rural (Figure 6 and 7), and a fine scaled gradient from developed to undeveloped, with variable levels of connectivity. We also conducted targeted sampling of prominent city parks in Bangalore that vary in size and location within the city.

Figure 5: Area sampled for presence of Psammophilus dorsalis and Calotes versicolor. Inner and outer red circles indicate an area that is 30 km and 60 km in radius from Bangalore city centre. Yellow pins indicate sampling locations (Map

Source: Google Earth).

Figure 6: Rural landscape commonly seen around the city of Bangalore. These rocky outcrops are the preferred and natural habitat for Psammophilus dorsalis

(Images courtesy: Anuradha Batabyal)

Figure 7: Representative images of urban areas sampled. Small pockets of rocky and vegetated areas in the city are often home to Psammophilus dorsalis (Images courtesy: Anuradha Batabyal)

At each of the plots, the following parameters were measured:

Table 2: Parameters measured in each plot

Parameter measured Method Description

Thermal gun was employed to measure the

environmental temperature around the Ambient temperature Thermal gun lizard

The presence and demography of lizards

(number of males, females and juveniles) Density of lizards Visual estimation was visually counted.

Wherever lizards were found, the height of

the perch occupied was measured using a Perch height Measuring tape measuring tape.

Wherever lizards were found, the

temperature of the perch occupied was Substrate temperature Thermal gun measured using a thermal gun.

The proportion of each substrate type was

quantified from digital analysis of photos Visual estimation, taken using a camera on a drone. Proportion of each digital Wherever not possible, the proportions substrate type quantification from were estimated visually by two aerial image independent observers.

Figure 8: Aerial images of the 20 x 20 sampling plots (below) were taken with a drone (above) built by Sree Sai Aerotech (Chennai, India)

Objective 2: Geographical variation in thermal limit and optimal body temperature

In the first half of June, a subset of P. dorsalis lizards from urban (from inside the city of Bangalore) and rural areas (Kolar district), which represented the two extreme environments along the urban-rural gradient, were brought to the laboratory. We measured their maximum and minimum temperature limits, also termed as critical maximum and critical minimum temperature respectively. There are multiple protocols that are typically utilized to measure critical maximum temperature of ectotherms. The two major methods to study thermal tolerance are the static method and the dynamic method. In the static method, the time to death of an organism is measured at a constant test temperature while in the dynamic method, test temperatures are increased until an end point is reached. The static method uses pharmacology techniques, where the equivalent of an LD50 from time is determined, wherein mortality curves are generated such that the "dosage" is the time during which animals are exposed to a constant test temperature until death (Fry et al. 1942). Median lethal high and low temperatures plotted against acclimation temperatures for a species form a polygon that delimits the zone of resistance from the zone of tolerance.

The dynamic method represented by concepts of the critical maximum and critical minimum temperature was introduced by Cowles and Bogert (1944). The critical maximum or minimum temperature is defined as the “thermal point at which locomotory activity becomes disorganized and the animal loses its ability to escape from conditions that will promptly lead to its death”. As an animal is subjected to increasingly high temperatures, it typically displays certain responses like loss of the righting response, the sudden commencement of muscular spasms and finally heat rigor, coma or death. Often, both static and dynamic methods are combined to study thermal tolerance.

We employed the dynamic method of studying critical maximum and critical minimum temperature to prevent any lethal loss of individuals during our study. We considered the failure to show righting response as criterion for determining these limits. Wild-caught lizards were housed in individual glass tanks (60 X 30 X 25 cm) lined with paper towel substrate. These tanks have basking lights and rocks where the lizards were fed crickets every alternate day and provided with water ad libitum.

For measurement of thermal limits, heated and cooled water baths were used. Lizards were placed in containers that were partially immersed in water baths. The temperature of the water bath was increased or decreased in small increments gradually. To check the righting response of the lizard, it was flipped on its back. If the lizard could right itself, it was placed back in the water bath and the procedure was repeated. The temperature at which lizards fail to show a righting response when considered the critical temperature limit. The body temperature at the critical point was measured by inserting a thermometer in the lizard’s cloaca. This protocol has been modified from Ackley et al 2015.

Preferred or optimal body temperatures were measured for lizards in a thermal gradient in the laboratory. This thermal gradient was built using a metal frame and Plexiglas and lined with sandy substrate. The gradient was created by placing two 60W and one 120W incandescent bulbs at one end of the setup while a frozen ice pack was placed on the other end.

The temperature in the gradient ranged from 21°C to 46°C (Balakrishna and Thaker unpublished). An individual lizard was placed in a cloth bag and introduced into the gradient, where it was allowed to move freely. The cloacal temperature of the lizard was taken every 20 minutes for an hour. All lizards were returned to site of capture within 1 day to minimize stress.

Figure 9: Thermal gradient setup in the lab of Dr. Maria Thaker to measure optimal body temperature of P. dorsalis (Images courtesy: Shashank Balakrishna)

Objective 3: Changes in distribution patterns under climate change scenarios

Using the data on microhabitat preferences and thermal limits obtained above, we first estimated and cross referenced the current geographical distribution for both species using the

Maximum Entropy (MaxEnt) ecological niche model. MaxEnt is a free modelling software that utilizes two sets of information: geographic coordinates of areas where the species has been found to be present and values of different environmental variables, such as maximum temperature, minimum temperature, precipitation, and elevation etc. across a wider geographic region of interest (Phillips et al 2004). MaxEnt also extracts a sample of background locations that it contrasts against the presence locations. Based on these information, it generates a predictive distribution map for the wider geographical area. It does so using the two principles of constraint and entropy. The presence locations and the value of environmental variables at those locations are constraints on where the species can be found and will determine the probable distributions of the species. Maximizing entropy is one way to choose among the many possible species distributions (Phillips et al 2004).

WorldClim Data on environmental variables annual mean temperature max temperature of warmest month min temperature of coldest month annual precipitation

Predictive distribution map Species Longitude Latitude P dorsalis x1 y1

C. versicolor x2 y2

Figure 10: Schematic work flow of the MaxEnt process

For this study, the presence locations of P. dorsalis and C. versicolor, obtained as part of Objective I, were entered into MaxEnt as geographical coordinates of the presence locations of our species of interest. The environmental layers were obtained from WorldClim, an international online repository of world climatic data. This repository stores the annual mean, maximum and minimum temperature and precipitation, along with 19 derived bioclimatic variables. This data is produced through interpolation of climatic data generated at a scale of 1 sq. km by weather stations the world over (Hijmans et al 2005).

As the sampling locations in this study lay within nine districts of Karnataka (Bangalore urban, Bangalore rural, Chamrajnagar, Chikkaballapur, Kolar, Mandya, Mysore, Ramanagara and Tumkuru), the environmental variable layers were clipped to the same geographical extent.

The layers used were annual mean temperature, maximum temperature of the warmest month, minimum temperature of the coldest month and annual precipitation. A datasheet containing presence locations of lizards and the clipped environmental layers were provided as input to

MaxEnt which generated predictive distribution maps for both the species.

Projected climatic scenarios

We then also projected climatic scenarios for the year 2070 using the climate projection model BCC-CSM1-1 to estimate future potential distributions in the same geographical area.

The Beijing Climate Center Climates System Model (BCC-CSM1-1) has been developed by the Beijing Climate Center (BCC) and takes into account the interaction of energy, water, momentum and carbon cycles (Wu et al 2012a). It can incorporate the sources of carbon emissions caused due to human activity, thereby stimulating the human induced effects of climate change. Under different emission scenarios, air temperature has been projected to increase continuously while precipitation is projected to increase intermittently. The BCC CSM model predicts a global increase in average temperature from 2.2°C to 4°C (Global Terrestrial

Network for Permafrost). Studies have found that the future projections from BCC CSM model are in accordance with those from other climate models (Zhang et al 2012).

RESULTS AND DISCUSSION

Objective 1: Quantify distribution, abundance, and microhabitat preferences

Aerial photographs of 44 grids, each with plot types A, B and C have been taken. Hence a total of 132 plots were photographed for digital analysis to estimate vegetation composition and proportion occupied (Table 3). At the other plots, it was not feasible to take aerial images; hence the proportions of substrate types in these plots were estimated visually. For a subset of plots, we corroborated the visual estimation method against the digital analysis of aerial photographs.

Digital analysis for calculating the proportions of different substrates present in the plots were carried out using ArcGIS software. Briefly, each pixel in the image or a polygon of such pixels was assigned to one of the substrate types (see example below), followed by calculating the percentage of area occupied by each substrate in the whole image. The substrate types considered included forbs or grass (vegetation), bare soil, concrete (constructed area), trees, small rocks, boulders and sheet rocks.

Table 3: Location coordinates of sampling grids

Site/Grid Coordinates Site/Grid Coordinates

1.1 13.08282, 77.58644 28.3 13.200955,77.31921

1.2 13.08296, 77.58654 29.1 13.176085,77.32226

1.3 13.08337, 77.58679 29.2 13.200955,77.31921

2.1 13.21361°, 77.58361° 29.3 13.204802,77.315863

2.2 13.21429, 77.58385 30.1 13.110028,77.329431

2.3 13.2142, 77.58564 30.2 13.110317,77.332747

3.1 13.34556, 77.5825 30.3 13.104512,77.327913

3.2 13.34476, 77.59935 31.1 13.188287,77.227783

4.2 13.43963, 77.61509 31.2 13.319767,77.3123

4.3 13.43366,77.62401 31.3 13.346063,77.32532

5.1 12.81061, 77.58659 32.1 13.335795,77.424723

5.2 12.81137, 77.58672 32.2 13.33589,77.425762

5.3 12.81218, 77.58768 32.3 13.331388,77.423549

6.1 12.67511, 77.56134 33.1 13.496269,77.507604

6.2 12.67465, 77.56087 33.2 13.491923,77.50662

6.3 12.67538, 77.56131 35.1 13.079507,77.447574

9.1 12.95526, 77.71507 35.2 13.079656,77.447675

9.2 12.95435, 77.71663 35.3 13.079274,77.447463

9.3 12.95455, 77.71299 36.1 13.079305,77.713128

10.1 12.938535,77.840695 36.2 13.079071,77.713189

10.2 12.937045,77.841167 36.3 13.079265,77.712989

10.3 12.938358,77.837516 37.1 12.806487,77.706953

11.1 12.955665,77.979248 37.2 12.807676,77.707756

11.2 12.955939,77.979414 37.3 12.807647,77.707854

11.3 12.955423,77.977258 38.1 12.693141,77.7104

12.1 12.947428,78.110628 38.2 12.692645,77.710649

12.2 12.952214,78.110992 38.3 12.692688,77.710855

12.3 12.953256, 78.111922 39.1 13.087806, 77.854386

17.1 12.812419, 77.182002 39.2 13.087208, 77.853931

17.2 12.812433, 77.181818 39.3 13.087889, 77.853764

19.2 12.808668, 77.316032 40.1 13.213912,77.868128

19.3 12.804836, 77.314670 40.2 13.214111,77.867008

20.1 12.692191,77.302414 40.3 13.214389,77.867079

20.2 12.691511,77.302093 43.1 13.232778,77.949605

20.3 12.685656,77.309401 43.2 13.227783,77.960727

21.1 12.546905,77.311796 43.3 13.22787,77.962089

21.2 12.543932,77.315441 44.1 13.077083, 77.987228

21.3 12.547532,77.310607 44.2 13.07595, 77.987042

22.1 12.792523,77.434887 44.3 13.076131, 77.987061

22.2 12.792523,77.434887 46.1 13.082606, 77.116936

22.3 12.79397,77.438364 46.2 13.084533, 77.118044

23.1 12.704992,77.450395 46.3 13.091086, 77.119117

23.2 12.703277,77.451906 47.1 13.212873,77.736728

23.3 12.703059,77.451677 47.2 13.215183,77.735437

24.1 12.550705,77.447825 47.3 13.214807,77.734331

24.2 12.550467,77.4477 48.1 13.340206,77.718825

24.3 12.550117,77.450333 48.2 13.340348,77.718327

25.1 12.407024,77.450229 48.3 13.339464,77.717845

25.2 12.407024,77.450229 47.1 13.212873,77.736728

25.3 12.407573,77.449701 47.2 13.215183,77.735437

26.1 13.068956,77.057403 47.3 13.214807,77.734331

26.2 13.069071,77.057523 48.1 13.340206,77.718825

26.3 13.068874,77.054357 48.2 13.340348,77.718327

27.1 13.082259,77.173637 48.3 13.339464,77.717845

27.2 13.082791,77.174707 28.1 13.209769,77.167757

27.3 13.088047,77.179598 28.2 13.207455,77.173277

Figure 11: Representative raw images taken using a drone at different sampling locations

Figure 12. Raw (left) and processed (right) images for plot types A, B, and C

Table 4: Example of the proportional differences in substrate types in one plot calculated from aerial image processing

Forbs / Bare Small Sheet Site Trees Concrete Boulders Coordinates grass Soil Rocks Rock

12.938535, 10.1 75.00 0.00 0.00 25.00 0.00 0.00 0.00 77.840695

12.937045, 10.2 5.00 0.00 29.00 64.00 2.00 0.00 0.00 77.841167

12.938358, 10.3 17.00 0.00 34.00 49.00 0.00 0.00 0.00 77.837516

Lizard abundance and demography was quantified for all the plots. Psammophilus dorsalis was found present in 19 of these plots (Figure 13), while C. versicolor was found present in 14 plots

(Figure 14). Psammophilus dorsalis (N = 111 total) was a far more common species in the

Greater Bangalore region than Calotes versicolor (N = 16 total), which is expected given the field guide descriptions of the habitats that are preferred for these species. Psammophilus dorsalis is known to prefer rocky habitats in semi-arid regions, while Calotes versicolor prefers forests and plains.

Figure 13: Sites surveyed in 2016 (blue dots) and sites where P. dorsalis was found present (yellow dots).

Figure 14: Sites surveyed in 2016 (blue dots) and sites where C. versicolor was found present (green dots).

Given the constraints for time and funding, and the fact that the surveys for lizards had to be completed within the peak breeding season and activity period, only five parks in the city were surveyed to determine lizard presence (Table 5). Presence of lizards were extremely low in these highly vegetated parks. Results from park surveys should be considered preliminary at this stage.

Table 5: Survey of city parks

Lizard Lizard S.No. Park Name Longitude Latitude Abundance Abundance (2015) (2016) 1 Lalbagh 12°56’26’’ 77°34’27’’ 0 0 2 Cubbon park 12°34’49’’ 77°21’5’’ 0 0 3 Kodigehalli 13 °04’12’’ 77 °35’02’’ -- 0 4 Mathikere 13 °01’56’’ 77 °35’52’’ -- 0 5 Sahakara Nagar 13 °03’38’’ 77 °35’21’’ -- 2

Figure 15: Aerial view of Cubbon Park, Bangalore, showing extensive vegetation and a lack of large boulders (Image source: Google Earth)

Abundance as a function of urbanization

We first determined whether abundance of agamid lizards species was influenced by urbanization, as measured by the distance from the city centre. This analysis was only conducted with P. dorsalis as densities of this species were high enough to statistically analyse. We find that the abundance of P. dorsalis increases with distance from city centre, such that there are more lizards farther from the core of Bangalore (Figure 16). Sex ratio of this species does not seem to be affected by distance from city centre (Figure 17).

Figure 16: Abundance of P. dorsalis increases with distance from city centre

(R2=0.04, p=0.03).

Figure 17: Sex ratio (female / male) of P. dorsalis in unaffected by distance from city centre (R2=0.01, p=0.5).

