vol. 182, no.

3 the american naturalist september 2013

E-Article

Quantifying Behavioral Changes in Territorial Animals Caused

by Sudden Population Declines

Jonathan R. Potts,1,2 Stephen Harris,2,* and Luca Giuggioli1,2,3

1. Bristol Centre for Complexity Sciences, University of Bristol, Bristol, United Kingdom; 2. School of Biological Sciences, University of
Bristol, Bristol, United Kingdom; 3. Department of Engineering Mathematics, University of Bristol, Bristol, United Kingdom
Submitted January 29, 2013; Accepted March 19, 2013; Electronically published July 12, 2013
Online enhancements: appendix, videos. Dryad data: http://dx.doi.org/10.5061/dryad.5dn48

idemiology (Beier 1993; Lewis Murray 1993; Kenkre et al.

abstract: Although territorial animals are able to maintain exclu-
sive use of certain regions of space, movement data from neighboring
2007; McCarthy and Destefano 2011). Though such ap-
individuals often suggest overlapping home ranges. To explain and plications often assume that territories are roughly sta-
unify these two aspects of animal space use, we use recently developed tionary, population density can change rapidly in a variety
mechanistic models of collective animal movement. We apply our of situations, such as when a population is suffering an
approach to a natural experiment on an urban red fox (Vulpes vulpes) epizooty of a terminal disease, thereby affecting the ter-
population that underwent a rapid decline in population density due
ritorial structure and behavior of the individual animals.
to a sarcoptic mange epizooty. By extracting details of movement
and interaction strategies from location data, we show how foxes
During 1994–1996, a sarcoptic mange epizooty decimated
alter their behavior, taking advantage of sudden population-level the red fox (Vulpes vulpes) population in Bristol, causing
changes by acquiring areas vacated due to neighbor mortality, while changes in both the movement of the foxes and territory
ensuring territory boundaries remain contiguous. The rate of ter- sizes (Baker et al. 2000). This provided support to the idea
ritory border movement increased eightfold as the population de- that territories deform elastically, which was first observed
clined and the foxes’ response time to neighboring scent reduced by
nearly 80 years ago by Huxley (1934) in coot (Fulica atra)
a third. By demonstrating how observed, fluctuating territorial pat-
terns emerge from movements and interactions of individual animals, populations. While elastic territories have since been de-
our results give the first data-validated, mechanistic explanation of tected in a variety of species, such as martins (Progne subis;
the elastic disc hypothesis, proposed nearly 80 years ago. Stutchbury 1991), warblers (Acrocephalus arundinaceus;
Ezaki 1995), and lizards (Anolis aeneus; Stamps and Krish-
Keywords: animal movement, home range, red fox (Vulpes vulpes),
nan 1998), construction of a mechanistic theory that un-
epizooty, territoriality, theoretical ecology.
derpins this elasticity has tended to remain elusive.
In the context of scent-marking animals, the process
Introduction with which animals respond to the information present in
scent deposited by a conspecific is key to the correct quan-
Much has been written about the factors affecting territory tification of territorial dynamics. From the movement per-
size, such as allometry, resource dispersion and availability,
spective of the individual animal, this is a binary choice
and population density (Kruuk and Parish 1982; Grant
between ignoring foreign scent, if the information con-
and Kramer 1990; Jetz et al. 2004; Moorcroft and Barnett
tained in it is either old or uninformative, or retreating.
2008; Van Moorter et al. 2009; Schradin et al. 2010). De-
A given location thus either has or does not have an “active
spite this, we know remarkably little about how territories
scent,” that is, a scent that is responded to by conspecifics
form and change shape (Adams 2001) and how animals
interact to maintain these territories (Börger et al. 2008). as a fresh territory cue. The presence/absence nature of
Providing this understanding is of great importance to the scent implies that the system is intrinsically stochastic,
many areas of ecology, from conservation biology to wild- so deterministic representation via reaction-diffusion for-
life management and from predator-prey dynamics to ep- malisms may be unable to account for the discrete nature
of the interaction events (e.g., see Durrett and Levin 1994;
* Corresponding author; e-mail: s.harris@bristol.ac.uk.
McKane and Newman 2004). A recent modeling frame-
Am. Nat. 2013. Vol. 182, pp. E73–E82. 䉷 2013 by The University of Chicago.
work (Giuggioli et al. 2011a, 2011b, 2012; Potts et al. 2011,
0003-0147/2013/18203-54436$15.00. All rights reserved. 2012) that accounts for the discrete nature of the scent-
DOI: 10.1086/671260 mediated interaction events is employed here to interpret

The American Naturalist 2013.182:E73-E82.

Downloaded from www.journals.uchicago.edu by 115.178.255.6 on 03/22/20. For personal use only.
E74 The American Naturalist

