Nathan Tennies, “Where Will it Grow? Bull Kelp Distribution Modeling in the Salish Sea”
Mentor: Filipe Alberto, Biological Sciences
The bull kelp, Nereocystis luetkeana, is a foundational species for aquatic ecosystems in the Salish Sea on the border of Washington and British Columbia. Its ecological importance, population contractions, and the present environmental pressure in these bodies of water require an increased research effort on bull kelp dynamics in the Salish Sea. We’re approaching this issue by developing a species distribution model (SDM) to predict, in a spatially explicit way, the probability of N. luetkeana occurrence across the Salish Sea. SDM models are built from presence and absence species occurrence data (the response variable) and its association with a set of spatially explicit environmental data (the predictor variables). Environmental data were obtained from the Aqua-MODIS NASA satellite, the Bio-ORACLE global marine dataset, and local digital elevation models. Model development started summer 2019 and was finished fall 2020. The main technique used was generalized linear modeling, a form of multiple regression analysis better suited for presence-absence responses. Our top four predictors of presence are depth, photosynthetically active radiation, maximum spring sea surface temperature, and maximum primary productivity. An SDM model will be valuable to guide current restoration efforts by helping allocate resources more effectively to populations with long-term promise.
Synchronous Presentation April 16th at 2:45pm
Phone Conference ID: 655 025 361#
Nice job, Nate. I am curious about the shallow photic zone depth (40m) and wonder if that area is pretty eutrophic from proximity to civilization. You took several very relevant components into account. If I remember correctly, kelp absorb nutrients through the fronds, so how does benthic P and N play in? Very stimulating to see a modeler need more data.
Thank you! I think the photic depth may have been more of a rule of thumb thing, but I’m not sure I would have to ask my PI about that. He’s the one who recommended that specific depth. I looked at some satellite images from landsat of the area once and it definitely seemed pretty eutrophic in near Seattle and Vancouver, but I don’t know something more concrete than that offhand. The images I viewed could have been after a big rainfall or something. Max benthic nitrate was not included in the end model, which is interesting in this context. It seems that I had to filter out max benthic phosphorus because of correlations with other predictor variables unfortunately. I don’t know offhand the nature of those correlations, I filtered those out quite some time ago. I definitely would not mind more data particularly presence/absence data that I would not be worried about the reliability for our uses. There were some other more complicated predictor variables that I thought would be interesting to include, like some of wave action or wave exposure, but we ended up working with what was shown in the first predictor table.
Nice job, Nate! It was great to see the work that you have been able to with the Alberto Lab.