As the human footprint on natural ecosystems continues to expand, urbanization has become one of the greatest challenges for biodiversity (McKinney, 2008). Approximately 55% of human population is concentrated in urban areas and this proportion is expected to rise to nearly 68% by the year 2050 (United Nations, 2019). The conversion of natural habitats to cities involves drastic and often permanent changes to the physical structure of environments (Elmqvist et al., 2013; Uchida et al., 2021). For organisms living in these urbanized areas, such structural changes have downstream effects and can influence selection pressures (Sih et al., 2011). For example, the reduction in the diversity of plants in cities negatively impacts insect communities (Vergnes et al., 2014; Sánchez-Bayo and Wyckhuys, 2019), which has consequences for the diet, foraging strategies, and even survival probability of insectivorous birds and reptiles. Even though examples of declining biodiversity in urban areas are now prevalent globally (McKinney, 2008; Huang et al., 2018), what is increasingly being recognised is that the impact of urbanization on organismal responses varies considerably. For instance, species richness of plants can increase in suburban areas, consistent with the intermediate disturbance hypothesis, whereas richness of vertebrate communities often reduces in urbanized areas (McKinney, 2006). Furthermore, the size of a city and intensity of urbanization can synergistically influence how organisms are impacted (McKinney, 2006; Schneider and Woodcock, 2008; Uchida et al., 2021). The term “urban”, therefore, is applied broadly and can obscure the complexity and variation of the underlying ecological processes.

The survival probability and therefore presence of biodiversity in urban areas is directly influenced by how well species respond to the various environmental changes that arise from urbanization. Habitat loss alone accounts for most of biodiversity loss in many cities around the world (Marzluff, 2001; Lim and Sodhi, 2004; Scolozzi and Geneletti, 2012; Paul and Nagendra, 2015; Teitelbaum et al., 2020). Fragmentation or the lack of connectivity of the remaining natural habitats further reduces species richness, especially for species that require large areas for resource acquisition (McKinney, 2008; Vergnes et al., 2014; Beninde et al., 2015) or have low dispersal ability (Stefanescu et al., 2004; Evans et al., 2017). Habitat loss and fragmentation can also result in changes to animal movement patterns (Tucker et al., 2018), habitat use, and distribution (Weller and Ganzhorn, 2004). Given that urbanisation can negatively affect the survival of some species, species composition and community structure in urban areas are different from those in natural ecosystems (Jellinek et al., 2004; Olivier et al., 2020). As species composition changes along urban gradients, predation pressure may also vary (McKinney, 2006; Eötvös et al., 2018; Piontek et al., 2021). Increased predation pressure, especially from feral commensals such as dogs and cats have resulted in significant reductions in small mammal, bird, and reptile populations in urban areas (Iverson, 1978; Baker et al., 2005; Sudhira and Nagendra, 2013; French et al., 2018). Reduced or no change in predation pressure in some cities are also common (Balakrishna et al., 2021), and this can increase prey survival probability, but also increase intra-specific or inter-specific competition within the fragmented urban habitat. Among the abiotic factors that change with urbanization, temperature and artificial light are the most pervasive. Land Surface Temperature (LST) is typically higher in urban areas, resulting in what is known as the urban heat island effect (Peng et al., 2012; Lee et al., 2014). Several studies associate the reduction of vegetation and increase in LST as an immediate effect of urbanization (Nalini, 2021; Yamak et al., 2021). An increase in LST could, however, be beneficial to some species; for example, the increase in grey-headed flying foxes Pteropus poliocephalus that roost in Melbourne has been attributed to higher urban temperatures (Parris and Hazell, 2005). The ecological effects of artificial light at night (ALAN) are abundant (Frank et al., 2006; Gaston et al., 2013) and include changes to animal behaviour, physiology and reproduction (Newport et al., 2014), as well as species interactions (Knop et al., 2017). Overall, these numerous changes to environmental and ecological factors result in changes to species responses which in turn can affect biodiversity in urban environments.