observations of the Bristol fox population before and dur- sation of vocal cues from recently dead neighbors
ing the 1994–1996 mange epizootic. (Newton-Fisher et al. 1993).
Some territorial animals have a strong drift tendency
toward the locations of their den sites and Moorcroft et
al. (2006) examined how these territorial patterns change Methods
when a coyote (Canis latrans) pack is removed from a Stochastic Simulations
population. Making use of a reaction-diffusion formalism
where the position densities of two neighboring packs are Stochastic simulations were performed based on the two-
coupled with density profiles of the scent marks and with dimensional (2-D) territorial random walk model of Giug-
a predetermined knowledge of the locations of the coyote gioli et al. (2011a) but employing one of two different
den sites, Moorcroft and Lewis (2006) compared the steady movement processes: nearest-neighbor random walks
state position distributions of the animals before and after (NNRWs) and ballistic walks (BWs). The NNRW process
the removal. This approach, however, cannot be used in is described in Giuggioli et al. (2011a), whereas ballistically
our context because the tendency to drift toward the den moving animals will always continue in a straight line,
site is either not present in Bristol’s foxes or not sufficiently unless they encounter foreign scent, causing them to turn
strong to generate steady state position distributions, since at random. For each simulation run, 25 animals were
the mean square displacement of the animals increases placed on a 2-D square grid size M # M with lattice spac-
with time and never settles (Giuggioli et al. 2011a). Neither ing a and population density r p 25/M 2, so that the grid
was the habitat spatially confined, as in Briscoe et al. is approximately 25 times the size of a territory. While the
(2002). One of the fundamental advancements of the sto- dominant pair share the same territory (Saunders et al.
chastic framework proposed by Giuggioli et al. (2011a) is 1993) and territory configuration could be calculated from
the ability to quantify the (discrete) longevity of scent cues any individual within a group (Baker et al. 2000), the
and the movements in territory borders, a key notion im- simulations only modeled one animal per territory, the
plicit in the elastic disc hypothesis. See Potts et al. (2012) dominant male, so when fitted to the data, r was the
for a detailed comparison of the two approaches. population density of dominant males.
By using the more recent approach, we quantify elasticity All simulated animals moved with the same movement
in territorial patterns by the use of a single parameter K, process, NNRW or BW, constrained by the fact that each
the diffusion constant of the territory border, measuring the animal could not enter a square that contained fresh scent
rate at which the variance of border positions increases over of a different animal, and so turned away from the square
time. Variations in mean territory size, the amount of di- in a random direction. Each animal deposited scent at
rectional persistence in the animal movement process, and every square it visited, which remained for a finite time
the animal velocity are also quantified, enabling behavioral TAS, the active scent time. After this time period had
changes due to a sudden population decline to be assessed. elapsed, the scent was no longer recognized by conspecifics
Additionally, we analyze agent-based simulations of systems as a fresh scent message and so was no longer present in
of moving and interacting animals (Giuggioli et al. 2011a), the simulation. The squares that contained active scent of
to determine the longevity of territorial scent marks, the an animal constituted its territory and the locations where
active scent time TAS, from the information provided in the two contiguous territories met made a territory border.
parameter K. Animals are modeled to move at random The speed of each animal was v, and t p a/v was the time
(Okubo and Levin 2002) but constrained to roam within it took for an animal to move distance a.
areas that do not contain scent of conspecifics. As each While certain animals are so-called borderland markers,
animal moves, it deposits scent, but once the scent has been for example, badgers (Meles meles; Hutchings et al. 2001),
present for time TAS, it is no longer considered by others who actively patrol their borders to discourage invaders,
to be fresh and so is ignored. In the field, while scent marks foxes are hinterland markers (Macdonald 1980), meaning
cannot persist after the chemicals have decayed or dispersed, they deposit scent marks evenly throughout their territory.
it may be beneficial for animals to intrude into a neigh- Our previous studies have shown that greater patrolling
boring territory if the odor of the scent mark they detect of the borders can reduce the amount they shift (Giuggioli
is old, suggesting that the territory may no longer be de- et al. 2011a), and we conjectured that foxes might be
fended. To test whether this happens in the Bristol fox pop- adopting such a strategy. However, closer analysis of the
ulation, we showed that the active scent time decreased after relative amount of time foxes spend near their territory
the outbreak of mange, demonstrating that TAS must arise borders reveals that this is not the case (see “Methods for
from a behavioral strategy rather than being solely a con- Inferring Time Spent at the Territory Border,” available
sequence of the persistence of the chemicals in the envi- online). Therefore, we have not included active border
ronment. This decrease might also be influenced by a ces- patrolling in this model.

The American Naturalist 2013.182:E73-E82.

Downloaded from www.journals.uchicago.edu by 115.178.255.6 on 03/22/20. For personal use only.
 Behavior Changes in Population Declines E75

A Reduced Analytic Approximation of Territory,” available online for the full analytic expression
the Simulation Model of P(x, y, tFV) and its derivation).

