Marine food webs are structured through a combination of top-down and bottom-up processes. In coral reef ecosystems, fish size is related to life-history characteristics and size-based indicators can represent the distribution and flow of energy through the food web. Thus, size spectra can be a useful tool for investigating the impacts of both fishing and habitat condition on the health and productivity of coral reef fisheries. In addition, coral reef fisheries are often data-limited and size spectra analysis can be a relatively cost-effective and simple method for assessing fish populations. Abundance size spectra are widely used and quantify the relationship between organism size and relative abundance. Previous studies that have investigated the impacts of fishing and habitat condition together on the size distribution of coral reef fishes, however, have aggregated all fishes regardless of taxonomic identity. This leads to a poor understanding of how fishes with different feeding strategies, body size-abundance relationships, or catchability might be influenced by top-down and bottom-up drivers. To address this gap, we quantified size spectra slopes of carnivorous and herbivorous coral reef fishes across three regions of Indonesia representing a gradient in fishing pressure and habitat conditions. We show that fishing pressure was the dominant driver of size spectra slopes such that they became steeper as fishing pressure increased, which was due to the removal of large-bodied fishes. When considering fish functional groups separately, however, carnivore size spectra slopes were more heavily impacted by fishing than herbivores. Also, structural complexity, which can mediate predator-prey interactions and provisioning of resources, was a relatively important driver of herbivore size spectra slopes such that slopes were shallower in more complex habitats. Our results show that size spectra slopes can be used as indicators of fishing pressure on coral reef fishes, but aggregating fish regardless of trophic identity or functional role overlooks differential impacts of fishing pressure and habitat condition on carnivore and herbivore size distributions.

README for the paper:

Fishing and habitat condition differentially affect size spectra slopes of coral reef fishes

Paul G. Carvalho*, Fakhrizal Setiawan, Karizma Fahlevy, Beginer Subhan, Hawis Madduppa, Guangyu Zhu, Austin T. Humphries. Ecological Applications. 2021.

*Corresponding author: 9 East Alumni Ave, University of Rhode Island, Kingston, RI 02881, USA; paulcarvalho@uri.edu

size_spectra_analysis.r (last updated 01/20/2021) contains code developed in R version 4.0.3 that completely replicates the analyses conducted for the paper. The file “functions.r” is loaded in this script and required for the code to run properly (see description of this file below). In addition, three GitHub packages are required for these analyses and are loaded on lines 20-22. This code quantifies size spectra slopes, runs Generalized Additive Models, and conducts multi-model inference. Code to create plots in the main paper and supplemental information are also included in this file.

functions.r (last updated 01/20/2021) contains code developed in R version 4.0.3. The statistical and plotting functions defined in this file are used in “size_spectra_analysis.r”. Many of the functions were adopted from https://github.com/andrew-edwards/fitting-size-spectra. The statistical functions calculate the log-likelihood for fitting size spectrum slope and calculate 95% CI. The plotting functions trophic.plots, slope_regAndFunc, and slope_reg calculate size spectrum slopes based on trophic role (carnivore vs. herbivore) and region of Indonesia where data were collected (Raja Ampat, Wakatobi, and Lombok). The primary purpose of these functions was to condense the code in “size_spectra_analysis.r”.

fish_size_spectra_data.csv contains underwater visual census data collected in Indonesia (regions: Raja Ampat, Wakatobi, and Lombok) from January 2018 to July 2018. Data include site names, transect number, species identifications, estimated size (cm), functional groups, trophic roles (i.e., carnivore, herbivore, and other), and parameter values for converting length (cm) to mass (g and kg).

drivers_size_spectra_data.csv contains site and habitat data (collected from January 2018 to July 2018) that were used in the Generalized Additive Models to explain site differences in size spectra slopes. Data include site names, site coordinates, site population gravity values (a metric of human population density), hard coral cover (percentage), algae cover (percentage), and structural complexity (see methods section for details on the classification).

https://github.com/andrew-edwards/fitting-size-spectra contains the original code for fitting size spectra slopes.

https://github.com/m-clark/visibly contains code for color palettes and other plotting aesthetics in R.