Data and code from: Goby gummies: a customizable and edible assay to quantify predation in aquatic ecosystems
Data files
Apr 02, 2026 version files 86.40 KB
-
Goby_Gummy_analyses.zip
69.02 KB
-
Methods_assay_survivor.csv
5.54 KB
-
README.md
8.44 KB
-
Squidpop_assay_survivor.csv
3.40 KB
Abstract
Predation is a critical ecosystem process that shapes the structure and functioning of biological communities. However, due to its intermittent nature, fast pace, and general unpredictability, predation is difficult to observe and quantify. Therefore, we commonly rely on indirect metrics or proxies of predation, which reflect the outcome of predation events but do not allow for inference about the predator’s decision-making process or predation rates. In terrestrial ecosystems, lifelike prey replicas have allowed ecologists to gain a broad understanding of predator choice, predation intensity, and their drivers. Yet in aquatic ecosystems, few scalable, interactive predation assays have been developed.
We introduce Goby Gummies, a customizable, edible prey model developed for aquatic ecology. Gummies are constructed using an inert, edible medium that can be cast into any desired shape (in our case, a goby fish), dyed various colors, and be supplemented with edible material to introduce variation in nutritional profiles. As such, Goby Gummies are a cheap, sustainable, high-throughput assay that can be tailored to a variety of aquatic ecosystems.
We performed two pilot studies to test Goby Gummies in a natural setting on coral reefs in Belize during which gummies were reliably consumed by a range of predatory fishes. First, we show that gummies supplemented with fish meal were preferred by predators over agar-only gummies, but the strength of this preference was dependent on their coloration, suggesting an intriguing interplay between external appearance and internal composition. Second, we compared fish-supplemented gummies to squidpops, a previously developed predation assay for marine systems. Goby gummies were consumed first more frequently and eaten at quicker rates than Squidpops and consistently attracted carnivorous predators whereas squidpops were frequently consumed by herbivorous parrotfishes.
Our results highlight that Goby Gummies provide a new predation assay tool in aquatic ecosystems that permits the exploration of many exciting questions surrounding prey and predator traits and their interplay. We envision goby gummies to kindle a diverse range of impactful studies across disciplines that mirror those conducted in terrestrial ecosystems.
Dataset DOI: 10.5061/dryad.xd2547dxb
Description of the data and file structure
Included are two datasets and a zipped folder containing the R project and code in which the data is analyzed. Both datasets (.csv) contain the outcomes from manually scoring videos of two separate experiments. The first experiment (data: Methods_assay_survivor.csv) compared an agar-only goby gummy to a gummy that was supplemented with dried Caranx ruber fish meal; creating a nutritional treament. Four gummy colors were tested (black, white, red, yellow), although the two gummies presented in a single assay were always the same color. In the second experiment (data: Squidpop_assay_survivor.csv), a black, C. ruber supplemented goby gummy was pitted against a squidpop: a circular piece of dried, squid mantle.
Files and variables
File: Methods_assay_survivor.csv
Description: dataset containing the scored results from assays comparing agar-only gummies to fish-supplemented gummies. The gummies in each assay were the same color, but the colors varied between assays.
Variables
- trial: the unique identifer for each trial
- model_eaten: indicates which prey model from the video was eaten as viewed from the position of the GoPro. left = left gummy, right = right gummy, neither = neither gummy was consumed in the 15 minute deployment period. discard = that video was "thrown out" for a variety of reasons including but not limited to: the gummy ripping apart on descent, the assay board flipping over during the deployment period, etc.
- color_check: the color of the gummies compared in a given assay. Both gummies are this color.
- type: the type of gummy that was first consumed in that assay. ruber indicates the fish-supplemented gummy (containing C. ruber fish meal) was consumed first. agar indicates that the agar-only gummy was consumed first. NA (not applicable) indicates neither gummy was consumed during the deployment period.
- predator_sp: the predator species that consumed the first gummy. NA (not applicable) indicates there was no predator as neither gummy was eaten during the deployment period.
- deploy_start: the time (in minutes:seconds) within the video recording at which the deployment begins. Please note this is not the start of the video, as the recording began while the assays were on the boat which then had to be swam out and placed in location by snorkelers. The deploy_start is when the assay had been placed on the reef and was stationary after the snorkeler swam away.
