Animals That Mark Their Territory

i. Introduction

Territoriality occurs widely throughout the animal kingdom, observed in taxa as diverse as mammals, birds, insects and fishes. Territories are spatial regions, defended against conspecifics, for the purpose of using resources and providing mating opportunities. Different species employ a wide variety of tactics to defend territories, such every bit degradation of visual or olfactory cues, fighting or ritualistic displays, with records of such behaviour dating dorsum as far every bit the seventeenth century [1,ii]. While theoretical biological science has a rich history of analysing pattern generation from microscale interactions to macroscale [3], it was non until about 20 years ago that population-level territorial patterns were modelled analytically as emerging from individual-level interaction events between similar animals [4] (although meet [5] for an early example of segregation emergence between animals with highly differing behavioural traits).

Historically, much modelling of space utilize has been based on phenomenological descriptions of the areas used by animals, such as drawing a minimum convex polygon around location fixes to construct a plausible home range [6], or assuming that space use will correspond with food availability, as with resource selection analysis [7]. These approaches have become increasingly sophisticated over the years, through models such as kernel density estimators and Brownian bridges, leading to realistic descriptions of spatial patterns [8,ix].

While accurate description is valuable, prediction requires a solid understanding of the individual-level mechanisms that requite rise to observed spatial patterns. The construction of quantitative, predictive environmental that can foresee the touch of environmental modify on species' ability to survive and reproduce is a vital challenge for twenty-first century science. The rapid changes currently occurring in many of the Globe's ecosystems force animals to respond before they can adapt, attempting as best they can to make apply of the new environments they find themselves in. By uncovering the behavioural mechanisms underlying their selection of expanse as they move and interact with one another, it may become possible to predict the effects of habitat variation on the spatial structure of a population.

On the more theoretical side, mechanistic territorial models provide a central step towards constructing a statistical mechanics for ecological systems [ten]. This plan seeks to discover quantitative theories explaining how 'macroscopic' ecosystem patterns derive from 'microscopic' individual processes, in analogy with the laws linking macroscopic physical backdrop, such equally pressure and temperature, to the behaviour of the underlying organisation of molecules. Although ecosystems containing living creatures are more circuitous than collections of molecules, the general principle of showtime with a random walk model, then using mathematical assay to derive properties of the system, has already borne much fruit in movement ecology research [11,12]. Therefore, scientists are gradually moving towards the goal of building a predictive ecological theory based on the concept of statistical mechanics, by constructing the jigsaw puzzle one piece at a time.

The specific puzzle-piece that relates to building models of space apply from territorial interactions, which is the focus of this review, began when Lewis & Murray [4] constructed and analysed ane-dimensional advection–diffusion equations based on scent deposition of wolves and the subsequent avoidance response of neighbouring packs. Since and then, this formulation has been refined to take business relationship of the precise details of movement and interaction events, generalized to the biologically realistic two-dimensional case and extended to account for environmental effects [xiii,14]. It has been successfully used to exam hypotheses about the underlying causes of territory size and shape and demonstrate the effects of population alter on territorial structure [15]. Recently, it has also been practical in the sociological context of human gang territories [16]. This wide variety of applications both demonstrates that there be general mathematical structures behind the multifarious territorial interaction mechanisms residing in the natural globe, and shows the effectiveness of mechanistic modelling at answering hitherto unsolved biological questions.

Three years ago, in that location was a farther cardinal advocacy in our agreement of territory formation and dynamics [17]. The authors stripped down the model of scent-mediated conspecific abstention to a very unproblematic, individual-based model (IBM) of territory formation, which was analysed from the footing up, without taking the mean-field limits used in the advection–diffusion arroyo [fifteen]. Although qualitatively similar territory patterns emerged in this model, information technology likewise displayed a number of features not present in the advection–diffusion approach, virtually notably details of the timescale over which territory boundaries shift, as well every bit an ability to quantify the longevity of scent mark cues purely by examining the evolution of animal locations over time [18,19].

The purpose of this paper is to requite a detailed review of the progress in both approaches to territorial modelling, which relate mainly to terrestrial mammalian carnivores, but have recently been extended to birds [xx]. We explain how a recently proposed formalism can unify the two frameworks and chronicle them to the fields of resources selection analysis and commonage animate being behaviour, two areas of ecology that take separately evolved rich histories of modelling and data analysis. Finally, we give a perspective into the future of mechanistic territory modelling and how we see its place in helping to respond pressing questions in current ecological research.

Throughout, we are conscientious to distinguish territorial formation from the related concept of home range emergence [21,22]. While territorial interactions are often primal in the formation of dwelling ranges, they are non a necessary mechanism, with dwelling ranges often forming in the absence of conspecific avoidance. The underlying localization processes in these cases may, for example, consequence from resource attraction or site fidelity owing to memory [23]. Although the examination of dwelling range formation in the absence of conspecific avoidance interactions is beyond the scope of this review, we include a section ('Home range emergence') where we explain how mechanistic territory models fit into the general try of agreement home range formation. For a adept contempo review of abode range analysis, instead see [22], or [24] in the context of mechanistic modelling.

2. The dynamical systems set-up

Mathematical models for creature movement have a variety of forms. Commonly, one might think of tracking an animate being'due south location at stock-still time intervals. In this instance, a move kernel would describe the distribution of footstep lengths and movement directions from one fourth dimension step to the next [eleven,12]. However, local environmental conditions, such as terrain or prey, or territorial signals, such equally odor marks, would as well feed into the move kernel [15,25]. An animal may bias movement abroad from steep terrain or might accept shorter steps in regions with high prey density. Territorial signals, such every bit odor marks, could also affect the direction of movement through abstention of foreign odour marks or allure towards familiar scent marks [15].

