Data from: Combining past and contemporary species occurrences with ordinal species distribution modeling to investigate responses to climate change
Data files
Jan 10, 2025 version files 48.49 KB
-
Pika_Patch-level_Data_Ordinal_Modeling_.csv
45.56 KB
-
README.md
2.94 KB
Abstract
Many organisms leave evidence of their former occurrence, such as scat, abandoned burrows, middens, ancient eDNA, or fossils, which indicate areas from which a species has since disappeared. However, combining this evidence with present-day occurrences within a single modeling framework remains challenging. Traditional binary species distribution modeling reduces occurrence to two temporally vague states (present/absent), so thus cannot leverage the information inherent in temporal sequences of evidence of past occurrence. In contrast, ordinal modeling can use the natural time-varying order of states (e.g., never occupied vs. occupied in the past vs. currently occupied) to provide greater insights into range shifts. We demonstrate the power of ordinal modeling for identifying the major influences of biogeographic and climatic, variables on current and past occupancy of the American pika (Ochotona princeps), a climate-sensitive mammal. Sampling over 5 years across the species’ southernmost, warm-edge range limit, we tested the effects of these variables at 570 habitat patches where occurrence was classified either as binary or ordinal. The two analyses produced different top models and predictors – ordinal modeling highlighted chronic cold as the most-important predictor of occurrence, whereas binary modeling indicated primacy of average summer-long temperatures. Colder wintertime temperatures were associated in ordinal models with higher likelihood of occurrence, which we hypothesize reflect longer retention of insulative and meltwater-provisioning snowpacks. Our binary results mirrored those of several other past pika investigations employing binary analysis. Because both ordinal- and binary-analysis top models included climatic and biogeographic factors, results constitute important considerations for climate-adaptation planning. Cross-time evidences of species occurrences are common, yet remain underutilized for assessing responses to climate change. Compared to multi-state occupancy modeling, which presumes all states occur in the same time period, ordinal models enable use of historical evidence of species’ occurrence to identify factors driving species’ distributions more finely across time.
README: Data from: Combining past and contemporary species occurrences with ordinal species distribution modeling to investigate responses to climate change
https://doi.org/10.5061/dryad.z612jm6n0
Description of the data and file structure
Our dataset includes n = 570 habitat patches surveyed by >1 of the authors and/or field technicians. We identified talus or broken-rock habitat patches to survey using high-resolution imagery in CalTopo.com. In the field patches were determined to be currently occupied, previously occupied, or having no evidence of occupancy by the study organism, the American pika (Ochotona princeps). The dataset spans 5 years, from 2016 to 2020 inclusive. We utilized this data to assess how analyzing binary (currently occupied vs unoccupied) and ordinal (occupied vs historically occupied vs no evidence) may indicate different factors as being important to describe species distributions, over different timescales. We analyzed the importance of climate variables downloaded from PRISM AN81d data, 2006-2021 (Daly et al. 2002).
Files and variables
File: Pika_Patch-level_Data_Ordinal_Modeling_.csv
Description: Ordinal occupancy status of surveyed habitat patches.
Variables
siteName: The name of the patch, designated by the authors and technicians
longitude: The geographic longitude recorded by handheld GPS units (rounded to two decimals)
latitude: The geographic latitude recorded by handheld GPS units (rounded to two decimals)
surveyYear: The year of the most recent survey for each patch
lastSurveyDate: The date of the most recent survey for each patch
ordinalStatus: Parameterized as 0 (no evidence of pika occurrence), 1 (old evidence of pika occurrence) or 2 (evidence of current pika occurrence)
binaryStatus: Given as either 0 (not currently pika-occupied) or 1 (current pika occurrence detected)
sub-region: Indicates northeast, southeast, northwest, and southwest “sub-regions” of mountain ranges that are separated from each other by ~45-100 km of lower-elevation topography in our study region.
elevation_m: Elevation, given in meters, was recorded by handheld GPS units
numHomeRanges: The number of 20 meter diameter potential pika home ranges estimated in situ by observers, aided by laser rangefinders.
meanDistToClosest4Patches_m: the mean geographic distance to each of four nearest patches
Code/software
Analyzed with R software. To analyze occupancy, we used ordinal models along continuous climate variables and a logistic error distribution (Package OrdinalNet version 2.12 in R, Wurm et al. 2021). We also used binary models that assumed a binomial error distribution (function 'glm' within 'lme4', Bates et al. 2015).
Access information
Other publicly accessible locations of the data:
- NA
Data was derived from the following sources:
- NA
Methods
Our dataset includes n = 570 habitat patches surveyed by >1 of the authors and/or field technicians and determined to be currently occupied, previously occupied, or having no evidence of occupancy by the study organism, the American pika (Ochotona princeps). The dataset spans 5 years, from 2016 to 2020 inclusive. Latitude and longitude were recorded using hand-held GPS devices accurate to 2-7 m (WAAS-enabled), and “survey Year” is the year in which the site was most recently surveyed. “Ordinal status” is parameterized as 0 (no evidence of pika occurrence), 1 (old evidence of pika occurrence), or 2 (evidence of current pika occurrence). Binary status is given as either 0 (not currently pika-occupied) or 1 (current pika occurrence detected). Our study domain is divided into northeast, southeast, northwest, and southwest “sub-regions” of mountain ranges that are separated from each other by ~45-100 km of lower-elevation topography. We recorded “elevation” (in meters) using handheld GPS units. The number of home ranges (“numHomeRanges”) is given for each site and was estimated in situ by observers, aided by laser rangefinders. If more than one value was given for the number of home ranges by different observers, then we used the mean value. The mean distance to the closest four patches is given in meters (m) and was calculated as the mean geographic distance to each of four nearest patches, which we used to determine patch-level isolation.