Remotely-sensed cloud cover predicts biodiversity & climate change vulnerability

The above image is of the Sahel in central Africa (creative commons license).

The modern world is in the midst of an extinction crisis.  Some estimates put current extinction rates at 10-1000x above the background rates observed before human domination of the earth’s systems.  Human manipulation of the landscape and global climate change are both contributing to high rates of global and local biodiversity loss.  In the past several years, the magnitude of the problem has led many ecologists (including here on the PLoS blog) to call for larger scale global approaches.  Biodiversity monitoring using satellite imagery, LIDAR data, and other remote sensing methods may be the only way to efficiently address such problems at appropriate scales.  Work last month in PLoS Biology has taken a unique look at the problem.  The authors use medium-grain (<1 km) global cloud cover from MODIS data (Moderate Resolution Imaging Spectroradiometer satellites) to show a remarkable connection between cloud cover, specific species distributions, biodiversity hotspots, and cloud forest biome delineations.  These relationships are finer resolution and more accurate than ever before.

The authors show that the cloudiest places in the world are in equatorial South America, the Congo River basin, and Southeast Asia (Fig 1a below).  

This is in comparison to some of the least cloudy places in the world: in baja Mexico, Saudi Arabia, and northern Africa.  These trends fit our expectations, but the authors go on to use intra and inter-annual cloud frequency to demonstrate even more meaningful ecological information.

Intra-annual variability in cloud cover is associated with intense rainfall seasonality: climates that experiences extreme drought during some period of the year, and extreme rainfall during other periods.  The authors show that the most severe intra-annual variability in cloud cover occurs in the monsoonal biomes of India and the arid area of the Sahel in Africa.  

This map acts almost as a heatmap for climate change and land use vulnerability.  Areas of the Sahel and monsoonal India are some of the hardest hit by land degradation and desertification due to intensive agricultural practices and climate change.  

The Wilson & Jetz (2016) cloud cover dataset indicates these vulnerable areas at high resolution and with a mechanistic underpinning.

The authors also demonstrate that the opposite trend (low annual seasonality in cloud cover) may be associated with climatic stability, the potential for lower rates of extinction, increased endemism, and regional maintenance of species diversity.  While there are alternative theories that argue the opposite (sometimes a stable climate leads to lower diversity), the authors demonstrate several known biodiversity hotspots (Congo River Basin, Indonesia, the Andes) are correlated with higher rates of both intra- and inter-annual stability in climate (Fig. 3 below).

Finally, the authors offer substantial improvements in our ability to predict particular species distributions based on remote-sensing data.  While past vegetation maps were mostly based on interpolated temperature and precipitation data (using weather stations), the authors show the integrated and higher-resolution value of cloud cover frequency data.  Using two species (Lepidocolaptes lacrymiger and Protea cynaroides) from two biomes and two different continents,

the authors demonstrate that our ability to accurately predict species distributions (verified using presence/absence data collected at each site) can improve by up to 30% (Figure 5f below).

The cloud cover dataset offers the highest resolution insight into species level data that has ever been shown at global scales.  While LIDAR and higher-resolution imagery may offer even greater insights into finer-scale processes, Wilson & Jetz (2016) are demonstrating data that help us make decisions about climate change vulnerability, species vulnerability, and biodiversity monitoring today.  Beyond this, the MODIS dataset is updated twice daily and will offer rapid assessment of climate and landuse changes into the future.  The authors also lay the groundwork for interpolating species and biome-level patterns using LIDAR data at higher resolutions in the future.  

The density and diversity of insights that will likely arise as a result of these data are just beginning.  To explore the data more extensively and/or use the data to ask other important questions about species, biomes, land use, or climate, the authors have made the entire dataset available for exploration here: http://www.earthenv.org/cloud.