Habitual Route Analysis Method (HRAM) and Geographic technologies to evaluate animal navigation

Abstract

Wild animal navigation usually provides complex data about the ecological aspects of space usage. Developments in Geography technologies play an important role in animal ecology and behavioral studies. The significant improvement in Global Positioning Systems (GPS) and analytical power of Geographic Information System (GIS) has furthered the discussion about patterns of navigation in wild animals. In the wild animals constantly revisit locations that provide important resources within their home range. In doing so, different species use diverse patterns of navigation to reach repeated locations. Yet, repetition of routes and route segments seems to be a default system in many species. Here we suggest an automated process, the habitual route analysis method (HRAM), which detects the use of habitual routes by wild animals. The HRAM it is a tool with substantial advantage to investigate the daily vector of navigation to assess if animals travel over repeated routes during the period of data collection. We analyzed the capabilities of the HRAM by test existing data of 58 days traveled by wild black capuchin monkeys in the rain forest and compared the automated method to the previously used manual method based only on GIS.

Full text

1
Habitual Route Analysis Method (HRAM) and Geographic technologies to evaluate animal
1
navigation
2
3
4
Andrea Presottoa,* Caitlin Currya, Patricia Izarb
5
a Department of Geography and Geosciences, Salisbury University, Salisbury, MD 21801, USA
6
b Department of Experimental Psychology, University of Sao Paulo, Sao Paulo, SP 08550,
7
Brazil
8
9
* Corresponding author
10
Email address: axpresotto@salisbury.edu
11
12
13
14
15
16
17
18
19
20
21
22
23

2
Abstract
24
Wild animal navigation usually provides complex data about the ecological aspects of space usage.
25
Developments in Geography technologies play an important role in animal ecology and behavioral
26
studies. The significant improvement in Global Positioning Systems (GPS) and analytical power
27
of Geographic Information System (GIS) has furthered the discussion about patterns of navigation
28
in wild animals. In the wild animals constantly revisit locations that provide important resources
29
within their home range. In doing so, different species use diverse patterns of navigation to reach
30
repeated locations. Yet, repetition of routes and route segments seems to be a default system in
31
many species. Here we suggest an automated process, the habitual route analysis method (HRAM),
32
which detects the use of habitual routes by wild animals. The HRAM it is a tool with substantial
33
advantage to investigate the daily vector of navigation to assess if animals travel over repeated
34
routes during the period of data collection. We analyzed the capabilities of the HRAM by test
35
existing data of 58 days traveled by wild black capuchin monkeys in the rain forest and compared
36
the automated method to the previously used manual method based only on GIS.
37
38
Keywords: Python, Habitual routes, Animal navigation, Geographic Information System (GIS),
39
Spatial analysis
40
41
42
43
44
45
46

3
Introduction
47
The ability to associate wild animals to its geographic location is a breakthrough to understanding
48
the ecological aspects of space usage of different species. Recent developments in ecological
49
informatics and geographic technologies play an increasingly important role in animal ecology
50
and behavioral studies (Brooks, Bonyongo, & Harris, 2008). The significant improvement in
51
analytical power of Global Positioning Systems (GPS) data, and Geographic Information System
52
(GIS) analytics further the discussion of how wild animal travel (Blake, Douglas‐Hamilton, &
53
Karesh, 2001; Bohrer, Beck, Ngene, Skidmore, & Douglas-Hamilton, 2014; Cagnacci, Boitani,
54
Powell, & Boyce, 2010).
55
GPS is the first improvement in geographic technologies to overcome the limitations when
56
studying wild animal navigation (Kie et al., 2010). It facilitates the development of new research
57
designs on animal navigation along with the GIS that allows for large data storage and spatial and
58
temporal measurements. Yet, when comes to large amount of data the visual approach and limited
59
operations of GIS can be time consuming. The quantification of repetition of routes and route
60
segments using only visual analysis may affect the accuracy of the results. It is widely spread that
61
wild animals travel far and come back to specific locations in searching for resources. The
62
repetition of travel routes or route segment is a pattern consistent with animals navigating through
63
a route network system, what has been describe for various species (Collett, 2010; Di Fiore &
64
Suarez, 2007; Noser & Byrne, 2007; Presotto et al., 2018; Wystrach & Graham, 2012). This pattern
65
of navigation is associated with the use of a sequence of landmarks, which in turn represents a
66
habitual route. The use of habitual routes may be a default mechanism to navigate.
67