Microhabitat preferences

To understand the microhabitat preferences of the agamid lizards, we used a general linear model to calculate the relative importance of each abiotic predictor in determining lizard presence in the microhabitat. A linear model is a statistical model that can be employed when the independent variables of our interest (e.g. vegetation height, proportion of substrate etc. in this case) follow a normal distribution. The general form of a linear model can be represented as: � = �� + �; where y is the dependent variable, x is the independent variables, a is the parameter associated with x and U is the error term.

The abundance of lizards in a single plot was considered as the response variable or y and the various parameters measured in each plot were considered as the independent predictors of lizard abundance. As the data was normally distributed, the error term was

Gaussian.

The following abiotic predictors were used:

1. Proportion of the area of plot comprising of:

a. Vegetation

b. Bare soil

c. Trees

d. Boulders

e. Small rocks/sheet rocks

2. Soil compaction

3. Average vegetation height

Separate linear models were constructed to explain the presence of P. dorsalis and C. versicolor respectively. All the analysis was carried out using the statistical platform R (version

0.99.489). In case of P. dorsalis, the presence of boulders was the most important predictor of lizard presence (Figure 18).

Figure 18: Abundance of P. dorsalis as a function of the proportion of area (400 sq. m) occupied by boulders (R2=0.14, p<0.001).

For C. versicolor, height of the vegetation present and soil compaction are more important for their presence (Figure 17).

Figure 19: Abundance of C. versicolor as a function of (a) average height of vegetation (R2=0.06, p=0.01) and of (b) soil compaction (R2=0.05, p=0.01) in a

400 sq. m area.

Objective 2: Geographical variation in thermal limit and optimal body temperature

The critical maximum temperature was attained by lizards under environmental temperature of

◦ 40 - 50 C while the critical minimum temperature was attained under environmental

◦ temperature of -5 - 5 C.

We find that P. dorsalis from urban areas have a higher critical maximum and minimum temperature than lizards from rural area (Figure 20). This indicates that urban lizards may not be able to function at lower temperatures like the rural lizards, but are able to function at higher temperatures than rural lizards. Consequently there is a shift in the range of tolerance limits levels of temperature between urban and rural lizards.

Figure 20: (a) Critical maximum and (b) critical minimum temperatures of rural a P. dorsalis indicate a shift in the range of tolerance levels in urban males compared to rural males

Preferred or optimal body temperatures were also measured for lizards in a thermal gradient in the laboratory. When placed in a thermal gradient, lizards from both urban and rural areas attain similar optimal body temperatures (Figure 21). However, urban lizards show a sharper rate of increase in their cloacal temperature as their initial cloacal temperature is lower than that of rural lizards.

38 C)

◦ 36

30 R_female R_male 28 U_female

Optimal bbody temperature ( 26 U_male 24 0 20 40 60 Time intervals (min)

Figure 21: Body temperature of rural (R) and urban (U) lizards in a thermal gradient. Note that when provided with a range of environmental temperatures,

Optimal Body Temperature was obtained within 20 min and ranged from 34.5 to

35.5°C.

Objective 3: Changes in distribution patterns under climate change scenarios

Using the data on lizard presence and location, as well as environmental variables including temperature and precipitation, the current geographical distribution of both the species for nine districts of Karnataka was estimated. These districts (Bangalore urban, Bangalore rural,

Chamrajnagar, Chikkaballapur, Kolar, Mandya, Mysore, Ramanagara and Tumkuru) include the sampling locations as well as area surrounding the sampling locations. Environmental variables used to generate the distribution models were annual mean temperature, maximum temperature of the warmest month, minimum temperature of the coldest month and annual precipitation. We then also projected climatic scenarios for the year 2070 to estimate future potential distributions in the same geographical area.

(a)

(b)

Figure 22: (a) Current and (b) Future (2070) projected geographical distribution of P. dorsalis

(a)

(b)

Figure 23: (a) Current and (b) Future (2070) projected geographical distribution of C. versicolor

Projected environmental temperature conditions in 2070 are expected to lie within the critical thermal limits for these agama species, despite the prediction that most of the area in our geographical region of interest is expected to experience consistently high temperatures (Figure

24). These temperatures are greater than the critical limit of the rural lizards and hence may shift their tolerance levels as can be seen in the case of urban lizards.

Figure 24: Projection for maximum temperature of warmest month in the year

◦ ◦ 2070. The temperature varies from 26.9 C to 38.3 C.

Although all lizards use behavioural thermoregulation to modulate their body temperatures, consistently high or low temperature conditions can still be beyond the capacity of thermoregulatory strategies. Temperatures are projected to rise steadily in the coming years

(Figure 25) and this coupled with the fact that temperatures inside the city are higher than the surroundings (Figure 26) can lead to decrease in the extent of temperature tolerance by lizards

(as seen from the results of Objective 2). Such temperature shifts can affect the presence, abundance and distribution of agama lizards.

Typical distribution models have one major drawback in that they are forced to utilize abiotic data that are often mismatched in scale from the information of animal presence. We utilized temperature data from WorldClim which it at a scale of 1 sq km, while our lizard location information is at a far finer spatial resolution. Furthermore, at this stage, we have assumed no further development of urbanized areas in our predictive model. This is the first survey of P. dorsalis and C. versicolor to determine distribution and survival under climate change. In the next phase (2016-2017), we aim to obtain finer-scaled temperature data and projections for our area of interest (Karnataka state), include urban growth predictions, and conduct a second survey in the following activity/breeding season at all sites. These additional data layers would provide a more accurate prediction about lizard distribution patterns in the future.

Figure 25: Annual average temperature in Bangalore, Karnataka. Courtesy:

Goddard Institute for Space Studies

24 C) ° 22

18 Temperature (

5 Kms 7 Kms 10 Kms 14 Kms 21 Kms 24 Kms

Distance from city centre

Figure 26: Monthly temperature in Bangalore as a function of distance from city centre

CONCLUSIONS AND RECOMMENDATIONS

Agamid lizards living in urban environments face additional challenges to survival that are beyond what is typical seen for species in undisturbed environments. The city of Bangalore supports two agamid species, the rock agama, P. dorsalis, and the eastern garden lizard C. versicolor. This project was designed to study the effects of urbanization and climate change at three levels of Biodiversity:

Organisms through Physiology: The critical temperature limits, and optimal body temperatures for urban and rural lizards of P. dorsalis were measured under controlled laboratory conditions. We found that the temperature tolerances for urban and rural populations were different, indicating that urbanization itself and the urban heat island effect have altered the physiological tolerance boundaries for populations of the species.

Temperature tolerance levels of the more common agamid species (P. dorsalis), however, seem to be within the environmental range predicted under current global warming climate change scenarios. Existing climate predictions for India and at the smaller scale, Karnataka, are very coarse and rely entirely on interpolated information from global weather stations. Thus, our prediction for environmental temperature conditions in this region in the future is unreliable.

We therefore caution against concluding that the agamid species, P dorsalis and C. versicolor that live in and around the city of Bangalore will be resistant to climate change-induced global warming. A combination of fine-scaled environmental monitoring at regional and national levels are necessary in order to generate climate projections, which can then be used to predict the persistence of thermally sensitive species.

Populations through Dynamics: Extensive surveys across the urban-rural gradient from from Bangalore city center to a 60 km radius into undeveloped habitat were conducted to target the location and population size of agamid lizards. The two agamid species that live in this region differ dramatically in abundance: P dorsalis are more common than C. versicolor.

As expected, abundance of P. dorsalis was also directly influenced by urbanization as the number of lizards increased with distance from city centre. No urbanization-related pattern was detected for C. versicolor given that they were generally rare. Sex ratio of P. dorsalis was not affected by urbanization, however repeated surveys using the same methods are necessary to determine whether the patterns of abundance and demography are consistence across space and time. Thus, the effects of urbanization and climate change on population dynamics is still to be determined with longitudinal data following the baseline data obtained here.

Species through Distributions: Extensive surveys across the urban-rural gradient in

Bangalore revealed key microhabitat preferences for the two agamid species. Although there are species specific requirements, the highest densities of both species are found in areas with boulders, low vegetation height, and adequate soil compaction. These microhabitat requirements are consistent with the basic ecology for agamid lizards as they provide defendable perch sites, vegetation for cover and support for insect prey, and soil for egg laying.

City parks that are highly vegetated and lack large boulders and bare soil are therefore not suitable habitats for these species.

OVERALL CONCLUSION

Biodiversity is being negatively affected worldwide by climate change and urbanization. While most studies have focused on studying the impact of urbanization on taxa such as birds and butterflies, our study demonstrates the effects of urbanization on agamid lizards. The survival of the two largest agamid species in and around the city of Bangalore seem to be more pressingly affected by rapid habitat destruction due to urbanization than global warming. A concerted effort to provide suitable microhabitats (boulders, low vegetation, soil) that are well connected will significantly improve the survival chance of these important mesocarnivore reptiles, P. dorsalis and C. versicolor. We provide here the first survey of agamid lizards in this region, which illustrates the need for repeated monitoring of reptiles in general. Active engagement with urban planners is imperative to ensure that suitable microhabitats are available to support these species in the long run. The protocols for agamid lizard surveys that we describe here in this report can be expanded to survey any urban-rural gradient, which is particularly challenging to assess remotely using other standardized methods. The potentially negative effects of global warming on these species cannot be fully discounted at this stage, given the scale mismatch between current environmental data and animal abundance and microhabitat information.

REFERENCES

1. Ackley, J. W., Wu, J., Angilletta, M. J., Myint, S. W., & Sullivan, B. (2015). Rich lizards:

How affluence and land cover influence the diversity and abundance of desert reptiles

persisting in an urban landscape. Biological Conservation, 182, 87-92.

2. Adolph, S. C. (1990). Influence of behavioral thermoregulation on microhabitat use by

two Sceloporus lizards. Ecology, 71(1), 315-327.

3. Angilletta, M. J., & Werner, Y. L. (1998). Australian geckos do not display diel variation

in thermoregulatory behavior. Copeia, 1998(3), 736-742.

4. Arnfield, A. J. (2003). Two decades of urban climate research: a review of turbulence,

exchanges of energy and water, and the urban heat island. International journal of

climatology, 23(1), 1-26.

5. Asbury, D. A., & Adolph, S. C. (2007). Behavioural plasticity in an ecological generalist:

microhabitat use by western fence lizards.

6. Ayers, D. Y., & Shine, R. (1997). Thermal influences on foraging ability: body size,

posture and cooling rate of an ambush predator, the python Morelia spilota. Functional

Ecology, 11(3), 342-347.

7. Baird, O. E., & Krueger, C. C. (2003). Behavioral thermoregulation of brook and rainbow

trout: comparison of summer habitat use in an Adirondack River, New

York. Transactions of the American Fisheries Society, 132(6), 1194-1206.

8. Baker, L. A., Brazel, A. J., Selover, N., Martin, C., McIntyre, N., Steiner, F. R., &

Musacchio, L. (2002). Urbanization and warming of Phoenix (Arizona, USA): Impacts,

feedbacks and mitigation. Urban ecosystems, 6(3), 183-203.

9. Balakrishna, S., Batabyal, A., & Thaker, M. (2016). Dining in the City: Dietary Shifts in

Indian Rock Agamas across an Urban–Rural Landscape.Journal of Herpetology.

10. Bartholomew, G. A. (1981). A matter of size: an examination of endothermy in insects

and terrestrial vertebrates.

11. Bartholomew, G. A. (1982). Physiological control of body temperature.Biology of the

Reptilia, 12, 167-211.

12. Beaugrand, G., Reid, P. C., Ibanez, F., Lindley, J. A., & Edwards, M. (2002).

13. Bauwens, D., Hertz, P. E., & Castilla, A. M. (1996). Thermoregulation in a lacertid lizard:

the relative contributions of distinct behavioral mechanisms.Ecology, 77(6), 1818-

1830.Reorganization of North Atlantic marine copepod biodiversity and

climate.Science, 296(5573), 1692-1694.

14. Beitinger, T. L., & Fitzpatrick, L. C. (1979). Physiological and ecological correlates of

preferred temperature in fish. American Zoologist, 19(1), 319-329.

15. Bellard, C., Bertelsmeier, C., Leadley, P., Thuiller, W., & Courchamp, F. (2012). Impacts

of climate change on the future of biodiversity. Ecology letters, 15(4), 365-377.

16. Bicego, K. C., Barros, R. C., & Branco, L. G. (2007). Physiology of temperature

regulation: comparative aspects. Comparative Biochemistry and Physiology Part A:

Molecular & Integrative Physiology, 147(3), 616-639.

17. Bishop, J. A., & Armbruster, W. S. (1999). Thermoregulatory abilities of Alaskan bees:

effects of size, phylogeny and ecology. Functional Ecology,13(5), 711-724.

18. Block, B. A., & Finnerty, J. R. (1994). Endothermy in fishes: a phylogenetic analysis of

constraints, predispositions, and selection pressures. Environmental Biology of

Fishes, 40(3), 283-302.

19. Brattstrom, B. H. (1979). Amphibian temperature regulation studies in the field and

laboratory. American Zoologist, 19(1), 345-356.

20. Brown, J. H., Gillooly, J. F., Allen, A. P., Savage, V. M., & West, G. B. (2004). Toward

a metabolic theory of ecology. Ecology, 85(7), 1771-1789.

21. Bronikowski, A. M., Bennett, A. F., & Lenski, R. E. (2001). Evolutionary adaptation to

temperature. VIII. Effects of temperature on growth rate in natural isolates of Escherichia

coli and Salmonella enterica from different thermal environments. Evolution, 55(1), 33-

40.

22. Buckley, L. B. (2008). Linking traits to energetics and population dynamics to predict

lizard ranges in changing environments. The American Naturalist,171(1), E1-E19.

23. Chappell, M. A., & Whitman, D. W. (1990). Grasshopper thermoregulation. Biology of

grasshoppers. Wiley, New York, 143-172.

24. Chown, S. L., Le Lagadec, M. D., & Scholtz, C. H. (1999). Partitioning variance in a

physiological trait: desiccation resistance in keratin beetles (Coleoptera,

Trogidae). Functional Ecology, 13(6), 838-844.

25. Christian, K. A., Bedford, G., Green, B., Schultz, T., & Newgrain, K. (1998). Energetics

and water flux of the marbled velvet gecko (Oedura marmorata) in tropical and temperate

habitats. Oecologia, 116(3), 336-342.

26. Christian, K. A., & Weavers, B. W. (1996). Thermoregulation of monitor lizards in

Australia: an evaluation of methods in thermal biology. Ecological monographs, 66(2),

139-157.

27. Ciais, P., Reichstein, M., Viovy, N., Granier, A., Ogée, J., Allard, V., ... & Chevallier, F.

(2005). Europe-wide reduction in primary productivity caused by the heat and drought in

2003. Nature, 437(7058), 529-533.

28. Clusella-Trullas, S., Terblanche, J. S., Blackburn, T. M., & Chown, S. L. (2008). Testing

the thermal melanism hypothesis: a macrophysiological approach. Functional

ecology, 22(2), 232-238.

29. Coelho, J. R. 2001. Behavioral and physiological thermoregulation in male cicada killers

(Sphecius speciosus) during territorial behavior. Journal of Thermal Biology 26:109–116.

30. Cooper, V. S., Bennett, A. F., & Lenski, R. E. (2001). Evolution of thermal dependence

of growth rate of Escherichia coli populations during 20,000 generations in a constant

environment. Evolution, 55(5), 889-896.

31. Cooper, W. E., & Greenberg, N. (1992). Reptilian coloration and behavior.Biology of the

Reptilia, 18, 298-422.