To complicate matters, some of these factors, such as terrain, are external to the animals, whereas others, such as casualty density or smell marker density, involve negative or positive feedbacks every bit animals modify their local environments, a procedure sometimes referred to as stigmergy [26]. One fashion of accounting for these furnishings is to create an IBM with reasonable behavioural rules that involve changes to the local environs and responses to environmental atmospheric condition. The IBM could exist faux to explore the suite of possible outcomes. A detail case of this approach is given later, in the section 'An private-based approach'. Showtime, however, nosotros examine an alternative arroyo, which attempts to judge the probability density role (PDF) of the animal or group of animals.

3. Probability density role approach

Although the precise location of the animal at whatsoever point in time is uncertain, the range of possible locations of the animal can notwithstanding be described using a PDF. This modelling approach simulates the time evolution of the PDF equally territorial interactions reshape it [fifteen]. Thus, rather than requiring multiple stochastic simulations, the PDF arroyo involves a unmarried simulation that tracks the expected infinite employ of the animal over time.

The basic tool for moving from individual descriptions to PDFs is the chief equation [11,12]. This is an iterative equation that describes the PDF of the brute at a time t + τ in terms of the PDF at time t and a dispersal kernel describing the individual movement patterns. The kernel is based on an private-level description of the brute'southward movement between t and t + τ. Many individual-level processes have been proposed for motility between successive locations, such every bit step option functions [27], Brownian motility [9] and land–space models [28]. The master equation provides the bridge between these individual-level descriptions of movement and more 'macroscopic' space-use patterns described by the PDF.

The classical method for analysing primary equations involves using the Fokker–Planck equation to guess the motion model. This allows one to go from the kernel-based description of movement for individuals, including environmental conditions and feedbacks, via the master equation, to a system of advection–diffusion equations that track the expected space apply over time. The resulting system of equations can be simulated on a computer or analysed mathematically to predict the emergence of territorial patterns.

Early applications of the Fokker–Planck approach focused on determining the behavioural ingredients needed for territorial blueprint formation [4]. They asked: what behavioural interaction terms, including scent-marking, volition requite rise to the spontaneous formation of territories? The simplest model involved ii packs interacting in i spatial dimension, each producing scent marks that cause abstention movement by the other pack. Each pack was modelled equally moving back towards its den site when it encountered foreign scent marks. It turns out that this simple model is sufficient to generate territories [4]. The improver of positive feedback, through enhanced scent-marking over foreign olfactory property marks, gives rising to basin-shaped patterns of scent marks, with the edges of the bowl describing heightened olfactory property densities found at the border of the territories [29]. It besides gives rise to buffer zones betwixt territories, where neither pack would go. Both these features have been studied extensively in wolf (Canis lupus) territories in northeastern Minnesota, and the fact that elementary behavioural rules give ascension to such realistic emergent patterns is a persuasive argument as to the model'southward validity [four].

Realistic models for brute territories must include multiple spatial dimensions, likewise as the spatial distribution of external factors, such as resource and topography. A second generation of sophisticated two-dimensional advection–diffusion models has been developed so as to include these factors [xv]. By using the method of maximum-likelihood to connect the models with data, hypotheses about the factors driving territorial design formation tin can be tested from the infinite-utilise patterns every bit measured by radiotelemetry data. This method was applied to test the role of scent-marking on coyote (Canis latrans) territorial patterns in the Hanford Arid Lands Ecosystem [13] (effigy 1) and additional impacts of topography and prey distribution on these patterns in the Lamar Valley region of Yellowstone [14]. Hither, the connection between advection–improvidence models for territorial patterns and classical hypothesis testing is new, and information technology provides a powerful approach for connecting mechanistic move models with information.

Figure 1. Mechanistic territorial model applied to coyote populations. These relocation data for coyote from different packs, denoted by different colours, are fitted using the method of maximum-likelihood. The model posits that animals movement randomly and avert foreign sent marks past moving dorsum towards their den site or organizing centre (triangles). The scent marks (non shown) have their own dynamics where there is a constant low level of mark, with foreign scent marks causing an over-marking response. Full details of the model are given in Moorcroft *et al*. [thirteen]. Reproduced with permission from Moorcroft *et al*. [13]. (Online version in colour.)

As these advection–diffusion models have become more mainstream, new applications take extended the modelling theory. For example, the process of shifting territories, as new groups grade and erstwhile groups divide, has been very recently explored using the advection–diffusion approach applied to an extensive dataset for territorial meerkats in South Africa [30]. These mechanistic models have also been reapplied in a new context, to the formation of gang territories in the Hollenbeck region of Los Angeles [16]. Hither, natural barriers to gang movement, including rivers and freeways, replaced the topography component in the models.

four. An individual-based approach

(a) The modelling framework

An alternative approach to advection–diffusion modelling was proposed in reference [17], whereby the animals were modelled on a detached lattice, and assay was performed without first taking a mean-field limit. It is well known that when interactions are rare, equally is ofttimes the case with territorial animals, continuum models can give very dissimilar results to the underlying IBM [31]. Therefore, it is important to examine whether there are aspects of territoriality that exist in an individual-based approach, but are not nowadays in reaction–advection–improvidence systems.