To fit data to the model efficiently, we used an analytic

approximation from Giuggioli et al. (2012) that models Calculating Home Range Overlap from
the movement of a correlated random walker (CRW; an- Territory Border Movement
imal) inside a moving square territory of average width The amount of overlap between neighboring home ranges
L, which is equal to 1/(r)1/2 (see fig. 1a for a pictorial was calculated by assuming that the two territories share
explanation). To measure the amount of time an animal a common edge. In the model, this edge is continuously
has directional persistence, we used the correlation time moving, creating an overlap in the space used by the two
T (Viswanathan et al. 2005). This has the following prac- adjacent animals when measured over a time interval T*.
tical interpretation. If position fixes from a CRW with For example, if an edge between two territories, each mod-
correlation time T were taken at time intervals greater than eled as a square of side L (see fig. 1), moves a distance of
T, analyzing the turning angles between the position fixes DL in the perpendicular direction during the time T*, then
would suggest that the animal was moving in an uncor- the overall area shared by these animals in two territories
related fashion. Conversely, if the fixes were taken at time during this time is L # DL. In practice, T* represents the
intervals less than T, turning angle analysis would suggest time window during which location data are collected. As
that the movement was correlated. the territory borders move during this time window, we
As territories exclude one another, the time dependence observe overlaps in the home ranges measured from the
of the territory border mean square displacement (MSD), locations of animals in neighboring territories (fig. 1b).
that is, the variance of the occupation probability, is In our model, the MSD in the perpendicular direction
slightly sublinear. This occurs due to an exclusion process, of the shared edge between two adjacent territories is
well studied in the statistical physics literature, for ex- Kt/ ln (t/t). Therefore, the width of the overlapping strip
ample, Liggett (1985), but only recently introduced into between the two neighboring home ranges is equal to the
the literature on animal territoriality (Giuggioli et al. mean absolute displacement, which is [KT* / ln (T* /t)]1/2.
2011a). An exclusion process is one where there are mul- The width of the home range is then 1/(r)1/2 ⫹
tiple moving objects, in our case territories, that cannot [KT* / ln (T* /t)]1/2, owing to the fact that the width of each
either overlap or occupy the same place at the same time. territory is 1/(r)1/2. Therefore, the fraction of a home range
Since the territories hamper each others’ movement, the that overlaps with this particular neighbor is
MSD does not grow linearly in time (free diffusion). In
fact, Landim (1992) has indicated that 2-D exclusion pro- 1
HRO p 1 ⫺ . (1)
cesses have an asymptotic (i.e. long-time) MSD propor- 冑
1 ⫹ KT* r/ ln (T* /t)
tional to t/ ln (t). Therefore, we assume that the output of
the territory border movement in the simulation model Notice that as T* increases, the fraction of overlap increases
is asymptotically equal to 2Kt/ ln (t/t). The reason for di- toward the theoretical maximum value of 1, where the two
viding by the constant t here is to ensure K has the units home ranges coincide. However, since territory border
of a diffusion constant (space2/time), which becomes con- movement is typically very slow, the time it would take
venient later when we compare it with the diffusion con- to get close to this situation is likely to be far longer than
stant v 2 t of the animal (e.g., see fig. 2). The analytic ap- the lifetime of the animal. Therefore, complete overlap is
proximation model has the territory border diffusion highly unlikely to be observed in reality.
constant K as an input parameter.
Since larger-than-average territories in the simulation Data Collection and Analysis
model tend to shrink but smaller-than-average ones tend
to grow, the analytic model contains a rate parameter g Movement data were taken from a long-term study of the
that measures the strength of this tendency (fig. 1a). We red fox population in the Bristol urban area. Our data are
have summarized the various parameters in the model in available in the Dryad Digital Repository, http://dx.doi
table 1. The data were fitted to the probability density .org/10.5061/dryad.5dn48 (Potts et al. 2013). Radio fixes
function P(x, y, tFV) for the animal to be at coordinates with a spatial resolution of 25 m # 25 m were taken every
(x, y) relative to its home range center, defined as the 5 min between 20:00 and 04:00 GMT, which encompasses
centroid of all the measured positions of the animal over most fox activity (Saunders et al. 1993), so throughout
the study period, at time t, given the input parameters this article “1 day” is equal to 8 h of fox location data.
V p (v, K, T, g, L) (see “Analytic Expression of the Prob- Radio telemetry data from 22 different territorial adult
ability Distribution for an Animal in a Slowly Moving foxes (i.e., 11 year old) monitored between 1990 and 1995

The American Naturalist 2013.182:E73-E82.

Downloaded from www.journals.uchicago.edu by 115.178.255.6 on 03/22/20. For personal use only.
E76 The American Naturalist

Figure 1: Representation of the approximate analytic model of animal movement within a dynamic territory. Territory borders move
randomly with a mean square displacement of 2Kt/ ln (t/t) around an average width of L p 1/(r)1/2 , where r is the population density. The
process that keeps the territories at this average width is represented by two springs, one vertical and one horizontal, each having a spring
constant of g. The higher g, the greater the tendency for larger- or smaller-than-average territories to move back toward an average size.
The animal, represented by a filled circle in a, moves as a correlated random walker with speed v and correlation time T (see “Methods”).
Panel a represents this setup, while b demonstrates how overlaps between adjacent home ranges arise from this model as animal positions
are measured over time. In b, the mean position of the territory border is represented by the solid black line, while the average extent of
the movement of this border to the left and right is represented by the dashed gray lines.

were analyzed. Data from both males and females were gorithm requires a starting value for V, called V0 p
used because the space use distributions of a male and a (v0 , K 0 , T0 , g0 , L 0 ), that is expected to be close to the max-
female from the same group are very similar (Baker et al. imum. We set v0 to be the total distance moved by the
2000). Each fox was tracked during one, two, or three foxes divided by the total time moved. Term T0 was ob-
seasons (spring, March–May; summer, June–August; au- tained by the formula T0 p ⫺5/ ln [A cos (v)S] min, where
tumn, September–November; winter, December–Febru- A cos (v)S is the mean of the cosines of the turning angles
ary; see table A1, available online). Starting in the summer v from the data (Viswanathan et al. 2005). The factor of
of 1994, a mange epizootic spread through Bristol’s foxes 5 comes about since location measurements were taken
causing the population density to decline rapidly, even- every 5 min. Since the long-time MSD of the animal is
tually killing almost all the foxes in the city (Baker et al. taken to be 2Kt/ ln (t/T ) (Giuggioli et al. 2012, eq. 3.3),
2000). When analyzing the data, we split them into two K 0 was obtained by fitting a curve A ⫹ 2K 0t/ ln (t/T0 ) to
sets: premange, before summer 1994 when the population the fox MSD against time t for t 1 1 day, using the least
density was relatively stable, and postmange, after summer squares method, where A is a fitting constant.
1994, when the population was rapidly declining. The pre- Term L 0 was found by taking the square root of the
mange data set contained N p 8,693 data points from 18 mean 100% minimum convex polygon (MCP) home range
different foxes, postmange N p 2,313 from 4 foxes (see area (Harris et al. 1990). While the MCP method is in
table A1). The last fox in the study area died in spring general not the most accurate for finding home range sizes
1996 (Baker et al. 2000). (Fieberg and Börger 2012), we use it only to find a starting
The log maximum likelihood method was employed to point for running the Nelder-Mead algorithm. Though a
fit data to the theoretical probability distribution more accurate method, for example, kernel density esti-
P(x, y, tFV). In particular, the Nelder-Mead simplex al- mation (KDE; Laver and Kelly 2008), may cause the al-
gorithm (Lagarias et al. 1998) was used to find the max- gorithm to converge slightly more quickly, the estimation
imum of L(V) p 冘n ln [P(x n , yn , tnFV)] for each set of pa- method we use to obtain the algorithm’s initial condition
rameter values V, where the sum is taken over 99% of makes no difference to the outcome of the algorithm. Since
position-time locations (x n , yn , tn) that attain the highest MCP has the advantage of being very simple to measure
P(x n , yn , tnFV) values. We excluded 1% of outliers to ensure from the data and is known to be an accurate measure of
the results were not biased by anomalous behavior (Harris home range area and territory boundaries in this particular
et al. 1990; Kenward et al. 2001). The Nelder-Mead al- fox population (Saunders et al. 1993), this is the method

The American Naturalist 2013.182:E73-E82.

