Fall Graduate Portfolio Colloquium - Damla Cinoglu

The Fall 2023 Graduate Portfolio Colloquium will be held on December 5th from 10:30 a.m. to 1:30 p.m. Damla Cinoglu's talk will be from 12:00 p.m. - 12:30 p.m. This event is virtual via Zoom.    

Title: System Evaluation by Betting Mechanism Portfolio in Scientific Computation

Abstract: Tropical forests are hotspots of tree diversity. But, these important 1 ecosystems are the subject of natural and anthropogenic disturbances, leaving us mosaics of successional forests. We can use these forests to better understand the strategies of trees within them that allow for their coexistence, enabling us to preserve their diversity better. Insect herbivory has shown to be maintaining tree diversity in the tropics, possibly as a fundamental mechanism. However, its variation within and among species and demographic groups, with individual height, and along forest successional age and understory light availability gradients has not yet been fully explored. I have carried out a field campaign to measure herbivory damage, individual height, and understory light availability on tree seedling across succession and by species (and demographic group) at Neotropical rainforests of Panama in the summers of 2022 and 2023. I built two Bayesian hierarchical models to model the incidence of herbivory (Binomial model) and % herbivory damage (Beta model). I found that when the data is organized by tree demographic groups (a coarser yet informed categorization) instead of species, the Binomial model has a better fit to the data (DIC score: 14140.57 vs. 34772.02) whereas the same pattern does not apply to the Beta model (DIC score: 42705.94 vs. 42470.64). At the community level, all covariates (height, light, height x light, forest age) had a minor positive influence on the incidence of herbivory (μ₁ = 0.10, μ₂ = 0.05, μ₃ = 0.05, μ₄ = 0.11) and the % herbivory damage (μ₁ = 0.08, μ₂ = 0.003, μ₃ = 0.04, μ₄ = 0.06). In explaining the incidence of herbivory, height as a covariate had explained greater variation than light for all demographic groups except for the fast growing trees (β_1,1 vs. β_2,1 = 1, β_1,2 vs. β_2,2 = 0.09,  β_1,3 vs. β_2,3 = 0.44 β_1,4 vs. β _2,4 = 1, β_1,5 vs. β_2,5 = 0.79). Whereas in the beta model, all demographic groups had higher % herbivory for their taller individuals (β_1,1 = 0.12, β_1,2 = 0.06, β_1,3 = 0.09,  β_1,4 = 0.08,  β_1,5 = 0.07) and for individuals sampled in later successional forests (β_4,1 = 0.07,  β_4,2 = 0.05,  β_4,3 = 0.1,  β_4,4 = 0.0008,  β_4,5 = 0.06). This research combines fieldwork and large-scale, Bayesian data analysis to tackle the question of how successional dynamics contribute to the rich diversity we see in tropical forests