32. Cossins, A. R., & Bowler, K. (1987). Rate compensations and capacity adaptations.

In Temperature biology of animals (pp. 155-203). Springer Netherlands.

33. Cowles, R. B., & Bogert, C. M. (1944). A preliminary study of the thermal requirements

of desert reptiles. Iguana, 83, 53.

34. Daniel, J. C. (1983). Book of Indian reptiles.

35. Danks, H. V. (2004). Seasonal adaptations in arctic insects. Integrative and Comparative

Biology, 44(2), 85-94.

36. Duellman, W. E., & Trueb, L. (1986). Cytogenetic molecular, and genomic evolution.

Biology of amphibians.

37. Dzialowski, E. M., & O’Connor, M. P. (2001). Physiological control of warming and

cooling during simulated shuttling and basking in lizards.Physiological and Biochemical

Zoology, 74(5), 679-693.

38. Dzialowski, E. M., & O’Connor, M. P. (2004). Importance of the limbs in the

physiological control of heat exchange in Iguana and Sceloporus undulatus. Journal of

Thermal Biology, 29(6), 299-305.

39. Ehleringer, J., & Forseth, I. (1980). Solar tracking by plants. Science,210(4474), 1094-

1098.

40. Eppley, R. W. (1972). Temperature and phytoplankton growth in the sea.Fish.

Bull, 70(4), 1063-1085.

41. Fitter, A. H., & Fitter, R. S. R. (2002). Rapid changes in flowering time in British

plants. Science, 296(5573), 1689-1691.

42. Forchhammer, M. C., & Post, E. (2000). Climatic signatures in ecology.Trends in

Ecology and Evolution, 15(7), 286.

43. Franco, A., Hill, J. K., Kitschke, C., Collingham, Y. C., Roy, D. B., Fox, R. I. C. H. A.

R. D., ... & Thomas, C. D. (2006). Impacts of climate warming and habitat loss on

extinctions at species' low-latitude range boundaries. Global Change Biology, 12(8),

1545-1553.

44. Frazier, M. R., Huey, R. B., & Berrigan, D. (2006). Thermodynamics constrains the

evolution of insect population growth rates:“warmer is better”.The American

Naturalist, 168(4), 512-520.

45. Fry, F. E. J., Brett, J. R., & Clawson, G. H. (1942). Lethal limits of temperature for young

goldfish. Rev. can. biol, 1, 50-56.

46. Gilchrist, G. W., Huey, R. B., & Partridge, L. (1997). Thermal sensitivity of Drosophila

melanogaster: evolutionary responses of adults and eggs to laboratory natural selection

at different temperatures. Physiological and Biochemical Zoology, 70(4), 403-414.

47. Gillooly, J. F., Brown, J. H., West, G. B., Savage, V. M., & Charnov, E. L. (2001). Effects

of size and temperature on metabolic rate. science,293(5538), 2248-2251.

48. Gillooly, J. F., Charnov, E. L., West, G. B., Savage, V. M., & Brown, J. H. (2002). Effects

of size and temperature on developmental time. Nature,417(6884), 70-73.

49. Gordo, O., & Sanz, J. J. (2005). Phenology and climate change: a long-term study in a

Mediterranean locality. Oecologia, 146(3), 484-495.

50. Grant, B. W., & Dunham, A. E. (1988). Thermally imposed time constraints on the

activity of the desert lizard Sceloporus merriami. Ecology, 69(1), 167-176.

51. Grigg, G. C., Beard, L. A., & Augee, M. L. (2004). The evolution of endothermy and its

diversity in mammals and birds. Physiological and Biochemical Zoology, 77(6), 982-

997.

52. Grigg, G. C., & Seebacher, F. (1999). Field test of a paradigm: hysteresis of heart rate in

thermoregulation by a free-ranging lizard (Pogona barbata).Proceedings of the Royal

Society of London B: Biological Sciences,266(1425), 1291-1297.

53. Grimm, N. B., Faeth, S. H., Golubiewski, N. E., Redman, C. L., Wu, J., Bai, X., & Briggs,

J. M. (2008). Global change and the ecology of cities. science,319(5864), 756-760.

54. Harley, C. D. (2011). Climate change, keystone predation, and biodiversity

loss. Science, 334(6059), 1124-1127.

55. Harrington, R., Woiwod, I., & Sparks, T. (1999). Climate change and trophic

interactions. Trends in Ecology & Evolution, 14(4), 146-150.

56. Harrison, J. F., Fewell, J. H., Roberts, S. P., & Hall, H. G. (1996). Achievement of thermal

stability by varying metabolic heat production in flying honeybees. Science, 274(5284),

88.

57. Hazel, W. N. (2002). The environmental and genetic control of seasonal polyphenism in

larval color and its adaptive significance in a swallowtail butterfly. Evolution, 56(2), 342-

348.

58. Heath, J. E. (1964). Reptilian thermoregulation: evaluation of field

studies.Science, 146(3645), 784-785.

59. Heinrich, B. (1974). Thermoregulation in endothermic insects. Science,185(4153), 747-

756.

60. Heinrich, B. (1981). Insect thermoregulation. Wiley Interscience John Wiley & Sons

Inc..

61. Heinrich B. (1990). Is ‘Reflectance’Basking Real?. Journal of experimental

biology, 154(1), 31-43.

62. Hertz, P. E., Huey, R. B., & Nevo, E. (1982). Fight versus flight: body temperature

influences defensive responses of lizards. Animal Behaviour,30(3), 676IN3677-IN4679.

63. Hertz, P. E., Huey, R. B., & Stevenson, R. D. (1993). Evaluating temperature regulation

by field-active ectotherms: the fallacy of the inappropriate question. American Naturalist,

796-818.

64. Hijmans, R.J., S.E. Cameron, J.L. Parra, P.G. Jones and A. Jarvis, 2005. Very high

resolution interpolated climate surfaces for global land areas. International Journal of

Climatology 25: 1965-1978.

65. Hochochka, P. W., & Somero, G. N. (2002). Biochemical adaptation: Mechanism and

process in biochemical evolution.

66. Holland, K. N., Brillf, R. W., Randolph, K. C., & Changf, J. R. S. (1992). Physiological

and behavioural thermoregulation in bigeye. Nature, 358, 30.

67. Huey, R. B. (1974, May). Behavioral thermoregulation in lizards: importance of

associated costs. American Association for the Advancement of Science.

68. Huey, R. B. (1991). Physiological consequences of habitat selection.American Naturalist,

S91-S115.

69. Huey, R. B., & Slatkin, M. (1976). Cost and benefits of lizard

thermoregulation. Quarterly Review of Biology, 363-384.

70. Huey, Raymond B., et al. "Why tropical forest lizards are vulnerable to climate

warming." Proceedings of the Royal Society of London B: Biological Sciences (2009):

rspb-2008.

71. Huey, R. B. (1982). Temperature, physiology, and the ecology of reptiles. InBiology of

the Reptilia.

72. Huey, R. B., & Pianka, E. R. (1977). Seasonal variation in thermoregulatory behavior

and body temperature of diurnal Kalahari lizards. Ecology, 58(5), 1066-1075.

73. Hughes, L. (2000). Biological consequences of global warming: is the signal already

apparent?. Trends in ecology & evolution, 15(2), 56-61.

74. James, A. C., & Partridge, L. (1995). Thermal evolution of rate of larval development in

Drosophila melanogaster in laboratory and field populations.Journal of Evolutionary

Biology, 8(3), 315-330.

75. Jeffery, W. R. (2006). Regressive evolution of pigmentation in the cavefish

Astyanax. Israel Journal of Ecology & Evolution, 52(3-4), 405-422.

76. Ji, X., Qiu, Q. B., & Diong, C. H. (2002a). Sexual dimorphism and female reproductive

characteristics in the oriental garden lizard, Calotes versicolor, from Hainan, Southern

China. Journal of Herpetology, 36(1), 1-8.

77. Ji, X., Qiu, Q. B., & Diong, C. H. (2002b). Influence of incubation temperature on

hatching success, energy expenditure for embryonic development, and size and

morphology of hatchlings in the oriental garden lizard, Calotes versicolor

(Agamidae). Journal of Experimental Zoology, 292(7), 649-659.

78. Kammer, A. E. (1981). Physiological mechanisms of thermoregulation.Insect

thermoregulation. edited by Bernd Heinrich.

79. Kemp, D. J., & Krockenberger, A. K. (2004). Behavioural thermoregulation in

butterflies: the interacting effects of body size and basking posture in Hypolimnas bolina

(L.)(Lepidoptera: Nymphalidae). Australian journal of Zoology, 52(3), 229-239.

80. Kingsolver, J. G., & Huey, R. B. (1998). Evolutionary analyses of morphological and

physiological plasticity in thermally variable environments.American Zoologist, 38(3),

545-560.

81. Kirschbaum, M. U. F. (2004). Direct and indirect climate change effects on

photosynthesis and transpiration. Plant Biology, 6(03), 242-253.

82. Kreuger, B., & Potter, D. A. (2001). Diel feeding activity and thermoregulation by

Japanese beetles (Coleoptera: Scarabaeidae) within host plant canopies.Environmental

entomology, 30(2), 172-180.

83. Lacey, E. P., & Herr, D. (2005). Phenotypic plasticity, parental effects, and parental care

in plants? I. An examination of spike reflectance in Plantago lanceolata

(Plantaginaceae). American Journal of Botany, 92(6), 920-930.

84. Lagerspetz, K. Y., & Vainio, L. A. (2006). Thermal behaviour of crustaceans.Biological

Reviews, 81(2), 237-258.

85. Licht, P., W. R. Dawson, V. H. Shoemaker, and A. R. Main. 1966b. Observations on the

thermal relations of western Australian lizards. Copeia 1966:97–110.

86. Lillywhite, H. B., & Navas, C. A. (2006). Animals, energy, and water in extreme

environments: perspectives from Ithala 2004. Physiological and Biochemical

Zoology, 79(2), 265-273.

87. Lima, F. P., Ribeiro, P. A., Queiroz, N., Hawkins, S. J., & Santos, A. M. (2007). Do

distributional shifts of northern and southern species of algae match the warming

pattern?. Global Change Biology, 13(12), 2592-2604.

88. Loo, Yen Yi, Lawal Billa, and Ajit Singh. "Effect of climate change on seasonal monsoon

in Asia and its impact on the variability of monsoon rainfall in Southeast

Asia." Geoscience Frontiers (2014).

89. Lustick, S. I. (1983). Cost-benefit of thermoregulation in birds: influences of posture,

microhabitat selection, and color. Behavioral energetics: the cost of survival in

vertebrates. Ohio State University Press, Columbus, 265-294.

90. May, M. (1995). Simultaneous control of head and thoracic temperature by the green

darner dragonfly Anax junius (Odonata: Aeshnidae). Journal of Experimental

Biology, 198(11), 2373-2384.

91. Mejía-Ortíz, L. M., & Hartnoll, R. G. (2005). Modifications of eye structure and

integumental pigment in two cave crayfish. Journal of Crustacean Biology, 25(3), 480-

487.

92. Menéndez, R., Gonzales-Meglas, A. D. E. L. A., Lewis, O. T., Shaw, M. R., & Thomas,

C. D. (2008). Escape from natural enemies during climate-driven range expansion: a case

study. Ecological Entomology, 33(3), 413-421.

93. Miller, S. R., & Castenholz, R. W. (2000). The evolution of thermotolerance in hot spring

cyanobacteria of the genus Synechococcus. Journal of Phycology, 36(s3), 48-48.

94. Mongold, J. A., Bennett, A. F., & Lenski, R. E. (1999). Evolutionary adaptation to

temperature. VII. Extension of the upper thermal limit of Escherichia coli. Evolution,

386-394.

95. Muñoz, J. L., Finke, G. R., Camus, P. A., & Bozinovic, F. (2005). Thermoregulatory

behavior, heat gain and thermal tolerance in the periwinkle Echinolittorina peruviana in

central Chile. Comparative Biochemistry and Physiology Part A: Molecular &

Integrative Physiology, 142(1), 92-98.

96. Nelson, K. C., & Palmer, M. A. (2007). Stream Temperature Surges Under Urbanization

and Climate Change: Data, Models, and Responses1.

97. Newell, J. C., & Quinn, T. P. (2005). Behavioral thermoregulation by maturing adult

sockeye salmon (Oncorhynchus nerka) in a stratified lake prior to spawning. Canadian

Journal of Zoology, 83(9), 1232-1239.

98. Oke, T. R. (1973). City size and the urban heat island. Atmospheric Environment

(1967), 7(8), 769-779.

99. Parker, D. E. (2004). Climate: large-scale warming is not urban. Nature,432(7015), 290-

290.\

100. Parmesan, C. (2006). Ecological and evolutionary responses to recent climate

change. Annual Review of Ecology, Evolution, and Systematics, 637-669.

101. Parmesan, C. (2007). Influences of species, latitudes and methodologies on estimates of

phenological response to global warming. Global Change Biology, 13(9), 1860-1872.

102. Parmesan, C., & Galbraith, H. (2004). Observed impacts of global climate change in the

US (Vol. 12). Arlington, VA, USA: Pew Center on Global Climate Change.

103. Parmesan, C., Ryrholm, N., Stefanescu, C., Hill, J. K., Thomas, C. D., Descimon, H., ...

& Tennent, W. J. (1999). Poleward shifts in geographical ranges of butterfly species

associated with regional warming. Nature,399(6736), 579-583.

104. Parmesan, C., & Yohe, G. (2003). A globally coherent fingerprint of climate change

impacts across natural systems. Nature, 421(6918), 37-42.

105. Parris, K. M., & Hazell, D. L. (2005). Biotic effects of climate change in urban

environments: The case of the grey-headed flying-fox (Pteropus poliocephalus) in

Melbourne, Australia. Biological Conservation, 124(2), 267-276.

106. Parry, M. L., Canziani, O. F., Palutikof, J. P., van der Linden, P. J., & Hanson, C. E.

(2007). IPCC, 2007: climate change 2007: impacts, adaptation and vulnerability.

Contribution of working group II to the fourth assessment report of the intergovernmental

panel on climate change.

107. Parzefall, J. (2001). A review of morphological and behavioural changes in the cave

molly, Poecilia mexicana, from Tabasco, Mexico. In The biology of hypogean fishes (pp.

263-275). Springer Netherlands.

108. Patiño, S., Grace, J., & Bänziger, H. (2000). Endothermy by flowers of Rhizanthes lowii

(Rafflesiaceae). Oecologia, 124(2), 149-155.

109. Peterson, A. T., Ortega-Huerta, M. A., Bartley, J., Sánchez-Cordero, V., Soberón, J.,

Buddemeier, R. H., & Stockwell, D. R. (2002). Future projections for Mexican faunas

under global climate change scenarios.Nature, 416(6881), 626-629.

110. Phillips, S. J., Dudík, M., & Schapire, R. E. (2004, July). A maximum entropy approach

to species distribution modeling. In Proceedings of the twenty-first international

conference on Machine learning (p. 83). ACM.

111. Prange, H. D., & Hamilton, G. B. (1992). Humidity selection by thermoregulating

grasshoppers. Journal of thermal biology, 17(6), 353-355.

112. Prinzinger, R., Pressmar, A., & Schleucher, E. (1991). Body temperature in

birds. Comparative Biochemistry and Physiology Part A: Physiology, 99(4), 499-506.

113. Riley, P. A. (1997). Melanin. The international journal of biochemistry & cell

biology, 29(11), 1235-1239.

114. Roberts, S. P., & Harrison, J. F. (1998). Mechanisms of thermoregulation in flying

bees. American zoologist, 38(3), 492-502.