- time_eaten_agar: the time (in minutes:seconds) within the video that the agar-only gummy was eaten by a predator. If the time is 15:00, this indicates the gummy was not eaten during this trial. The time was formatted in this manner to satisfy requirements for the survival curve analyses.
- time_eaten_ruber: the time (in minutes:seconds) within the video that the fish-supplemented gummy was eaten by a predator. If the time is 15:00, this indicates the gummy was not eaten during this trial. The time was formatted in this manner to satisfy requirements for the survival curve analyses.
- time_eaten: the time (in minutes:seconds) which the first gummy was eaten, regardless of composition. Does not discriminate between what type of gummy it was. NA (not applicable) indicates neither gummy was consumed during the deployment period.
File: Squidpop_assay_survivor.csv
Description: dataset containing the scored results from assays comparing fish-supplemented gummies to squidpops. The gummy in each assay were always black and contained dried material from C. ruber.
Variables
- trial: the unique identifer for each trial
- model_eaten: indicates which model (in this case, gummy or squidpop) from the video was eaten as viewed from the position of the GoPro. left = left gummy/squidpop, right = right gummy/squidpop, neither = neither gummy nor squidpop was consumed in the 15 minute deployment period.
- type: the type of model that was first consumed in that assay. gummy indicates the fish-supplemented gummy (containing C. ruber fish meal) was consumed first. squid indicates the squidpop was consumed first. NA (not applicable) indicates neither gummy was consumed during the deployment period.
- predator_gummy: the predator species that consumed only the gummy (not inclusive of the squidpop). NA (not applicable) indicates there was no predator that consumed the gummy during the deployment period.
- deploy_start: the time (in minutes:seconds) within the video recording at which the deployment begins. Please note this is not the start of the video, as the recording began while the assays were on the boat which then had to be swam out and placed in location by snorkelers. The deploy_start is when the assay had been placed on the reef and was stationary after the snorkeler swam away.
- time_eaten_gummy: the time (in minutes:seconds) within the video that the fish-supplemented gummy was eaten by a predator. If the time is 15:00, this indicates the gummy was not eaten during this trial. The time was formatted in this manner to satisfy requirements for the survival curve analyses.
- predator_squid: the predator species that consumed only the squidpop (not inclusive of the gummy). NA (not applicable) indicates there was no predator that consumed the squidpop during the deployment period.
- time_eaten_squid: the time (in minutes:seconds) within the video that the squidpop was eaten by a predator. If the time is 15:00, this indicates the squidpop was not eaten during this trial. The time was formatted in this manner to satisfy requirements for the survival curve analyses.
File: Goby_Gummy_analyses.zip
Description: a zipped folder containing the R project and sub-folders needed to reproduce the figures and analyses used in the publication. There are three sub folders. data contains the two previously described .csv files, figures is an empty folder where pdfs of plots can be saved to from R, and scripts contains the five scripts used to run the analyses.
Components
- data (folder): folder with two csv files which contain the data
- Methods_assay_survivor.csv
- Squidpop_assay_survivor.csv
- figures (folder): empty folder in which plots from R can be generated and saved to. Without this folder, plots created from the R scripts will not save.
- Goby_Gummy_analyses.Rproj (R project file): a configuration file used by R Studio to define a self-contained working environment for this specific set of analyses.
- scripts (folder): folder containing five scripts used for data handling, plotting, and analysis. These scripts are as follows:
- 01_selection_probability_MEE.R (R script): compares the selection preference for agar vs fish-supplemented gummies across the four colors
- 02_survivor_analysis_MEE.R (R script): modeling and comparing survival curves of agar vs fish-supplemented gummies.
- 03_predator_breakdown_MEE.R (R script): summarizing the distribution of gummy predators from the methods assay (data analyzed in the two, previous scripts)
- 04_predator_breakdown_SP_comparison_MEE.R (R script): comparing the distribution of predators for both fish-supplemented gummies and squidpops
- 05_survivor_analysis_SP_MEE.R (R script): modeling and comparing survival curves of fish-supplemented gummies vs squidpops.
