The importance of quantifying selection climates: Predictable and unpredictable variation in predation risk and the implications for prey responses
Data files
Jun 20, 2025 version files 158.60 KB
-
Code.R
15.50 KB
-
Oviposition_responses.csv
105.69 KB
-
Predation_risk.csv
34.38 KB
-
README.md
3.04 KB
Abstract
Prey behavioural traits within a population are commonly variable, unexpectedly so, given that predation is thought to be a strong selection pressure. This variation likely arises from complex, variable selection pressures, but experimental evaluations of prey responses to natural variability in selection pressures remain limited. We focus on the rock-pool breeding mosquito Aedes vexans, which can influence the predation risk to its offspring through site selection for oviposition. We quantified the nature of selection pressures in the wild, i.e., the spatial and temporal variation in larval predation risk, by measuring densities of predatory dragonfly nymphs in rock pools along the mosquito breeding season. To examine the implication of selection pressure regimes for the evolution of oviposition site selection, we conducted manipulative experiments and measured female oviposition responses to variation in predation risk. Predation risk varied extensively over space and time; this variation showed both predictable and unpredictable elements. Females avoided large pools with permanent predators but appear to show variable responses to medium-sized pools with unpredictable predation risk. We suggest that while it might be challenging to quantify complex selection environments and their impact on wild prey responses, more such studies can help explain puzzling variation in anti-predator responses.
Dataset DOI: 10.5061/dryad.h18931zz2
Description of the data and file structure
Oviposition responses were measured only from experimental pools. Oviposition responses were counted as the number of eggs laid in the pools on the ovistrips placed along the walls of the pools. Pools were either given a treatment of control or predator. The experiment was done during the mosquito breeding season from July to December.
Predation risk was measured across pools along the mosquito breeding season. For sampling predators (nymphs of Bradinopyga geminata), we removed them from the pools and placed them in a container. The nymphs were counted and then added back to the pools. Details can be found in the ESM. Please note that the empty cells in the predator column indicate that the sampling could not be conducted as the pool was dried. These are treated as "NA" values.
Files and variables
File: Oviposition_responses.csv
Description:
Variables
- pool_ID: Identity of the pool - modelled as a random effect
- date: date of egg collection
- session: each session represented a block, which consisted of multiple days during which eggs were collected
- predator: number of predators added to the pool
- perimeter_cm: perimeter of the pool calculated using ImageJ
- ovistrip_ID: identity of the ovistrip, as pools were lined with multiple ovistrips covering half the perimeter
- eggcount: number of eggs laid by Aedes vexans on each ovistrip, i.e., main response variable
- egg_sum: sum of eggs from ovistrips in the same pool
- season: categorical variable with two levels - pre-diapause and diapause season
- background_predation: categorical variable with two levels - low and high
- egg_den: egg_sum divided by total length of ovistrips, i.e., half of the pool perimeter
File: Predation_risk.csv
Description:
Variables
- pool_id: Identity of pools
- depth_inch: pool depth measured by dipping a scale in the pool
- sa_inch: pool surface area calculated using ImageJ
- peri_inch: perimeter of the pool calculated using Image J
- vol_inch: pool volume estimated by multiplying depth_inch and sa_inch - derived column from depth and sa
- peri_cm: perimeter of the pool converted to cm from inches
- month: month of sampling
- date: date of sampling
- predator: number of predators
Code/software
All analyses were done in R version 4.2.2. The Code.R file has all the supporting code for statistical modelling of predation risk and oviposition responses. This file also has the code to generate all the figures in the main manuscript. The code file is split into three sections - section 1 is on modelling adult behavioural responses using the Oviposition*_*responses.csv file; section 2 is on modelling the predation risk using the Predation_risk.csv file; and section 3 is code for making all the figures.