Data from: Evaluating biotic and abiotic drivers of avian community mobbing responses along urban gradients in Southern California
Data files
Dec 03, 2024 version files 942.24 KB
-
Ewing_mobbing_study_Rcode.R
21.68 KB
-
owl_dataset_full.csv
913.17 KB
-
README.md
7.40 KB
Abstract
Urbanization is a significant pressure affecting wildlife, and has the potential to greatly alter behavioral responses in animal communities. A behavioral response that is potentially affected by urbanization is the mobbing of predators by potential avian prey species. We explored the effect of various abiotic and biotic factors in influencing avian mobbing responses along an urban-rural gradient. We conducted predator simulations by using playback of the vocalizations of the Western Screech-owl, Megascops kennicottii, which is a predatory species that elicits a mobbing response from other birds. These vocalizations, accompanied by stuffed models of the screech-owls, were broadcast at a variety of points along an urban-rural gradient in Los Angeles and Orange Counties in southern California. We used an experimental approach using playback, i.e., vocalization and models, to investigate whether mobbing responses of birds change in areas where predators may be naturally present (high vegetation density) or absent (high impervious cover). We recorded the number of individual birds and species that exhibited mobbing behavior at experimental sites, as well as various biotic and abiotic factors that may influence avian mobbing, including noise level, impervious surface cover, avian community turnover across the urban-to-rural gradient, and the structure of local vegetation, which we assumed may be important for either hosting roosting screech-owls or providing cover for mobbing bird species. For both the number of mobbing individuals and species, we showed that mobbing responses decreased with increasing noise levels and percentage of impervious surfaces and increased with increasing woody vegetation. There was some evidence that predator presence influenced mobbing responses. Our results show that the changes associated with urbanization can significantly alter anti-predator behavior in birds, and that these changes can alter avian social eavesdropping networks.
README: Evaluating biotic and abiotic drivers of avian community mobbing responses along urban gradients in Southern California
https://doi.org/10.5061/dryad.sf7m0cgd7
The following repository is associated with Ewing, Benjamin, Wood, Eric M. and Martinez, Ari. “*Evaluating biotic and abiotic drivers of avian community mobbing responses along urban gradients in Southern California.” *Ecosphere. In press. It contains a csv file containing the raw data that was collected during the mobbing trials conducted for my Master's thesis at CSULB, as well as a R document containing the code used to generate the models, predictions, and graphs for the manuscript.
Data was collected during point counts at various sites throughout Los Angeles and Orange Counties, California, United States. I recorded the all the birds seen or heard within 100m of the point for 10 minutes. This was followed by 5 minutes of playback of a vocalization, either of a predator (Western Screech-owl) or harmless control species (Mourning Dove). Following playback I continued to record species detected for 3 minutes. Once playback started I started over with counting species and individuals. I recorded any mobbing behavior observed during this time. Following the trials variables such as average noise and temperature were collected. Vegetation data and impervious cover data were collected for each point later on.
File Descriptions:
Ewing_mobbing_study_Rcode.R
This R document contains the code used in my analyses. We subset the data and ran a series of General Linearized Mixed Models to test the hypotheses in our manuscript. Following model selection we generated model predictions for the top models, and graphed those predictions. It also contains the code for testing correlations between our variables.
owl_dataset_full.csv
This csv file contains the data that we collected during the course of our experiments. The columns in the dataset are described in detail below. Not all columns were used in the data analysis for this particular manuscript, though all are described below.
- Date: The date of the trial
- Season: Whether the trial took place in the breeding season (SUMMER; April-June) or nonbreeding season (WINTER; October-February)
- Observers: Initials of observers (BJE-Benjamin James Ewing; YD-Ynez Diaz)
- Trial: A number and letter corresponding to which round of playback the trial was a part of
- Model_ID: Each owl or dove model had a unique ID. These were randomly assigned to each predator or control trial, as appropriate.
- Audio_ID: Each of our recordings had a unique ID. These were randomly assigned to the appropriate type of trial as well.
- Playback: Whether the trial was an owl recording or a dove (control) recording.