The and then-chosen 'territorial random walk' models animals equally nearest-neighbour lattice random walkers, each of whom deposits aroma every bit information technology moves, which lasts for a finite amount of time, the 'active scent fourth dimension' (T _As), after which other (conspecific) animals no longer respond to the mark as fresh (effigy ii). They are able to motion to any nearest-neighbour lattice site unless the site contains active scent of a conspecific, in other words unless that site is in the conspecific's territory [nineteen].

Figure 2. — Effigy 2. The private-based territoriality model with case output. The left-hand console represents a hypothetical snapshot in time of the position of 2 territorial random walkers (animals), the red and blue dots, and their territories, represented by the blood-red and blue open circles, respectively. If a red (blue) open circle is present at a lattice site, it means that the red (blue) beast has been at that place quondam in the past T _{Every bit} timesteps. The absence of any scent marks at coordinates (v,1), (ii,three) and (ii,4) implies that no brute has occupied those coordinates within a time T _As, i.e. this is interstitial area. The side by side time the blue animate being moves, it can go to any of the four side by side lattice sites with equal probability, whereas the reddish animal is constrained to move either upwards or right. The right-hand panel demonstrates the sort of dwelling house range patterns that tin can arise from such a model. Reproduced from Giuggioli *et al*. [17]. (Online version in colour.)

An reward of this approach is that it provides a natural definition of the animal's territory at any point in time: the set of lattice sites containing active odor of that animal. This readily corresponds to the definition from Burt [21] of a territory being 'any defended area'. This definition is contrasted with that of a home range, the latter beingness 'that area traversed by the individual in its normal activities of food gathering, mating and caring for young' [21]. In more than mathematical terminology, this might exist called the utilization distribution of an animal as measured over a period of time spent engaging in such 'normal' daily activities.

The territories that emerge from these lattice models are non static, only alter slowly over time, typically much slower than the movement of the animals themselves. Every bit a outcome, when measured over a finite fourth dimension window, the utilization distributions (habitation ranges) of animals in adjacent territories will overlap slightly. Such overlapping dwelling house ranges are common in territorial systems, just contrast with the concept of contiguous territories or territories separated by buffer zones [4]. That both home ranges and territories sally in conceptually carve up, simply clearly divers means from this model enables rigorous qualification of the traditional descriptive differences [21].

If the movement of the animals has no intrinsic localization process, so home range overlap will steadily enlarge as the fourth dimension window is increased, without ever stabilizing. The urban foxes (Vulpes vulpes) studied past Potts et al. [xix] lack such a central place attraction, but many animals do take a bias in their movement towards a den or nest site [15]. Incorporating this bias into the IBM arroyo causes stable home ranges to emerge, despite the territory borders remaining in constant flux [18].

The main majority of work on individual-based territorial models has so far been based on full territorial exclusion, where animals completely avoid areas containing conspecific territory marks. However, it is typical for animals to exhibit a sure amount of curiosity and probing on the territory border, pushing into recently marked areas a pocket-size amount before subsequently retreating. Indeed, such a process has recently been shown to occur in populations of Amazonian birds [20]. In reference [26], the patterns emerging from a procedure of partial exclusion in an IBM were studied, giving qualitatively realistic patterns of overlapping dwelling ranges.

(b) Mathematical assay

An advantage of the individual-based arroyo is that it explains the phenomenon of moving territory borders, sometimes called the 'elastic disc hypothesis', which has been observed in species from a variety of taxa (run across references in Potts et al. [nineteen]), ever since the seminal paper of Huxley [32]. A disadvantage is that it is highly computationally intensive to fit stochastic IBMs to data.

To circumvent this result, judge analytic versions of the simulation models that describe the motility of animals in side fluctuating territory borders were constructed in one-dimension [33] and ii-dimensions [eighteen,34]. These were solved exactly, giving expressions that are readily fitted to information on beast move [19]. The models are based on the observation from simulation output that territory borders exhibit slow random motion that constrain the animals' intrinsic diffusive motion. Equally such, parametrizing them requires knowledge of the territory border movement and they do non, in themselves, contain information about the scent-marking process. Therefore, fitting data to these models does not give any information about the agile scent fourth dimension.

However, in that location turns out to be a 'parameter collapse' of the simulation output to a universal curve relating the generalized improvidence constant of the territory border, K, to a dimensionless input parameter Z, and then that Inline Formula for detail constants α and β reported in reference [33] in one-dimension and [19] in ii-dimensions. Here, in one-dimension and Inline Formula in two-dimensions, where ρ is the population density and D is the intrinsic diffusion constant of the animal. This enables users of this modelling approach to excerpt the active scent time from details of the border movement that, in turn, tin can exist extracted from movement information via the approximate analytic model (effigy iii).

Figure 3. — Effigy 3. Using private-based territoriality models to extract smell longevity from location data. Location data tin be fitted to the analytic model of [34] using the methods of [19] to give data almost the territory border motility, K, the creature'southward intrinsic diffusion abiding, D, and the population density ρ. Assay of the IBM from [17] then gives a universal curve K = αDexp(*−βDρT* _As) [19], which tin be used, together with the information on K, D and ρ, to obtain an estimate of the agile scent time, T _AS.