115. Roetzer, T., Wittenzeller, M., Haeckel, H., & Nekovar, J. (2000). Phenology in central

Europe–differences and trends of spring phenophases in urban and rural

areas. International Journal of Biometeorology, 44(2), 60-66.

116. 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.

117. Roth, M. (2002, January). Effects of cities on local climates. In Proceedings of Workshop

of IGES/APN Mega-City Project (pp. 23-25). Kitakyushu, Japan.

118. Rustad, L. E. J. L., Campbell, J., Marion, G., Norby, R., Mitchell, M., Hartley, A.. &

Gurevitch, J. (2001). A meta-analysis of the response of soil respiration, net nitrogen

mineralization, and aboveground plant growth to experimental ecosystem

warming. Oecologia, 126(4), 543-562.

119. Samietz, J., Salser, M. A., & Dingle, H. (2005). Altitudinal variation in behavioural

thermoregulation: local adaptation vs. plasticity in California grasshoppers. Journal of

evolutionary biology, 18(4), 1087-1096.

120. Sanborn, A. F., Villet, M. H., & Phillips, P. K. (2004). Endothermy in African

Platypleurine cicadas: the influence of body size and habitat (Hemiptera:

Cicadidae). Physiological and Biochemical Zoology, 77(5), 816-823.

121. Santos, M. (2007). Evolution of total net fitness in thermal lines: Drosophila subobscura

likes it ‘warm’. Journal of evolutionary biology, 20(6), 2361-2370.

122. Scholander, P. F., Hock, R., Walters, V., Johnson, F., & Irving, L. (1950a). Heat

regulation in some arctic and tropical mammals and birds. The Biological Bulletin, 99(2),

237-258.

123. Scholander, P. F., Walters, V., Hock, R., & Irving, L. (1950b). Body insulation of some

arctic and tropical mammals and birds. The Biological Bulletin,99(2), 225-236.

124. Seebacher, F., & Grigg, G. (2001). Changes in heart rate are important for

thermoregulation in the varanid lizard Varanus varius. Journal of Comparative

Physiology B, 171(5), 395-400.

125. Sinervo, B., Mendez-De-La-Cruz, F., Miles, D. B., Heulin, B., Bastiaans, E., Villagrán-

Santa Cruz, M., & Gadsden, H. (2010). Erosion of lizard diversity by climate change and

altered thermal niches. Science, 328(5980), 894-899.

126. Shoemaker, V. H., McClanahan, L. L., Withers, P. C., Hillman, S. S., & Drewes, R. C.

(1987). Thermoregulatory response to heat in the waterproof frogs Phyllomedusa and

Chiromantis. Physiological zoology, 365-372.

127. Smith, E. N. (1979). Behavioral and physiological thermoregulation of

crocodilians. American Zoologist, 19(1), 239-247.

128. Snucins, E. J., & Gunn, J. M. (1995). Coping with a warm environment: behavioral

thermoregulation by lake trout. Transactions of the American Fisheries Society, 124(1),

118-123.

129. Somero, G. N. (2004). Adaptation of enzymes to temperature: searching for basic

“strategies”. Comparative Biochemistry and Physiology Part B: Biochemistry and

Molecular Biology, 139(3), 321-333.

130. Stevenson, R. D. (1985). Body size and limits to the daily range of body temperature in

terrestrial ectotherms. American Naturalist, 102-117.

131. Talloen, W., Dyck, H. V., & Lens, L. (2004). The cost of melanization: Butterfly wing

coloration underenvironmental stress. Evolution, 58(2), 360-366.

132. Thomas, C. D., Cameron, A., Green, R. E., Bakkenes, M., Beaumont, L. J., Collingham,

Y. C., & Hughes, L. (2004). Extinction risk from climate change. Nature, 427(6970),

145-148.

133. Torti, V. M., & Dunn, P. O. (2005). Variable effects of climate change on six species of

North American birds. Oecologia, 145(3), 486-495.

134. Tsai, M. L., Dai, C. F., & Chen, H. C. (1998). Desiccation resistance of two

semiterrestrial isopods, Ligia exotica and Ligia taiwanensis (Crustacea) in

Taiwan. Comparative Biochemistry and Physiology Part A: Molecular & Integrative

Physiology, 119(1), 361-367.

135. Van Damme, R., Bauwens, D., Castilla, A. M., & Verheyen, R. F. (1989). Altitudinal

variation of the thermal biology and running performance in the lizard Podarcis

tiliguerta. Oecologia, 80(4), 516-524.

136. Van Damrne, R., & Verheyen, R. F. (1990). Comparative thermal ecology of the

sympatric lizards Podarcis tiliguerta and Podarcis sicula. Acta Œbologica, 11(4), 503-

512.

137. Verdu, J. R., Diaz, A., & Galante, E. (2004). Thermoregulatory strategies in two closely

related sympatric Scarabaeus species (Coleoptera: Scarabaeinae). Physiological

Entomology, 29(1), 32-38.

138. Visser, M. E., & Both, C. (2005). Shifts in phenology due to global climate change: the

need for a yardstick. Proceedings of the Royal Society of London B: Biological

Sciences, 272(1581), 2561-2569.

139. Visser, M. E., Holleman, L. J., & Gienapp, P. (2006). Shifts in caterpillar biomass

phenology due to climate change and its impact on the breeding biology of an

insectivorous bird. Oecologia, 147(1), 164-172.

140. Walther, G. R., Post, E., Convey, P., Menzel, A., Parmesan, C., Beebee, T. J., ... &

Bairlein, F. (2002). Ecological responses to recent climate change.Nature, 416(6879),

389-395.

141. Walton, B. M., & Bennett, A. F. (1993). Temperature-dependent color change in Kenyan

chameleons. Physiological Zoology, 270-287.

142. Walvoord, M. E. (2003). Cricket frogs maintain body hydration and temperature near

levels allowing maximum jump performance. Physiological and Biochemical

Zoology, 76(6), 825-835.

143. Weng, Q., Lu, D., & Schubring, J. (2004). Estimation of land surface temperature–

vegetation abundance relationship for urban heat island studies.Remote sensing of

Environment, 89(4), 467-483.

144. Whitaker, B. D., & Poff, K. L. (1980). Thermal adaptation of thermosensing and negative

thermotaxis in Dictyostelium. Experimental cell research,128(1), 87-93.

145. White, M. A., Nemani, R. R., Thornton, P. E., & Running, S. W. (2002). Satellite

evidence of phenological differences between urbanized and rural areas of the eastern

United States deciduous broadleaf forest. Ecosystems,5(3), 260-273.

146. Wu T., Li W., Ji J., Xin X., et al. (2012a) The 20th century global carbon cycle from the

Beijing Climate Center Climate System Model (BCC_CSM). Submitted to J. Climate

147. Zhang, L., Wu, T., Xin, X., Dong, M., & Wang, Z. (2012). Projections of annual mean

air temperature and precipitation over the globe and in China during the 21st century by

the BCC Climate System Model BCC_CSM1. 0.Acta Meteorologica Sinica, 26, 362-375.

APPENDIX A

Presence, Abundance and Microhabitat Data for Calotes versicolor and Psammophilus dorsalis

Presence, abundance and microhabitat data for Calotes versicolor

Project: Urbanization and its effects on lizards: a study from a climate change perspective

etation heigh etation Grid Species Presence No. trees No. Trees (%) Trees Abundance veg

Bare soil (%) soil Bare Concrete (%) Concrete Boulders (%) Boulders . Sheet rock (%) rock Sheet Vegetation (%) Vegetation Small rocks (%) rocks Small Soil compaction Soil Canopy cover (%) cover Canopy Avg 1.1 15 10 70 5 0 0 0 calotes 0 0 0.5 2 110 0

