Author (Your Name)

Nam LeFollow

Date of Award


Document Type

Honors Thesis (Open Access)


Colby College. Geology Dept.


Bruce Rueger

Second Advisor

Robert Gastaldo

Third Advisor

Struan Smith


Harrington Sound, Bermuda, is a nearly enclosed lagoon acting as a subtropical/tropical, carbonate-rich basin in which carbonate sediments, reef patches, and carbonate-producing organisms accumulate. Here, one of the most important calcareous groups is the Foraminifera. Analyses of common benthic orders, including miliolids (Quinqueloculina and Triloculina spp.) and rotaliids (Homotrema rubrum, Elphidium spp., and Ammonia beccarii), are essential in understanding past and present environmental conditions affecting the island's coastal environment. These taxa have been studied previously; however, factors explaining their individual patterns of abundance in the Sound are not well detailed. The goal of this study is to understand foraminiferal assemblage trends by assessing parameters including depth, test-size, temperature, salinity, dissolved oxygen, pH, and variable interactions. The results of this study can serve as a baseline model for Neogene carbonate lagoons that accumulated Foraminifera under different climate conditions.

Benthic sediment samples were collected at depths ranging from 3 m to 24 m using an Ekman dredge, while water-quality data were collected using an In-Situ SMAR Troll MP probe. Due to the limited test-size of these taxa, four size fractions were examined (1/2+ mm, 1/4 -1/2 mm, 1/8 - 1/4 mm, 1/16 - 1/8 mm). Dead-test and fragment counts were performed after randomizing sample selections with a sediment splitter. Using RStudio and ArcMap, basic, advanced statistical, and geospatial analyses were conducted on each taxon (segregated by test and fragment size) to examine depth-profile and spatial relationships between count data, physical, and geochemical factors.

Results from multiple quantile regression, nonparametric MANOVA, and geospatial modeling show that each taxon, under identical depth and spatial constraints, can be accounted for by different functional combinations of variables. In testing multivariate response and factorial influence, initial relative effect estimations and subset algorithms from nonparametric MANOVA effectively quantify abundance patterns. Spatial maps of all test-size fractions of Quinqueloculina & Triloculina spp. showed that these species were concentrated in deeper and colder regions in Harrington Sound, which disagreed with previously reported studies. Fragment counts of H. rubrum increased in shallower and warmer regions, with lower salinity, higher dissolved oxygen, and higher pH. Both Elphidium spp. and A. beccarii test counts were found be concentrated in areas with high salinity, although there were several anomalies across test-size fractions for both taxa.

Although there was a limited number of sampling points and measurements, the study's statistical methods proved to be effective in modeling the distribution of test-size concentrations within a complex system. Thus, as one of the few pioneering studies to cumulatively use predictive, multivariate, and geospatial models, this study urges future research studies to push beyond conventional, quantitative methodological boundaries regarding similar inquiries.


Foraminifera, Statistical Modeling, Climate Research, Geology, Bermuda