An important, unsolved effect from this approach is to understand analytically why the parameter collapse to Inline Formula is observed, and whether it holds for all parameter values or simply those analysed in references [nineteen,33]. Some initial steps towards this stop were fabricated in reference [35], where the authors noted that this tendency is related to the drift probability of one territory boundary into its neighbour, via a showtime passage time argument. This drift probability can be thought of as the amount of pressure one territory exerts on a neighbor. Although this surprisingly challenging mathematical trouble provided a key step forward, much more needs to be done to empathise fully this parameter collapse.

(c) Ecological and epidemiological lessons

Applying this model to animal location data enables quantification of both the interaction process, that is the agile odour time, and the corporeality of intrinsic flexibility in the territorial structure, that is the border diffusion constant K. Past using information earlier and after an outbreak of mange in Bristol's red fox population [19], it was possible to quantify how both the territorial structure and the behaviour of foxes changed as the illness spread through the population.

These changes turned out to be quite dramatic, having of import consequences for modelling epizootics in territorial populations. The study showed that it is non authentic to assume that the animals, even those that do not take the disease, volition necessarily maintain their behavioural patterns. Large amounts of government money rely on practiced agreement of such disease spread, notably the contempo decision to choose badgers by the Uk regime to stop the spread of bovine tuberculosis [36]. This decision itself was based upon the controversial notion that badgers will not modify their territorial structures as a result of disturbing the population through culling [37, §§3.6.ix–x]. The approach of [19] gives perhaps the first mechanistic theory that explains why such assumptions are probable to be false, so the underlying modelling framework could show useful in helping governments make better-informed decisions.

five. Fit to the movement process or the territorial pattern?

When applying mechanistic territory models to data, researchers have generally tended to fit the emergent territorial patterns to relocation data, regardless of whether they take used advection–diffusion or IBM approaches [fourteen,19]. A unlike approach fits models to the fine-scale move and interaction processes, then uses them to derive the resulting space-use patterns [12]. An advantage of the quondam arroyo is that it does not rely on the availability of detailed movement data. A disadvantage is that the fitting procedure, typically based on a maximum-likelihood approach [14], requires that animal locations exist independent samples of the utilization distribution. Obtaining sets of points that are approximately independent ordinarily requires using a small subsample of the data, which can mean discarding a lot of information [xv].

Fitting a model directly to the underlying movement and interaction processes, on the other manus, allows one to make utilise of all the location data bachelor. Attributable to advances in global positioning satellite applied science over recent years, fine-scaled creature motility data are becoming increasingly mutual, making such model fitting possible. Once such a model has been parametrized, it is possible to use either simulation or mathematical analysis to derive the resulting territorial patterns [xx]. Because these patterns are non themselves fitted to the positional data, every bit in previous approaches, this approach is far more conservative in answering whether a model is sufficient to produce territorial patterns (figure four).

Figure 4. — Figure four. Fit to the process or the pattern? (a) gives a schematic of the modelling scheme for mechanistic models where we fit information to the territorial design [fifteen,19]. (b) How this process differs when we fit the data to the movement and interaction processes [20]. (Online version in colour.)

The procedure used in this assay is based around the notion of a step selection role [27], which gives the probability Inline Formula of moving from position y to ten, given information near the surrounding landscape . Moorcroft & Barnett [38] noted that this is precisely equivalent to the motion kernel of mechanistic territory and dwelling house range models. Therefore, past fitting pace choice functions to data, using methods such equally in reference [39], information technology is possible to parametrize a mechanistic movement model, which can, in turn, exist used to derive space-utilize patterns in a mathematical and not-speculative manner, using techniques developed in reference [xv]. By coupling together step option functions for unlike animals [20], interactions tin also be explicitly incorporated in this modelling approach.

half-dozen. Optimality and game theory

Although a mechanistic model with stock-still parameters, may, on boilerplate, describe fauna movement behaviours, individuals may modulate their behavioural responses, responding to local conditions so as to optimize fettle [40]. Mathematically, this could be achieved past a modification of parameters in the mechanistic model. Still, when more than one group is simultaneously involved in optimizing, the advisable framework to depict interactions is actually in terms of a game [41].

It turns out the issue of buffer zones betwixt wolf territories provides a fascinating context for the application of game theory. This is because in that location is a strong positive correlation betwixt the locations of the buffer zones and heightened densities of the primary casualty species for wolves in northeastern Minnesota, white-tailed deer. The deer appear to thrive in these buffer zones owing to reduced predation force per unit area. This begs the question equally to why the territorial wolves do non simply trespass into the buffer zones between territories and eat the precious prey species before neighbouring packs take the opportunity. After all, animals are seldom mindless automata, obeying fixed behavioural rules, and it is natural to enquire how these rules might conform then as to maximize fitness.

The idea that territorial movement behaviour tin be modified so as to improve a wolf pack's fettle is quite reasonable biologically simply is a challenge to address quantitatively [41]. An early try to model optimal behavioural responses of territorial wolves in this complex spatial predator–casualty dynamic used the theory of differential games to testify weather condition under which buffer zones would persist and why they might break down [42]. Packs were assumed to modulate their movement behaviour so as to effort to maximize food intake while minimizing the gamble of hostile altercations with neighbours. A central result from this analysis showed that buffer zones can persist equally evolutionarily stable outcomes, providing the punishment for interpack altercation is loftier, and, crucially, providing there e'er remains a random component of movement, describing the uncertainty inherent to wolf motility. This area of coupling spatially explicit territorial models to game theory is in its infancy, and there is a real opportunity for new analysis.

seven. The related concept of dwelling range

Any animal that maintains a territory volition ipso facto have a home range. However, the converse is non true. Many animals exhibit home range behaviour without actively defending a territory, for example caribou herds [43]. Consequently, much try has gone into examining the mechanisms that cause the formation of home ranges in the absence of conspecific abstention processes (e.m. encounter Grimm & Railsback [44] for various IBM approaches to this). Although nosotros focus here on models that incorporate territorial interactions, information technology is worth giving a brief overview of other home range models as they are often closely related. Detailed reviews can be found elsewhere [22,24].

