by Siddharth Unnithan Kumar, Jonathon Turnbull, Oscar Hartman Davies, Samuel A. Cushman, and Timothy Hodgetts
In recent decades, ecologists have paid increasing attention to questions of how and why animals move. The interlinked biodiversity crises of the Anthropocene are, in several respects, linked to the capacity of wildlife to move: to adapt to changing climates; to navigate fragmented habitats; and to manoeuvre through landscapes marked by anthropogenic alteration. Ecological scientists track, model, and predict the movement of wildlife using a diversity of digital technologies to understand how animals live their lives. A large part of this work involves modelling connectivity: the extent to which landscapes linked together and enable organismal movement. Contemporary connectivity models, however, are often reductive as they are based on oversimplified renderings of the wildlife and landscapes they seek to model. Environmental historian Etienne Benson (2016, 39) argues that the tracking systems and computer programs that underpin much contemporary wildlife research have contributed to the emergence of the ‘minimal animal’: in other words, “an animal that is nothing but a stochastic pattern traced across a blank page.” Similarly, environmental writer Charles Bergman (2005, 266) makes comparable observations in his work on animal telemetry, in which he argues that through the use of tracking technologies the “animal seems to have become a beast with no body, a simulation of itself.”
The conversations we report on here, among a group of conservation scientists and geographers, arise from questioning what is lost when connectivity models simplify not just the animal, but whole ecologies – or webs of relations. We discuss how existing models in the current paradigm of ecological science can be improved, whilst also reflecting on what they exclude. As a result, we identify several questions that require cross-disciplinary collaboration – a research agenda of sorts, which seeks to revitalise ecological models of movement with a fuller conceptualisation of ‘life’ itself.
To begin, we would like to share how this interdisciplinary conversation came into being. Modelling connectivity involves the use of a range of digital technologies. Particularly widespread is the use of ‘resistance surfaces’, which ecological scientists use to quantify landscape structure, in order to predict how an animal will move according to various ‘push and pull’ factors in its environment. These factors are represented as numerical ‘resistance values’ and are assigned to tiles that surround the ‘digital animal’ (Adams, 2019). Sid, a mathematical ecologist, perhaps ecological mathematician, tasked with developing such models, felt dissatisfied with their inadequacy for capturing the vibrancy of the worlds they are deployed to study. He was not the first ecologist to have such concerns. Nevertheless, within the field of connectivity modelling, Sid could not find literature that studied the various mechanistic and binary assumptions in these models with a much-needed critical or inquisitive lens. In the process of seeking more sensuous and embodied approaches, he stumbled upon the field of more-than-human geography after several spontaneous and mind-opening conversations with colleagues and friends (particularly his housemate Kayla). Thus, Sid, joined by his PhD supervisor – Sam, an ecological scientist – initiated conversations with Jonny, Oscar, and Tim who are more-than-human geographers with interests in animal mobilities and digital ecologies. Together, we began to consider how we might question (and look deeper into) the paradigm of resistance surfaces in connectivity modelling, bringing our respective approaches together to explore how they may lead in exciting new directions for us and others.
In this essay, we trace a brief history of resistance surfaces and examine in more detail the problems they present to researchers interested in animal movement, connectivity, and wildlife management. We develop the concept of ‘minimal ecologies’ to account for the sparse depictions of organismal lives, relations and experiences in these models, and reflect on how such models might be re-imagined through cross-disciplinary collaborations.
CONNECTIVITY AND RESISTANCE SURFACES: A BRIEF HISTORY
Ecological connectivity has become a focal subject of research and action in conservation science and policy in recent decades. Analytical methods and predictive algorithms concerning connectivity have proliferated since the 1990s. The concepts underpinning these approaches, however, can be traced further back to the transport geography of the late 1950s and 1960s. William Warntz, whose work focused on using mathematical approaches to spatial analysis in geography, is often credited with first expressing the concept of least-cost paths (Warntz, 1957). A least-cost path is the route on a map between two locations which minimises the ‘cost of movement’ between these two places. It is calculated by assigning, to each pixel on a map, a number which represents the presumed cost to movement, and then applying the widely-used Dijkstra’s algorithm to find the route which results in the smallest cost. With this economic vernacular, Warntz and his colleagues in theoretical geography conceptually connected the cost of movement through any place in a landscape to the slope of a terrain surface (Lindgren 1967a; Lindgren 1967b). Intuitively, it is obvious that moving through steep or varied topography is more energetically demanding than moving through flat terrain. Other features of a landscape – such as roads, rivers, different vegetation or land use types – offer varying costs (or ‘resistances’) to movement. This concept was elaborated in the late 1960s to generalise the idea of resistance in relation to various geographic features (Huff and Jenks, 1968). For a given landscape, the resulting two-dimensional pixelated map, together with a movement cost value assigned to each pixel, is known as a ‘resistance surface’.
