Out With The Trash Developing Biomass Estimates for Marine Debris Estimates

Turner, Amanda; Tucker, Ean; Payton, Tokea; Sims, Randi; Childress, Michael

Background

Marine debris estimate is equivalent in weight to more than 50 thousand double-decker buses

It is estimated that 10% of global marine litter (sometimes estimated at 640,000 metric tons) consisting of abandoned or lost fishing equipment, enters the ocean every year. This fishing equipment includes rope, nets, traps, and trap debris (Macfadyen et al 2009; FAO 2009).

In 2017, the Florida Keys National Marine Sanctuary established an organization called Goal Clean Seas (GCS). GCS trains local dive shops to become permitted Blue Star Operators. This status allows for shops to conduct Dive Against Debris events in partnership with PADI and Project Aware. Initially, GCS set out to remove debris distributed by Hurricane Irma; however, after removing over 10,000 lbs in their first year it was evident that much of the non-storm related debris consisted of gear from recreational and commercial fishing.

Matching removal efforts with the amount of marine debris entering in the ocean has so far been impossible, allowing for debris to age and accumulate biomass ultimately becoming artificial habitat, refuges, and entrapment risks.

Photo (left): de Carvalho-Souza et al. 2018

Goals

Develop biomass estimates of organisms removed from pictures and weights
Use this data to make models that can be used to predict biomass from images of an organism
Make conclusions on the positive or negative effects debris removal has on the health of coral reef ecosystems

Primary Research Question

When does the cost of debris removal due to loss of biodiversity exceed the benefit of debris removal in terms of hazards to marine life?

To assess the costs and benefits of marine debris removals in the Florida Keys National Marine Sanctuary Given the potential for significant removal of biomass; organisms have been measured and recorded following debris removal events...

Research Methods

Part One: Trash Treasure Hunt

To begin, our team went down to the Florida Keys to collect and photograph benthic marine debris. The images collected were used to get a clear idea of the size of the debris, the organism that could be affected, and the average biomasses of organisms attached to each category of debris.

Part Two: Sort and Organize

Once the debris was brought up to the boat, the debris was sorted by material and scaled images were taken of both the debris and any associated organisms found with the debris.

Part Three: ImageJ Analysis

ImageJ was used to collect length, width, area measurements of the organisms found with the debris in order to create statistical models that relate biomass to each parameter.

It's a trashy job but someone has to do it! Depicted here are some of the awesome volunteers that help keep the Florida Keys coral reefs clean.

Marine Debris Critters

To date . . .

. . . we have participated with 5 dive shops for 16 dives with an average volunteer count of 12 persons per dive. From these dives and the debris collected we have identified over 1185 individual organisms noting a max genus richness of 56 and a min of 1 per debris dive.

A variety of organisms are found with debris

A subset of organisms from all genus identified and measured for length, width, and area in centimeters were weighed to gain estimates of biomass.

Weighing a small octopus

Biomass Estimate Data

To determine the best measurement (length, width, area) predictor for weight in grams of each phylum recorded multiple regression analyses were used.

The multiple regression analysis of the annelida (worms) data indicated that width, length, and area explained 65% of the weight with an adjusted R-squared of 0.6513 and F-statistic of 28.39. However, only width was significant (p = 0.0283) suggesting that width may be the best predictor for worm mass. A separate linear model for the effect of width indicated an adjusted R-squared of 0.3894 and 29.06 F-statistic. Indicating a relatively weak positive correlation.

The best predictor for this model seemed to be the measured area of the arthropod carapace. The multiple regression analysis of the phylum arthropoda (consists of crustaceans) data indicated that width, length, and area explained 52% of the weight with an adjusted R-squared of 0.528 and F-statistic of 21.14. However, only area was significant (p = 0.0002). A linear model for the effect of area alone indicated an adjusted R-squared of 0.5273 and an F-statistic of 61.24. Indicating a positive correlation.

The multiple regression analysis comparing length, width, and area as predictors for algae biomass indicated significant values for both width (p < 0.01) and area (p < 0.001) for phylum Chlorophyta. All variables combined accounted for 62% of the model (adjusted R-squared = 0.6213) with an F-statistic of 21.23. When adjusting for each significant variable in the model, the effect of area appeared to be most influential. A separate linear model for the area effect (not shown) indicated an adjusted R-squared of 0.8089 and a F-statistic of 157.6 indicating a positive correlation. 47% of the model was explained by width suggesting that area is the best predictor for biomass in this model.

The best predictor for this model seemed to be the measured area of the phylum Porifera (sponges). The multiple regression analysis comparing length, width, and area as predictors for sponge biomass indicated significant values for both width (p < 0.01) and area (p < 0.001). All variables combined accounted for 62% of the model (adjusted R-squared = 0.6213) with an F-statistic of 21.23. When adjusting for each significant variable in the model, the effect of area appeared to be most influential. A separate linear model for the effect of area (not shown) indicated an adjusted R-squared of 0.9384 and a F-statistic of 549.9 indicating a positive correlation. 2% of the model was explained by width and was no longer significant (p > 0.05) suggesting that area is the best predictor for biomass in this model.

Conclusions

These preliminary data will allow for sampling and biomass estimates to best determine how much biomass is being removed during debris removal events. The linear equations found for each of the regression models can be used to estimate biomass removed for each represented phyla. Once more data is collected, phyla measurements may be divided up into lower taxonomic categories for a more detailed and potentially more accurate estimate.

These data will be added as parameters in a spatially explicit agent based model to be used as a marine debris management tool in the Florida Keys National Marine Sanctuary. This model will account for species richness and biomass estimates, loss from ghost fishing and physical damage, as well as the human dimension centered around dive shops and their volunteers.

References

de Carvalho-Souza, G. F., Llope, M., Tinoco, M. S., Medeiros, D. V., Maia-Nogueira, R., and C. L. S. Sampaio. 2018. Marine litter disrupts ecological processes in reef systems. Marine Pollution Bulletin. 133: 464-471. doi: 10.1016/j.marpolbul.2018.05.049

FAO, Report of the Thirty-third Session of the Committee on Fisheries, Rome, Italy 9–13 July 2018. FAO Fisheries and Aquaculture, Report No. 1249, Rome, 2019. License: CC BY-NC-SA 3.0 IGO.

G. Macfadyen, T. Huntington, R. Cappell, Abandoned, lost or otherwise discarded fishing gear. UNEP Regional Seas Reports and Studies No. 185. FAO Fisheries and Aquaculture Technical Paper, No. 523. Rome, UNEP/FAO, 2009, p. 115.

R Code PDF

Scan for pdf of R code used to make graphs and statistical analyses

Appreciate