Models of home range emergence in the absence of territorial interactions typically involve fidelity to a detail identify or places. To generate this allegiance, models oft assume that there is an underlying memory process [45]. As animals movement, they will retrieve where they have gone in the recent past and modify their future movements accordingly. These modifications may crusade biases towards sites that they have recently visited [23], towards patches of particularly arable resources that they recall visiting [46], or away from places where predators take been recently encountered [46].

Once this exploratory phase is over and the home range established, the animate being's motion mechanisms may merely be described as a bias towards desirable sites. This idea naturally leads to the utilize of site allegiance models equally skillful way to estimate home ranges from data, the and then-chosen motion kernel density figurer (MKDE) [47]. Past explicitly incorporating movement processes, such every bit Brownian bridges [nine], these models tin can give better interpretation of dwelling range distributions than traditional methods such as (ordinary) kernel density estimation [viii] or minimum convex polygons [6]. It remains an interesting open question as to whether MKDE tin be improved further past the inclusion of territorial interactions.

8. 'Not-mechanistic' territory models

Over a decade ago, Adams [48] made a thorough review of territorial models, including mechanistic models. Withal, the term 'mechanistic' was used in a much broader sense than in this paper, and included 'geometric models' of territory borders, whereby the territory is assumed to exist a priori, only its size and shape are affected by the behaviour of its inhabitants. For example, Adams [49] describes a model of fire ants (Solenopsis invicta), where the force per unit area on a territory border increases with biomass and decreases with the square of the distance to the nest site. This is used to predict the relative sizes and shapes of neighbouring territories. However, the reasons behind the option of these item determinants of territory pressure are purely descriptive, and not derived from underlying processes. Models of the ants' movements and interactions such as reviewed in this newspaper could potentially assist parametrize this model in a more than mechanistic, and less speculative, way.

Adams [48] besides reviews game-theoretic toll–benefit models and models of territory establishment. While the former have since been integrated into the mechanistic framework [42], the latter have yet to be understood from detailed descriptions of individual movements and interactions. The phenomenon of dispersal and re-institution, ofttimes by adolescent animals, is very of import for agreement population dynamics, disease spread and range expansion. Models that have been so far proposed in this regard tend to exist based around the work of Fretwell & Lucas [50] which posits that an animal will establish a territory wheresoever its fettle is maximized, often using an economic cost–do good framework [51]. Although some models take considered the costs of movement and interactions, e.g. Stamps & Krishnan [52], and more recent studies have modelled motion on a form scale of approximately x fourth dimension steps per lifetime [53], to the all-time of our noesis, none explicitly model the fine-scale movement and interactions that accept place during territory establishment. Incorporating these ideas may requite a more authentic agreement of the territorial dynamics that occur during dispersal and re-establishment.

ix. Unsolved problems and time to come directions

Although mechanistic models have been successfully used to examination hypotheses nigh the processes that cause territorial patterns to form, due east.yard. [13,14], the approach is typically based around testing which model fits the data best out of a set of hypothesized models, without seeking to empathize how shut the best model is to empirical reality. This is a major shortcoming for two reasons. First, though the best model may be significantly better than the others, this does not mean that it is sufficient to describe the data with enough accurateness to brand accurate predictions about possible future scenarios. Second, without a quantitative measurement of closeness of a model to the information, information technology is not possible to tell when the model is circuitous enough to have identified all the key processes underlying territory formation. If mechanistic models are going to assist turn ecology into a truly predictive science, so at that place is a pressing demand to fill up this gap.

Some other challenge is to understand better how and when the IBM and advection–improvidence approaches differ, and when each should be applied. Typically, mean-field fractional differential equation (PDE) approaches piece of work well when there are large numbers of individuals. In other circumstances, as is often the case with territorial animals where a single individual or pack is defending the territory, it makes sense to bank check results of PDE studies against the underlying IBM to ensure that the predictions are accurate.

The chief advantage of the PDE approach is that it gives analytic formulae that obviate the need for excessive simulation analysis. Thus, as long as the results are like to the underlying IBM, such analysis is very user-friendly. While at that place be accurate analytic approximations to the IBM territory models proposed then far, they practise not explicitly incorporate the territorial interaction parameter, T _AS [18,33,34]. To remedy this, information technology is necessary either to create an analytic model that links the edge movement to the interaction process, a programme that was initiated in [35], or to construct more accurate deterministic approximations than traditional hateful-field methods permit. One possible avenue in the latter direction might exist to utilize van Kampen's methods [54], which have successfully been used to find analytic reasons for disparities betwixt mean-field and IBM approaches in biological systems [31].