Initially, the application of the resistance surface to connectivity models in conservation science was limited by the lack of computing power available. However, this was soon addressed during the GIS revolution of the 1980s (Eastman, 1989; Douglas, 1994). Into the 21st century, connectivity analyses utilising landscape resistance surfaces and accompanying algorithms to predict terrestrial animal movement routes across these surfaces became increasingly prevalent. The first such applications in conservation science aimed to compute a least cost path between two locations to predict the importance of ecological corridors to wildlife movement (Adriaensen et al., 2003). This idea was later extended by Cushman (2009) to produce a network of least-cost paths between all combinations of any number of source points. More sophisticated resistance-based algorithms for predicting connectivity have since been developed, but all still use the framework of landscape resistance. Presently, the dominant connectivity models in use are the resistant kernel (Compton et al., 2007), a cost-distance algorithm which uses a resistance surface to calculate landscape connectivity without requiring known destination locations, and CircuitScape, which models connectivity across the resistance surface using classical circuit theory from physics (McRae et al., 2008). In the latter case, the resistance surface is treated as a big circuit and animal movement is simulated as current flowing through this circuit. Here, individual animals are modelled by electrons in which “mortality can be represented by resistors connected to ground, with their conductances reﬂecting probabilities of mortality” (see Figure 1, taken from McRae et al., 2008).
THE TROUBLE WITH RESISTANCE SURFACES, OR, GETTING BACK TO THE WRONG LANDSCAPE
These conceptual tools have extensive limitations, not just for attending to the liveliness of animal movement, but also in accounting for the complexity of the entire landscape and its creative processes (see Abram, 1996). We must question what is lost here – both of the animal and of the encompassing ecologies – through the use of these methodologies. To begin to understand this, and the resulting consequences across research and practice, we start by exploring how resistance surfaces are produced. We note here that these models have been used specifically to understand terrestrial mammals. As such, we focus our attention on these animals and their spatial and temporal scales of movement.
First, movement data relating to individual animals’ movements are aggregated into a big data set of latitude-longitude pairs. High-frequency radio-telemetry data is preferred, but genetic and camera trap data are also used. Second, this dataset is given to a mathematical ‘conditional logistic regression’ model, which then outputs 5-10 numbers, each specifying the degree to which a particular environmental variable (such as elevation, or the presence of roads) assists or impedes movement. Here, it is important to note that only those variables for which there are reliable datasets can be incorporated. Choices must therefore be made as to what variables count in determining movement. Finally, by adding together several GIS layers, each representing the chosen environmental variables weighted according to their expected influence on movement, a pixelated map is produced. This is the resistance surface. Figure 2 illustrates a resistance surface used in Elliott et al. 2014, created from GIS layers centred around the Hwange National Park in Zimbabwe.
As a mix of interested observers and resistance surface practitioners, we find common threads across our own dissatisfactions with this approach. Broadly, the problems we identify are of three types: (1) internal to the models themselves (i.e. they could be improved on their own terms using techniques common to modelling); (2) external to these specific models but nevertheless operating within the current connectivity modelling paradigm (i.e. different models could be developed that rest on similar underlying assumptions); and (3) problems which fundamentally trouble the paradigmatic assumptions of connectivity modelling itself. Herein, we elaborate these concerns and point to why interdisciplinary collaboration can offer significant improvements to study in this field.
- Internal: spatiotemporal non-stationarity
Resistance surfaces render landscapes static and unchanging and fail to account for their dynamism. For instance, these models do not account for how animals move and behave differently depending on the time of day, nor due to the several other temporal cycles that govern their inner and outer worlds, such as their life histories or the changing seasons. Animal movement behaviour is also assumed to remain constant through space. This means the presumed decisions made by animals do not vary according to their location or the path they took to get to any specific point. What is lost in these assumptions is referred to by ecological scientists as ‘temporal and spatial non-stationarity’. By drawing inspiration from scientific models in other fields, there is an understanding in this community that algorithmic details could be added within these models which would better approximate this spatiotemporal dynamism. As such, recent work in connectivity modelling has drawn on quantitative techniques to tweak the models as necessary, but has stayed within the bounds of the deeper assumptions underlying these models (e.g. Kaszta et al., 2021; Zeller et al., 2020).