4
We believe that travel over an advantage point, when a panoramic view is possible allow at least
68
beard capuchins to visually detect routes, ting them with landmarks, and constantly repeat routes
69
or route segments to find their way around (Presotto et al. 2018; Suarez, 2014).
70
Thus, we developed a tool that detects the use of repeated routes and route segments comprising
71
the habitual routes traveled when a species revisit resource site, which seems a significant
72
advantage to travel in the wild.
73
Long-term studies generate increasing amount of data (Fig. 1) Thus, we argue that there is a
74
substantial advantage to investigate the routes and route segments along with frequently of visit
75
resources by using an automated method. Here, we demonstrate the advantages of the HRAM over
76
the manual approach in GIS only.
77
78
79
80
Fig. 1 Elephant collar data showing GPS fixes
81
82

5
Habitual Route Network Analysis (HRAM)
83
HRAM identifies repeated route and route segments. It uses the daily travel vector to identify the
84
repetitions. Thus, the input data is a line shapefile. The method is a combination of python and
85
structured query language (SQL) within Pgadmin by PostGres. Pgadmin is an open source object-
86
relational database management system, which utilizes a spatial extension called PostGIS. The
87
PostGIS extension enables spatial queries based on the geometry of the projected data. HRAM
88
detects habitual routes based on the hypothesizes that animals using habitual routes: (a) repeat
89
routes and route segments more often than they create new segments when revisiting resources
90
and the usage of (b) intersections along the routes that could be used as landmarks.
91
Datasets used in conjunction with the tool are organized into number sequences that
92
chronologically represent each month. For example, January is 01, February is 02, and so on for
93
the rest of the months. Daily routes are stored within their corresponding month's folder (Fig. 2).
94
This organization system is critical for the execution of the tool. The initial inputs include the path
95
to the data, the Spatial Reference System Identifier (SRID), the buffer distance, and the database
96
credentials. The tool has seven functions throughout the script and each executes an individual
97
task.
98
99
Fig 2. Example of data storage
100

6
HRAM isolates daily routes one by one and creates a buffer around the routes. The buffer created
101
around the route represents the visual field an animal can see given the factors of their
102
environment. This visual distance is defined by the user based on an expert's knowledge about how
103
far the species can see in its environment. For instance, studies with capuchin monkeys assume
104
the buffer distance based on the monkeys’ sight range. Capuchin monkeys can see within a forested
105
environment 50 meters (Janson & Di Bitetti, 1997). The unit of measurement that determines the
106
buffer’s distance is based on the data’s projection. In this study, the data is projected using
107
WGS_1984_UTM_Zone_23S, which uses meters as its unit of measurement. Thus, when the user
108
inputs ‘50’ for the buffer distance, the tool interprets it as 50 meters. For accurate results, the data
109
must be projected. If data are projected in feet results will be presented in feet.
110
Once the daily route buffer is created, all other daily routes generated from the same dataset are
111
overlaid onto the buffer for the entire study period. For example, for black capuchins, if August
112
19th is the first entire day in the dataset, a buffer of 50 m is created around August 19th’s vector
113
trajectory. Then all the other days within the study are compared against the area of August 19’s
114
buffer (Fig. 3).
115
116

7
Fig. 3: a) All September 2007 routes and August 19th, 2007; b) all September 2007 days and
117
segments that repeated the location used August 19th, 2007; c) Route segments repeated and
118
intersected in September previously used August 19th, 2007.
119
120
When another route is found within that area then it is considered a repeated route. HRAM outputs
121
two shapefiles for each month. The first shapefile shows the shapes of the segments which are
122
repeated for each daily route within the month. The second shapefile has the amount of repetitions
123
for each route within the month.
124
HRAM excludes routes and route segment repetition within the same month. For example, if
125
capuchins revisit a specific resource April 1st and repeat the route segment on April 15th, then this
126
segment was excluded to eliminate the possibility that the monkeys travel using habitual routes
127
because the same resources happen within the same month following the same principle applied
128
at Presotto and Izar (2010). For example, April and May routes are plotted together, and then for
129
both months, another layer representing only repeated routes and route segments between these
130
two months is created. This route represents the repeated routes or route segments for April_May.
131
April is then compared to June, generating the repeated route and route segments for April_June.
132
Once all routes and route segments are generated, all segments within the 50 m buffer of all daily
133
routes compose the habitual routes used by the studied species (Fig. 4).
134
HRAM output results allow further analysis using GIS. HRAM is freely available on Github
135
(Curry and Presotto 2018).
136
Building Daily Routes
137
The daily routes can be easily created using an open source software that we tested prior to run the
138
HRAM. The GME – Geospatial Modelling Environment creates point to lines, converting the point
139