All mechanistic models and then far have been based effectually what might exist called 'stigmergent' interactions [26]. That is, interactions that are mediated by modification of the environment. A classic example is odour or pheromone deposition. One animal deposits odour, adding to the environmental cues at that betoken. Sometime subsequently, another animal responds to this cue by altering its behaviour. Other stigmergent processes include visual cues or song cues. The latter practice not persist in the surround per se just rather exist in other animals' cognitive maps of the surround, who hear the cue and may respond several days after to the memory of it by avoiding the area from whence information technology came [55].

While near applications of mechanistic territory models so far have been regarding scent-marker mammals, information technology is straightforward to translate the ideas to other stigmergent processes, every bit evidenced by the use of this concept to model vocal cues in birds [20]. However, it is not and then obvious how one might construct mechanistic models that incorporate direct interactions such as fighting and ritual displays, as observed in a diverseness of species [56]. In some bird populations, for case, neighbours may actively move every then often to a specific identify on the territory border, whereupon they claiming the neighbouring flock to a territorial boxing, which often consists of an aggressive display rather than actual physical contact. The outcome of such a battle may decide whether or not one of the flocks is able to advance its boundary and increase its territory [55]. Such complex behaviour is perchance tricky to model and analyse from a mechanistic perspective, just is a necessary aspect to examine in society to sympathise fully how territories form and change.

We cease past reiterating the idea that dwelling house range germination appears to be largely governed past one or both of two factors: territorial interactions and a cognitive map of the environment [57]. The latter may include various aspects of noesis, such every bit those about resource availability, predation probability or other environmental covariates. One of the most important challenges for the hereafter will exist integrating these two of import aspects of spatial localization to course an authentic, predictive theory of how space-apply patterns emerge from the detailed, varied and complicated behaviours of interacting animals.

Acknowledgements

We are grateful to members of the Lewis Enquiry Group for helpful discussions and two anonymous reviewers who helped improve the newspaper.

Funding statement

This study was partly supported by NSERC Discovery and Accelerator grants (Chiliad.A.Fifty., J.R.P.). Thousand.A.L. as well gratefully acknowledges a Canada Research Chair and a Killam Research Fellowship.

Footnotes

© 2014 The Authors. Published by the Royal Society under the terms of the Artistic Commons Attribution License http://creativecommons.org/licenses/past/3.0/, which permits unrestricted use, provided the original writer and source are credited.