Here, we quantify the ways in which urbanization changes the environment and examine how these potential environmental factors in turn influence the presence of the Peninsular rock agama (Psammophilus dorsalis) in one of the fastest growing cities in India. Among the reptile species found in the city of Bengaluru, P. dorsalis) is the most prominent lizard. This species is found across southern India, in semi-arid areas that are typically characterised by rocky habitat with boulders and sheet rocks interspersed with scrub vegetation (Radder et al., 2005). Despite considerable evidence of this species’ ability to be an urban utilizer (Amdekar et al., 2018; Batabyal and Thaker, 2019b), what is unclear is whether environmental factors influence their distribution and abundance in urban areas. For example, habitat composition, including vegetation cover is an important factor that influences insect diversity and density, the primary source of food for the insectivorous P. dorsalis (Balakrishna et al., 2016). The availability of rocks and boulders, or artificial substrates with similar structural properties, are used as active and sleep sites, and therefore, are also critical resources for this agamid lizard (Radder et al., 2005; Mohanty et al., 2021). Similarly, higher LST could be beneficial for efficient thermoregulation, whereas ALAN would be disruptive for sleep (Aulsebrook et al., 2018). Therefore, the presence and relative abundance of P. dorsalis is likely to be determined by processes operating at both the landscape and microhabitat scales.

We started by characterizing urbanization in Bengaluru, based on a range of environmental factors at the landscape scale. These factors include habitat composition, habitat connectivity and diversity, rate of habitat change, avian predation pressure, LST and ALAN, which were derived from remote sensed and citizen science data. We then examined how different landcover types and environmental factors change across two measures of urbanization: ‘distance from the city centre’, which captures the somewhat concentric pattern of urban expansion in Bengaluru, and ‘proportion of built-up’, which directly captures urbanization. We predict that these two measures of urbanization will be tightly correlated and that habitat connectivity, habitat diversity, rate of habitat change, and avian predation pressure would decrease with increasing degree of urbanization, whereas LST and ALAN would increase with urbanization. In addition to the landscape-level information, we captured microhabitat level data, including microhabitat composition and vegetation height, using targeted drone imagery and on-ground measurements in a smaller subset of locations across the study area. These data enabled us to examine the form and pattern of urbanization and to use a multi-scale approach to predict the presence of the agamid lizard species at the landscape scale and their abundance at the microhabitat scale. We predict that these various measures of urbanization will negatively impact both presence and abundance of the lizard in a rapidly growing metropolis in India.

Urbanization alters several biotic and abiotic components of an ecosystem. The patterns of change in these components greatly differ from one city to another, depending on various factors, including the size of the city, magnitude of change, and form of urban development (Uchida et al., 2021). Our study examines the magnitude and direction of change in different environmental factors in response to two measures of urbanization in Bengaluru, distance from city center, which reflects the mode of urban expansion, and proportion of built-up, which captures the degree of urbanization. We find that the relationship of only two factors, ALAN and habitat diversity, scaled in predictable but non-linear ways with these measures of urbanization. All other factors, such as LST, avian predation pressure, habitat contiguity (rocky-scrub), and rate of change (rocky-scrub) were highly variable and showed no consistent scaling across the extent of urbanization. For the iconic Peninsular rock agama, P. dorsalis, LST at the landscape scale and the proportion of rocky substrate at the microhabitat scale predicted their distribution in the city. Thus, although urbanization results in a range of human-induced rapid environmental change (Sih et al., 2011), many of these potential selection pressures may not directly influence species presence at relatively large or fine scales.

Our examination of environmental factors across varying degrees of urbanization in Bengaluru revealed several interesting patterns that reflect high heterogeneity and non-uniform scaling. Given the rapid expansion in the field of urban ecology in the last few decades, it is not surprising that measures of ‘urbanization’ vary between studies (Szulkin et al., 2020). For example, many studies consider the distance from the center of the city as a measure of urbanization while others use anthropogenic development or human population density. Different measures, however, capture different aspects of urbanization and can influence the interpretation of the impacts on ecosystems and biodiversity (Szulkin et al., 2020). Despite the strong correlation between ‘distance from the city center’ and ‘proportion of built-up’ for Bengaluru, we find that most ecological factors showed no discernable patterns or associations with either of these measures of urbanization.

