A new approach to geostatistical synthesis of historical records reveals capuchin spatial responses to climate and demographic change
Cite this dataset
Perry, Susan; Jacobson, Odd; Barrett, Brendan (2024). A new approach to geostatistical synthesis of historical records reveals capuchin spatial responses to climate and demographic change [Dataset]. Dryad. https://doi.org/10.5068/D1QT17
Abstract
The recent proliferation of GPS technology has transformed animal movement research. Yet, time-series data from this recent technology rarely span beyond a decade, constraining longitudinal research. Long-term field sites hold valuable historic animal location records, including hand-drawn maps and semantic descriptions. Here, we introduce a generalized workflow for converting such records into reliable location data to estimate home ranges, using 30 years of sleep-site data from 11 white-faced capuchin (Cebus imitator) groups in Costa Rica. Our findings illustrate that historic sleep locations can reliably recover home range size and geometry. We showcase the opportunity our approach presents to resolve open questions that can only be addressed with very long-term data, examining how home ranges are affected by climate cycles and demographic change. We urge researchers to translate historical records into usable movement data before this knowledge is lost; it is essential to understand how animals are responding to our changing world.
README: A new approach to geostatistical synthesis of historical records reveals capuchin spatial responses to climate and demographic change
Raw sleep site data:
Here we present two chunks of data that were created in the process of converting early historical data to something that could be linked to GPS locations so that the full process can be understood. We did not provide the more recent data for conservation reasons, as we want to protect the locations of habituated primates. Scientists with a legitimate reason to see the full data set can contact Susan Perry for more information.
Sleepsites_1997_oddshift.csv
SleepsitesAA1990-95_bjbshift.csv
The file Sleepsites_1997_oddshift.csv
is from the first season in which electronic devices (Psion Observers) were used to record data in the field (though these were still supplemented with field notes having hand-drawn maps to help us find sleep sites the following morning, as this was our first field season working with RR group, so its home range had not yet been mapped). The file 1993AAsubsetData.xlsx
is a chunk of the larger file SleepsitesAA1990-95_bjbshift.csv
that is mentioned in the R code. This is from the first 3 years of the project when we focused exclusively on the AA group, recording data on microcassette recorders and writing additional notes about sleep site locations in field notebooks.
Originally there were more columns, including (a) values for a 50 m grid system (which we abandoned when it became clear that we could rarely pinpoint the precise grid cell at that level of granularity), (b) columns indicating whether we saw the monkeys curl up in their sleep trees, or at least were with them during the last hour before sundown, (c) extensive comments about how certain we were about the specific grid cell assigned and listing other possible values (when we were considering listing all possible grid cell values), and (d) comments containing notes from Susan about which sleep sites needed to be double checked via older versions of our maps, and then checked in the field by finding those points and taking a GPS reading. When this became possible, we corrected the Grid100 value. The versions presented here use only the columns that are used in the present analysis, after we had refined our data processing workflow as described in the manuscript.
- date: the date representing the evening on which we watched the monkeys settle into their sleep tree for the night.
- location: a description of where the monkeys slept
- group: the group of monkeys monitored on that date
- Grid100: the number assigned to the grid cell in the 100 m2 grid. Note that the numbers in the entire grid published in the code repository have been systematically altered to protect these threatened monkeys, so that their sleep sites cannot be easily located by non-researchers.
- certainty?: As described in the manuscript, a score of 0 indicates that we can confidently assign the sleep site to that grid cell. A score of 1 means that the sleep site lies within a radius of 1 away from the grid cell assigned, i.e. could be within that cell or one of the 8 neighboring grid cells (corresponding to 300 meters squared). A score of 2 means that the sleep site was within 2 grid cells of the one chosen (corresponding to 500 meters squared), etc.
Demographic data:
annual_group_sizes.csv
This file reports the group size (including members of all age-sex classes) for each social group in each year, derived from the Lomas Barbudal Monkey Project demographic database, using only census days in which trained observers were present in the group for at least 6 hours.
Raw GPS data:
For conservation reasons, i.e., to protect the animals from the pet trade and poachers, these data will not be made public. They are archived at Movebank, and can be accessed here, pending permission of the curator (Susan Perry):
https://www.movebank.org/cms/webapp?gwt_fragment=page=studies,path=study3389013696
Description of the relationship between the data and the code used to analyze it:
To analyze these data, please refer to the code found at the following location:
https://doi.org/10.17617/3.HUKMS6
The descriptions of sleep sites were transformed to location data using the quadrant centroids in which they fell, using a map that had a 100x100m2 grid. The code for this and other procedures described below can be found in 01_georeferencing/scripts/00_hist_slp_data_processing.R.
All location data (georeferenced data, GPS-tracking data, and GPS sleep-site data) were used to estimate home ranges using the ctmm package in R. To do this, one can follow 02_validation/scripts
for annotated steps within the 01 and 02 scripts.
Next, the home ranges using sleep site data were validated, using the scripts 03-06, by comparing them to home ranges estimated from tracking data. Then home ranges estimated from sleep site data, once validated, were used in the case study models to draw predictions about the effects of ENSO and group size on home range area. To generate the tables necessary for these models, run the 07 script within the 02_validation/scripts
folder. The statistical models and plots for the case study are found in the 03_case_study_models
folder. For more details please see the README.md file within the Edmond code repository linked above.
Code/Software
The code used to process these raw data is housed in the following repository:
Funding
National Science Foundation, Award: BCS-1638428, Biological Anthropology
National Science Foundation, Award: BCS-0613226, Biological Anthropology
National Science Foundation, Award: BCS-0848360, Biological Anthropology
National Science Foundation, Award: DDIG 1232371
National Science Foundation, Award: 9633991
National Science Foundation, Award: SES-99870429
National Science Foundation, 2 Graduate Fellowships
National Geographic Society, Award: 7968-06
National Geographic Society, Award: 8671-09
National Geographic Society, Award: 20113909
National Geographic Society, Award: 9795-15
National Geographic Society, Award: 45176R-18
Leakey Foundation, Award: (9 grants)
Templeton World Charity Foundation, Award: 0208
Wenner-Gren Foundation, Award: 8409 plus one unnumbered grant
National Geographic Society, Award: 1 unnumbered grant (co-PI Barbara Smuts)
Wild Capuchin Foundation
University of California, Los Angeles
University of Michigan–Ann Arbor
Max Planck Institute for Evolutionary Anthropology
American Society of Primatologists
ARCS Foundation
Sigma Xi