1.2 80 10 0 10 0 0 0 calotes 0 0 0.5 1 70 0

1.3 55 5 20 20 0 0 0 calotes 0 0 2 1 70 0

2.1 65 0 0 34 1 0 0 calotes 0 0 2.5 0 20 0

2.2 22 0 0 78 0 0 0 calotes 0 0 5 0 0 0

3.1 9 0 0 91 0 0 0 calotes 0 0 5 0 0 0

3.2 17 0 0 83 0 0 0 calotes 0 0 3.5 0 55 0

4.1 12 0 6 82 0 0 0 calotes 0 0 2 4 90 0

4.2 44 0 10 46 0 0 0 calotes 0 0 3.5 0 35 0

4.3 18 0 0 82 0 0 0 calotes 0 0 2.5 0 30 0

5.1 95 0 5 0 0 0 0 calotes 0 0 1 0 100 0

5.2 25 0 50 25 0 0 0 calotes 0 0 4 0 30 0

5.3 50 0 50 0 0 0 0 calotes 0 0 3.5 0 80 0

6.1 15 0 0 30 1 54 0 calotes 0 0 1 2 30 0

6.2 3 0 18 37 0 0 42 calotes 0 0 1 4 90 0

6.3 22 11 0 21 0 46 0 calotes 0 0 4 1 20 0

7.1 58 0 0 42 0 0 0 calotes 0 0 4 0 80 0

7.2 79 0 0 21 0 0 0 calotes 0 0 4 0 80 0

7.3 73 0 0 27 0 0 0 calotes 0 0 5 0 130 0

8.1 82 0 0 18 0 0 0 calotes 0 0 0.5 3 65 30

8.2 77 0 0 19 4 0 0 calotes 0 0 4.5 0 20 0

8.3 5 17 15 63 0 0 0 calotes 0 0 5 2 70 5

9.1 5 5 90 0 0 0 0 calotes 0 0 0 1 0 0

9.2 25 5 20 50 0 0 0 calotes 0 0 0.5 2 60 30

9.3 0 0 100 0 0 0 0 calotes 0 0 0 0 0 0

10.1 75 0 0 25 0 0 0 calotes 0 0 3 0 5 0

10.2 5 0 29 64 2 0 0 calotes 0 0 3 0 5 0

10.3 17 0 34 49 0 0 0 calotes 1 3 5 0 10 0

11.1 47 0 0 53 0 0 0 calotes 0 0 5 4 5 0

11.2 30 0 0 70 0 0 0 calotes 0 0 5 0 5 0

11.3 1 0 71 28 0 0 0 calotes 0 0 0.5 3 0 0

12.1 45 0 0 19 36 0 0 calotes 0 0 5 2 100 0

12.2 21 0 53 26 0 0 0 calotes 0 0 5 3 100 0

12.3 0 0 55 45 0 0 0 calotes 1 2 5 0 0 0

13.1 10 10 0 80 0 0 0 calotes 0 0 0.5 8 0 0

13.2 10 10 10 70 0 0 0 calotes 0 0 0.5 6 0 0

13.3 50 10 40 0 0 0 0 calotes 0 0 0.5 7 50 0

14.1 14 0 0 27 5 53 0 calotes 0 0 1 0 50 0

14.2 32 0 0 49 32 0 0 calotes 0 0 1 1 100 5

14.3 15 0 0 85 0 0 0 calotes 0 0 2.5 0 25 0

15.1 6 0 0 49 24 0 21 calotes 0 0 2 4 170 10

15.2 3 0 36 52 9 0 0 calotes 0 0 5 0 45 20

15.3 10 0 0 76 14 0 0 calotes 0 0 1 2 80 0

16.1 22 0 0 78 0 0 0 calotes 0 0 3 0 85 0

16.2 13 0 0 87 0 0 0 calotes 0 0 3 0 85 0

16.3a 15 0 22 63 0 0 0 calotes 0 0 3 0 85 0

16.3b 15 0 22 63 0 0 0 calotes 1 1 3 0 85 0

17.1 35 52 0 13 0 0 0 calotes 0 0 2.5 4 70 20

17.2 38 59 0 13 0 0 0 calotes 0 0 2.5 5 70 20

17.3 34 0 0 66 0 0 0 calotes 0 0 5 1 40 5

18.1 3 0 0 34 63 0 0 calotes 0 0 1 0 90 0

18.2 3 0 81 17 0 0 0 calotes 0 0 4.5 0 20 0

18.3 2 0 48 50 0 0 0 calotes 0 0 3.5 4 20 5

19.1 14 0 0 86 0 0 0 calotes 0 0 3 0 80 0

19.2 26 12 32 30 0 0 0 calotes 0 0 3 1 40 0

19.3 0 5 95 0 0 0 0 calotes 0 0 2.5 5 60 0

20.1 5 0 26 7 0 0 62 calotes 1 1 3.5 0 0 0

20.2 1 0 48 6 0 45 0 calotes 0 0 3.5 0 0 0

20.3 1 0 0 51 2 0 46 calotes 0 0 0.5 0 70 0

21.1 58 0 0 40 2 0 0 calotes 0 0 1.5 1 35 2

21.2 2 0 37 61 0 0 0 calotes 0 0 5 0 25 0

21.3 0 13 0 26 9 0 52 calotes 0 0 1 0 60 0

22.1 41 0 39 20 0 0 0 calotes 0 0 4.5 0 90 0

22.2 4 1 92 3 0 0 0 calotes 0 0 0 2 40 0

22.3 42 0 9 49 0 0 0 calotes 0 0 0 0 110 0

23.1 59 0 17 24 0 0 0 calotes 1 1 5 1 5 0

23.2 26 32 5 36 1 0 0 calotes 0 0 1 1 60 0

23.3 5 0 77 18 0 0 0 calotes 0 0 5 2 65 0

24.1 10 0 39 51 0 0 0 calotes 0 0 3 1 25 10

24.2 10 0 0 32 0 0 58 calotes 0 0 2.5 7 30 0

24.3 12 0 57 31 0 0 0 calotes 0 0 2.5 1 20 0

25.1 30 0 45 25 0 0 0 calotes 0 0 5 0 60 0

25.2 9 0 0 91 0 0 0 calotes 0 0 5 2 30 0

25.3 10 0 48 42 0 0 0 calotes 0 0 5 3 40 0

26.1 33 0 0 47 20 0 0 calotes 0 0 0 3 5 30

26.2 13 0 69 19 0 0 0 calotes 0 0 1.5 1 5 0

27.1 1 2 44 53 0 0 0 calotes 0 0 5 1 0 0

27.2 42 0 11 47 0 0 0 calotes 0 0 2 0 30 0

27.3 38 0 0 62 0 0 0 calotes 1 1 5 0 30 0

28.1 4 0 70 26 0 0 0 calotes 0 0 1 0 55 0

28.2 89 0 0 89 0 0 0 calotes 0 0 4 0 70 0

28.3 28 0 0 30 0 42 0 calotes 0 0 3 1 130 25

29.1 1 0 58 41 0 0 0 calotes 0 0 0 0 20 0

29.2 17 0 0 83 0 0 0 calotes 0 0 4.5 0 65 0

29.3 5 0 30 65 0 0 0 calotes 0 0 0 0 80 0

30.1 7 0 41 52 0 0 0 calotes 0 0 5 0 70 0

30.2 37 8 0 55 0 0 0 calotes 0 0 1.5 0 0 0

30.3 0 0 21 79 0 0 0 calotes 1 1 5 0 0 0

31.1 16 0 41 42 0 0 0 calotes 0 0 3 0 0 0

31.2 10 0 19 71 0 0 0 calotes 0 0 2 1 25 0

31.3 56 0 0 44 0 0 0 calotes 0 0 2.5 0 130 0

32.1 61 0 18 20 1 0 0 calotes 0 0 5 0 60 0

32.2 19 0 43 34 3 0 0 calotes 0 0 5 0 35 0

32.3 37 0 0 22 0 41 0 calotes 0 0 2.5 1 90 5

33.1 0 0 40 60 0 0 0 calotes 0 0 4 0 30 0

33.2 72 0 0 28 0 0 0 calotes 0 0 4.5 0 110 0

33.3 3 0 0 71 26 0 0 calotes 0 0 1.5 4 50 0

34.1 49 0 24 24 2 0 0 calotes 1 1 5 2 80 5

34.2 89 0 0 11 0 0 0 calotes 0 0 2 0 55 0

34.3 39 0 0 61 0 0 0 calotes 0 0 2.5 0 15 0

35.1 59 0 0 41 0 0 0 calotes 0 0 5 0 20 0

35.2 51 0 9 40 0 0 0 calotes 0 0 5 0 20 0

35.3 45 0 24 31 0 0 0 calotes 0 0 5 0 20 0

36.1 20 0 4 20 0 0 0 calotes 0 0 5 0 35 0

36.2 17 0 57 26 0 0 0 calotes 0 0 1 0 35 0

36.3 71 0 9 20 0 0 0 calotes 0 0 2.5 0 35 0

37.1 20 0 54 26 0 0 0 calotes 0 0 0 0 0 0

37.2 3 0 65 33 0 0 0 calotes 0 0 0 0 0 0

37.3 1 0 0 99 0 0 0 calotes 0 0 5 0 10 0

38.1 1 0 36 62 0 0 0 calotes 0 0 3 0 80 0

38.2 12 0 39 49 0 0 0 calotes 0 0 3 0 60 0

38.3 0 0 33 67 0 0 0 calotes 0 0 1.5 0 40 0

39.2 1 0 84 15 0 0 0 calotes 0 0 5 2 5 0

39.3 9 0 34 57 0 0 0 calotes 1 1 5 1 5 0

40.1 2 0 4 94 0 0 0 calotes 0 0 2.5 1 100 0

40.2 25 0 0 75 0 0 0 calotes 1 1 0.5 0 0 0

41.1 11 0 0 89 0 0 0 calotes 0 0 5 2 80 0

41.2 17 0 0 83 0 0 0 calotes 0 0 5 2 80 0

41.3 2 0 7 91 0 0 0 calotes 0 0 3 0 85 0

43.1 25 16 0 44 15 0 0 calotes 0 0 1 0 150 0

43.2 35 0 23 32 10 0 0 calotes 0 0 1.5 0 0 0

43.3 10 0 0 26 0 64 0 calotes 0 0 3 0 50 0

44.1 7 0 0 93 0 0 0 calotes 1 1 5 0 10 0

44.2 9 0 10 81 0 0 0 calotes 0 0 5 2 10 0

44.3 13 0 0 87 0 0 0 calotes 1 1 5 0 10 0

46.1 9 0 0 83 8 0 0 calotes 0 0 5 0 10 0

46.2 4 0 18 75 3 0 0 calotes 1 1 5 0 10 0

46.3 6 3 86 5 0 0 0 calotes 0 0 5 0 100 0

47.1 40 0 40 20 0 0 0 calotes 0 0 5 2 0 0

47.2 6 0 0 94 0 0 0 calotes 0 0 2 0 40 0

47.3 21 0 26 53 0 0 0 calotes 0 0 0 0 0

48.1 4 0 0 69 0 27 0 calotes 0 0 3 0 30 0

48.2 3 0 0 94 3 0 0 calotes 0 0 2 0 25 0

48.3 37 0 13 50 0 0 0 calotes 0 0 0.5 0 40 0

Presence, abundance and microhabitat data for Psammophilus dorsalis

Project: Urbanization and its effects on lizards: a study from a climate change perspective

eigh

Grid etation Species Presence No. trees No. Trees (%) Trees Abundance Bare soil (%) soil Bare Boulders (%) Boulders Sheet rock (%) rock Sheet Vegetation (%) Vegetation Small rocks (%) rocks Small Soil compaction Soil Construction (%) Construction Canopy cover (%) cover Canopy Avg veg Avg

1.1 15 10 70 5 0 0 0 dorsalis 0 0 0.5 2 110 0

1.2 80 10 0 10 0 0 0 dorsalis 0 0 0.5 1 70 0

1.3 55 5 20 20 0 0 0 dorsalis 0 0 2 1 70 0

2.1 65 0 0 34 1 0 0 dorsalis 0 0 2.5 0 20 0

2.2 22 0 0 78 0 0 0 dorsalis 0 0 5 0 0 0

3.1 9 0 0 91 0 0 0 dorsalis 0 0 5 0 0 0

3.2 17 0 0 83 0 0 0 dorsalis 0 0 3.5 0 55 0

4.1 12 0 6 82 0 0 0 dorsalis 0 0 2 4 90 0

4.2 44 0 10 46 0 0 0 dorsalis 1 2 3.5 0 35 0

4.3 18 0 0 82 0 0 0 dorsalis 1 3 2.5 0 30 0

5.1 95 0 5 0 0 0 0 dorsalis 1 4 1 0 100 0

5.2 25 0 50 25 0 0 0 dorsalis 1 5 4 0 30 0

5.3 50 0 50 0 0 0 0 dorsalis 0 0 3.5 0 80 0

6.1 15 0 0 30 1 54 0 dorsalis 1 4 1 2 30 0

6.2 3 0 18 37 0 0 42 dorsalis 1 2 1 4 90 0

6.3 22 11 0 21 0 46 0 dorsalis 0 0 4 1 20 0

7.1 58 0 0 42 0 0 0 dorsalis 0 0 4 0 80 0

7.2 79 0 0 21 0 0 0 dorsalis 0 0 4 0 80 0

7.3 73 0 0 27 0 0 0 dorsalis 0 0 5 0 130 0

8.1 82 0 0 18 0 0 0 dorsalis 1 3 0.5 3 65 30

8.2 77 0 0 19 4 0 0 dorsalis 0 0 4.5 0 20 0

8.3 5 17 15 63 0 0 0 dorsalis 1 2 5 2 70 5

9.1 5 5 90 0 0 0 0 dorsalis 0 0 0 1 0 0

9.2 25 5 20 50 0 0 0 dorsalis 0 0 0.5 2 60 30

9.3 0 0 100 0 0 0 0 dorsalis 0 0 0 0 0 0

10.1 75 0 0 25 0 0 0 dorsalis 0 0 3 0 5 0

10.2 5 0 29 64 2 0 0 dorsalis 0 0 3 0 5 0

10.3 17 0 34 49 0 0 0 dorsalis 0 0 5 0 10 0

11.1 47 0 0 53 0 0 0 dorsalis 0 0 5 4 5 0

11.2 30 0 0 70 0 0 0 dorsalis 0 0 5 0 5 0

11.3 1 0 71 28 0 0 0 dorsalis 1 6 0.5 3 0 0

12.1 45 0 0 19 36 0 0 dorsalis 0 0 5 2 100 0

12.2 21 0 53 26 0 0 0 dorsalis 1 2 5 3 100 0

12.3 0 0 55 45 0 0 0 dorsalis 1 1 5 0 0 0

13.1 10 10 0 80 0 0 0 dorsalis 0 0 0.5 8 0 0

13.2 10 10 10 70 0 0 0 dorsalis 0 0 0.5 6 0 0

13.3 50 10 40 0 0 0 0 dorsalis 0 0 0.5 7 50 0

14.1 14 0 0 27 5 53 0 dorsalis 0 0 1 0 50 0

14.2 32 0 0 49 32 0 0 dorsalis 1 2 1 1 100 5

14.3 15 0 0 85 0 0 0 dorsalis 0 0 2.5 0 25 0

15.1 6 0 0 49 24 0 21 dorsalis 1 3 2 4 170 10

15.2 3 0 36 52 9 0 0 dorsalis 1 2 5 0 45 20

15.3 10 0 0 76 14 0 0 dorsalis 1 2 1 2 80 0

16.1 22 0 0 78 0 0 0 dorsalis 0 0 3 0 85 0

16.2 13 0 0 87 0 0 0 dorsalis 0 0 3 0 85 0

16.3a 15 0 22 63 0 0 0 dorsalis 0 0 3 0 85 0

16.3b 15 0 22 63 0 0 0 dorsalis 0 0 3 0 85 0

17.1 35 52 0 13 0 0 0 dorsalis 0 0 2.5 4 70 20

17.2 38 59 0 13 0 0 0 dorsalis 0 0 2.5 5 70 20

17.3 34 0 0 66 0 0 0 dorsalis 0 0 5 1 40 5

18.1 3 0 0 34 63 0 0 dorsalis 0 0 1 0 90 0

18.2 3 0 81 17 0 0 0 dorsalis 0 0 4.5 0 20 0

18.3 2 0 48 50 0 0 0 dorsalis 0 0 3.5 4 20 5

19.1 14 0 0 86 0 0 0 dorsalis 0 0 3 0 80 0

19.2 26 12 32 30 0 0 0 dorsalis 0 0 3 1 40 0

19.3 0 5 95 0 0 0 0 dorsalis 0 0 2.5 5 60 0

20.1 5 0 26 7 0 0 62 dorsalis 0 0 3.5 0 0 0

20.2 1 0 48 6 0 45 0 dorsalis 1 2 3.5 0 0 0

20.3 1 0 0 51 2 0 46 dorsalis 0 0 0.5 0 70 0

21.1 58 0 0 40 2 0 0 dorsalis 0 0 1.5 1 35 2

21.2 2 0 37 61 0 0 0 dorsalis 0 0 5 0 25 0

21.3 0 13 0 26 9 0 52 dorsalis 0 0 1 0 60 0

22.1 41 0 39 20 0 0 0 dorsalis 0 0 4.5 0 90 0

22.2 4 1 92 3 0 0 0 dorsalis 0 0 0 2 40 0

22.3 42 0 9 49 0 0 0 dorsalis 0 0 0 0 110 0

23.1 59 0 17 24 0 0 0 dorsalis 0 0 5 1 5 0

23.2 26 32 5 36 1 0 0 dorsalis 0 0 1 1 60 0

23.3 5 0 77 18 0 0 0 dorsalis 0 0 5 2 65 0

24.1 10 0 39 51 0 0 0 dorsalis 0 0 3 1 25 10

24.2 10 0 0 32 0 0 58 dorsalis 1 2 2.5 7 30 0

24.3 12 0 57 31 0 0 0 dorsalis 0 0 2.5 1 20 0

25.1 30 0 45 25 0 0 0 dorsalis 0 0 5 0 60 0

25.2 9 0 0 91 0 0 0 dorsalis 0 0 5 2 30 0

25.3 10 0 48 42 0 0 0 dorsalis 0 0 5 3 40 0

26.1 33 0 0 47 20 0 0 dorsalis 0 0 0 3 5 30

26.2 13 0 69 19 0 0 0 dorsalis 0 0 1.5 1 5 0

27.1 1 2 44 53 0 0 0 dorsalis 0 0 5 1 0 0

27.2 42 0 11 47 0 0 0 dorsalis 0 0 2 0 30 0

27.3 38 0 0 62 0 0 0 dorsalis 0 0 5 0 30 0

28.1 4 0 70 26 0 0 0 dorsalis 0 0 1 0 55 0

28.2 89 0 0 89 0 0 0 dorsalis 0 0 4 0 70 0

28.3 28 0 0 30 0 42 0 dorsalis 1 1 3 1 130 25

29.1 1 0 58 41 0 0 0 dorsalis 0 0 0 0 20 0

29.2 17 0 0 83 0 0 0 dorsalis 0 0 4.5 0 65 0

29.3 5 0 30 65 0 0 0 dorsalis 0 0 0 0 80 0

30.1 7 0 41 52 0 0 0 dorsalis 0 0 5 0 70 0

30.2 37 8 0 55 0 0 0 dorsalis 0 0 1.5 0 0 0

30.3 0 0 21 79 0 0 0 dorsalis 0 0 5 0 0 0

31.1 16 0 41 42 0 0 0 dorsalis 1 1 3 0 0 0

31.2 10 0 19 71 0 0 0 dorsalis 0 0 2 1 25 0

31.3 56 0 0 44 0 0 0 dorsalis 0 0 2.5 0 130 0

32.1 61 0 18 20 1 0 0 dorsalis 0 0 5 0 60 0

32.2 19 0 43 34 3 0 0 dorsalis 1 2 5 0 35 0

32.3 37 0 0 22 0 41 0 dorsalis 1 2 2.5 1 90 5

33.1 0 0 40 60 0 0 0 dorsalis 1 4 4 0 30 0

33.2 72 0 0 28 0 0 0 dorsalis 0 0 4.5 0 110 0

33.3 3 0 0 71 26 0 0 dorsalis 1 2 1.5 4 50 0

34.1 49 0 24 24 2 0 0 dorsalis 0 0 5 2 80 5

34.2 89 0 0 11 0 0 0 dorsalis 0 0 2 0 55 0

34.3 39 0 0 61 0 0 0 dorsalis 0 0 2.5 0 15 0

35.1 59 0 0 41 0 0 0 dorsalis 0 0 5 0 20 0

35.2 51 0 9 40 0 0 0 dorsalis 0 0 5 0 20 0

35.3 45 0 24 31 0 0 0 dorsalis 0 0 5 0 20 0

36.1 20 0 4 20 0 0 0 dorsalis 0 0 5 0 35 0

36.2 17 0 57 26 0 0 0 dorsalis 0 0 1 0 35 0

36.3 71 0 9 20 0 0 0 dorsalis 0 0 2.5 0 35 0

37.1 20 0 54 26 0 0 0 dorsalis 0 0 0 0 0 0

37.2 3 0 65 33 0 0 0 dorsalis 0 0 0 0 0 0

37.3 1 0 0 99 0 0 0 dorsalis 0 0 5 0 10 0

38.1 1 0 36 62 0 0 0 dorsalis 1 2 3 0 80 0

38.2 12 0 39 49 0 0 0 dorsalis 0 0 3 0 60 0

38.3 0 0 33 67 0 0 0 dorsalis 1 4 1.5 0 40 0

39.2 1 0 84 15 0 0 0 dorsalis 0 0 5 2 5 0

39.3 9 0 34 57 0 0 0 dorsalis 0 0 5 1 5 0

40.1 2 0 4 94 0 0 0 dorsalis 0 0 2.5 1 100 0

40.2 25 0 0 75 0 0 0 dorsalis 0 0 0.5 0 0 0

41.1 11 0 0 89 0 0 0 dorsalis 0 0 5 2 80 0

41.2 17 0 0 83 0 0 0 dorsalis 0 0 5 2 80 0

41.3 2 0 7 91 0 0 0 dorsalis 0 0 3 0 85 0

43.1 25 16 0 44 15 0 0 dorsalis 1 5 1 0 150 0

43.2 35 0 23 32 10 0 0 dorsalis 1 3 1.5 0 0 0

43.3 10 0 0 26 0 64 0 dorsalis 1 10 3 0 50 0

44.1 7 0 0 93 0 0 0 dorsalis 0 0 5 0 10 0

44.2 9 0 10 81 0 0 0 dorsalis 0 0 5 2 10 0

44.3 13 0 0 87 0 0 0 dorsalis 0 0 5 0 10 0

46.1 9 0 0 83 8 0 0 dorsalis 1 7 5 0 10 0

46.2 4 0 18 75 3 0 0 dorsalis 0 0 5 0 10 0

46.3 6 3 86 5 0 0 0 dorsalis 0 0 5 0 100 0

47.1 40 0 40 20 0 0 0 dorsalis 0 0 5 2 0 0

47.2 6 0 0 94 0 0 0 dorsalis 0 0 2 0 40 0

47.3 21 0 26 53 0 0 0 dorsalis 0 0 0 0 0

48.1 4 0 0 69 0 27 0 dorsalis 1 4 3 0 30 0

48.2 3 0 0 94 3 0 0 dorsalis 0 0 2 0 25 0

48.3 37 0 13 50 0 0 0 dorsalis 0 0 0.5 0 40 0

SE1.C 50 0 50 0 0 0 0 dorsalis 0 0 1.75 0 70 0

SE1.A 0 0 100 0 0 0 0 dorsalis 0 0 0 0 0 0

SE1.E 100 0 0 0 0 0 0 dorsalis 1 1 1.75 0 80 0

S1.A 0 0 100 0 0 0 0 dorsalis 0 0 0 0 0 0

S1.C 50 0 50 0 0 0 0 dorsalis 0 0 5 0 300 0

S1.E 40 0 0 30 30 0 0 dorsalis 0 0 1.5 0 0 0

NW2.C 0 0 50 50 0 0 0 dorsalis 0 0 4.5 0 0 0

NW2.E 50 0 10 20 20 0 0 dorsalis 1 4 0 0 90 0

N2.A 0 0 100 0 0 0 0 dorsalis 0 0 0 0 0 0

N2.C 50 0 50 0 0 0 0 dorsalis 1 1 1 0 100 0

N2.E 30 0 0 40 10 0 0 dorsalis 1 5 3.5 0 170 0

APPENDIX B

A Brief Tutorial on Maxent

By Steven Phillips, Steven Phillips, Miro Dudik and Rob Schapire

AT&T Research

Center for Biodiversity and Conservation, American Museum of Natural History

For this project, we used MaxEnt to model species distributions for P. dorsalis and C. versicolor in and around Bangalore, Karnataka. All the following text in Appendix B have been taken from the manual that provides a brief introduction and tutorial for MaxEnt.

