Historic manatee use of Three Sisters and Jurassic/House Springs

Crystal River National Wildlife Refuge Complex

This report provides a preliminary summary of manatee observations during aerial surveys via fixed wing aircraft of Three Sisters Spring and the Jurassic Spring and House Spring vicinity during the winters (October through March) of 1983-1984 to 2017-2018. For additional survey details, see Kleen (2014).

Study area

This project focuses on manatee observations in the Kings Bay area of the Crystal River, Citrus County, Florida (Figure 1). In particular, it summarizes the observations of manatees within three springs in Kings Bay: Three Sisters Spring and the general vicinity around two smaller springs, Jurassic and House Springs (Figure 1).

Figure 1. Manatee observations (points) during winters (Oct - Mar) of 1983-1984 to 2018-2019 in Kings Bay, Crystal River, Citrus County, Florida. Three Sisters Spring and the vicinity of Jurassic and House Springs (polygons) are the focal areas of the current study.

Survey summary

Manatees were first observed during aerial surveys in Three Sisters Springs and the Jurassic and House Springs vicinity in the winter of 1993-1994 though more intensive access and use of the springs is more recent (Figure 2).

Figure 2. Manatee observations during winters (Oct - Mar) of 1983-1984 to 2017-2018 in Three Sisters Spring and the vicinity of Jurassic and House Springs.

Modelling manatee presence and abundance

For this preliminary analysis, in which we summarize manatee observations without reference to any potentially important covariates such as water level or water temperature, we focus on two basic aspects of manatee use of the springs: probability of observation within the springs during aerial surveys, and maximum observed count of manatees within the springs over the course of each winter period.

Probability of manatee presence

We used a binomial generalized additive model (GAM; Wood 2006) to estimate the probability at least one manatee was observed during aerial surveys in either Three Sisters Spring or the Jurassic/House Spring vicinity each winter. The GAM allows for a potentially nonlinear change in the probability of a manatee observation over time (i.e., from 1983-1984 to 2017-2018). We allowed any change in probability of manatee observation over time to vary by spring. Models were fit with the mgcv package (Wood 2019) within R [R Core Team (2019); Version 3.6.1].

The probability of observing manatees using Three Sisters Spring and in the vicinity of Jurassic and House Springs increased substantially in the early 2000s (Figure 3); the model estimated an approximately 25% annual increase in the odds of observing a manatee during a survey. There was little to suggest a difference in the probability of observering a manatee between the two survey areas. It is worth noting that these models do not account for imperfect detection in general, nor possibile differences in manatee detectability between springs.

Figure 3. Probability of observing a manatee in Three Sisters Spring or in the vicinity of Jurassic and House Springs during aerial winter surveys. Lines represent the average estimated probability over time; shaded ribbons indicated the pointwise standard error. Points indicate the observed proportion of aerial surveys during a winter with at least one manatee observed.

Maximum manatee count per winter period

We used a negative binomial GAM to estimate the annual maximum count of manatees observed during aerial surveys in either Three Sisters Spring or the Jurassic/House Spring vicinity. As with the probability of observation (above), the GAM allowed for a potentially nonlinear change in the observed maximum manatee count over time (i.e., from 1983-1984 to 2017-2018). We allowed any change in maximum observed manatee count to vary by spring.

The maximum number of manatees observed on any given survey in Three Sisters Spring and in the vicinity of Jurassic and House Springs has steadily increased since the early 2000s, particularly in Three Sisters Spring (Figure 4). However, there is substantial unexplained variation in maximum manatee counts, presumably because manatee use of the springs relates to environmental variables (e.g., water temperature and water level) that are not considered in these preliminary models.

Figure 4. Maximum number of manatees observed in Three Sisters Spring or in the vicinity of Jurassic and House Springs during aerial winter surveys. Lines represent the average estimated maximum count over time; shaded ribbons indicated the pointwise standard error. Points indicate the maximum count of manatees observed during a given winter’s aerial surveys.

Literature cited

Kleen, J. M., and A. D. Breland. 2014. Increases in seasonal manatee (Trichechus Manatus Latirostris) abundance within Citrus County, Florida. Aquatic Mammals 40:69–80. <http://www.aquaticmammalsjournal.org/index.php?option=com_content&view=article&id=676:increases-in-seasonal-manatee-trichechus-manatus-latirostris-abundance-within-citrus-county-florida&catid=54&Itemid=157>. Accessed 3 Aug 2020.

R Core Team. 2019. R: A language and environment for statistical computing. manual, R Foundation for Statistical Computing, Vienna, Austria. <https://www.R-project.org/>.

Wood, S. 2019. Mgcv: Mixed GAM computation vehicle with automatic smoothness estimation. <https://CRAN.R-project.org/package=mgcv>. Accessed 3 Aug 2020.

Wood, S. N. 2006. Generalized additive models: An introduction with R. Chapman and Hall/CRC, Boca Raton, FL.