Downloaded from www.journals.uchicago.edu by 115.178.255.6 on 03/22/20. For personal use only.
 Behavior Changes in Population Declines E77

0
10 Nearest Neighbour RW
Ballistic walk

Normalized territory border movement

−2
10

−4
10 Percentage overlap

60

−6 40
10
20

0
0 5
−8
10 Population density (km−2)

0 5 10 15
Normalized active scent time

Figure 2: Simulation output showing the dependence of territory boundary movement on the active scent time for nearest-neighbor random
walks (NNRWs) and ballistic walks (BWs). The vertical axis is K/v 2t , and the horizontal is Z p TASv2tr (see table 1 for definitions of terms).
The crosses and circles show values from the simulation output. The solid line is a best fit for the NNRW case log10 (K/v 2t) p 0.085 ⫺
0.247Z, and the dashed line is the best fit for the BW case log10 (K/v 2t) p 0.467 ⫺ 0.709Z. Inset, the percentage of each home range that
overlaps with neighboring ranges. The solid (NNRW) and dashed (BW) lines use fixed values of TAS , v , T* , and t p T taken from the
premange data, whereas the dotted (BW) and dot-dashed (NNRW) lines use postmange data.

we choose here. Finally, to find g0, the maximum of as Z increased. The rate of decrease was much greater in
L(V) was calculated for V p (v0 , K 0 , T0 , g, L 0 ) as g varies the BW case since each animal tended to move across the
across parameter space from g p 10⫺3 to g p 10 4. Error territory in a much shorter timescale than the NNRW case.
bars for the best fit were obtained using the bootstrap This caused each point on the territory border to be res-
algorithm for variance calculation (see, e.g., Wasserman cented at shorter time intervals so that the borders moved
2004) by resampling each data set 100 times. less. As an indicator of how these timescales differ, if the
territories were immobile and square, the territory width
would be 1/(r)1/2 and the area would be 1/r. Therefore,
Results
the respective timescales would be proportional to
The Effect of Movement Processes on Territorial Dynamics 1/(r)1/2 for the BW case, the time it takes for a ballistic
walker to move a distance of 1/(r)1/2, compared to 1/r for
For both of the movement processes simulated, NNRW
NNRW, the first passage time for such a walker to traverse
and BW, the borders of the emergent territories each had
a distance of 1/(r)1/2 (see, e.g., Redner 2007, section 2.4).
an MSD that increased asymptotically as 2Kt/ ln (t/t). The
magnitude of the diffusion constant K depended on the
directionality of the animal’s movement, the population Territorial Dynamics of a Red Fox Population before and
density r and the active scent time TAS (fig. 2). For each after an Outbreak of Sarcoptic Mange
movement process, K depended upon the ratio Z p
TAS /TC between the active scent time and TC p 1/(v 2 tr), During the 1994–1996 mange epizootic, the red fox pop-
the territory coverage time, representing how long it takes ulation density declined rapidly, causing the foxes to both
for an animal to move around its territory (Potts et al. extend their home ranges and move faster (Baker et al.
2012). For the NNRW case, TC is closely related to the 2000). To investigate further the effect of population de-
mean time for an animal in a confined region to return cline on the animals’ movements and interactions, we fit-
to the place from which it started (Condamin et al. 2007). ted the data to the model of animal movement in a ter-
For either movement process, K decreased exponentially ritory of fluctuating size and position (Giuggioli et al.

The American Naturalist 2013.182:E73-E82.

Downloaded from www.journals.uchicago.edu by 115.178.255.6 on 03/22/20. For personal use only.
E78 The American Naturalist

Table 1: Glossary of terms and best-fit values

Symbol Model Explanation Premange Postmange
K A Territory border diffusion constant (m /min) 2
3.97 Ⳳ .09 32.8 Ⳳ .6
T A Correlation time (min; see “Methods”) 14.9 Ⳳ .5 14.8 Ⳳ .4
L A Average territory width (m) 435 Ⳳ 37 926 Ⳳ 44
g A Territory renormalization rate (min⫺1): rate at which a 108 Ⳳ 2 140 Ⳳ 2
territory tends to return to an average area of 1/L2
v A, S Average animal velocity (m min⫺1) 7.39 Ⳳ .16 18.9 Ⳳ .3
r A, S Population density (km⫺2) 5.29 Ⳳ 1.02 1.17 Ⳳ .12
T* A, S Time window over which home range is measured (days) 84.0 84.0
TAS S Active scent time (days) 5.07 Ⳳ .55 3.37 Ⳳ .16
Tat S Territorial acquisition time (days) ... 4.89 Ⳳ .16
a S Lattice site separation (length) ... ...
t S t p a/v (time) ... ...
TC S TC p 1/(v 2tr) (time) ... ...
Z S Z p TAS/TC (dimensionless) 10.5 Ⳳ .2 10.0 Ⳳ .2
Note: Glossary of parameters for both the analytic model of animal movement inside slowly fluctuating territory borders and
the simulation model of territorial random walkers, together with their values for the fox data premange and postmange, if applicable.
Error range is 1 standard deviation using the bootstrap algorithm (see main text for details). The second column details whether
the parameter is used in the analytic model (A) or simulation model (S).