- External: Context-dependence and ‘minimal ecologies’
Whereas the lack of spatiotemporal dynamism in resistance surfaces can be remedied to a degree using existing quantitative techniques – by changing the internal workings of the algorithm – a more fundamental and confounding problem for these models lies in what ecological scientists refer to as ‘context-dependence’. This is a broad term which refers to all the complexities which pertain to the uniqueness and situatedness of any ecological phenomenon, and as a result often elude the usual tools deployed in scientific studies which typically search for aggregate and normative understandings of life processes. For example, resistance surfaces – and thus the connectivity models that use them – are unable to account for animal individuality. Rather, they presume a universal and singular animal that will always take the same decision in a given circumstance without reference to the animal’s age, fitness, mood, and so on.
This concern also registers in cross-disciplinary research. For example, lack of context-dependence and the epistemic disservice this performs on animals and landscapes is addressed by Benson’s concept of the ‘minimal animal’ (2014; 2016). He argues that this simulated, disembodied, and ultimately minimal animal stems from a methodological approach of ‘behavioural minimalism’ in ecology, which Steven Lima and Patrick Zollner (1996, 133) conceptualise as “a focus on only those few behavioural traits that are likely to be important to the question under study.” This, Benson (2016, 50) suggests, has become not just a common research strategy but an ontology of the animal, at risk of implying that “animals were just as simple as the simulations used to model them.” But animals are not only minimised, transformed into beasts with no bodies. They are also abstracted from their ecological contexts. All but erased in these movement ecology models is “the landscape and the presence of other animals or events within it” (Benson, 2016, 50). The resistance surface entered conservation science as an attempt to move beyond the absent or featureless landscape of movement models and account for spatial heterogeneity in constructing predictions of connectivity. By rendering landscapes static and pixelated in space and time, as well as relying on the behavioural minimalism of movement data, resistance surfaces perform an ontological reduction on landscapes and ecologies themselves. The minimal animal exists in the resistance surface, but perhaps more significantly, we suggest that through an equivalent approach of landscape or ecological minimalism, these models also give rise to a ‘minimal ecology’.
On one level, minimal ecologies are inherent to ecological modelling. Indeed, models cannot – and do not purport to – holistically capture the dynamism of ecosystems as a whole. This would not be useful for the conservation decision-making, which they are designed to inform. Thus, to critique this fundamental element of ecological minimalism – the extension of the methodological approach of behavioural minimalism to whole ecosystems – has little analytical traction. Rather, we need to consider what is being minimised, and to what effect. The notion of context-dependence points to a more tangible problem with the minimal ecologies of the resistance surface: that the diverse relations through which ecologies cohere are all but eviscerated. Reducing the ecological relations present to a set of single-species movement tracks, combined with a selection of static environmental and geo-physical features, results in too many relations being lost to make sense of the system.
Two examples illustrate the importance of considering context-dependence. First, accounting for human presence in landscapes is crucial in applied conservation science. Countless studies highlight the radically different behaviour of animals in the presence of humans. Aside from being dynamic in space and time, human-animal relationships are highly complex, encapsulating affects from joy to fear to ambivalence (Barua, 2014; Pooley et al., 2017). However, the cost-benefit framework of resistance surfaces collapses these generative interactions to a single number: a positive number if the human presence (represented by a population density value assigned to each map pixel) is deemed to ‘attract’ the animal, and negative otherwise. By funnelling these affective interchanges into a single number, the resulting connectivity models are unable to make any sense or use of movement data generated by interactions which do not neatly fit into the attraction-aversion binary.
The second example relates to quality of life. Habitats may have equal levels of connectivity in a model, but one set of aggregated paths could lead, for example, to higher levels of toxic exposures, anxiety, or human conflict. Certain paths might indeed be the least costly, but an animal living by these pathways might also be living a very minimal life. This minimal life describes a form of biopolitics that governs the sheer biological fact of life rather than being concerned with, or indeed prioritising, the way a life is lived, or the quality of a life lived (Buchanan, 2010). For example, in the UK, red squirrels are often found inhabiting conifer habitats. One explanation is that these are appropriate and desirable red squirrel habitats. Another explanation is that the larger grey squirrels are absent from these plantations, instead preferring the richer lives offered by mixed broadleaf woodlands. Where grey squirrels come to share habitat with red squirrels in the UK, the latter soon disappear. The greys carry a pox virus, to which they are immune, that decimates reds. Considering this, red squirrels then inhabit less-rich conifer plantations because they are the only habitats left available to them. Ecological minimalism and the least cost path approach do not account for this kind of context-dependence. For resulting modes of conservation governance, rather than contributing to their flourishing, connectivity models may in fact wrongly value the minimal lives of red squirrels living in conifer plantations, ignorant to the potential for prioritising richer lives.