Many studies have shown an ‘urban heat island effect’ as a common outcome of urbanization (Battles and Kolbe, 2019; Yamak et al., 2021). However, we find no evidence of elevated temperatures (LST) in the urban core or where anthropogenic built-up areas were highest. Lack of a positive relationship between urbanization and LST can be attributed to Bengaluru’s high altitude (~900m) location on the Deccan plateau and the relatively dense vegetation across the city (Siddiqui et al., 2021). As expected, ALAN was extremely high in the urbanized areas and this anthropogenic alteration to the environment rapidly reduced away from city center. Notably, intensity of ALAN was highly variable in the rural areas, where the ‘built-up’ landcover class was relatively low (ranging from 0-25%). For taxa that are negatively affected by ALAN, rural areas with high levels of night light (e.g., some villages, cropfields or plantations) may pose a challenge. We also found higher densities of avian predators closer to the city center as opposed to rural areas, suggesting that urbanized areas provide avian predators with sufficient refuge and prey to sustain abundant populations (McCabe et al., 2018). This pattern of higher avian predators in urban areas has also been documented in other Indian cities, such as New Delhi where raptors are subsidized by active feeding and the unrestricted access to food waste and their associated prey (Kumar et al., 2018). In addition to avian predators, feral dogs are known predators of P. dorsalis (Amdekar and Thaker, 2019), and although cities across India have relatively high populations of feral dogs (Home et al., 2018), information on the relative abundance and distribution patterns of dogs in Bengaluru is lacking. Thus, the spatial distribution of predation pressure for lizards may be incomplete.

The ongoing process of urbanization means that modifications to existing habitat structure can occur at different times and rates. Drastic changes to habitats seldom occur in highly urbanized areas due to the already saturated constructed spaces, whereas rapid changes may be evident in areas transitioning from rural to suburban spaces. In our study area, we find that the most urbanized area in Bengaluru is characterized by a highly developed ‘built-up’ core within 20 km of the city center. This core is surrounded by a matrix of cropland, plantations, rocky-scrub, and water bodies. The ‘rocky-scrub’ land cover class is among the most susceptible to anthropogenic alteration, but contrary to our expectation, the rate of change in this landcover type varied across our study site with no clear urban to rural gradient. In areas where a change in the proportion of habitat types was observed, the rocky-scrub habitat appeared to be converted to other land use classes, such as croplands or plantations, and to a lesser extent, into built-ups areas. Urbanization is typically characterized by a replacement of natural vegetation and terrain with artificial structures, leading to reduced habitat diversity and contiguity (McKinney, 2006; Uchida et al., 2021). Such changes to natural habitats were apparent in our study, as less urbanized areas showed higher habitat diversity and greater habitat contiguity. Despite the overall patterns of urban-induced change in the environmental factors that we quantified here, what was most apparent was the high variation across the entire study area. As others have indicated, urbanization develops non-uniformly, resulting in heterogenous patterns of important environmental drivers across the urban-rural gradient (McDonnell et al., 2012; Uchida et al., 2021). For example, a reduction of natural habitat contiguity combined with increasing LST could negatively impact insect abundance, which in turn can have bottom-up consequences for their predators. Landscape heterogeneity, therefore has the potential to result in emergent properties for the ecosystem, by disproportionately increasing the risk for some species, or the ecosystem, when particular environmental drivers interact (Newman et al., 2019).

For many species, urbanization increases selection pressures, but for others, urban-induced changes to the environment can relax the pressure. Relative importance of these selection pressures for species, however, may depend on their spatial scale. At the landscape scale, we find that LST predicted lizard presence across the city. As intuitively appealing as it is to attribute higher surface temperatures as a positive effect for an ectothermic species, such as P. dorsalis (Diamond et al., 2018; Diamond and Martin, 2020), this predictor explained limited variation in lizard presence at the landscape scale. LST is also positively correlated with the proportion of cropland, which suggests that these drivers may interact to influence the suitability of the habitat for lizards. In this region, agriculture is a mosaic of crops across small land holdings, and this heterogeneity of crops could support high insect diversity and availability within and around the cropland matrix (Sirami et al., 2019), thereby providing insect prey for lizards (Balakrishna et al., 2016). Alterations at the microhabitat scale may affect animal survival and distribution in a different way from those at the landscape scale. At the microhabitat scale, P. dorsalis was more likely to be found in sites with higher proportions of rocks, which is consistent with the general microhabitat preferences and geographic distribution for the species (Radder et al., 2005). None of the other large-scale or fine-scaled predictors adequately explained lizard abundance, suggesting that the environmental variation that we find at both spatial scales only weakly predict lizard presence, regardless of lizard density.