2012). Table 1 details the parameters V p (v, K, T, g, L) between turns was higher. The lack of change in turning
that gave the best fits of the premange and postmange data angle distribution may indicate an inbuilt species- or hab-
sets to the probability distribution P(x, y, tFV) for an an- itat-specific search strategy.
imal moving within its territory. Figure 3 shows all the The expected percentage of home range overlap, HRO,
premange fox positions, superimposed on the utilization was measured over a time window of T* p 84.0 days,
distribution of the model, which measures the expected which was the maximum time between the first and the
distribution of animal fixes across a season (see “Analytic last location fix for any of the seasons during which data
Expression of the Probability Distribution for an Animal were gathered. It was calculated using equation (1) to be
in a Slowly Moving Territory”). Video 1 shows the evo- HRO p 24.7% Ⳳ 3.3% premange and HRO p
lution of both the fox positions and the model’s probability 30.6% Ⳳ 2.5% postmange (error bars are 1 SD). This ap-
distribution over time. parent increase in overlap is not statistically significant
As well as the foxes having a much larger average velocity (P p .07). However, given that there were only four in-
after the mange outbreak, the value of K increased more dividuals tracked in the postmange period and that the
than eightfold, meaning that territory borders moved much data during that period were dominated by two of these
more rapidly after the population density declined. As the four (see table A1), this increase is not negligible. The
foxes died out, neighboring foxes took over the newly va- failure of this test to reject the null hypothesis that there
cated areas, causing the borders to move and the territories is no increase in overlap may therefore be subject to a
to enlarge, as evidenced by an increase in average territory Type II error (see, e.g., Wasserman 2004), so this result
size from L2 p 0.189 km2 premange to L2 p 0.857 km2 should be considered tentative. Previous studies using this
postmange. In table 1, we present values of the territory data set (Baker et al. 2000) used MCP techniques to mea-
width L, which is the square root of its area. The increase sure the overlaps directly. Though MCP techniques have
in g after the mange outbreak suggests that territories were in recent years been shown to be less than ideal (Laver
pushed toward an average size faster when the population and Kelly 2008), the study by Baker et al. (2000) also
density was less, in keeping with the idea that territories showed an apparent small increase in home range overlap
were more “elastic” during the postmange period. that was not statistically significant.
While the Bristol foxes increased v as the population Were TAS, v, and T to have remained constant as the
density dropped, surprisingly, they did not increase T. That population density decreased, there would have been a de-
is, the turning angle distribution had similar statistics both pendency of the home range overlap on the population
before and after the mange outbreak. However, due to density of the type shown in the inset of figure 2 (solid
their increased velocity, the distance for which they would curve). In such a case, unless the density is very low, the
persist in roughly the same direction, v T, was 2.5 times percentage of overlap decreases as the population density
higher postmange. In other words, the mean step length increases. However, for extremely low densities, neighboring

The American Naturalist 2013.182:E73-E82.

Downloaded from www.journals.uchicago.edu by 115.178.255.6 on 03/22/20. For personal use only.
 Behavior Changes in Population Declines E79

1.5

Normalized Y Position 1

0.5

−0.5

−1

−1.5
−1.5 −1 −0.5 0 0.5 1 1.5
Normalized X Position

Figure 3: The utilization distribution of the analytic model of animal movement inside a slowly moving territory, with location fixes of
Bristol foxes before the 1994 outbreak of sarcoptic mange (N p 8,693 ), and parameter values fitted to the premange data. The fixes have
been normalized so that the center of each seasonal home range is at position (0, 0) and the distances from the center have been divided
by L p 435 m, the average territory width. Different symbols denote different foxes. The contours are placed at deciles (1/10, 2/10, 3/10,
etc.) of the distribution height, except for the outer two, which are placed at 1/100 and 1/1,000 of the height. See video 1 to view the
evolution of the probability distribution over time.

animals would be so far apart that they are unlikely to move between adjacent lattice sites was set equal to the
encounter one another’s territorial borders, meaning the correlation time, that is, t p T.
ranges would overlap little. This trend in home range over- Though we analyzed both ballistic and random walks
lap implies that the home range size HRS p L2{1 ⫹ in our simulation model, the foxes we studied tended to
[KT* r/ ln (T* /t)]1/2}2 is proportional to r⫺a where a 1 1, as- make several turns between one visit to the border and
suming fixed TAS, v, and T. The home range size is estimated the next. This is evident from the correlation time T of
by taking the square of the home range width L{1 ⫹ just under 15 min (table 1), which is insufficient time for
[KT* r/ ln (T* /t)]1/2}. Using the premange values of these pa-
the foxes to traverse their territories at average speed (e.g.,
rameters, we fitted a straight line to the plot of log (HRS)
7.39 m min⫺1 premange, with a territory width of 435 m).
against log (r) (using linear least squares) to find that a ≈
Therefore, we restrict our data analysis to using the ran-
1.74.
dom walk version of the simulation model.
For the premange data, the dimensionless value K/v 2T
was 4.89 Ⳳ 0.54 # 10⫺3 (SD), whereas postmange
Inferring Active Scent Time from Location Data
K/v 2T p 6.19 Ⳳ 0.56 # 10⫺3 (SD). The best-fit curve
The value of TAS for the fox population was found by using from the NNRW simulation output, log10 (K/v 2T ) p
the best-fit values of K from the analytic model together 0.085 ⫺ 0.247Z, gives Z p 10.5 Ⳳ 0.2 (SD) premange and
with the NNRW trend curve from the simulation output Z p 10.0 Ⳳ 0.2 (SD) postmange. To link the simulation
(fig. 2). Since animals that move with a correlation time parameters to the analytic model, the population density
T appear to be uncorrelated random walkers when sampled was assumed to be r p 1/L2, and was 5.29 Ⳳ 1.02 km⫺2
at a temporal resolution lower than or equal to T (i.e., the (SD) fox territories premange and r p 1.16 Ⳳ 0.12 km⫺2
time between fixes is greater than T ), the time it takes to (SD) postmange. Therefore, TAS p Z/(v 2T r) was 5.07 Ⳳ

The American Naturalist 2013.182:E73-E82.