- The Resistance Paradigm and the Animal economicus
Questions arising from context-dependence point us toward a deeper problem with the very paradigm that underpins landscape resistance: the assumption that animal behaviour follows from rational, economistic cost-benefit calculations. We term this assumed calculated animal the Animal economicus.
Our aim here is not to interrogate the deep-seated and long-standing assumptions shared between the biological sciences and economics. Nor is it to fully expound the critiques of rational-choice economics, both from within that discipline and from the wider social sciences, whose implications have perhaps yet to fully resonate within the biological sciences. That is a broader, ongoing conversation beyond the scope of this essay. Similarly, we do not aim to critique the positivist tradition within ecological modelling; again, this has been much discussed elsewhere.
Instead, our aim is more specific. This exploration of the resistance surface and its limitations now provides us with an excellent lens to tease apart the difference between: (1) the scientific method of quantification and measurement, a process seemingly essential to all scientific modelling; and (2) the specific modelling, and (2) the specific form of rational-cost-benefit quantification occurring in the resistance paradigm. Doing so creates a creative space for scientists to overcome the limitations of the least-cost-choice assumption while still producing connectivity models important for conservation policy, and takes steps towards navigating the impasse between connectivity models and the sensuous worlds beyond their reach. Explicitly: quantification in scientific models is not itself the problem. Rather, the rational-least-cost mode of quantification leaves little room for liveliness, agency and non-conflictual human-animal relationships, and our sense is that moving beyond this framework provides scope for creatively developing quantitative approaches which can better attend to these aspects of animals’ lives.
Animal economicus – derived from economic models that assume the existence of rational human actors – has become so deeply entrenched and widely used across scientific thought that it is now intimately tied to the process of quantifying animal behaviours themselves. The field of conservation science is so plagued by this model of animal life that it is itself unable to offer the insights required to overcome the shortcoming in current connectivity models. Recent mainstream scientific publications (e.g. Elliot et al., 2014) highlight the importance of attending to animals’ movement more sensitively than just at the population level, and a foundational theoretical paper on animal movement (Nathan et al., 2008) proposes a ‘unifying paradigm’ to capture and quantify all aspects of movement. Yet, by aligning themselves with Animal economicus and perpetuating behavioural minimalism, these approaches are incapable of representing creativity and character, reducing animal liveliness into ‘random’ aspects of movement. It requires courage, humility, and curiosity to account for behaviours which have been rejected by scientific models. It asks that we step beyond the Animal economicus and attend to the quality of life lived by nonhumans, and their multiform interactions with humans and other beings.
Adjacent and overlapping disciplines have also struggled with this same problem of ecological minimalism. For example, research attempting to enliven ecologies has been central to developments in more-than-human and animal geographies in recent decades. Building on the ‘more-than-human’ tradition in geography (see Philo, 1995; Wolch and Emel, 1998; Philo and Wilbert, 2000; Wolch, 2002; Whatmore, 2002; and Buller, 2013; 2014; 2016), scholars have sought in various ways to attend to the space-making practices of animals, their lived spatialities and the multiplicity of animals’ experiences (Hodgetts and Lorimer, 2015). Of particular relevance to the problems of resistance surfaces, some work in this field has explored the experiential aspects of animals’ mobilities as embodied practices (Hodgetts and Lorimer, 2020), and the ways animal lives are shaped through relational ‘atmospheres’ (Lorimer et al., 2020). Inspired by ethological methods (Kirksey and Helmreich 2010; Lestel 2014), animal geographers have developed a variety of methodologies for studying how animals use space and move through it. Maan Barua’s (2021) ‘more-than-human ethnography’ is one such example.