Effects of urbanization on the distribution and abundance of lizards in other studies are similarly inconsistent (French et al., 2018). Several lizards, such as Lacerta agilis, readily occupy urbanized and natural habitats (Becker and Buchholz, 2016), whereas other species, such as Anolis cristatellus and Plestiodon reynoldsi have been found to prefer natural habitats, with the latter being absent from urbanized areas altogether (Pike and Roznik, 2009; Kolbe et al., 2016; Battles and Kolbe, 2019). Many have recognized that heterogeneity in the way cities develop and the resulting mosaic of natural and modified habitats may better represent urban-induced environment change (McDonnell et al., 2012; Ramalho and Hobbs, 2012; Szulkin et al., 2020). Our approach to capture this heterogeneity was to calculate a habitat diversity index and a habitat connectivity index, but these measures showed extremely high spatial variation. Thus, like in many developing cities, the pattern of urbanization in Bengaluru is spatially variable (McDonnell et al., 2012; Szulkin et al., 2020). Temporal variation in anthropogenic development is an additional feature of urbanization that captures environmental stability (Ramalho and Hobbs, 2012). This is particularly relevant for species that are territorial, and are consequently exposed to habitat development within their dispersal range over their lifetime. Temporal dynamics of landscape change, measured as the rate of rocky-scrub conversion over a nine year period, did not seem to influence lizard distribution across the city. In spatio-temporally dynamic cities such as Bengaluru, the presence of P. dorsalis may depend on several factors that we could not directly quantify, such as the availability of insect prey and microhabitat-scale thermal heterogeneity.

Overall, our multiscale sampling approach and use of remote sensing data from both drone imagery and satellite sensors, allowed us to generate several insights about urbanization and its effects on lizards. Not surprisingly, we find that spatial and temporal changes to the environment caused by urbanization are highly variable and non-linear, and thus, urban-rural ‘gradients’ are difficult to characterize and generalize. The fact that environmental factors scale in different ways depending on whether urban intensity is measured as the distance from the city center or as the proportion of anthropogenic built-up reinforces the challenge in defining urbanization. The ways in which the environmental effects of urbanization scale across space can influence the relative importance of these potential selection factors for species survival (Uchida et al., 2021). For an urban utilizer like P. dorsalis, which is known to exhibit high behavioral and physiological flexibility in response to urbanization (Amdekar et al., 2018; Batabyal and Thaker, 2019a; Mohanty et al., 2021), most features of urbanization showed little effect on its distribution and abundance. However, not all species are expected to respond as urban utilizers (Fischer et al., 2015), and studies that increase spatial and temporal resolution of sampling will be needed to understand the distribution patterns of those taxa and of biodiversity in general. Whether P. dorsalis can sustain continual habitat and environmental changes during the urbanization process remains uncertain. We emphasize the need to examine urbanization as a suite of shifted selection pressures that are appropriately scaled to the taxa of interest (Szulkin et al., 2020), as a way to understand persistence and distribution in this novel and now ubiquitous habitat.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The animal study was reviewed and approved by Institutional Animal Ethics Committee, Indian Institute of Science.

MT and MA designed the project, MA and AN collected the field data, KS, HP, NM, AN conducted the GIS and statistical analyses, MT lead the writing with contributions from AN, HP, NM, and KS. All authors contributed to the article and approved the submitted version.

This project was supported by funds from the Environmental Management and Policy Research Institute (RES/CR-10/NSCC/LIZARDS/2015-16). The authors are being supported by the Long-term Ecological Observatories Program of the Ministry of Environment, Forests and Climate Change for HP, the Indian Institute of Science Raman fellowship program to NM, the Inspire Fellowship by the Department of Science and Technology to KS, and the DBT-Wellcome Trust Fellowship to MT (IA/I/19/2/504639).

We thank Mihir Joshi, Anuradha Batabyal and Arka Pal for help in the collection of field data. We also thank Dr. Sudhira HS, who provided inputs about Bengaluru and helped KS with QGIS.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.