Downloaded from www.journals.uchicago.edu by 115.178.255.6 on 03/22/20. For personal use only.
E80 The American Naturalist

mated the Bristol fox population, territorial patterns

changed rapidly, affecting both the foxes’ movement and
interaction processes. Fluctuations of territory borders be-
came more pronounced, evidenced by an eightfold in-
crease in the territory diffusion constant K. The animals
increased their average speed of travel by around 2.5 times.
We also observed a decrease in the active scent time. These
behavioral changes enabled the foxes to increase their ter-
ritory size to maintain contiguous borders and thereby
exclude potential invaders (Baker et al. 2000), while en-
Video 1: Video 1, available online, shows how the probability dis- suring that the percentage of home range overlap remained
tribution P(x, y, tFV) evolves through time. The contours on the roughly constant.
video show P(x, y, tFV) for the values of V that best fit the premange The decrease in active scent time from 5 days to just
data (see main text). The dots on the video show cumulative locations
of foxes through time from the premange data set, normalized so over 3 days meant that foxes waited for a shorter time
that the centers of their home ranges are all at (0, 0) and the distances before attempting to acquire territorial area that they be-
from the center are divided by L p 435 m. lieved had been vacated. As well as scent, foxes use vocal
cues to inform neighbors of their presence (Newton-Fisher
0.55 days (SD) premange and 3.37 Ⳳ 0.16 days (SD) post- et al. 1993). The absence of these additional cues after the
mange, a statistically significant decrease (P p .002). death of a fox could suggest to neighbors that the territory
We also used the active scent time to infer how long it had been vacated, and so they may be more willing to
took for a neighbor to seize a territory once it had been venture into areas that contain scent that is only 3 or 4
vacated, the so-called territory acquisition time Tat. By days old. There may also have been a reduction in the
running NNRW simulations whereby one fox is removed number of scent marks left by foxes in the terminal stages
part way through, we measured the time it took for other of sarcoptic mange. While scent marks contain informa-
foxes to seize 90% of the dead fox’s territory (see video tion about health and status (Arnold 2009), a territory was
2). Except when the population density was very low, the not invaded until after the neighbors had died (Baker et
dimensionless value (Tat ⫺ TAS )rv 2 t tended to range be- al. 2000), so it appears that health did not influence active
tween about 3.5 and 5 (see “Calculating Territorial Ac- scent time in foxes.
quisition Time,” available online). For the postmange case Overlapping home ranges emerge in the model as a
of Z p 10.0, the value of (Tat ⫺ TAS )rv 2 t was found to be direct outcome of fluctuating territory borders, without
4.50, by averaging over various values of r, v, t, and TAS requiring animals to wander into neighboring territories.
such that Z p 10.0. This implies that the territory acqui- While the home range is a measurement of the utilization
sition time Tat was approximately 5 days. distribution across a period of time such as a day, month,
or season, the territory is the area being defended, for
Discussion
By applying recently developed agent-based models of ter-
ritory formation to a natural experiment in a red fox pop-
ulation, we have quantified the changes in both territorial
patterns and individual behavior elicited by a rapid pop-
ulation decline in a paradigmatic territorial species. Our
modeling framework enabled us to relate elasticity in ter-
ritory borders directly to the individual-level movement
and interaction mechanisms, allowing information about
territorial dynamics to be inferred from animal movement
data. We have constructed a program for making these Video 2: Video 2, available online, shows the dynamics of territorial
inferences, when interactions are scent mediated, by fitting acquisition. It shows 25 territorial random walkers on a 100 #
a time-evolving probability distribution to spatiotemporal 100-square lattice with periodic boundary conditions. Each territory
location data, and we have applied this to data on red fox is denoted by a different color. The white squares are interstitial
movements. This program gives a way of quantifying the regions, where there is no active scent. Partway through the video,
the simulated animal with the cyan territory is removed, shown by
mechanisms underpinning the elastic disc hypothesis the color of the territory turning black. As the scent of this animal
(Huxley 1934). becomes inactive, the black squares turn white, allowing the other
As the 1994–1996 epizootic of sarcoptic mange deci- animals to move in and acquire the territory.

The American Naturalist 2013.182:E73-E82.

Downloaded from www.journals.uchicago.edu by 115.178.255.6 on 03/22/20. For personal use only.
 Behavior Changes in Population Declines E81

