Predictions of macroalgae habitat shifts due to climate change impact in an Antarctic Peninsula fjord
Macroalgae are (among) the major primary producers in West Antarctic coastal environments and respond highly sensitive to environmental factors such as light conditions. In Potter Cove, King George Island (Isla 25 de Mayo), Antarctica, suspended particulate matter (SPM) from glacial melting leads to...
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Conference Object |
| Language: | unknown |
| Published: |
2016
|
| Subjects: | |
| Online Access: | https://epic.awi.de/id/eprint/47171/ https://zenodo.org/record/162116#.Wul9XcmpVaQ https://hdl.handle.net/10013/epic.e11c0c95-a3cd-4e57-a86a-7a6f7e9618f1 |
| Summary: | Macroalgae are (among) the major primary producers in West Antarctic coastal environments and respond highly sensitive to environmental factors such as light conditions. In Potter Cove, King George Island (Isla 25 de Mayo), Antarctica, suspended particulate matter (SPM) from glacial melting leads to shading of photosynthetic light during the summer growth season, with predictable effects on macroalgal bed extension on newly ice-free areas inside the cove. The R-package ‗biomod2‘ includes 10 different species distribution models (SDM) and 10 different evaluation methods to predict species or community occurrence based on statistical relationships with environmental conditions. In this study, we applied ‗biomod2‘ to macroalgae presence and absence data to test the suitability of SDMs and to assess the environmental response of macroalgae to glacial retreat. Four different scenarios of distribution shifts were modeled assuming different SPM conditions for varying climate change scenarios. According to the averaged evaluation scores of relative operating characteristics (ROC) and true scale statistics (TSS) by the models applied, those methods based on a multitude of decision trees such as Random Forest and Classification Tree Analysis, reached the highest predictive ability followed by generalized boosted models (GBM) and maximum-entropy approaches (MAXENT). The final ensemble model (EM) used 135 of 200 calculated models (TSS > 0.7) and identified hard substrate and SPM as the best predictors explaining more than 60 % of the distribution. These variables were followed by distance to glacier, total organic carbon (TOC), bathymetry and slope. The modeled present status of macroalgae spatial distribution in this study results in only 18.25 % of visually estimated areas colonized by macroalgae in Potter Cove. We coupled the EM with changing SPM conditions representing an increase of melt water input to model the light condition niche of macroalgae in Potter Cove and its potential response to projected climate ... |
|---|