Table of Contents

• Coverage
• Dataset Description
  • Methods & Sampling
  • Data Processing Description
  • BCO-DMO Processing Description
• Data Files
• Related Publications
• Parameters
• Instruments
• Project Information
• Funding

All references to figures and tables are from Yeager, M. E. and A.R. Hughes. 2025. Functional trait analysis reveals the hidden stability of multitrophic communities. Ecology. In press.

Study system
This study was conducted along the southern shore of Rhode Island across six coastal ponds: Green Hill (GH) Pond, Ninigret Pond (NP), Point Judith (PJ) Pond, Potter Pond (PP), Quonochontaug Pond (QP), and Winnapaug Pond (WP; Appendix S1: Fig. S1 of Yeager and Hughes, 2025). Fish communities in these ponds mostly consist of marine species connected to the ocean via a breachway that reassemble each year via larval dispersal from the ocean (Satchwill and Sisson 1991a, 1991b, Sisson and Satchwill 1991). The fish communities found throughout these ponds encompass a range of trophic levels (2 - 4.5) and feeding modes (Appendix S6: Table S1, Fig S1). Here, we examined 13 fish functional traits which can be categorized into three functional roles: (1) energy acquisition, (2) locomotion, and (3) nutrient recycling (Villéger et al. 2017). Fish communities are a good model system to ask these types of questions due to a strong foundation in the functional trait literature, offering a thorough understanding of the functional roles fish provide and functionally meaningful traits to measure (Dumay et al. 2004, Mason et al. 2007, Villéger et al. 2010, 2017, Albouy et al. 2011, Stuart-Smith et al. 2013, Mouillot et al. 2013, Yeager et al. 2017, McLean et al. 2019).

Fish community collections
From June to October 2018, we sampled six coastal pond fish communities monthly via 150-foot beach seine in conjunction with the Rhode Island Department of Environmental Management (RIDEM) fish and macroinvertebrate survey. This survey has been ongoing since 2010, with fish and macroinvertebrates identified to the lowest taxonomic level, enumerated and a subset measured to infer population size structure. Fish collections in 2018 for this study were collected via the same methods as the RIDEM survey. The fish in these communities span trophic level ranges from to 2.1 to 4.5, consisting of detritivores, invertivores, and piscivores. We targeted 38 species that account for 99.4% of total abundance across the survey. For each species, we aimed to collect 20 individuals evenly distributed across their size range, informed from past survey data. Upon collection, fish were either transferred into seawater containers for excretion incubations or euthanized immediately via a seawater-clove oil (Eugenol extract, Syzygium aromaticum) mixture (IACUC protocol #: 18-0622R). Once euthanized, fish were held on ice before returning to the lab to conduct morphometric analysis. We collected a total of 708 fish across 27 species and 23 families. We analyzed a subset of 200 fish for nutrient recycling traits, resulting in an average of 18.63 + 2.6 fish per species for energy acquisition and locomotion traits and an average of 7.48 + 0.91 fish per family for nutrient recycling traits.

Nutrient recycling traits
To quantify nutrient recycling traits, we conducted excretion incubations, targeting 27 fish families (N = 1-3 species per family), which account for 99.7% of total abundance across the survey. For each family, we targeted 10 individuals evenly distributed across their size range. Directly after fish were removed from the seine net, individuals were placed into separate 3-liter (L) sterile plastic bags of seawater directly taken from that site before seine collection and allowed to incubate for 30 minutes (Appendix S2: Fig. S1f). During incubations, plastic bags were placed in a large cooler to ensure minimal stress. Directly after incubations, fish were transferred to a seawater-clove oil mixture for euthanasia. Two 60-millilieter (mL) 0.7-micrometer (μm) filtered water samples were taken from each plastic bag directly before and after each incubation trial, resulting in a pre- and post-incubation water sample for both N and P concentrations. Water samples were placed on ice and frozen immediately once returning from the field and kept in a -20 degrees Celsius (°C) freezer until processing.

We analyzed all water samples for concentrations of ammonium (NH4+) and phosphate (PO43-) using two spectra-photometric assays of phenolhypochlorite (Solórzano 1969) and molybdenum blue (Murphy and Riley 1962) methods as modified by Whiles et al. (2009). To quantify the N and P contribution for each fish, we took the difference of N and P concentrations from the pre- and post-incubation water samples (Appendix S2: Fig. S1g). Lastly, we took the ratio of change of N to P for each fish to calculate the N:P functional trait.

