A new Double Observer based census framework to improve abundance estimations in mountain ungulates and other gregarious species with a reduced effort
Data files
Nov 26, 2024 version files 69.79 MB
-
DOAS_maindata.zip
69.79 MB
-
README.md
1.58 KB
Abstract
Estimating animal abundance has a key role to play in ecology and conservation, but survey methods are always challenged by imperfect detection. Among the techniques applied to deal with this issue, Double Observer (DO) is increasing in popularity due to its cost-effectiveness. However, the effort of using DO for surveying large territories can be significant. A DO-based survey method that allows accurate abundance estimations with reduced effort would increase the applicability of the method. This would have positive effects on the conservation of species which are challenging to survey such as mountain ungulates.
We used computer simulations based on real data and a field test to assess the reliability of the Double Observer (DO) and of a new proposed survey procedure, the Double Observer Adjusted Survey (DOAS). DOAS is based on total block counts adjusted with some DO surveys conducted in a proportion of the total area only. Such DO surveys are then used to estimate detection probability with a mark-recapture derived approach.
We found that full DO is much more accurate than simple block counts for abundance estimations. DOAS is a less demanding alternative to full DO and can produce comparable abundance estimates, at the cost of a slightly lower precision. However, in the DOAS overall detectability has to be estimated within a sufficient number of sites (around a quarter of the total) to obtain a higher precision and avoid large overestimations.
Practical implications. Double Observer methods could increase the reliability of abundance estimations in mountain ungulates and other gregarious species. Full DO in particular could allow researchers to obtain unbiased estimations with high precision and its usage is therefore suggested instead of block counts in wildlife monitoring. Given the high costs of full DO, the DOAS procedure could be a viable and cost-effective survey strategy to improve abundance estimates when resources are scarce.
README: A new Double Observer based census framework to improve abundance estimations in mountain ungulates and other gregarious species with a reduced effort
https://doi.org/10.5061/dryad.zkh1893ks
Description of the data and file structure
We provide the R script to perform computer simulations (DOAS script.R) on the reliability of the 3 census methods testes: block counts, full DO and our proposed DOAS procedure.
The "DOAS_maindata.csv" is a representative subset of the final dataset obtained running the provided script. Each row is a single simulation, the columns are (see paper for details):
- det.p = simulated detection probability
- pop.size= simulated population size
- mean.group = average simulated group size
- size.effect= multiplying factor for the detectability of a single group compared to a large group
- miscount= miscount effect (average group abundance counted compared to the real one)
- surv.sites= number of sites in which DO is performed for the DOAS
- obs.effect= observer effect (det.p of the second observer compared to the first)
- detp.var= coefficient of variation of detection probability across sites
- totalDO.groupdetect= detectability estimated with the total DO
- totalDO= ratio of individuals estimated with total DO
- census= ratio of individuals estimated with block counts
- groupadjust.censusSR= ratio of individuals estimated with DOAS
- group.detectionSR= detectability estimated with DOAS
Code/software
DOASscript.R is the R script to obtain all the data in DOAS_maindata.csv
Methods
Main data were obtained through computer simulations, with the script provided "DOAS script.R"
With the provided dataset we aimed to test the robustness of the full DO and our proposed DOAS (see related paper for more info), in different field conditions. The DOAS method is based on Double Observer surveys being performed in only a portion of the total target area to estimate the detection probability, that is then used to adjust the counts obtained with total block counts conducted in the whole area.
To test DO and DOAS methods we carried out simulated surveys on populations randomly built under real-case parameters, using as a case study the population of Alpine ibex (Capra ibex) in Gran Paradiso National Park (GPNP, Italy). This simulation results in the provided "MAIN data). Using these simulated surveys, we tested whether DO conducted in the entire area (full DO) and DOAS are able to provide reliable abundance estimates, compared to block counts, and the relative costs of the two DO methods. Moreover, we analyzed factors that can influence the proportion of individuals estimated with full DO and DOAS, including abundance, group size and detection probability