- Speaker_Dir: Which of the 4 cardinal directions the speaker was oriented towards (North, South, East or West). This was randomly assigned for each trial. We added this component shortly after beginning data collection, hence the NAs in the first few trials.
- Start: The time the trial began, in military time.
- End: The time the trial ended, in military time.
- Time: Whether the trial took place in the morning (AM) or evening (PM).
- Temp: The temperature in degrees Celsius. Some NAs when recording equipment malfunctioned.
- Humidity: Percent humidity at point. Some NAs when recording equipment malfunctioned.
- Cloud: Estimated percentage of cloud cover at the time of the trial.
- Noise0s-Noise50s: we recorded the noise in decibels (dB) at 10 second intervals for one minute. These values are recorded in these six columns. This method of calculating average noise was introduced a few trials in; hence the NAs during the first few.
- NoiseAvg: The average of the preceding six columns.
- NoiseMax: The highest noise measurement in decibels (dB). This was recorded by our noise meter.
- Transect: Which of our four transects the point belonged to (Pasadena, Orange County (OC), Palos Verdes, or Crystal Cove).
- Gradient: Whether the point was classified as urban, semiurban, or rural. This classification was replaced with the more precise percent impervious cover calculation in the final analyses.
- Tallest Tree: The tallest tree (in meters) within 20 meters of the point. Measured using a clinometer.
- Percent_Woody: The percentage of woody vegetation within 20m of the point. Percent woody cover was estimated for each of the four quadrants within the 20m radius. These were then averaged to get the value shown here.
- PredPres: Whether the focal predator (Western Screech-owl) was present (1) or absent (0) at the location. This was determined via data from ebird.org. Due to the heavy usage of ebird in the greater Los Angeles area, it was considered a fairly accurate estimate of where bird species are found. In cases of uncertainty, personal surveys were conducted to further establish presence/absence.
- Recording: A recording was taken during each trial. This column records the file name.
- Point: The name of the point. Names describe location (Pas-Pasadena; OC-Orange County; PV-Palos Verdes; CC-Crystal Cove), gradient type (R-rural; S-semiurban; U-urban). Each combination had 5 points, which correspond to the numbers 1-5 in the names.
- Bray_Curtis: The Bray-Curtis community similarity score for the point. These were calculated using the vegan package in R, with the rural site with the highest species count used as the reference point. Bray-Curtis scores were calculated for each site in both seasons.
- Percent_Impervious: The percent impervious cover within 2km of each point. Calculated from LARIAC data for points in Los Angeles County, and NLCD data for Orange County.
- Comment: Any comments on that particular trial.
The following columns describe the birds detected and any mobbing behavior displayed.
- Start_Time: The time that the bird was first detected
- Trial_Stage: Whether the bird was detected prior to (PRE) or after (POST) playback began
- Species: The four-letter code for the bird species. UNKN is an unknown species. Codes reflect the names of species as of 2023.
- Distance: Estimate of how far away the bird was, in meters.
- Detection: Whether to bird was seen (V) or heard only (A).
- Individual: How many individuals were observed.
Missing data: NA
*Mobbing Variables; many of these were created for subsetting purposes. *
* Mob_Order: If the bird mobbed, what order did it respond in. 1=first to respond; 0=did not mob.
* Mob_Ind: If the bird(s) mobbed, how many individuals were there.
* Mob_Bird: Whether the bird(s) mobbed (1) or did not mob (0).
* Mob_Species: If the bird(s) mobbed, what species were they. If mobbed, it displays the 4-letter code for the species. If they didn't; it displays 0.
* Mob_Trial: Whether birds mobbed during the overall trial (1) or did not (0).
* Mob_Approach: Whether the bird(s) approached the owl (or dove) model (1) or did not (0). Used to determine if mobbing occurred.
* Mob_Hop: Whether the bird(s) were hopping around in the vicinity of the model (1) or not (0). Used to determine if mobbing occurred.
* Mob_Call: Whether the birds(s) gave calls in the vicinity of the model (1) or not (0). Used to determine if mobbing occurred.
* Notes: Any notes on the point count and mobbing trial portion of the study.