Fig 1.
NOAA Coastal Buoys of Florida.
Re-rendered from https://www.ndbc.noaa.gov/ on 12/22/2021. This demonstrates the distribution of the relevant NOAA coastal buoys (represented by green circles) around the coast of Florida, which collect meteorological data on coastal conditions every six minutes. Each buoy is represented by a five-letter code that correspond to counties filled in the same color. Corresponding county names are listed in Table 1.
Table 1.
Names of NOAA buoys used in our analysis.
The Florida counties that correspond to meteorological data, and NOAA reference information for each buoy.
Fig 2.
Occurrences of V. vulnificus in Florida between 2008 and 2020.
Panel A depicts the relative occurrences of the 448 cases across each of Florida’s 50 reporting counties documented between 2008 and 2020. The data is in descending order of frequency by county, with the county of Hillsborough demonstrating the highest reported occurrences of V. vulnificus in the state. Panel B demonstrates the breakdown of V. vulnificuscases in Florida by month, between the years 2008 and 2020 in the state of Florida. July, August, and September demonstrate the highest incidence of reporting. Panel C breaks down our dataset of 448 cases according to season. Summer (June, July, and August) had the greatest number of reported cases, followed by fall (September, October, and November), with the fewest cases reported in the winter (December, January, and February), followed by spring (March, April, and May). Panel D represents our dataset of 448 cases broken down according to year. In 2008, the fewest cases were reported, while in 2017, the highest frequency of cases was reported.
Table 2.
Case fatality rates of V. vulnificus as reported by the state of Florida.
In general, case fatality rates did not exceed 30 percent, except for January and April.
Table 3.
Spearman’s correlation coefficients for our meteorological variables.
To determine the strength of the relationship between each meteorological variable, we performed a series of non-parametric Spearman’s Rank Correlation ρ coefficients to measure their association. We performed the analyses covering four conditions: all periods, e.g., during months when no cases were reported, during months when cases were reported, and during months when deaths were reported as one dataset. While many of the variables had statistically significant relationships according to the p-value, those correlations’ strengths were weak, except for ATMP and WTMP, where a strong relationship was defined as one with a ρ of greater than 0.80.
Fig 3.
Correlations between various environmental variables and cases of Vibrio vulnificus.
A) windspeed (WSPD) in meters/second versus the incidence of V. vulnificus cases reported in Florida, stratified by V. vulnificus incidence, specifically during months when no cases were reported, months when cases with a survival outcome were reported, and months when mortality due to V. vulnificus infection was reported. The mean WSPD for months when no cases were reported was 3.35 ± 0.02 SE m/s, while the mean WSPD for months when cases where survival was the outcome was a mean of 3.26 ± 0.05 SE m/s, and the mean WSPD for months when death was the outcome was 3.27 ± 0.10 SE m/s. The Mann-Whitney U test indicated that the differences between the two groups were statistically significant (W = 325379, p-value = 0.01527). B) sea-level pressure (PRES) is stratified by V. vulnificus incidence according to months reporting deaths, months reporting cases, and months reporting no cases. The mean PRES for months when no cases were reported was 1017.8 ± 0.005 SE hPa while the mean PRES for months when cases where survival was the outcome was a mean of 1016.5 ± 0.003 SE hPa, and the mean PRES for months when death was the outcome was 1016.5 ± 0.002 SE hPa. C) Water temperature stratified by the reporting of V. vulnificus, specifically during months when no cases were reported, months when cases with a survival outcome were reported, and months when mortality due to V. vulnificus was reported. We found that our WTMP Kruskal-Wallis test was significant for this analysis, (χ2 = 176.14, df = 2, p-value < 0.0001), however, similar to ATMP, a pairwise multiple comparisons test with a Bonferroni adjustment demonstrated no significant difference between WTMP when survival cases were reported, and WTMP when non-survival cases were reported (p-value = 0.25). D) Air temperature (ATMP) stratified by V. vulnificus incidence, specifically during months when no cases were reported, months when cases with a survival outcome were reported, and months when a mortality due to V. vulnificus was reported. As ATMP increases, so too does the likelihood of the severity of V. vulnificus cases. While this was a significant relationship (ATMP Kruskal-Wallis χ2 = 181.32, df = 2, p-value < 0.0001), a follow-up pairwise multiple comparison test with a Bonferroni adjustment demonstrated that the differences in ATMP between months when cases were reported and when cases were not reported was significant (p<0.0001), but not during months when survival cases were reported, and non-survival cases were reported (p = 0.72). Please see main text for additional details.
Table 4.
Listed in this table are the models used to assess the meteorological, spatial, and temporal variables in a series of logistic regressions.
Also listed are their corresponding AIC values, McFadden’s R2, model degrees of freedom, and null deviance. *Corresponds to the best-identified model.