References

1
Olina GP
. 1622 Uccelliera, overo Discorso della Natura e Proprietà di Diversi Uccelli, e in particolare di que che Cantano, con il Modo di Prendergli, Conoscergli, Allivargli e Matenergli . Rome, Italia. See: http://gdz.sub.uni-goettingen.de/dms/load/img/?PPN=PPN479740488&IDDOC=278056. Google Scholar
2
Nice MM
. 1941 The function of territory in bird life. Am. Midl. Nat. 26 , 441–487. (doi:x.2307/2420732). Crossref, Google Scholar
3
Levin SA& Segel LA
. 1985 Pattern generation in space and attribute. SIAM Rev. 27 , 45–67. (doi:10.1137/1027002). Crossref, ISI, Google Scholar
4
Lewis MA& Murray JD
. 1993 Modelling territoriality and wolf–deer interactions. Nature 366 , 738–740. (doi:10.1038/366738a0). Crossref, ISI, Google Scholar
5
Shigesada N, Kawasaki One thousand& Teramoto East
. 1979 Spatial segregation of interacting species. J. Theor. Biol. 79 , 83–99. (doi:x.1016/0022-5193(79)90258-3). Crossref, PubMed, ISI, Google Scholar
6
Harris S, Cresswell WJ, Forde PG, Trewhella WJ, Woollard T& Wray South
. 1990 Home-range analysis using radio-tracking information: a review of problems and techniques particularly as applied to the study of mammals. Mammal Rev. xx , 97–123. (doi:ten.1111/j.1365-2907.1990.tb00106.x). Crossref, ISI, Google Scholar
7
Manly BF, McDonald LL, Thomas DL, McDonald TL& Erikson WP
. 2002 Resource selection by animals: statistical blueprint and assay for field studies , 2nd edn. New York, NY: Chapman and Hall. Google Scholar
viii
Worton BJ
. 1989 Kernel methods for estimating the utilization distribution in home-range studies. Environmental 70 , 164–168. (doi:10.2307/1938423). Crossref, ISI, Google Scholar
9
Benhamou S
. 2011 Dynamic approach to space and habitat use based on biased random bridges. PLoS ONE vi , e14592. (doi:10.1371/periodical.pone.0014592). Crossref, PubMed, ISI, Google Scholar
10
Levin SA
. 2012 Towards the spousal relationship of theory and data. Interface Focus 2 , 141–143. (doi:10.1098/rsfs.2012.0006). Link, ISI, Google Scholar
11
Turchin P
. 1998 Quantitative analysis of motion: measuring and modeling population redistribution in animals and plants . Sunderland, MA: Sinauer Associates. Google Scholar
12
Okubo A& Levin SA
. 2002 Diffusion and ecological problems: mod perspectives , 2nd edn. New York, NY: Springer. Google Scholar
13
Moorcroft PR, Lewis MA& Crabtree RL
. 1999 Dwelling range analysis using a mechanistic home range model. Environmental 80 , 1656–1665. (doi:10.1890/0012-9658(1999)080[1656:HRAUAM]2.0.CO;2). Crossref, ISI, Google Scholar
14
Moorcroft PR, Lewis MA& Crabtree RL
. 2006 Mechanistic abode range models capture spatial patterns and dynamics of coyote territories in Yellowstone. Proc. R. Soc. B 273 , 1651–1659. (doi:10.1098/rspb.2005.3439). Link, ISI, Google Scholar
15
Moorcroft PR& Lewis MA
. 2006 Mechanistic home range analysis . Princeton, NJ: Princeton University Printing. Google Scholar
xvi
Smith LM, Bertozzi AL, Brantingham PJ, Tita GE& Valasik M
. 2012 Adaption of an ecological territorial model to street gang spatial patterns in Los Angeles. Discrete Contin. Dyn. Syst. 32 , 3223–3244. (doi:10.3934/dcds.2012.32.3223). Crossref, ISI, Google Scholar
17
Giuggioli 50, Potts JR& Harris Southward
. 2011 Animal interactions and the emergence of territoriality. PLoS Comput. Biol. 7 , 1002008. (doi:10.1371/journal.pcbi.1002008). Crossref, ISI, Google Scholar
18
Potts JR, Harris S& Giuggioli L
. 2012 Territorial dynamics and stable abode range germination for central place foragers. PLoS ONE vii , e34033. (doi:10.1371/journal.pone.0034033). Crossref, PubMed, ISI, Google Scholar
19
Potts JR, Harris S& Giuggioli L
. 2013 Quantifying behavioural changes in territorial animals acquired past sudden population declines. Am. Nat. 182 , E73–E82. (doi:x.1086/671260). Crossref, PubMed, ISI, Google Scholar
xx
Potts JR, Mokross Thou& Lewis MA
. Submitted. A unifying framework for quantifying the nature of animal interactions. Encounter http://arxiv.org/abs/1402.1802. Google Scholar
21
Burt WH
. 1943 Territoriality and abode range concepts equally applied to mammals. J. Mammal 24 , 346–352. (doi:10.2307/1374834). Crossref, Google Scholar
22
Börger Fifty, Dalziel B& Fryxell JM
. 2008 Are there general mechanisms of animal home range behavior? A review and prospects for future research. Ecol. Lett. 11 , 637–650. (doi:ten.1111/j.1461-0248.2008.01182.10). Crossref, PubMed, ISI, Google Scholar
23
Briscoe BK, Lewis MA& Parrish SE
. 2002 Home range formation in wolves due to odor marking. Bull. Math. Biol. 64 , 261–284. (doi:10.1006/bulm.2001.0273). Crossref, PubMed, ISI, Google Scholar
24
Moorcroft PR
. 2012 Mechanistic approaches to understanding and predicting mammalian space use: recent advances, time to come directions. J. Mammal 93 , 903916. (doi:x.1644/11-MAMM-S-254.ane). Crossref, ISI, Google Scholar
25
Potts JR, Bastille-Rousseau Thousand, Murray DL, Schaefer JA& Lewis MA
. 2014 Predicting local and non-local effects of resource on animal space use using a mechanistic footstep-pick model. Methods Ecol. Evol. v , 253–262. (doi:10.1111/2041-210X.12150). Crossref, PubMed, ISI, Google Scholar
26
Giuggioli L, Potts JR, Rubenstein DI& Levin SA
. 2013 Stigmergy, collective actions and animal social spacing. Proc. Natl Acad. Sci. USA 110 , xvi 904–16 909. (doi:10.1073/pnas.1307071110). Crossref, ISI, Google Scholar
27
Fortin D, Beyer HL, Boyce MS, Smith DW, Duchesne T& Mao JS
. 2005 Wolves influence elk movements: beliefs shapes a trophic cascade in Yellowstone National Park. Ecology 86 , 1320–1330. (doi:10.1890/04-0953). Crossref, ISI, Google Scholar
28
Patterson TA, Thomas Fifty, Wilcox C, Ovaskainen O& Matthiopoulos J
. 2008 State–space models of individual animal move. Trends Ecol. Evol. 23 , 87–94. (doi:ten.1016/j.tree.2007.x.009). Crossref, PubMed, ISI, Google Scholar