While we agree with scholars who have argued for the importance for researchers across the social and natural sciences to take seriously the lives and stories of individual animals (e.g. Bear, 2011), this is not quite what we’re arguing towards here. Rather, by excluding animals as subjects with rich lives in favour of focusing on the mechanistic aspects of animal movement and a limited set of measurable landscape qualities, we are troubled that current ecological models are unable to accurately understand and predict how, why, and ultimately where, animals move – as individuals yes, but also in the aggregate. This, we argue, is the result of ecological minimalism, or the minimal ecologies inherent to contemporary connectivity modelling.
BEGINNINGS: A RESEARCH AGENDA
Discussions between conservation and social scientists are not new, unique, or particularly unusual. In our shared experience, they are sometimes fraught, sometimes generative, and often useful. In this piece we have charted the beginnings of an interdisciplinary collaboration centred on a shared concern with a particular modelling technique – resistance surfaces – and how they produce what we call ‘minimal ecologies’.
What problems do minimal ecologies engender? We draw attention to two themes that thread through our discussion above. First, we share a concern that the minimal ecologies produced through resistance surface models do not capture the complexity of ecological life as lived through relations, events, and interactions. Our sense is that this absence matters and that it has wider effects beyond the confines of the models themselves. Second, we share a concern that these models do not do enough to capture the quality of life of the organisms they describe. Given the use of such models in environmental and conservation management, we sense that this matters too. Animal economicus, and minimal ecologies, are poor tools for understanding – and governing – the movement of lives.
This essay reports the start of a conversation, not its conclusion. Hence, we conclude with a set of questions that we aim to explore together: a research agenda, of sorts. First, we want to develop forms of modelling that are better able to incorporate spatio-temporal variability. Sometimes fields are easier to cross because the wheat provides cover from predators; sometimes the fox hides in the wheat; sometimes the wheat is harvested. Landscapes are complex and changing. Our models must reflect this. Second, we want to think through and develop forms of connectivity modelling that are better able to incorporate a broader suite of ecological relations. What relations are not captured that could or should be? How might we define the objectives of the model in ways that would allow for a more lively, less minimal, ecology? Third, we want to explore how to incorporate the quality of life into connectivity models – again, this requires both new objectives/parameters, and the innovative use of technologies, to proceed. Fourth, we are interested in further exploring the effects of digitising ecologies in the specific context of connectivity modelling, contributing to an existing, wider and effervescent cross-disciplinary conversation regarding the digitisation of ecological relations. And finally, we hope to contribute to thinking through alternatives to the dominant paradigm of the rational, economic, cost-benefit animal – be they human or squirrel.
Abram, D. 1996. The Spell of the Sensuous. Vintage Books.
Adams, W.M. 2019. Digital Animals. The Philosopher, 108(1). Available: https://www.thephilosopher1923.org/adams.
Adriaensen, F., Chardon, J.P., De Blust, G., Swinnen, E., Villalba, S., Gulinck, H. and Matthysen, E. 2003. The Application of ‘Least-Cost’ Modelling as a Functional Landscape Model. Landscape and Urban Planning, 64(4): 233-247.
Barua, M. 2014. Bio-Geo-Graphy: Landscape, Dwelling, and the Political Ecology of Human-Elephant Relations. Environment and Planning D: Society and Space, 32(5): 915-934.
Barua, M. 2021. Feral ecologies: the making of postcolonial nature in London. Journal of the Royal Anthropological Institute.
Bear, C. 2011. Being Angelica? Exploring individual animal geographies. Area, 43(3): 297-304.
Benson, E. 2014. Minimal Animal: Surveillance, Simulation, and Stochasticity in Wildlife Biology. Antennae, 30: 39-53.
Benson, E. 2016. Trackable life: Data, sequence, and organism in movement ecology. Studies in History and Philosophy of Science Part C: Studies in History and Philosophy of Biological and Biomedical Sciences, 57: 137-147.
Bergman, C. 2005. Inventing a beast with no body: Radio-telemetry, the marginalization of animals, and the simulation of ecology. Worldviews: Global Religions, Culture, and Ecology, 9(2): 255-270.
Buchanan, I. 2010. Agamben, Georgio. In: A Dictionary of Critical Theory. Oxford University Press.
Buller, H. 2013. Animal geographies I. Progress in Human Geography, 38(2): 308-318.
Buller, H. 2014. Animal geographies II: Methods. Progress in Human Geography, 39(3): 374-384.
Buller, H. 2016. Animal geographies III: Ethics. Progress in Human Geography, 40(3): 422-430.
Compton, B., McGarigal, K., Cushman, S. and Gamble, L. 2007. A Resistant-Kernel Model of Connectivity for Amphibians that Breed in Vernal Pools. Conservation Biology, 21(3): 788-799.