Morphological traits
To collect traits which inform energy acquisition and locomotion functional roles, we measured 15 morphometrics. For each fish, we took a series of five photos: (i) lateral full body, (ii) lateral head, (iii) lateral head with mouth protruded, (iv) ventral full body, and (v) anterior with mouth open, all with a ruler in shot for length standardization (Appendix S2: Fig. S1). Using ImageJ analysis (Schneider et al. 2012), we measured 15 morphometrics which were used to calculate five energy acquisition traits (oral gape surface, oral gape shape, oral gape position, protrusion, eye size) and five locomotion traits (eye position, body transverse surface, body transverse shape, pectoral fin position, caudal peduncle throttling) (Appendix S2: Table S1a-e; Albouy et al. 2011). The 10 continuous functional trait measurements are commonly used in morphological studies on fishes, have been connected to diet or movement (Sibbing and Nagelkerke 2000, Dumay et al. 2004, Mason et al. 2007, Villéger et al. 2010), and they are easily measured and broad enough to apply to any fish species.

All references to figures and tables are from Yeager, M. E. and A.R. Hughes. 2025. Functional trait analysis reveals the hidden stability of multitrophic communities. Ecology. In press.

Generalized additive model fitting
We quantified the functional traits of fish from the past RIDEM fish and macroinvertebrate survey from 2010-2015 within the RI coastal ponds by fitting generalized additive models (GAMs) to the 13 functional traits (response variables) with species and length as predictor variables. GAMs model a response variable (e.g., functional trait) as the sum of nonlinear functions from different predictor variables (e.g., species and fork length; Hastie and Tibshirani 1990). We utilized thin-plate penalized regression splines, which adds a penalty to the smoother function to avoid overfitting (Wood and Augustin 2002). Penalty weights were optimized using the restricted maximum likelihood (REML) score which minimizes the root mean square error of the model fit to the data and balances model complexity with the goodness of fit and tends to under smooth compared to other estimations like GCV. To run all models, we used the function 'gam' of the mgcv R package (R Development Core Team 2013). We used the GAMs fit to each functional trait (Appendix S3 for model fits) to infer the functional traits of 27 fish species for which we had fork length data across 6 years and 6 ponds of survey data collected from RIDEM. Specifically, each year from 2010-2015, RIDEM sampled communities monthly from May to October via 150-foot beach seine net. Individuals were counted, measured, and identified to species. For the purpose of this study, we examined species and their abundances averaged across sampling stations (Appendix S1) to account for dependence within each pond and averaged across months to account for seasonal differences of species presence. This yielded functional trait information for a total of 81,337 fish.

Relative stability analysis
We examined the temporal stability across six years (2010-2015) of the six coastal pond fish communities along with their functional traits in ordination space using non-metric multidimensional scaling (nMDS). We constructed the functional trait matrix by calculating the community-weighted values at each year by pond combination for each functional trait. We then used the 'betadisper' function in vegan R package on the dissimilarity matrix of both traits and species communities to calculate two metrics of community and trait stability: (1) average year-to-centroid distance, and (2) average year-to-year distance. For both metrics, a lower value equates to greater stability, via less community turnover. The average year-to-centroid distance is the average across all years of the Bray-Curtis dissimilarity between each pond community at each year and the pond community’s average (or centroid; Fig 1a, c). This metric measures how the community diverges from its average composition throughout years. The second metric, average year-to-year distance, is the pond average of the Bray-Curtis dissimilarity between consecutive years (Fig 1b, d). This metric identifies small incremental changes. This metric is useful for evaluating whether a community is undergoing a directional shift due to an environmental change by comparing the consecutive year distances with the total change across all years. To compare the two stability metrics across species and trait community composition, we calculated the relative distance by dividing stability by the maximum distance found across all ponds and years for each composition type. We then ran a two-way ANOVA for both year-to-centroid and year-to-year distance, testing whether relative stability differed across pond, trait versus species composition, or the interaction.