The material below is directly quoted from Phillips et al (2005).

Use and credit for this content must be cited as:

Phillips, S. J. (2005). A brief tutorial on Maxent. AT&T Research.

For more details on the theory maximum entropy modelling, explanation of the data used and the different types of statistical analysis, see:

Phillips, S. J., Dudik, M. & Schapire, R.E. 2004. A maximum entropy approach to species distribution modeling. Pages 655-662 in Proceedings of the 21st International Conference on

Machine Learning. ACM Press, New York

Steven J. Phillips, Robert P. Anderson and Robert E. Schapire, Maximum entropy modeling of species geographic distributions. Ecological Modelling, Vol 190/3-4 pp 231-259, 2006.

Steven J. Phillips and Miroslav Dudik, Modeling of species distributions with Maxent: new extensions and a comprehensive evaluation. Ecography, Vol 31, pp 161-175, 2008.

Getting started

Downloading

The software consists of a jar file, maxent.jar, which can be used on any computer running Java version 1.4 or later. Maxent can be downloaded, along with associated literature, from www.cs.princeton.edu/~schapire/maxent; the Java runtime environment can be obtained from java.sun.com/javase/downloads. If you are using Microsoft Windows (as we assume here), you should also download the file maxent.bat, and save it in the same directory as maxent.jar. The website has a file called “readme.txt”, which contains instructions for installing the program on your computer.

Firing up

If you are using Microsoft Windows, simply click on the file maxent.bat. Otherwise, enter "java - mx512m -jar maxent.jar" in a command shell (where "512" can be replaced by the megabytes of memory you want made available to the program). The following screen will appear:

To perform a run, you need to supply a file containing presence localities (“samples”), a directory containing environmental variables, and an output directory. In our case, the presence localities are in the file “samples\bradypus.csv”, the environmental layers are in the directory “layers”, and the outputs are going to go in the directory “outputs”. You can enter these locations by hand, or browse for them.

While browsing for the environmental variables, remember that you are looking for the directory that contains them – you don’t need to browse down to the files in the directory. After entering or browsing for the files for Bradypus, the program looks like this:

The file “samples\bradypus.csv” contains the presence localities in .csv format. The first few lines are as follows:

species,longitude,latitude bradypus_variegatus,-65.4,-10.3833 bradypus_variegatus,-65.3833,-10.3833 bradypus_variegatus,-65.1333,-16.8 bradypus_variegatus,-63.6667,-17.45 bradypus_variegatus,-63.85,-17.4

There can be multiple species in the same samples file, in which case more species would appear in the panel, along with Bradypus. Coordinate systems other than latitude and longitude can be used provided that the samples file and environmental layers use the same coordinate system. The “x” coordinate (longitude, in our case) should come before the “y” coordinate (latitude) in the samples file.

If the presence data has duplicate records (multiple records for the same species in the same grid cell), the duplicates are removed by default; this can be changed by clicking on the “Settings” button and deselecting “Remove duplicate presence records”.

The directory “layers” contains a number of ascii raster grids (in ESRI’s .asc format), each of which describes an environmental variable. The grids must all have the same geographic bounds and cell size

(i.e. all the ascii file headings must match each other perfectly). One of our variables, “ecoreg”, is a categorical variable describing potential vegetation classes. The categories must be indicated by numbers, rather than letters or words. You must tell the program which variables are categorical, as has been done in the picture above.

Doing a run

Simply press the “Run” button. A progress monitor describes the steps being taken. After the environmental layers are loaded and some initialization is done, progress towards training of the maxent model is shown like this:

The gain is closely related to deviance, a measure of goodness of fit used in generalized additive and generalized linear models. It starts at 0 and increases towards an asymptote during the run. During this process, Maxent is generating a probability distribution over pixels in the grid, starting from the uniform distribution and repeatedly improving the fit to the data. The gain is defined as the average log probability of the presence samples, minus a constant that makes the uniform distribution have zero gain. At the end of the run, the gain indicates how closely the model is concentrated around the presence samples; for example, if the gain is 2, it means that the average likelihood of the presence samples is exp(2) ≈ 7.4 times higher than that of a random background pixel. Note that Maxent isn’t directly calculating “probability of occurrence”. The probability it assigns to each pixel is typically very small, as the values must sum to 1 over all the pixels in the grid (though we return to this point when we compare output formats).

The run produces multiple output files, of which the most important for analyzing your model is an html file called “bradypus.html”. The end of this file gives pointers to the other outputs, like this:

Looking at a prediction

By default, the html output file contains a picture of the model applied to the given environmental data:

The image uses colors to indicate predicted probability that conditions are suitable, with red indicating high probability of suitable conditions for the species, green indicating conditions typical of those

99 where the species is found, and lighter shades of blue indicating low predicted probability of suitable conditions. For Bradypus, we see that suitable conditions are predicted to be highly probable through most of lowland Central America, wet lowland areas of northwestern South America, the Amazon basin, Caribean islands, and much of the Atlantic forests in south-eastern Brazil. The file pointed to is an image file (.png) that you can just click on (in Windows) or open in most image processing software. If you want to copy these images, or want to open them with other software, you will find the

.png files in the directory called “plots” that has been created as an output during the run.

The test points are a random sample taken from the species presence localities. The same random sample is used each time you run Maxent on the same data set, unless you select the “random seed” option on the settings panel. Alternatively, test data for one or more species can be provided in a separate file, by giving the name of a “Test sample file” in the Settings panel.

Output formats

Maxent supports three output formats for model values: raw, cumulative and logistic. First, the raw output is just the Maxent exponential model itself. Second, the cumulative value corresponding to a raw value of r is the percentage of the Maxent distribution with raw value at most r. Cumulative output is best interpreted in terms of predicted omission rate: if we set a cumulative threshold of c, the resulting binary prediction would have omission rate c% on samples drawn from the Maxent distribution itself, and we can predict a similar omission rate for samples drawn from the species distribution. Third, if c is the exponential of the entropy of the maxent distribution, then the logistic value corresponding to a raw value of r is c·r/(1+c·r). This is a logistic function, because the raw value is an exponential function of the environmental variables. The three output formats are all monotonically related, but they are scaled differently, and have different interpretations. The default

100 output is logistic, which is the easiest to conceptualize: it gives an estimate between 0 and 1 of probability of presence. Note that probability of presence depends on details of the sampling design, such as the plot size and (for vagile organisms) observation time; logistic output estimates probability of presence assuming that the sampling design is such that typical presence localities have probability of presence of about 0.5. This value of 0.5 is fairly arbitrary, and can be adjusted (using the “default prevalence” parameter) if information is available on the probability of presence at typical presence localities. The picture of the Bradypus model above uses the logistic format. In comparison, using the raw format gives the following picture:

101

Note that we have used a logarithmic scale for the colors. A linear scale would be mostly blue, with a few red pixels (you can verify this by deselecting “Logscale pictures” on the Settings panel) since the raw format typically gives a small number of sites relatively large values – this can be thought of as an artifact of the raw output being given by an exponential distribution.

102

Using the cumulative output format gives the following picture:

As with the raw output, we have used a logarithmic scale for coloring the picture in order to emphasize differences between smaller values. Cumulative output can be interpreted as predicting suitable conditions for the species above a threshold in the approximate range of 1-20 (or yellow through orange, in this picture), depending on the level of predicted omission that is acceptable for the application.

103

Statistical analysis

The “25” we entered for “random test percentage” told the program to randomly set aside 25% of the sample records for testing. This allows the program to do some simple statistical analysis. Much of the analysis used the use of a threshold to make a binary prediction, with suitable conditions predicted above the threshold and unsuitable below. The first plot shows how testing and training omission and predicted area vary with the choice of cumulative threshold, as in the following graph:

Here we see that the omission on test samples is a very good match to the predicted omission rate, the omission rate for test data drawn from the Maxent distribution itself. The predicted omission rate is a straight line, by definition of the cumulative output format. In some situations, the test omission line lies well below the predicted omission line: a common reason is that the test and training data are not independent, for example if they derive from the same spatially autocorrelated presence data.

104

The next plot gives the receiver operating curve for both training and test data, shown below. The area under the ROC curve (AUC) is also given here; if test data are available, the standard error of the AUC on the test data is given later on in the web page.

If you use the same data for training and for testing then the red and blue lines will be identical. If you split your data into two partitions, one for training and one for testing it is normal for the red (training) line to show a higher AUC than the blue (testing) line. The red (training) line shows the “fit” of the model to the training data. The blue (testing) line indicates the fit of the model to the testing data, and is the real test of the models predictive power. The turquoise line shows the line that you would expect if your model was no better than random. If the blue line (the test line) falls below the turquoise line then this indicates that your model performs worse than a random model would. The further towards the top left of the graph that the blue line is, the better the model is at predicting the presences contained in the test sample of the data. For more detailed information on the AUC statistic a good starting reference is: Fielding, A.H. & Bell, J.F. (1997) A review of methods for the assessment of prediction errors in conservation presence/ absence models. Environmental Conservation 24(1): 38-49.

105

Because we have only occurrence data and no absence data, “fractional predicted area” (the fraction of the total study area predicted present) is used instead of the more standard commission rate (fraction of absences predicted present). For more discussion of this choice, see the paper in Ecological Modelling mentioned on Page 1 of this tutorial. It is important to note that AUC values tend to be higher for species with narrow ranges, relative to the study area described by the environmental data. This does not necessarily mean that the models are better; instead this behavior is an artifact of the AUC statistic.

If test data are available, the program automatically calculates the statistical significance of the prediction, using a binomial test of omission. For Bradypus, this gives:

For more detailed information on the binomial statistic, see the Ecological Modelling paper mentioned above.

106

Which variables matter most?

A natural application of species distribution modeling is to answer the question, which variables matter most for the species being modeled? There is more than one way to answer this question; here we outline the possible ways in which Maxent can be used to address it.

While the Maxent model is being trained, it keeps track of which environmental variables are contributing to fitting the model. Each step of the Maxent algorithm increases the gain of the model by modifying the coefficient for a single feature; the program assigns the increase in the gain to the environmental variable(s) that the feature depends on. Converting to percentages at the end of the training process, we get the middle column in the following table:

107

These percent contribution values are only heuristically defined: they depend on the particular path that the Maxent code uses to get to the optimal solution, and a different algorithm could get to the same solution via a different path, resulting in different percent contribution values. In addition, when there are highly correlated environmental variables, the percent contributions should be interpreted with caution. In our Bradypus example, annual precipitation is highly correlated with October and

July precipitation. Although the above table shows that Maxent used the October precipitation variable more than any other, and hardly used annual precipitation at all, this does not necessarily imply that October precipitation is far more important to the species than annual precipitation.

The right-hand column in the table shows a second measure of variable contributions, called permutation importance. This measure depends only on the final Maxent model, not the path used to obtain it. The contribution for each variable is determined by randomly permuting the values of that variable among the training points (both presence and background) and measuring the resulting decrease in training AUC. A large decrease indicates that the model depends heavily on that variable.

Values are normalized to give percentages.

To get alternate estimates of variable importance, we can also run a jackknife test by selecting the “Do jackknife to measure variable important” checkbox. When we press the “Run” button again, a number of models are created. Each variable is excluded in turn, and a model created with the remaining variables. Then a model is created using each variable in isolation. In addition, a model is created using all variables, as before. The results of the jackknife appear in the “bradypus.html” files in three bar charts, and the first of these is shown below.

108

We see that if Maxent uses only pre6190_l1 (average January rainfall) it achieves almost no gain, so that variable is not (by itself) useful for estimating the distribution of Bradypus. On the other hand,

October rainfall (pre6190_l10) allows a reasonably good fit to the training data. Turning to the lighter blue bars, it appears that no variable contains a substantial amount of useful information that is not already contained in the other variables, because omitting each variable in turn did not decrease the training gain considerably.

The bradypus.html file has two more jackknife plots, which use either test gain or AUC in place of training gain, shown below.

109

Comparing the three jackknife plots can be very informative. The AUC plot shows that annual precipitation (pre6190_ann) is the most effective single variable for predicting the distribution of the

110 occurrence data that was set aside for testing, when predictive performance is measured using AUC, even though it was hardly used by the model built using all variables. The relative importance of annual precipitation also increases in the test gain plot, when compared against the training gain plot.

In addition, in the test gain and AUC plots, some of the light blue bars (especially for the monthly precipitation variables) are longer than the red bar, showing that predictive performance improves when the corresponding variables are not used.

This tells us that monthly precipitation variables are helping Maxent to obtain a good fit to the training data, but the annual precipitation variable generalizes better, giving comparatively better results on the set-aside test data. Phrased differently, models made with the monthly precipitation variables appear to be less transferable. This is important if our goal is to transfer the model, for example by applying the model to future climate variables in order to estimate its future distribution under climate change.

It makes sense that monthly precipitation values are less transferable: likely suitable conditions for

Bradypus will depend not on precise rainfall values in selected months, but on the aggregate average rainfall, and perhaps on rainfall consistency or lack of extended dry periods. When we are modeling on a continental scale, there will probably be shifts in the precise timing of seasonal rainfall patterns, affecting the monthly precipitation but not suitable conditions for Bradypus.

In general, it would be better to use variables that are more likely to be directly relevant to the species being modeled. For example, the Worldclim website (www.worldclim.org) provides “BIOCLIM” variables, including derived variables such as “rainfall in the wettest quarter”, rather than monthly values.

A last note on the jackknife outputs: the test gain plot shows that a model made only with January precipitation (pre6190_l1) results in a negative test gain. This means that the model is slightly worse

111 than a null model (i.e., a uniform distribution) for predicting the distribution of occurrences set aside for testing. This can be regarded as more evidence that the monthly precipitation values are not the best choice for predictor variables.

How does the prediction depend on the variables?

Now press the “Create response curves”, deselect the jackknife option, and rerun the model. This results in the following section being added to the “bradypus.html” file:

112

Each of the thumbnail images can be selected (by clicking on them) to obtain a more detailed plot, and if you would like to copy or open these plots with other software, the .png files can be found in the

“plots” directory. Looking at frs6190_ann, we see that the response is high for the smallest values of frs6190_ann (close to 0), and quickly drops toward 0. The value shown on the y-axis is predicted probability of suitable conditions, as given by the logistic output format, with all other variables set to their average value over the set of presence localities.