29
Lewis MA, White KAJ& Moorcroft PR
. 1997 Assay of a model for wolf territories. J. Math. Biol. 35 , 749–774. (doi:ten.1007/s002850050075). Crossref, ISI, Google Scholar
xxx
Bateman AW, Lewis MA, Gall Yard, Manser MB& Clutton-Brock Th
. Submitted. Territoriality and home-range dynamics in meerkats Suricata suricatta . . Google Scholar
31
McKane AJ& Newman TJ
. 2004 Stochastic models in population biological science and their deterministic analogs. Phys. Rev. East 70 , 041902. (doi:10.1103/PhysRevE.seventy.041902). Crossref, ISI, Google Scholar
32
Huxley JS
. 1934 A natural experiment on the territorial instinct. Brit. Birds 27 , 270–277. Google Scholar
33
Giuggioli L, Potts JR& Harris Due south
. 2011 Brownian walkers within subdiffusing territorial boundaries. Phys. Rev. E 83 , 061138. (doi:10.1103/PhysRevE.83.061138). Crossref, ISI, Google Scholar
34
Giuggioli L, Potts JR& Harris S
. 2012 Predicting oscillatory dynamics in the movement of territorial animals. J. R. Soc. Interface 9 , 1529–1543. (doi:10.1098/rsif.2011.0797). Link, ISI, Google Scholar
35
Potts JR, Harris S& Giuggioli L
. 2011 An anti-symmetric exclusion process for ii particles on an infinite 1D lattice. J. Phys. A Math. Theor. 44 , 485003. (doi:10.1088/1751-8113/44/48/485003). Crossref, ISI, Google Scholar
36
Department for Surroundings, Food and Rural Affairs. 2011The Authorities's policy on bovine TB and badger control in England. See http://www.defra.gov.great britain/. Google Scholar
37
Krebs JR
. 1997 Bovine tuberculosis in cattle and badgers . London, UK: Ministry building of Agronomics, Fisheries and Nutrient (MAFF) Publications. Google Scholar
38
Moorcroft PR& Barnett A
. 2008 Mechanistic home range models and resources selection analysis: a reconciliation and unification. Environmental 89 , 1112–1119. (doi:10.1890/06-1985.1). Crossref, PubMed, ISI, Google Scholar
39
Forester JD, Im HK& Rathouz PJ
. 2009 Accounting for beast motion in estimation of resource selection functions: sampling and data analysis. Ecology xc , 3554–3565. (doi:10.1890/08-0874.i). Crossref, PubMed, ISI, Google Scholar
xl
Morrell LJ& Kokko H
. 2005 Bridging the gap between mechanistic and adaptive explanations of territory formation. Behav. Ecol. Sociobiol. 57 , 381–390. (doi:ten.1007/s00265-004-0859-5). Crossref, ISI, Google Scholar
41
Lewis MA& Moorcroft PR
. 2001 ESS analysis of mechanistic domicile range models: the value of signals in spatial resource partitioning. J. Theor. Biol. 210 , 449–461. (doi:10.1006/jtbi.2001.2323). Crossref, PubMed, ISI, Google Scholar
42
Hamelin FM& Lewis MA
. 2010 A differential game theoretical analysis of mechanistic models for territoriality. J. Math. Biol. 61 , 665–694. (doi:10.1007/s00285-009-0316-i). Crossref, PubMed, ISI, Google Scholar
43
Mahoney SP& Virgl JA
. 2003 Habitat selection and census of a nonmigratory woodland caribou population in Newfoundland. Can. J. Zool. 81 , 321–334. (doi:ten.1139/z02-239). Crossref, ISI, Google Scholar
44
Grimm V& Railsback SF
. 2005 Individual-based modeling and ecology . Princeton, NJ: Princeton Academy Press. Crossref, Google Scholar
45
Fagan WF, et al.
2013 Spatial memory and animal motion. Ecol. Lett. sixteen , 1316–1329. (doi:10.1111/ele.12165). Crossref, PubMed, ISI, Google Scholar
46
Avgar T, Deardon R& Fryxell JM
. 2013 An empirically parameterized individual based model of fauna movement, perception, and retentiveness. Ecol. Model. 251 , 158–172. (doi:10.1016/j.ecolmodel.2012.12.002). Crossref, ISI, Google Scholar
47
Benhamou Southward& Riotte-Lambert L
. 2012 Beyond the utilization distribution: identifying habitation range areas that are intensively exploited or repeatedly visited. Ecol. Model. 227 , 112–116. (doi:ten.1016/j.ecolmodel.2011.12.015). Crossref, ISI, Google Scholar
48
Adams ES
. 2001 Approaches to the written report of territory size and shape. Annu. Rev. Ecol. Syst. 32 , 277–303. (doi:10.1146/annurev.ecolsys.32.081501.114034). Crossref, Google Scholar
49
Adams ES
. 1998 Territory size and shape in fire ants: a model based on neighborhood interactions. Ecology 79 , 1125–1134. (doi:10.1890/0012-9658(1998)079[1125:TSASIF]ii.0.CO;two). Crossref, ISI, Google Scholar
50
Fretwell SD& Lucas HL
. 1969 On territorial behavior and other factors influencing habitat distribution in birds. I. Theoretical development. Acta Biotheor. nineteen , sixteen–36. (doi:10.1007/BF01601953). Crossref, Google Scholar
51
Loehle C
. 2013 Differential sorting of individuals in territorial species affects apparent habitat quality. J. Wildl. Manage. 77 , 1166–1169. (doi:10.1002/jwmg.574). Crossref, ISI, Google Scholar
52
Stamps JA& Krishnan VV
. 1990 The effect of settlement tactics on territory sizes. Am. Nat. 135 , 527–546. (doi:10.1086/285060). Crossref, ISI, Google Scholar
53
Barraquand F& Murrell DJ
. 2012 Evolutionarily stable consumer home range size in relation to resource demography and consumer spatial organisation. Theor. Ecol. five , 567–589. (doi:ten.1007/s12080-011-0148-7). Crossref, ISI, Google Scholar
54
van Kampen NG
. 1992 Stochastic processes in physics and chemistry . Amsterdam, Holland: Elsevier. Google Scholar
55
Jullien M& Thiollay JM
. 1998 Multi-species territoriality and dynamic of neotropical understory bird flocks. J. Anim. Ecol. 67 , 227–252. (doi:10.1046/j.1365-2656.1998.00171.ten). Crossref, ISI, Google Scholar
56
Maynard Smith J
. 1974 The theory of games and the evolution of beast conflicts. J. Theor. Biol. 47 , 209–221. (doi:10.1016/0022-5193(74)90110-6). Crossref, PubMed, ISI, Google Scholar
57
Spencer WD
. 2012 Home ranges and the value of spatial information. J. Mammal 93 , 929–947. (doi:ten.1644/12-MAMM-Due south-061.one). Crossref, ISI, Google Scholar