Dissimilarity analysis
To test for statistical differences in community and trait composition across pond and year, we used a permutational multivariate analysis of variance (PERMANOVA). We then explored which traits and fish species were driving the dissimilarity across both year-to-centroid and year-to-year 'distance' by conducting similarity percentage (SIMPER) analysis. For the year-to-centroid simper test, we first calculated the average species and trait values through time for each pond, then ran the simper analysis grouping by pond and selecting the comparisons of each year to the average. For the year-to-year simper test, we ran the simper analysis grouping by year and only selected the consecutive year comparisons. We used the functions 'metaMDS', 'adonis', and 'simper' of the vegan R package to plot nMDS, calculate PERMANOVA, SIMPER and the biplot correlation test respectively (R Development Core Team 2013). To examine how specific groups related to dissimilarity across ponds, we categorized species by feeding mode using FishBase.org (Froese and Pauly, 2023). Fish were assigned to the following 4 feeding modes: (1) detritivore, (2) planktivore, (3) hunting meiofauna, (4) hunting macrofauna (Appendix S6: Table S6).

Biplot correlation analysis
Lastly, we examined the correlation of communities between functional traits and species and trophic composition. Using the function 'envfit' of the vegan R package (R Development Core Team 2013), we correlated the functional trait ordination with the species abundance matrix as well as a feeding mode abundance matrix. Plotting the species or trophic feeding mode whose abundances were significantly correlated with the ordination axes allowed us to identify species and trophic-trait associations based on the overlap or the close proximity of species vectors with functional trait scores.

- Imported original file "Functional_Trait_Data_Rhode_Island_Salt_Pond_Survey.xlsx" into the BCO-DMO system.
- Converted dates to YYYY-MM-DD format.
- Adjusted column names to comply with BCO-DMO naming conventions.
- Flagged 'NA' and '#DIV/0!' as missing data values (missing data are empty/blank in the final CSV file).
- Replaced positive "Narrow River" longitude values with negative values to correct for error in original spreadsheet.
- Saved final file as "functional_trait_data_rhode_island_salt_pond_survey-1.csv".

Version history:
v1 (date: 2022-03-11) - dataset originally published at BCO-DMO.
v2 (date: 2025-02-16) - significant updates/edits to the "Methods & Sampling" and "Data Processing Description" sections of the metadata; title also revised slightly.

NSF Award Abstract:
Disease outbreaks in the ocean are increasing, causing losses of ecologically important marine species, but the factors contributing to these outbreaks are not well understood. This 5-year CAREER project will study disease prevalence and intensity in two marine foundation species - the seagrass Zostera marina and the Eastern oyster Crassostrea virginica. More specifically, host-disease relationships will be explored to understand how genetic diversity and population density of the host species impacts disease transmission and risk. This work will pair large-scale experimental restorations and smaller-scale field experiments to examine disease-host relationships across multiple spatial scales. Comparisons of patterns and mechanisms across the two coastal systems will provide an important first step towards identifying generalities in the diversity-density-disease relationship. To enhance the broader impacts and utility of this work, the experiments will be conducted in collaboration with restoration practitioners and guided by knowledge ascertained from key stakeholder groups. The project will support the development of an early career female researcher and multiple graduate and undergraduate students. Students will be trained in state-of-the-art molecular techniques to quantify oyster and seagrass parasites. Key findings from the surveys and experimental work will be incorporated into undergraduate courses focused on Conservation Biology, Marine Biology, and Disease Ecology. Finally, students in these courses will help develop social-ecological surveys and mutual learning games to stimulate knowledge transfer with stakeholders through a series of workshops.

The relationship between host genetic diversity and disease dynamics is complex. In some cases, known as a dilution effect, diversity reduces disease transmission and risk. However, the opposite relationship, known as the amplification effect, can also occur when diversity increases the risk of infection. Even if diversity directly reduces disease risk, simultaneous positive effects of diversity on host density could lead to amplification by increasing disease transmission between infected and uninfected individuals. Large-scale field restorations of seagrasses (Zostera marina) and oysters (Crassostrea virginica) will be utilized to test the effects of host genetic diversity on host population density and disease prevalence/intensity. Additional field experiments independently manipulating host genetic diversity and density will examine the mechanisms leading to dilution or amplification. Conducting similar manipulations in two marine foundation species - one a clonal plant and the other a non-clonal animal - will help identify commonalities in the diversity-density-disease relationship. Further, collaborations among project scientists, students, and stakeholders will enhance interdisciplinary training and help facilitate the exchange of information to improve management and restoration efforts. As part of these efforts, targeted surveys will be used to document the perceptions and attitudes of managers and restoration practitioners regarding genetic diversity and its role in ecological resilience and restoration.