Note that if the environmental variables are correlated, as they are here, the marginal response curves can be misleading. For example, if two closely correlated variables have response curves that are near opposites of each other, then for most pixels, the combined effect of the two variables may be small.

As another example, we see that predicted suitability is negatively correlated with annual precipitation

(pre6190_ann), if all other variables are held fixed. In other words, once the effect of all the other variables has already been accounted for, the marginal effect of increasing annual precipitation is to decrease predicted suitability. However, annual precipitation is highly correlated with the monthly precipitation variables, so in reality we cannot easily hold the monthly values fixed while varying the annual value. The program therefore produces a second set of response curves, in which each curve is made by generating a model using only the corresponding variable, disregarding all other variables:

113

In contrast to the marginal response to annual precipitation in the first set of response curves, we now see that predicted suitability generally increases with increasing annual precipitation.

114

Feature types and response curves

Response curves allow us to see the difference among different feature types. Deselect the “auto features”, select “Threshold features”, and press the “Run” button again. Take a look at the resulting feature profiles – you’ll notice that they are all step functions, like this one for pre6190_l10:

If the same run is done using only hinge features, the resulting feature profile looks like this:

115

The outlines of the two profiles are similar, but they differ because different feature types allow different possible shapes of response curves. The exponent in a Maxent model is a sum of features, and a sum of threshold features is always a step function, so the logistic output is also a step function

(as are the raw and cumulative outputs). In comparison, a sum of hinge features is always a piece-wise linear function, so if only hinge features are used, the Maxent exponent is piece-wise linear. This explains the sequence of connected line segments in the second response curve above. (Note that the lines are slightly curved, especially towards the extreme values of the variable; this is because the logistic output applies a sigmoid function to the Maxent exponent.) Using all classes together (the default, given enough samples) allows many complex responses to be accurately modeled. A deeper explanation of the various feature types can be found by clicking on the help button.

Interactive exploration of predictions: the Explain tool

116

This interactive tool allows you to investigate how Maxent’s prediction is determined by the predictor variables across a study area. Clicking on a point on the map shows its location in each response curve. The top right graph shows how much each variable contributes to the logit of the prediction

(pointing at a bar on the graph gives the variable name and numerical contribution). By observing the contributions to the logit, you will see how the Maxent prediction is driven by different variables in different parts of the region.

The tool requires the model to be additive (without interactions between variables), so it should only be run on the output of a runs without product features. Your computer needs enough memory to hold all predictor variables at once. If you do a run without product features, a clickable link to the

Explain tool is included after the main picture of the model.

SWD Format

Another input format can be very useful, especially when your environmental grids are very large. For lack of a better name, it’s called “samples with data”, or just SWD. The SWD version of our Bradypus file, called “bradypus_swd.csv”, starts like this:

species,longitude,latitude,cld6190_ann,dtr6190_ann,ecoreg,frs6190_ann,h_dem,pre6190_ann,pre6190_l10,pre6190_l1,pre6190_l4,pre61

90_l7,tmn6190_ann,tmp6190_ann,tmx6190_ann,vap6190_ann bradypus_variegatus,-65.4,-10.3833,76.0,104.0,10.0,2.0,121.0,46.0,41.0,84.0,54.0,3.0,192.0,266.0,337.0,279.0 bradypus_variegatus,-65.3833,-10.3833,76.0,104.0,10.0,2.0,121.0,46.0,40.0,84.0,54.0,3.0,192.0,266.0,337.0,279.0 bradypus_variegatus,-65.1333,-16.8,57.0,114.0,10.0,1.0,211.0,65.0,56.0,129.0,58.0,34.0,140.0,244.0,321.0,221.0 bradypus_variegatus,-63.6667,-17.45,57.0,112.0,10.0,3.0,363.0,36.0,33.0,71.0,27.0,13.0,135.0,229.0,307.0,202.0 bradypus_variegatus,-63.85,-17.4,57.0,113.0,10.0,3.0,303.0,39.0,35.0,77.0,29.0,15.0,134.0,229.0,306.0,202.0

117

It can be used in place of an ordinary samples file. The difference is only that the program doesn’t need to look in the environmental layers (the ascii files) to obtain values for the variables at the sample points, instead it reads the values for the environmental variables directly from the table. The environmental layers are thus only used to read the environmental data for the “background” pixels – pixels where the species hasn’t necessarily been detected. In fact, the background pixels can also be specified in a SWD format file. The file “background.csv” contains 10,000 background data points.

The first few look like this:

background,-61.775,6.175,60.0,100.0,10.0,0.0,747.0,55.0,24.0,57.0,45.0,81.0,182.0,239.0,300.0,232.0 background,-66.075,5.325,67.0,116.0,10.0,3.0,1038.0,75.0,16.0,68.0,64.0,145.0,181.0,246.0,331.0,234.0 background,-59.875,-26.325,47.0,129.0,9.0,1.0,73.0,31.0,43.0,32.0,43.0,10.0,97.0,218.0,339.0,189.0 background,-68.375,-15.375,58.0,112.0,10.0,44.0,2039.0,33.0,67.0,31.0,30.0,6.0,101.0,181.0,251.0,133.0 background,-68.525,4.775,72.0,95.0,10.0,0.0,65.0,72.0,16.0,65.0,69.0,133.0,218.0,271.0,346.0,289.0

We can run Maxent with “bradypus_swd.csv” as the samples file and “background.csv” (both located in the “swd” directory) as the environmental layers file. Try running it – you’ll notice that it runs much faster, because it doesn’t have to load the large environmental grids. Another advantage is that you can associate different records with environmental conditions from different time periods. For example, two occurrences recorded 100 years apart from the same grid cell probably reflect considerable variation in environmental conditions, but unless you use SWD format, both records would be given the same environmental variables values. The downside is that it can’t make pictures or output grids, because it doesn’t have all the environmental data. The way to get around this is to use a “projection”, described below.

118

Batch running

Sometimes you need to generate multiple models, perhaps with slight variations in the modeling parameters or the inputs. Generation of models can be automated with command-line arguments, obviating the need to click and type repetitively at the program interface. The command line arguments can either be given from a command window (a.k.a. shell), or they can be defined in a batch file. Take a look at the file “batchExample.bat” (for example, right click on the .bat file inWindows

Explorer and open it using Notepad). It contains the following line:

java -mx512m -jar maxent.jar environmentallayers=layers togglelayertype=ecoreg samplesfile=samples\bradypus.csv outputdirectory=outputs redoifexists autorun

The effect is to tell the program where to find environmental layers and samples file and where to put outputs, to indicate that the ecoreg variable is categorical. The “autorun” flag tells the program to start running immediately, without waiting for the “Run” button to be pushed. Now try double clicking on the file to see what it does.

Many aspects of the Maxent program can be controlled by command-line arguments – press the

“Help” button to see all the possibilities. Multiple runs can appear in the same file, and they will simply be run one after the other. You can change the default values of parameters by adding command-line arguments to the “maxent.bat” file. Many of the command-line arguments also have abbreviations, so the run described in batchExample.bat could also be initiated using this command:

java -mx512m -jar maxent.jar –e layers –t eco –s samples\bradypus.csv –o outputs –r -a Replication

119

The "replicates" option can be used to do multiple runs for the same species. The most common uses for this flag are for repeated subsampling and for cross-validation. Replication can be controlled either from the Settings panel, or using command line arguments. By default, the form of replication used is cross-validation, where the occurrence data is randomly split into a number of equal-size groups called “folds”, and models are created leaving out each fold in turn. The left-out folds are then used for evaluation. Cross-validation has one big advantage over using a single training/test split: it uses all of the data for validation, thus making better use of small data sets. As an example, doing a run with the number of replicates set to 10 creates 10 html pages, plus a page that summarizes statistical information for the cross-validation. For example, we get ROC curves with error bars and average AUC across models, and summary response curves with one standard deviation error bars. For Bradypus, the cross-validated

ROC curve shows some variability between models:

120

The single-variable response of Bradypus to annual precipitation shows little variation (on the left, below), while the marginal response to annual precipitation is more variable (below, right).

Two alternative forms of replication are supported: repeated subsampling, in which the presence points are repeatedly split into random training and testing subsets, and bootstrapping, where the training data is selected by sampling with replacement from the presence points, with the number of samples equaling the total number of presence points. With bootstrapping, the number of presence points in each set equals the total number of presence points, so the training data sets will have duplicate records.

With all three forms of replication, you may want to avoid eating up disk space by turning off the

“write output grids” option, which will suppress writing of output grids for the replicate runs, so that you only get the summary statistics grids (avg, stderr etc.).

121

Regularization.

The “regularization multiplier” parameter on the settings panel affects how focused or closely-fitted the output distribution is – a smaller value than the default of 1.0 will result in a more localized output distribution that is a closer fit to the given presence records, but can result in to overfitting (fitting so close to the training data that the model doesn’t generalize well to independent test data). A larger regularization multiplier will give a more spread out, less localized prediction. Try changing the multiplier, and examine the pictures produced and changes in the AUC. As an example, setting the multiplier to 3 makes the following picture, showing a much more diffuse distribution than before:

122

The potential for overfitting increases as the model complexity increases. First try setting the multiplier very small (e.g. 0.01) with the default set of features to see a highly overfit model. Then try the same regularization multiplier with only linear and quadratic features.

Projecting

A model trained on one set of environmental layers (or SWD file) can be “projected” by applying it to another set of environmental layers (or SWD file). Situations where projections are needed include modeling species distributions under changing climate conditions, applying a model of the native distribution of an invasive species to assess invasive risk in a different geographic area, or simply evaluating the model at a set of test locations in order to do further statistical analysis. Here we’re going to use projection for a simplistic climate change prediction, and to give a taste of the difficulties involved in making reliable predictions of distributions under climate change.

The directory “hotlayers” has the same environmental data as the “layers” directory, with two changes: the annual average temperature variable (tmp6190_ann.asc) has all values increased by 30, representing a uniform 3 degree Celsius increase, while the maximum temperature variable

(tmx6190_ann.asc) has all values increased by 40, representing a 4 degree Celsius increase. These changes represent a very simplified estimate of future climate, with higher average temperature and higher temperature variability, but with no change in precipitation. To apply a model of Bradypus to this new climate, enter the samples file and current environmental data as before, using either grids or

SWD format, and enter the “hotlayers” directory in the “Projection Layers Directory”, as pictured below.

123

The projection layers directory (or SWD file) must contain variables with the same names as the variables used for training the model, but describing a different conditions (e.g., a different geographic region or different climatic model). For both the training and projection data, each variable name is either the column title (if using an SWD format file) or the filename without the .asc file ending (if using a directory of grids).

When you press “Run”, a model is trained on the environmental variables corresponding to current climate conditions, and then projected onto the ascii grids in the “hotlayers” directory. The output ascii grid is called “bradypus_variegatus_hotlayers.asc”, and in general, the projection directory name is appended to the species name, in order to distinguish it from the standard (un-projected) output. If

124

“make pictures of predictions” is selected, a picture of the projected model will appear in the

“bradypus.html” file. In our case, this produces the following picture:

We see that the predicted probability of presence is drastically lower under the warmer climate. The prediction is of course dependent on the parameters of the model we’re projecting. If we use only hinge and categorical features, rather than the default set of features, the projected distribution is more substantial:

125

Two different models that look very similar in the area used for training may look very different when projected to a new geographic area or new climate conditions. This is especially true if there are correlated variables that allow a variety of ways to fit similar-looking models, since the correlations between the variables may change in the area you’re projecting to.

126

Is the predicted range reduction of Bradypus under climate change reasonable? If we look at the marginal response curves for the model made with default features, we see that the maximum temperature is exerting a much stronger influence on the prediction:

Looking at a histogram of maximum temperature values at the known occurrences for Bradypus, we see that most occurrences (about 80%) have maximum temperature between 30 and 34 degrees

Celsius. Only a single occurrence is above 34 degrees, even though a significant fraction of the background is between 34 and 35 degrees.

127

Under our climate change prediction, all 80% of the Bradypus locations currently above 30 degrees will have the maximum temperature increase to above 34 degrees. Therefore it may indeed be reasonable to predict that such locations will not be suitable for Bradypus, so Bradypus might not survive in most of its current range. Note that it is difficult to make any conclusions about why such conditions are not suitable: it may be that Bradypus is intolerant to heat, or it may be that higher maximum temperatures would allow fire to cause widespread replacement of rainforest by fire-tolerant tree species, eliminating most suitable habitat for Bradypus. To further investigate the prospects of

Bradypus under climate change, we could do physiological studies to investigate the species’ tolerance for heat, or study the fire ecology of rainforest boundaries in the region.

Note: histograms like the two above are useful tools for investigating your data. They were made in R using the following commands: swdPresence <- read.csv("swd/bradypus_swd.csv") hist(swdPresence$tmx6190_ann, probability=TRUE, breaks=c(5:37*10), xlab="Annual maximum temp * 10", main="Bradypus presence points") swdBackground <- read.csv("swd/background.csv") hist(swdBackground$tmx6190_ann, probability=TRUE, breaks=c(5:37*10), xlab="Annual maximum temp * 10", main="Background points")

We can see from the histograms that Bradypus can occasionally tolerate high temperatures, as evidenced by the single record with maximum temperature of 35 degrees. On the other had, there are extremely few points in the background with temperatures of 36 or above, so we have no evidence of whether or not Bradypus can tolerate even higher temperatures, which will be widespread under the future climate prediction. This is known as the problem of novel climate conditions: when projecting, the predictor variables may take on values outside the range seen during model training. The primary

128 way Maxent deals with this problem is “clamping”, which treats variables outside the training range as if they were at the limit of the training range. This effect can be seen in the response curves described above, as the response is held constant outside the training range. Whenever a model is projected,

Maxent makes a picture that shows where clamping has had a large effect. Projecting the Bradypus model made with all features gives this clamping picture, where the values depicted are the absolute difference between predictions with and without clamping.

Clamping has clearly had little effect in this case – in particular, the response curve for maximum temperature above shows that the prediction had already leveled off near zero at the hot end of the scale, so clamping has little effect.

We also compare the environmental variables used for projection to those used for training the model.

After the clamping map, we see the following two pictures:

129

130

The leftmost picture is a multivariate similarity surface (MESS), as described in Elith et al., Methods in Ecology and Evolution, 2010. It shows how similar each point is in hotlayers to conditions seen during model training. Negative values (shown in red) indicate novel climate, i.e., hotlayers values outside the range in layers. The value shown is the minimum over the predictors of how far out of range the point is, expressed as a fraction of the range of that predictor’s values in layers. Positive values (shown in blue) are similar to BIOCLIM values, with a score of 100 meaning that a point is not at all novel, in the sense that its hotlayers values are all exactly equal to the median value in layers.

The picture on the right shows the most dissimilar variable (MoD), and as we would expect, it shows that novel climate conditions in hotlayers are due to average temperature (mauve, mostly north of the

Amazon River) or maximum temperature (teal blue, mostly south of the Amazon) being outside of the training range.

Masks

A mask variable is useful if you want to train a model using only a subset of the region. For example, we may want to train a model for Bradypus based on occurrences in Central America, and then project the model onto South America. To do this, we create a new “predictor” variable (called mask.asc, for example) with the same dimensions, cell size and projection as the environmental variables, containing a constant value (1, for example) throughout Central America and no-data values everywhere else.

