Limits...
A Geographic Mosaic of Climate Change Impacts on Terrestrial Vegetation: Which Areas Are Most at Risk?

Ackerly DD, Cornwell WK, Weiss SB, Flint LE, Flint AL - PLoS ONE (2015)

Bottom Line: The model was then projected for 54 future climate scenarios, spanning a representative range of temperature and precipitation projections from the CMIP3 and CMIP5 ensembles.Surprisingly, sensitivity to climate change is higher closer to the coast, on lower insolation, north-facing slopes and in areas of higher precipitation.The greater sensitivity of moist and low insolation sites is an unexpected outcome that challenges views on the location and stability of climate refugia.

View Article: PubMed Central - PubMed

Affiliation: Department of Integrative Biology, University of California, Berkeley, California, United States of America; Jepson Herbarium, University of California, Berkeley, California, United States of America.

ABSTRACT
Changes in climate projected for the 21st century are expected to trigger widespread and pervasive biotic impacts. Forecasting these changes and their implications for ecosystem services is a major research goal. Much of the research on biotic responses to climate change has focused on either projected shifts in individual species distributions or broad-scale changes in biome distributions. Here, we introduce a novel application of multinomial logistic regression as a powerful approach to model vegetation distributions and potential responses to 21st century climate change. We modeled the distribution of 22 major vegetation types, most defined by a single dominant woody species, across the San Francisco Bay Area. Predictor variables included climate and topographic variables. The novel aspect of our model is the output: a vector of relative probabilities for each vegetation type in each location within the study domain. The model was then projected for 54 future climate scenarios, spanning a representative range of temperature and precipitation projections from the CMIP3 and CMIP5 ensembles. We found that sensitivity of vegetation to climate change is highly heterogeneous across the region. Surprisingly, sensitivity to climate change is higher closer to the coast, on lower insolation, north-facing slopes and in areas of higher precipitation. While such sites may provide refugia for mesic and cool-adapted vegetation in the face of a warming climate, the model suggests they will still be highly dynamic and relatively sensitive to climate-driven vegetation transitions. The greater sensitivity of moist and low insolation sites is an unexpected outcome that challenges views on the location and stability of climate refugia. Projections provide a foundation for conservation planning and land management, and highlight the need for a greater understanding of the mechanisms and time scales of potential climate-driven vegetation transitions.

No MeSH data available.


Related in: MedlinePlus

Correlates of projected sensitivity of vegetation to climate change.Scatterplots of sensitivity of vegetation to climate change (based on slopes of Bray-Curtis distances, as shown in Fig 7) vs. four landscape factors: a) DJF minimum temperature; b) precipitation; c) wind speed; d) equinox solar radiation. Light blue lines show partial regression slopes from multiple regression on these four factors.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4482696&req=5

pone.0130629.g010: Correlates of projected sensitivity of vegetation to climate change.Scatterplots of sensitivity of vegetation to climate change (based on slopes of Bray-Curtis distances, as shown in Fig 7) vs. four landscape factors: a) DJF minimum temperature; b) precipitation; c) wind speed; d) equinox solar radiation. Light blue lines show partial regression slopes from multiple regression on these four factors.

Mentions: We evaluated spatial variation in sensitivity as a function of the baseline values for the seven predictor variables using multiple regression. Across the entire region, a multiple linear regression explained 19.4% of the variation. While all variables were statistically significant (given the very large sample size), we were able to drop the three variables with the lowest contributions (soil depth, CWD and JJA), and R2 was reduced only slightly to 19.0%. Of the remaining four variables, the direction of the relationship with sensitivity was in some cases surprising. Vegetation was predicted to be more sensitive to climate change in sites with warmer winter (i.e. near the coast), higher rainfall, lower wind, and lower solar insolation (i.e. north-facing slopes) (Fig 10). To illustrate the effect of aspect and insolation, we selected four pairs of sites in different parts of the SFBA, with one on a north-facing site with high vegetation sensitivity, and the other on a south-facing slope with lower sensitivity. For each one, we then created an artificial gradient with MAT increasing up to 4°C, CWD, JJA, and DJF increasing proportionally, as shown in Fig 2, and PPT held constant. These graphs illustrate the greater vegetation change on the north-facing slopes, which in each case go from a high probability of one type (redwood or grassland in these cases) and rapidly shift to a high probability of another (shrubland or blue oak woodland, respectively) (S9 Fig). In contrast, the south-facing slopes have probabilities more evenly distributed over several different types, and the distribution does not change much up to 4° of warming.