example, by fresh scent marks, at any point in time. As Acknowledgments

the locations of an animal are measured over time, its
This work was partially supported by Engineering and
territory varies in both position and size. These fluctua-
Physical Sciences Research Council grants EP/E501214/1
tions may be difficult to detect in ecological studies, since
(J.R.P.) and EP/I013717/1 (L.G.) and by the Dulverton
by the time sufficient location fixes have been obtained to
Trust (S.H). We thank the editors and two anonymous
measure territory size, using either KDE (Worton 1989) reviewers, whose comments helped improve the article.
or other techniques (as reviewed in Fieberg and Börger
2012), the borders may have changed. Consequently, if the
fluctuating territories are contiguous, the utilization dis- Literature Cited
tributions measured over a period of time will overlap
(Giuggioli et al. 2011a). So overlap may simply be an Adams, E. S. 2001. Approaches to the study of territory size and shape.
artifact of the timescale for the data collection, due to Annual Review of Ecology, Evolution, and Systematics 32:277–303.
Arnold, J. 2009. Olfactory communication in red foxes (Vulpes
home range size increasing over time (Börger et al. 2006),
vulpes). PhD thesis, University of Bristol.
and not an implicit biological phenomenon, although Baker, P., S. M. Funk, S. Harris, and P. C. L. White. 2000. Flexible
there may be circumstances where it is related to com- spatial organization of urban foxes, Vulpes vulpes, before and during
petition for resources (Buchmann et al. 2010) or territorial an outbreak of sarcoptic mange. Animal Behaviour 59:127–146.
intrusion (Burt 1943). Beier, P. 1993. Determining minimum habitat areas and habitat cor-
Though we used random walks to model fox movement ridors for cougars. Conservation Biology 7:94–108.
Benhamou, S. 2011. Dynamic approach to space and habitat use
in Bristol’s resource-rich environment, movement pro-
based on biased random bridges. PLoS ONE 6:e14592.
cesses in a resource-poor environment probably exhibit a Benhamou, S., and L. Riotte-Lambert. 2012. Beyond the utilization
certain amount of heterogeneity, dependent on the spatial distribution: identifying home range areas that are intensively ex-
distribution of food resources (Tremblay et al. 2007). It is ploited or repeatedly visited. Ecological Modelling 227:112–116.
evident from the comparison of two simple animal move- Börger, L., B. Dalziel, and J. M. Fryxell. 2008. Are there general
ment processes, ballistic motion and diffusive random mechanisms of animal home range behavior? a review and pros-
pects for future research. Ecology Letters 11:637–650.
walks, that the nature of the animal movement greatly
Börger, L., N. Franconi, F. Ferretti, F. Meschi, G. De Michele, A.
affects the territory border dynamics (fig. 2). Recent stud- Gantz, and T. Coulson. 2006. An integrated approach to identify
ies by Benhamou (2011) and Benhamou and Riotte-Lam- spatiotemporal and individual-level determinants of animal home
bert (2012) use movement-based kernel density estima- range size. American Naturalist 168:471–485.
tions and Brownian bridge techniques to determine the Briscoe, B. K., M. A. Lewis, and S. E. Parrish. 2002. Home range
heterogeneous movement processes that arise from ani- formation in wolves due to scent marking. Bulletin of Mathe-
matical Biology 64:261–284.
mals searching for food within their home ranges. In the
Buchmann, C. M., F. M. Schurr, R. Nathan, and F. Jeltsch. 2010. An
future, more complicated movement processes, caused ei- allometric model of home range formation explains the structuring
ther by resource distribution or other ecological phenom- of animal communities exploiting heterogeneous resources. Oikos
ena (e.g., Morales et al. 2004; Reynolds 2010), could be 120:106–118.
built into our modeling framework to quantify territorial Burt, W. H. 1943. Territoriality and home range concepts as applied
dynamics in populations where the movement cannot be to mammals. Journal of Mammalogy 24:346–352.
Condamin, S., O. Bénichou, and M. Moreau. 2007. Random walks
realistically modeled by a simple random walk.
and Brownian motion: a method of computation for first-passage
As well as providing insights into behavioral ecology, times and related quantities in confined geometries. Physical Re-
these results have important consequences for modeling view E 75:021111.
epizootic spread, particularly when the disease is terminal Durrett, R., and S. Levin. 1994. The importance of being discrete
or restricts movement and hence scent-marking behavior. (and spatial). Theoretical Population Biology 46:363–394.
Rather than assuming constant territory or home range Ezaki, Y. 1995. Establishment and maintenance of the breeding ter-
ritory in the polygynous great reed warbler. Ecological Research
sizes (e.g., Smith and Harris 1991; Smith and Wilkinson
10:359–368.
2002; Kenkre et al. 2007; Salkeld et al. 2010), our results Fieberg, J., and L. Börger. 2012. Could you please phrase “home
demonstrate that the rapid decline of population density range” as a question? Journal of Mammalogy 93:890–902.
elicited by terminal disease spread causes large changes in Giuggioli, L., J. R. Potts, and S. Harris. 2011a. Animal interactions
the territorial dynamics. This in turn affects the behavior and the emergence of territoriality. PLoS Computational Biology
of the animals and the interanimal contact rates. Our ap- 7:1002008.
———. 2011b. Brownian walkers within subdiffusing territorial
proach provides the necessary basis to enable future epi-
boundaries. Physical Review E 83:061138.
demiological studies to take such behavioral and territorial ———. 2012. Predicting oscillatory dynamics in the movement of
fluctuations into account, allowing for improved predic- territorial animals. Journal of the Royal Society Interface 9:1529–
tions of disease spread. 1543.

The American Naturalist 2013.182:E73-E82.

Downloaded from www.journals.uchicago.edu by 115.178.255.6 on 03/22/20. For personal use only.
E82 The American Naturalist

