Testing the sentinel method: live and artificial prey display contrasting patterns of predation across an urban gradient
Data files
Dec 15, 2025 version files 15.79 KB
-
README.md
6.20 KB
-
Yu_et_al._2025._Ecol_Evol.csv
9.59 KB
Abstract
Assessing changes in the intensity of biotic interactions across environmental gradients is a central issue in ecology. The sentinel method has been widely adopted to study predator-prey interactions by establishing patches of prey under different conditions that predators can attack. Sentinels, proxies for prey, are frequently worm-shaped prey resembling caterpillars and are specifically used to assess predation by arthropod-feeding predators, with predation measured as the rate of disappearance or evidence of predation after a certain period of exposure. While it has been suggested that artificial sentinel prey might produce divergent results to live prey, previous studies showed mixed results in the difference between these two prey types. Results are likely to vary with context, and the assessment of different prey types along urban gradients is still lacking. Here, we performed an experiment at ten sites across a natural-to-urban gradient in Suzhou (East China) combining live prey and artificial prey to determine differences in predation intensity between these prey types. We released 2,575 artificial prey and 3,825 live prey, either separately (artificial or live prey alone) or combined, in a randomized sequence. We found a positive relationship between our index of predation and the level of urbanization using both types of prey. However, predation rate using artificial prey was lower than with live prey and showed a different pattern with urbanization. The predation rate using live prey was higher for avian predators and lower for insect predators with increasing urbanization rate. Our results suggest that artificial and live prey produce strongly divergent estimates of predation intensity. Thus, while artificial prey might be used as a rapid-screening tool, live prey should be favored in comprehensive studies to assess this fundamental ecosystem service.
Dataset DOI: 10.5061/dryad.t76hdr8dz
Description of the data and file structure
This dataset contains the raw observational data from a field experiment designed to compare predation intensity using two methods—live prey and artificial plasticine prey—across a natural-to-urban gradient in Suzhou, Jiangsu Province, East China.
The study tested the hypothesis that artificial sentinel prey (plasticine caterpillars) produce divergent estimates of predation intensity and patterns compared to live prey. Data was collected at ten primary sites, with some sites containing multiple sub-location replicates, resulting in a total of 16 location IDs. Predation events were identified and attributed to either avian or insect predators based on distinct attack marks (e.g., beak pecks vs. mandible marks).
Methodology Summary
- Sites: Ten sites were selected along a gradient from natural/wetland to urban park in Suzhou.
- Prey: Two prey types were used: (1) Live prey (mealworms), and (2) Artificial prey (plasticine models shaped like caterpillars).
- Design: Prey were deployed in a randomized sequence, both separately and in combination, on vegetation.
- Census: After retrieval, each prey item was examined. A predation event was recorded if clear attack marks were present. Marks were classified as from avian predators or invertebrate/insect predators.
- Urbanization Metric: The
UrbanizationRatewas derived from remote sensing data (e.g., the proportion of impervious surface within a buffer around the site)
Files and variables
File: Yu_et_al._2025._Ecol_Evol.csv
Description: This dataset contains the raw observational data from a field experiment designed to compare predation intensity using two methods—live prey and artificial plasticine prey—across a natural-to-urban gradient in Suzhou, Jiangsu Province, East China.
Variables
| Column Name | Data Type | Description |
|---|---|---|
| Site | String | Unique identifier for the study location (e.g., DushuhuPark, YangjiaVillage). Sites with numbers (e.g., DushuhuPark1) denote sub-location replicates within a primary site. |
| UrbanizationRate | Numeric (Decimal) | A continuous index quantifying the level of urbanization based on impervious surface area for the location. Higher values indicate more urbanized environments. |
| DayN | Integer | The experimental day or trial sequence number (1 to 4) within the location. |
| PredationRate | Numeric (Decimal) | The proportion of prey attacked. Calculated as P / Total. Range: 0 to 1. |
| PreyType | String | Type of prey deployed: Live prey or Plasticine prey. |
| PredatorType | String | Broad category of predator based on attack marks: Avian predators or Insect predators. |
| P | Integer | Count of prey items that were attacked/predated upon (successes). |
| N | Integer | Count of prey items that were not attacked (failures). |
| Total | Integer | Total number of prey items deployed in that experimental unit (P + N). |
Code/software
All the analyses were carried out in R v 4.2.3 (R Core Team, 2023), using the packages lme4 (Douglas et al., 2015), glmmTMB (Mollie et al., 2017), MuMin v 1.47.5 (Bartoń, 2023), car v 3.1.2 (Fox & Weisberg, 2019), DHARMa v 0.4.6 (Hartig, 2022) and the figures were produced using ggplot2 (Wickham, 2015).
Bartoń, K. (2023). MuMIn: Multi-Model Inference (R package version 1.47.5). https://CRAN.R-project.org/package=MuMIn
Douglas, B., Martin, M., Ben, B., & Steve, W. (2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software, 67(1), 1–48. https://doi.org/10.18637/jss.v067.i01
Fox, J., & Weisberg, S. (2019). An R Companion to Applied Regression (3rd ed.). Sage Publications. https://socialsciences.mcmaster.ca/jfox/Books/Companion/
Hartig, F. (2022). DHARMa: Residual Diagnostics for Hierarchical (Multi-Level / Mixed) Regression Models (R package version 0.4.6). https://CRAN.R-project.org/package=DHARMa
Mollie, E. B., Kasper, K., Koen, J. van B., Arni, M., Casper, W. B., Anders, N., Hans, J. S., Martin, M., & Benjamin, M. B. (2017). glmmTMB balances speed and flexibility among packages for zero-inflated generalized linear mixed modeling. The R Journal, 9(2), 378–400. https://doi.org/10.32614/RJ-2017-066
Wickham, H. (2016). ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York.
Access information
Other publicly accessible locations of the data:
- N/A
Data was derived from the following sources:
- N/A