Cushman, S. A., McKelvey, K. S., & Schwartz, M. K. (2009). Use of empirically derived source‐destination models to map regional conservation corridors. Conservation Biology, 23(2), 368-376.
Douglas, D. 1994. Least-cost Path in GIS Using an Accumulated Cost Surface and Slopelines. Cartographica. 31(3): 37–51.
Eastman, J.R. 1989. Pushbroom algorithms for calculating distances in raster grids. Proceedings, AutoCarto 9: 288-97.
Elliot, N., Cushman, S., Macdonald, D. and Loveridge, A. 2014. The devil is in the dispersers: predictions of landscape connectivity change with demography. Journal of Applied Ecology, 51(5): 1169-1178.
Hodgetts, T. and Lorimer, J. 2015. Methodologies for animals’ geographies. cultural geographies, 22(2): 285-295.
Hodgetts, T. and Lorimer, J. 2020. Animals’ mobilities. Progress in Human Geography, 44(1): 4-26.
Huff, D. and Jenks, G. 1968. Graphic interpretation of the friction of distance in gravity models. Annals of the Association of American Geographers. 58(4): 814.
Kaszta, Z, Cushman, S. and Slotow, R. 2021. Temporal Non-stationarity of Path-Selection Movement Models and Connectivity: An Example of African Elephants in Kruger National Park. Frontiers in Ecology and Evolution.
Kirksey, E. and Helmreich, S. 2010. The emergence of multispecies ethnography. Cultural Anthropology, 25(4): 545-576.
Lestel, D. 2014. Towards an ethnography of animal worlds. Angelaki: Journal of Theoretical Humanities, 19(3): 75-89.
Lima S. and Zollner, P. 1996. Towards a behavioral ecology of ecological landscapes. Trends in Ecology and Evolution, 11: 131-135.
Lindgren, E. 1967a. Proposed solution for the minimum path problem. Harvard Papers in Theoretical Geography, Geography and the Properties of Surface Series, 4.
Lindgren, E. 1967b. A minimum path problem reconsidered. Harvard Papers in Theoretical Geography, Geography and the Properties of Surface Series, 28.
Lorimer, J., Hodgetts, T. and Barua, M. 2020. Animals’ Atmospheres. Progress in Human Geography, 43(1): 26-45.
McRae, B., Dickson, B., Keitt, T. and Shah, V. 2008. Using circuit theory to model connectivity in ecology, evolution, and conservation. Ecology, 89(10): 2712-2724.
Nathan, R., Getz, W., Revilla, E., Holyoak, M., Kadmon, R., Saltz, D. and Smouse, P. 2008. A movement ecology paradigm for unifying organismal movement research. PNAS, 105(49): 19052-19059.
Philo, C. 1995. Animals, geography, and the city: Notes on inclusions and exclusions. Environment and Planning D: Society and Space, 13(6): 655-681.
Philo, C. and Wilbert, C. 2000. Animal Spaces, Beastly Places. Routledge: London.
Pooley, S., Barua, M., Beinart, W. and Dickman, A. 2017. An interdisciplinary review of current and future approaches to improving human-predator relations. Conservation Biology, 31(3): 513-523.
Warntz, W. 1957. Transportation, Social Physics, And The Law Of Refraction. The Professional Geographer 9, no. 4 (1957): 2-7.
Warntz, William (1965) “A note on surfaces and paths and applications to geographical problems, IMaGe Discussion Paper #6, Ann Arbor: Michigan Inter-University Community of Mathematical Geographers.
Whatmore, S. 2002. Hybrid Geographies: Natures, Cultures, Spaces. SAGE: London.
Wolch, J. 2002. Animal Urbis. Progress in Human Geography, 26(6): 721-742.
Wolch, J. and Emel, J. 1998. Animal Geographies: Place, Politics, and Identity in the Nature-culture Borderlands. Verso: London.
Zeller, K., Lewison, R., Fletcher, R., Tulbure, M. and Jennings, M. 2020. Understanding the Importance of Dynamic Landscape Connectivity. Land, 9(9): 303.
Roboto Regular Font Entry (Not displayed, do not remove)
Roboto Regular Italic Font Entry (Not displayed, do not remove)
Roboto Bold Font Entry (Not displayed, do not remove)
Roboto Bold Italic Font Entry (Not displayed, do not remove)
Arbutus Font Entry (Not displayed, do not remove)