Grant, J. W. A., and D. L. Kramer. 1990. Territory size as a predictor Okubo, A., and S. A. Levin. 2002. Diffusion and ecological problems:
of the upper limit to population density of juvenile salmonids in modern perspectives. 2nd ed. Springer, New York.
streams. Canadian Journal of Fisheries and Aquatic Sciences 47: Potts, J. R., S. Harris, and L. Giuggioli. 2011. An anti-symmetric
1724–1737. exclusion process for two particles on an infinite 1-D lattice Journal
Harris, S., W. J. Cresswell, P. G. Forde, W. J. Trewhella, T. Woollard, of Physics A 44:485003.
and S. Wray. 1990. Home-range analysis using radio-tracking data: ———. 2012. Territorial dynamics and stable home range formation
a review of problems and techniques particularly as applied to the for central place foragers PLoS ONE 7:e34033.
study of mammals. Mammal Review 20:97–123. ———. 2013. Data from: Quantifying behavioral changes in terri-
Hutchings, M. R., K. M. Service, and S. Harris. 2001. Defecation and torial animals caused by sudden population declines. American
urination patterns of badgers Meles meles at low density in south- Naturalist, Dryad Digital Repository, http://dx.doi.org
west England. Acta Theriologica 46:87–96. /10.5061/dryad.5dn48.
Huxley, J. S. 1934. A natural experiment on the territorial instinct. Redner, S. 2007. A guide to first-passage processes. Cambridge Uni-
British Birds 27:270–277. versity Press, Cambridge.
Jetz, W., C. Carbone, J. Fulford, and J. H. Brown. 2004. The scaling Reynolds, A. M. 2010. Animals that randomly reorientate at cues left
of animal space use. Science 306:266–268. by correlated random walkers do the Lévy walk. American Nat-
Kenkre, V. M., L. Giuggioli, G. Abramson, and G. Camelo-Neto. uralist 175:607–613.
2007. Theory of hantavirus infection spread incorporating local- Salkeld, D. J., M. Salathé, P. Stapp, and J. H. Jones. 2010. Plague
ized adult and itinerant juvenile mice. European Physical Journal outbreaks in prairie dog populations explained by percolation
B 55:461–470. thresholds of alternate host abundance. Proceedings of the Na-
Kenward, R. E., R. T. Clarke, K. H. Hodder, and S. S. Walls. 2001. tional Academy of Sciences of the USA 107:14247–14250.
Density and linkage estimators of home range: nearest-neighbor Saunders, G., P. C. L. White, S. Harris, and J. M. V. Rayner. 1993.
clustering defines multinuclear cores. Ecology 82:1905–1920. Urban foxes (Vulpes vulpes): food acquisition, time and energy
Kruuk, H., and T. Parish. 1982. Factors affecting population density, budgeting of a generalized predator. Symposia of the Zoological
group size and territory size of the European badger, Meles meles. Society of London 65:215–234.
Journal of Zoology (London) 196:31–39. Schradin, C., G. Schmohl, R. G. Rödel, I. Schoepf, S. M. Treffler, J.
Lagarias, J. C., J. A. Reeds, M. H. Wright, and P. E. Wright. 1998. Brenner, M. Bleeker, M. Schubert., B. König, and N. Pillay. 2010.
Convergence properties of the Nelder-Mead simplex method in Female home range size is regulated by resource distribution and
low dimensions. SIAM Journal of Optimization 9:112–147. intraspecific competition: a long-term field study. Animal Behav-
Landim, C. 1992. Occupation time large deviations for the symmetric iour 79:195–203.
simple exclusion process. Annals of Probability 20:206–231. Smith, G. C., and S. Harris. 1991. Rabies in urban foxes (Vulpes
Laver, P. N., and M. J. Kelly. 2008. A critical review of home range vulpes) in Britain: the use of a spatial stochastic simulation model
studies. Journal of Wildlife Management 72:290–298. to examine the pattern of spread and evaluate the efficacy of dif-
Lewis, M. A., and J. D. Murray. 1993. Modelling territoriality and ferent control regimes. Philosophical Transactions of the Royal
wolf-deer interactions. Nature 366:738–740. Society B: Biological Sciences 334:459–479.
Liggett, T. M. 1985. Interacting particle systems. Springer, New York. Smith, G. C., D. Wilkinson. 2002. Modelling disease spread in a
Macdonald, D. W. 1980. Patterns of scent marking with urine and novel host: rabies in the European badger Meles meles. Journal of
faeces amongst carnivore communities. Symposia of the Zoological Applied Ecology 39:865–874.
Society of London 45:107–139. Stamps, J. A., and V. V. Krishnan. 1998. Territory acquisition in
McCarthy, K. P., and S. Destefano. 2011. Effects of spatial disturbance lizards. IV. Obtaining high status and exclusive home ranges. An-
on common loon nest site selection and territory success. Journal imal Behaviour 54:461–472.
of Wildlife Management 75:289–296. Stutchbury, B. J. 1991. Floater behavior and territory acquisition in
McKane, A. J., and T. J. Newman. 2004. Stochastic models in pop- male purple martins. Animal Behaviour 42:435–443.
ulation biology and their deterministic analogs. Physical Review Tremblay, Y., A. J. Roberts, and D. P. Costa. 2007. Fractal landscape
E 70:041902. method: an alternative approach to measuring area-restricted
Moorcroft, P. R., and A. Barnett. 2008. Mechanistic home range searching behavior. Journal of Experimental Biology 210:935–945.
models and resource selection analysis: a reconciliation and uni- Van Moorter, B., D. Visscher, S. Benhamou, L. Börger, M. S. Boyce,
fication. Ecology 89:1112–1119. and J. M. Gaillard. 2009. Memory keeps you at home: a mecha-
Moorcroft, P. R., and M. A. Lewis. 2006. Mechanistic home range nistic model for home range emergence. Oikos 118:641–652.
analysis. Princeton University Press, Princeton, NJ. Viswanathan, G. M., E. P. Raposo, F. Bartumeus, J. Catalan, and M.
Moorcroft, P. R., M. A. Lewis, and R. L. Crabtree. 2006. Mechanistic G. E. da Luz. 2005. Necessary criterion for distinguishing true
home range models capture spatial patterns and dynamics of coy- superdiffusion from correlated random walk processes. Physical
ote territories in Yellowstone. Proceedings of the Royal Society B: Review E 72:011111.
Biological Sciences 273:1651–1659. Wasserman, L 2004. All of statistics: a concise course in statistical
Morales, J. M., D. T. Haydon, J. Friar, K. E. Holsinger, and J. M. Fryxell. inference. 2nd ed. Springer, New York.
2004. Extracting more out of relocation data: building movement Worton, B. J. 1989. Kernel methods for estimating the utilization
models as mixtures of random walks. Ecology 85:2436–2445. distribution in home-range studies. Ecology 70:164–168.
Newton-Fisher, N., S. Harris, P. White, and G. Jones. 1993. Structure
and function of red fox Vulpes vulpes vocalisations. Bioacoustics 5: Associate Editor: Peter Nonacs
1–31. Editor: Judith L. Bronstein

The American Naturalist 2013.182:E73-E82.

Downloaded from www.journals.uchicago.edu by 115.178.255.6 on 03/22/20. For personal use only.