8
daily point shapefile to daily line shapefile. This task can also be done using python code or Arcpy
140
library. The connection of geographic coordinates results in a mathematical vector trajectory.
141
After HRAM detects the habitual routes the analysis can be expanded using GIS in many different
142
ways. We create buffers (zones of predetermined distance) around the habitual routes and extract
143
a list of all coordinate points that are locate within these zones. Then we calculate the proportion
144
of locations that fall around the habitual routes, such as: a) total geographic coordinates of the
145
animal locations, b) location of each food sources used by the animals, and c) location of any
146
sleeping sites (e.g. Presotto et al. 2018; Presotto and Izar 2010).
147
148
149
Discussion
150
The representation of daily routes associated with behavioral data is rich in details about locations
151
and animal activities in time-space. It is time consuming using GIS manual method analysis but
152
combined with python script it can significantly reduce the time while giving the power to discern
153
patterns of navigation among species. The output of HRAM showing where segments overlap
154
allow information extraction that can fast show how many times the location was used, the first
155
visit and the subsequent visits of that location, and if combined with a satellite image, the
156
predominant habitat information of the habitual routes.
157
As much as we agree with the aspect that even when the data collection is not enough to show
158
environmental or assuming behavioral details, patterns of travel as basic information can be
159
important to build databases find preliminary results, and provide additional information,
160
maintaining a standardized data set to future time-space analysis.
161
162

9
Acknowledgements
163
We thank the GIS graduate program at Salisbury University for supporting this project. We
164
thank Noah Krach for the technical assistance along with Stuart Hamilton for his valuable
165
suggestions. We thank Connect Conservation for the elephant data and the Instituto Florestal de
166
São Paulo, especially the Parque Estadual Carlos Bothelho’s manager, José Carlos Maia, for
167
permission to collect capuchin data. Data collection was supported by FAPESP grant 06/56059-0
168
to P.I. and CNPq grants to P.I. and A.P. and Connect Conservation.
169
References
170
Blake, S., Douglas‐Hamilton, I., & Karesh, W. B. (2001). GPS telemetry of forest elephants in
171
central africa: Results of a preliminary study Wiley Online Library.
172
Bohrer, G., Beck, P. S., Ngene, S. M., Skidmore, A. K., & Douglas-Hamilton, I. (2014).
173
Elephant movement closely tracks precipitation-driven vegetation dynamics in a Kenyan
174
forest-savanna landscape BioMed Central.
175
Brooks, C., Bonyongo, C., & Harris, S. (2008). Effects of global positioning system collar weight
176
on zebra behavior and location error Wiley Online Library.
177
Cagnacci, F., Boitani, L., Powell, R. A., & Boyce, M. S. (2010). Animal ecology meets GPS-
178
based radiotelemetry: A perfect storm of opportunities and challenges The Royal Society.
179
Collett, M. (2010). How desert ants use a visual landmark for guidance along a habitual route
180
National Acad Sciences.
181

10
Di Fiore, A., & Suarez, S. A. (2007). Route-based travel and shared routes in sympatric spider
182
and woolly monkeys: Cognitive and evolutionary implications Springer.
183
Janson, C. H., & Di Bitetti, M. S. (1997). Experimental analysis of food detection in capuchin
184
monkeys: Effects of distance, travel speed, and resource size Springer.
185
Kie, J. G., Matthiopoulos, J., Fieberg, J., Powell, R. A., Cagnacci, F., Mitchell, M. S., . . .
186
Moorcroft, P. R. (2010). The home-range concept: Are traditional estimators still relevant
187
with modern telemetry technology? The Royal Society.
188
Noser, R., & Byrne, R. W. (2007). Travel routes and planning of visits to out-of-sight resources
189
in wild chacma baboons, Papio ursinus Elsevier.
190
Presotto, A., Verderane, M. P., Biondi, L., Mendonca-Furtado, O., Spagnoletti, N., Madden, M.,
191
& Izar, P. (2018). Intersection as key locations for bearded capuchin monkeys (Sapajus
192
libidinosus) traveling within a route network Springer.
193
Suarez, S. A. (2014). Ecological factors predictive of wild spider monkey (Ateles belzebuth)
194
foraging decisions in Yasuni
,
Ecuador Wiley Online Library.
195
Wystrach, A., & Graham, P. (2012). What can we learn from studies of insect navigation?
196
Elsevier.
197
198