Doctor of Philosophy (PhD)


Biological Sciences

Document Type



Tropical cyclones (TC) are intense, localized disturbances that can potentially cause extensive damage to ecosystems. This dissertation is motivated by a need for interdisciplinary study of TC-ecosystem dynamics with applications to ecology, geography, and analytical methodology. I explore concept linkages between TC dynamics and ecosystems to characterize TC-ecosystem relationships as ecological disturbance regimes. This dissertation tests hypotheses and relationships of TC effects on estuarine fish assemblages in the north-western Gulf of Mexico. I explore concepts in TC-ecosystem relationships to link biotic dynamics with physical disturbance.

In Chapter 2, I compared spatial performance of generalized linear models (GLMs) and generalized additive models (GAMs) using abundances of 127 fish species (from SEAMAP-GOM data, 1985-2016). Overall, GAMs that explicitly fit functions of latitude and longitude had much lower residual spatial autocorrelation, retained more information from the data, and were more parsimonious than all other models.

Chapter 3 first explicitly defined concepts of spatiotemporal variation in TC patterns as ecological disturbance regimes. I then identified three critical properties of individual and recurrent TCs at any particular location, months within years, and among years: maximum sustained wind speed (Wmax), forward speed (FS) and probability of strike. Wmax and FS models estimated expected mean values across the study region as well as the distributions around the means. Additionally, I established probabilities of strike between the left and right sides of hurricanes with weakest and strongest winds, respectively. I showed that the interactions of these properties were explicitly defined in space and varied over spatial scale, and clearly delineated geographically defined TC regimes. Wmax and FS were further influenced by localized geomorphology and bathymetry. Further, I applied concepts of spatial processes and error under the modified areal unit problem (MAUP) to refine inferences about spatial scale. I showed how MAUP can help interpret ambiguous maps by comparing continuous spatially-explict GAMs to discrete Markov random field models (MRF, or geographically-weighted regression); interpretations of each model were mutually informative and complementary. To examine the relevance of FS to TC regimes, I reviewed literature that included articles reporting either TC effects on fishes or no effect and calculated the FS of those studies’ storms. For all papers reporting an effect, FS was slower than the mean, when for all papers reporting no effect, FS was greater than the mean.

Chapter 4 tested the theory developed in Chapter 3 by examining responses of fish populations over 35 yr of TC activity among four regions within the GOM. I examined the effects of three hurricane characteristics. Life history traits related to growth and timing, as well as estuarine-coastal spatial and habitat use, were strongly related to different TC characteristics.



Committee Chair

Hellberg, Michael



Available for download on Saturday, July 05, 2025