Code/software
R version 4.5.0
R packages needed are:
- dplyr ver 1.2.0
- brms ver 2.23.0
- modelr ver 0.1.11
- ggplot2 ver 4.0.2
- tidybayes ver 3.0.7
- patchwork ver 1.3.2
- survival ver 3.8-3
- lubridate ver 1.9.4
- ggsurvfit ver 1.2.0
- gtsummary ver 2.4.0
- viridis ver 0.6.5
- forcats ver 1.0.1
R is required to open 01_selection_probability_MEE.R, 02_survivor_analysis_MEE.R, 03_predator_breakdown_MEE.R, 04_predator_breakdown_SP_comparison_MEE.R, and 05_survivor_analysis_SP_MEE.R
R Studio is required to open Goby_Gummy_analyses.Rproj
Microsoft Excel can be used to view Methods_assay_survivor.csv and Squidpop_assay_survivor.csv
We conducted two field experiments to test the performance of Goby Gummies, presenting binary choice assays to reef fishes in situ on Belizean coral reef habitats. The first experiment entailed deploying gummies to assess the frequency at which gummies were consumed and how nutritional supplement (i.e., adding ground fish meal) may alter predator choice. The second test involved comparing Goby Gummies to Squidpops, a previously established standardized assay that was designed to assess ecosystem-wide predation rates. All aspects of this study were conducted in accordance with the University of Texas Institutional Animal Care and Use Committee (AUP-2024-00071) and under approved field research permits (Belizean Fisheries Department permit number: 0040-25).
For the first experiment, we deployed a total of 86 binary choice assays that offered an agar-only gummy and a gummy that had been supplemented with 0.25g of dried whole-body fish meal obtained from the bar jack (Caranx ruber). This style of assay was chosen to deliberately present resident predators with one of two choices, yielding Bernoulli trial data that can be robustly analyzed using generalized linear models. Gummies were dyed one of four different colors (black, white, red, and yellow), but only gummies of the same color were compared experimentally in one assay (e.g., a red agar-only gummy to a red C. ruber-containing gummy, as described previously). Two gummies were fastened to a 300 mm × 300 mm clear acrylic assay board containing a GoPro that was deployed by a snorkeler in coral reef habitat ranging from 2 – 4 m depth. 6-12 assays were deployed simultaneously in batches with each assay board spaced a minimum of 10 m apart to minimize the chance of predators encountering multiple assays. The assays were left undisturbed for 15 min before being recollected, which yielded ample time for predators to consume the gummies while providing enough buffer to accommodate deployment disturbance from the snorkelers. The videos were then watched and scored for predator preference, which was determined by the first prey gummy to have 25% of its mass removed through a clear biting motion. We chose this threshold as it more accurately reflects the size-based constraints on fish predation, which demand that predators are generally able to engulf or capture prey that fits the horizontal gape of their mouths in one feeding motion. This threshold has two important consequences: first, if a gummy was mouthed but did not have a substantial portion of its material removed, it was not scored as eaten. Second, if smaller, non-predatory fishes only take a small “nibble” on the prey to obtain nutrition, it is not scored as predation (thus reflecting their general inability to consume the prey in real life). Last, Bayesian regression models using a Bernoulli distribution were then used to assess predator preference for selecting fish-supplemented gummies (coded as a 1 for the response variable) against agar-only gummies (coded as a 0 for the response variable)
We performed a second field deployment in which Goby Gummies were compared to Squidpops: an established predation assay protocol in which a circular disk of dried squid mantle (13mm in diameter) is deployed in batches of 25 to gauge ambient predation pressure. For the second field experiment, we compared fish-supplemented Goby Gummies to Squidpops using the previously described binary-choice assay approach to determine: 1) which assay type was preferred, 2) the frequency of consumption during a 15-minute period, and 3) if different predatory species targeted the respective assays. We deployed 48 assays in coral reef habitat (depth: ~2m), in which a black, C. ruber supplemented Goby Gummy was pitted against a Squidpop. The orientation of Gummy and Squidpop were reversed for every other assay (i.e., on which side of the board the item was placed) to remove any location bias. Identical to the previous experiment, videos were watched and scored using the same 25% volumetric consumption threshold.