A Geographic Mosaic of Climate Change Impacts on Terrestrial Vegetation: Which Areas Are Most at Risk?

Ackerly DD, Cornwell WK, Weiss SB, Flint LE, Flint AL - PLoS ONE (2015)

Correlates of projected sensitivity of vegetation to climate change.Scatterplots of sensitivity of vegetation to climate change (based on slopes of Bray-Curtis distances, as shown in Fig 7) vs. four landscape factors: a) DJF minimum temperature; b) precipitation; c) wind speed; d) equinox solar radiation. Light blue lines show partial regression slopes from multiple regression on these four factors.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4482696&req=5

pone.0130629.g010: Correlates of projected sensitivity of vegetation to climate change.Scatterplots of sensitivity of vegetation to climate change (based on slopes of Bray-Curtis distances, as shown in Fig 7) vs. four landscape factors: a) DJF minimum temperature; b) precipitation; c) wind speed; d) equinox solar radiation. Light blue lines show partial regression slopes from multiple regression on these four factors.
Mentions: We evaluated spatial variation in sensitivity as a function of the baseline values for the seven predictor variables using multiple regression. Across the entire region, a multiple linear regression explained 19.4% of the variation. While all variables were statistically significant (given the very large sample size), we were able to drop the three variables with the lowest contributions (soil depth, CWD and JJA), and R2 was reduced only slightly to 19.0%. Of the remaining four variables, the direction of the relationship with sensitivity was in some cases surprising. Vegetation was predicted to be more sensitive to climate change in sites with warmer winter (i.e. near the coast), higher rainfall, lower wind, and lower solar insolation (i.e. north-facing slopes) (Fig 10). To illustrate the effect of aspect and insolation, we selected four pairs of sites in different parts of the SFBA, with one on a north-facing site with high vegetation sensitivity, and the other on a south-facing slope with lower sensitivity. For each one, we then created an artificial gradient with MAT increasing up to 4°C, CWD, JJA, and DJF increasing proportionally, as shown in Fig 2, and PPT held constant. These graphs illustrate the greater vegetation change on the north-facing slopes, which in each case go from a high probability of one type (redwood or grassland in these cases) and rapidly shift to a high probability of another (shrubland or blue oak woodland, respectively) (S9 Fig). In contrast, the south-facing slopes have probabilities more evenly distributed over several different types, and the distribution does not change much up to 4° of warming.

Bottom Line: The model was then projected for 54 future climate scenarios, spanning a representative range of temperature and precipitation projections from the CMIP3 and CMIP5 ensembles.Surprisingly, sensitivity to climate change is higher closer to the coast, on lower insolation, north-facing slopes and in areas of higher precipitation.The greater sensitivity of moist and low insolation sites is an unexpected outcome that challenges views on the location and stability of climate refugia.

View Article: PubMed Central - PubMed

Affiliation: Department of Integrative Biology, University of California, Berkeley, California, United States of America; Jepson Herbarium, University of California, Berkeley, California, United States of America.

ABSTRACT
Changes in climate projected for the 21st century are expected to trigger widespread and pervasive biotic impacts. Forecasting these changes and their implications for ecosystem services is a major research goal. Much of the research on biotic responses to climate change has focused on either projected shifts in individual species distributions or broad-scale changes in biome distributions. Here, we introduce a novel application of multinomial logistic regression as a powerful approach to model vegetation distributions and potential responses to 21st century climate change. We modeled the distribution of 22 major vegetation types, most defined by a single dominant woody species, across the San Francisco Bay Area. Predictor variables included climate and topographic variables. The novel aspect of our model is the output: a vector of relative probabilities for each vegetation type in each location within the study domain. The model was then projected for 54 future climate scenarios, spanning a representative range of temperature and precipitation projections from the CMIP3 and CMIP5 ensembles. We found that sensitivity of vegetation to climate change is highly heterogeneous across the region. Surprisingly, sensitivity to climate change is higher closer to the coast, on lower insolation, north-facing slopes and in areas of higher precipitation. While such sites may provide refugia for mesic and cool-adapted vegetation in the face of a warming climate, the model suggests they will still be highly dynamic and relatively sensitive to climate-driven vegetation transitions. The greater sensitivity of moist and low insolation sites is an unexpected outcome that challenges views on the location and stability of climate refugia. Projections provide a foundation for conservation planning and land management, and highlight the need for a greater understanding of the mechanisms and time scales of potential climate-driven vegetation transitions.

No MeSH data available.


Related in: MedlinePlus