The mask variable is placed in the same directory as the environmental variables, and is treated the same way as the other environmental variables. Because it is constant, it is never used in the model, but the no-data values serve to restrict model training to Central America.

To project the resulting model onto South America, we would create a new directory containing copies of all the environmental variables, together with a new mask variable (also called mask.asc), that is

131 equal to 1 throughout South America, and has no-data values elsewhere. This new directory is given as a “projection layers” argument to Maxent.

Bias grids

By default, when using Maxent we make the assumption that species occurrence data are unbiased, independent samples from the distribution of the species. The assumption of lack of bias is easily violated, for example if sample collection effort is biased towards more easily accessed areas such as areas close to roads or population centers. If you believe that your species occurrence data constitute a biased sample, and you have a good understanding of the spatial pattern of sample collection effort that produced your occurrence data, you can provide Maxent with a “bias grid” which is then used to correct for the bias. The bias grid should have the same dimensions, cell size and projection as the environmental variables, and should be positive (or no-data) everywhere. The values should indicate relative sampling effort, so if two cells have values 1 and 2, that means the probability of having visited the second cell is twice as high as the first. Note that the bias grid gives a priori relative sampling probabilities; it does not indicate where sampling actually happened.

Additional command-line tools

The Maxent jar file contains a number of tools that can be accessed from the command line. For

Microsoft users: the features described here can be used in a batch file, like maxent.bat. As an alternative, Start->run->cmd gets you a shell for running commands interactively; cygwin (available free online) is a good alternative with a much more powerful shell that offers many unix utilities.

132

Quick visualization of grid file

Grid files in .asc, .grd and .mxe format, and some files in .bil format, can be viewed using the following command:

java -mx512m -cp maxent.jar density.Show filename

As with all the commands described below, you may need to add the path to the maxent.jar file and/or the file you want to view. For example, you might use:

java -mx1000m -cp C:\maxentfiles\maxent.jar density.Show C:\mydata\var1.asc

Show can take some optional arguments (immediately after density.Show):

-s sampleFile gives a file with presences to be shown in white dots

-S speciesname says which species in the sampleFile to show with dots

-r radius controls the size of the white and purple dots for occurrence records

-L removes the legend

-o writes the picture to a file in .png format

With a little Windows wizardry, you can make Show be invoked just by clicking on .asc, .grd or .mxe files. Make a batch file, say called showFile.bat, with the following single line in it:

java -mx512m -cp "c:\maxentfiles\maxent.jar" density.Show %1

133 then associate files of type .asc, .grd or .mxe with the batch file: from a windows explorer (a.k.a. "My

Computer"), Tools->Folder Options->File Types... You may need to make the batch file executable: right click on it and follow directions.

Making an SWD file

To make an SWD-format file from a non-SWD file:

java -cp maxent.jar density.Getval samplesfile grid1 grid2 ...

where samplesfile is .csv file of occurrence data and grid1, grid2, etc. are grids in .asc, .mxe, .grd or

.bil format. The output is written to "standard output", which means it appears in the command window. To write the output to a file, use a "redirect":

java -cp maxent.jar density.Getval samplesfile grid1 grid2 ... > outfile

If all the grids are in a directory you can avoid having to list them all by name by using a "wildcard":

java -cp maxent.jar density.Getval samplesfile directory/*.asc ... > outfile

because the wildcard (*) gets expanded to a list of all files that match.

Making an SWD background file

134

To pick a collection of background points uniformly at random from your study area:

java -cp maxent.jar density.tools.RandomSample num grid1 grid2 ...

where "num" is the number of background points desired.

Calculating AUC

The following command:

java -cp maxent.jar density.AUC testpointfile predictionfile

will calculate a presence-background AUC, where the presence points are given in the testpointfile and background points are drawn randomly from the predictionfile. The testpointfile is a .csv file (which may optionally be swd format), while the predictionfile is a grid file, typically representing the output of a species distribution model.

Projection

This tool allows you to apply a previously-calculated Maxent model to a new set of environmental data:

java -cp maxent.jar density.Project lambdaFile gridDir outFile [args]

135

Here lambdaFile is a .lambdas file describing a Maxent model, and gridDir is a directory containing grids for all the predictor variables described in the .lambdas file. As an alternative, gridDir could be an swd format file. The optional args can contain any flags understood by Maxent -- for example, a

"grd" flag would make the output grid of density.Project be in .grd format.

File conversion

To convert a directory full of grids in one format to another:

java -cp maxent.jar density.Convert indir insuffix outdir outsuffix

where indir and outdir are directories and insuffix and outsuffix are one of asc, mxe, grd or bil.

136

Analyzing Maxent output in R

Maxent produces a number of output files for each run. Some of these files can be imported into other programs if you want to do your own analysis of the predictions. Here we demonstrate the use of the free statistical package R on Maxent outputs: this section is intended for users who have experience with R. We will use the following two files produced by Maxent:

bradypus_variegatus_backgroundPredictions.csv

bradypus_variegatus_samplePredictions.csv

The first file is only produced when the “writebackgroundpredictions” option is turned on, either by using a command-line flag or by selecting it from Maxent’s settings panel. The second file is always produced. Make sure you have test data (for example, by setting the random test percentage to 25); we will be evaluating the Maxent outputs using the same test data Maxent used. First we start R, and install some packages (assuming this is the first time we’re using them) and then load them by typing

(or pasting):

install.packages("ROCR", dependencies=TRUE)

install.packages("vcd", dependencies=TRUE)

library(ROCR)

library(vcd)

library(boot)

Throughout this section we will use blue text to show R code and commands and green to show R outputs. Next we change directory to where the Maxent outputs are, for example:

137

setwd("c:/maxent/tutorial/outputs")

and then read in the Maxent predictions at the presence and background points, and extract the columns we need:

presence <- read.csv("bradypus_variegatus_samplePredictions.csv")

background <- read.csv("bradypus_variegatus_backgroundPredictions.csv")

pp <- presence$Logistic.prediction # get the column of predictions

testpp <- pp[presence$Test.or.train=="test"] # select only test points

trainpp <- pp[presence$Test.or.train=="train"] # select only test points

bb <- background$logistic

Now we can put the prediction values into the format required by ROCR, the package we will use to do some ROC analysis, and generate the ROC curve:

combined <- c(testpp, bb) # combine into a single vector

label <- c(rep(1,length(testpp)),rep(0,length(bb))) # labels: 1=present, 0=random

pred <- prediction(combined, label) # labeled predictions

perf <- performance(pred, "tpr", "fpr") # True / false positives, for ROC curve

plot(perf, colorize=TRUE) # Show the ROC curve

performance(pred, "auc")@y.values[[1]] # Calculate the AUC

The plot command gives the following result:

138

while the “performance” command gives an AUC value of 0.8677759, consistent with the AUC reported by Maxent. Next, as an example of a test available in R but not in Maxent, we will make a bootstrap estimate of the standard deviation of the AUC.

AUC <- function(p,ind) {

pres <- p[ind]

combined <- c(pres, bb)

label <- c(rep(1,length(pres)),rep(0,length(bb)))

predic <- prediction(combined, label)

return(performance(predic, "auc")@y.values[[1]])

}

139

b1 <- boot(testpp, AUC, 100) # do 100 bootstrap AUC calculations

b1 # gives estimates of standard error and bias

This gives the following output:

ORDINARY NONPARAMETRIC BOOTSTRAP

Call:

boot(data = testpp, statistic = AUC, R = 100)

Bootstrap Statistics :

original bias std. error t1* 0.8677759 -0.0003724138 0.02972513

and we see that the bootstrap estimate of standard error (0.02972513) is close to the standard error computed by Maxent (0.028). The bootstrap results can also be used to determine confidence intervals for the AUC:

boot.ci(b1)

gives the following four estimates – see the resources section at the end of this tutorial for references that define and compare these estimates.

Intervals :

140

Level Normal Basic

95% ( 0.8099, 0.9264 ) ( 0.8104, 0.9291 )

Level Percentile BCa

95% ( 0.8064, 0.9252 ) ( 0.7786, 0.9191 )

Those familiar with use of the bootstrap will notice that we are bootstrapping only the presence values here. We could also bootstrap the background values, but the results would not change much, given the very large number of background values (10000).

As a final example, we will investigate the calculation of binomial and Cohen’s Kappa statistics for some example threshold rules. First, the following R code calculates Kappa for the threshold given by the minimum presence prediction:

confusion <- function(thresh) {

return(cbind(c(length(testpp[testpp>=thresh]), length(testpp[testpp

c(length(bb[bb>=thresh]), length(bb[bb

}

mykappa <- function(thresh) {

return(Kappa(confusion(thresh)))

}

mykappa(min(trainpp))

141 which gives a value of 0.0072. If we want to use the threshold that minimizes the sum of sensitivity and specificity on the test data, we can do the following, using the true positive rate and false positive rate values from the “performance” object used above to plot the ROC curve:

fpr = [email protected][[1]]

tpr = [email protected][[1]]

sum = tpr + (1-fpr)

index = which.max(sum)

cutoff = [email protected][[1]][[index]]

mykappa(cutoff)

This gives a kappa value of 0.0144. To determine binomial probabilities for these two threshold values, we can do:

mybinomial <- function(thresh) {

conf <- confusion(thresh)

trials <- length(testpp)

return(binom.test(conf[[1]][[1]], trials, conf[[1,2]] / length(bb), "greater"))

}

mybinomial(min(trainpp))

mybinomial(cutoff)

This gives p-values of 5.979e-09 and 2.397e-11 respectively, which are both slightly larger than the p- values given by Maxent. The reason for the difference is that the number of test samples is greater

142 than 25, the threshold above which Maxent uses a normal approximation to calculate binomial p- values.

R Resources

Some good introductory material on using R can be found at:

http://spider.stat.umn.edu/R/doc/manual/R-intro.html, and other pages at the same site.

http://www.math.ilstu.edu/dhkim/Rstuff/Rtutor.html

143

APPENDIX C

Tutorial on running linear models in R

Written by

The design was inspired by the S function of the same name described in

Chambers (1992). The implementation of model formula by Ross Ihaka was based on Wilkinson & Rogers (1973).

144

lm {stats} R Documentation

Fitting Linear Models

DESCRIPTION

lm is used to fit linear models. It can be used to carry out regression, single stratum analysis of variance and analysis of covariance (although aov may provide a more convenient interface for these).

USAGE lm(formula, data, subset, weights, na.action,

method = "qr", model = TRUE, x = FALSE, y = FALSE, qr = TRUE,

singular.ok = TRUE, contrasts = NULL, offset, ...)

ARGUMENTS

formula an object of class "formula" (or one that can be coerced to that class): a

symbolic description of the model to be fitted. The details of model

specification are given under ‘Details’.

data an optional data frame, list or environment (or object coercible

by as.data.frame to a data frame) containing the variables in the model. If

not found in data, the variables are taken fromenvironment(formula), typically

the environment from which lm is called.

145

subset an optional vector specifying a subset of observations to be used in the

fitting process.

weights an optional vector of weights to be used in the fitting process. Should

be NULL or a numeric vector. If non-NULL, weighted least squares is

used with weights weights (that is, minimizingsum(w*e^2)); otherwise

ordinary least squares is used. See also ‘Details’,

na.action a function which indicates what should happen when the data

contain NAs. The default is set by the na.action setting of options, and

is na.fail if that is unset. The ‘factory-fresh’ default isna.omit. Another

possible value is NULL, no action. Value na.exclude can be useful.

method the method to be used; for fitting, currently only method = "qr" is

supported; method = "model.frame" returns the model frame (the same as

with model = TRUE, see below).

model, x, logicals. If TRUE the corresponding components of the fit (the model y, qr frame, the model matrix, the response, the QR decomposition) are

returned.

singular.ok logical. If FALSE (the default in S but not in R) a singular fit is an error.

contrasts an optional list. See the contrasts.arg of model.matrix.default.

offset this can be used to specify an a priori known component to be included

in the linear predictor during fitting. This should be NULL or a numeric

vector of length equal to the number of cases. One or more offset terms

146

can be included in the formula instead or as well, and if more than one

are specified their sum is used. See model.offset.

... additional arguments to be passed to the low level regression fitting

functions (see below).

DETAILS

Models for lm are specified symbolically. A typical model has the form response ~ terms where response is the (numeric) response vector and terms is a series of terms which specifies a linear predictor forresponse. A terms specification of the form first + second indicates all the terms in first together with all the terms in second with duplicates removed. A specification of the form first:second indicates the set of terms obtained by taking the interactions of all terms in first with all terms in second. The specification first*second indicates the cross of first and second. This is the same as first

+ second + first:second.

If the formula includes an offset, this is evaluated and subtracted from the response.

If response is a matrix a linear model is fitted separately by least-squares to each column of the matrix.

See model.matrix for some further details. The terms in the formula will be re-ordered so that main effects come first, followed by the interactions, all second-order, all third-order and so on: to avoid this pass a terms object as the formula

(see aov and demo(glm.vr) for an example).

147

A formula has an implied intercept term. To remove this use either y ~ x - 1 or y ~ 0 + x. See formula for more details of allowed formulae.

Non-NULL weights can be used to indicate that different observations have different variances (with the values in weights being inversely proportional to the variances); or equivalently, when the elements ofweights are positive integers w_i, that each response y_i is the mean of w_i unit-weight observations (including the case that there are w_i observations equal to y_i and the data have been summarized).

lm calls the lower level functions lm.fit, etc, see below, for the actual numerical computations. For programming only, you may consider doing likewise.

All of weights, subset and offset are evaluated in the same way as variables in formula, that is first in data and then in the environment of formula.

VALUE

lm returns an object of class "lm" or for multiple responses of class c("mlm", "lm").

The functions summary and anova are used to obtain and print a summary and analysis of variance table of the results. The generic accessor functions coefficients, effects, fitted.values and residualsextract various useful features of the value returned by lm.

An object of class "lm" is a list containing at least the following components:

148

coefficients a named vector of coefficients

residuals the residuals, that is response minus fitted values.

fitted.values the fitted mean values.

rank the numeric rank of the fitted linear model.

weights (only for weighted fits) the specified weights.

df.residual the residual degrees of freedom.

call the matched call.

terms the terms object used.

contrasts (only where relevant) the contrasts used.

xlevels (only where relevant) a record of the levels of the factors used in fitting.

offset the offset used (missing if none were used).

y if requested, the response used.

x if requested, the model matrix used.

model if requested (the default), the model frame used.

na.action (where relevant) information returned by model.frame on the special

handling of NAs.

In addition, non-null fits will have components assign, effects and (unless not requested) qr relating to the linear fit, for use by extractor functions such as summary and effects.

149

USING TIME SERIES

Considerable care is needed when using lm with time series.

Unless na.action = NULL, the time series attributes are stripped from the variables before the regression is done. (This is necessary as omitting NAs would invalidate the time series attributes, and if NAs are omitted in the middle of the series the result would no longer be a regular time series.)

Even if the time series attributes are retained, they are not used to line up series, so that the time shift of a lagged or differenced regressor would be ignored. It is good practice to prepare a data argument byts.intersect(..., dframe = TRUE), then apply a suitable na.action to that data frame and call lm with na.action = NULL so that residuals and fitted values are time series.

NOTE

Offsets specified by offset will not be included in predictions by predict.lm, whereas those specified by an offset term in the formula will be.

REFERENCES

Chambers, J. M. (1992) Linear models. Chapter 4 of Statistical Models in S eds

J. M. Chambers and T. J. Hastie, Wadsworth & Brooks/Cole.

150

Wilkinson, G. N. and Rogers, C. E. (1973) Symbolic descriptions of factorial models for analysis of variance. Applied Statistics , 22, 392–9.