Land Surface Temperature and Miombo forest canopy phenophases: what induces leaf fall and leaf flush?

Miombo woodland is the most widespread tropical seasonal woodland and dry forest formation in Africa covering between 2.7 and 3.6 million km² in eleven countries. Leaf fall and leaf flush during the dry season is a major characteristic feature of the various Miombo species. However, the question on what induces the leaf fall process is by far inconclusive. Different studies indicate either moisture or temperature or both elements as inducers for leaf fall. Knowing what induces leaf fall is important for studying the consequence of e.g., climate change on the Miombo forest. To better understand the driver of leaf fall in Miombo forest we employed a simple remote sensing and statistical analysis approach using long term averages (2009 – 2018) of Land Surface Temperature (LST) of the Miombo forest, various vegetation indices (VI), actual evaporation (E_a), and root zone soil moisture (SM). The vegetation indices (VI) included the Normalised Difference Water Index (NDWI) as indicator of vegetation water content and the Normalised Difference Vegetation Index (NDVI) as indicator of plant photosynthetic activities and leaf cover. Results showed that the NDWI, NDVI, E_a and SM begun to decline immediately following the end of the rainy season in early April while the LST remained relatively constant before it began to decline in May when leaf fall in some Miombo species begins. Hysteresis graphs revealed that vegetation water content (i.e. NDWI) responded quicker to changes in both LST and SM. Furthermore, high rates of decrease in NDWI and NDVI values were observed between July and September the same period when LST increased. This is also the same period when leaf fall intensifies in Miombo forest. Correlation analysis revealed strong season-dependent LST relationship with VI and SM with the rainy season exhibiting strong negative linear correlations (R² = 0.77, 0.91, 0.88; for the NDWI, NDVI and SM respectively). In the dry season relatively weaker negative correlations (R² = 0.52, 0.60, 0.55; for NDWI, NDVI and SM respectively) were observed. On the other hand SM showed strong positive linear correlations (R² > 0.6) with NDWI and NDVI (for the rainy and dry seasons respectively). The correlations imply that in Miombo forest soil water content (i.e. SM), vegetation water content (i.e. NDWI) and the photosynthetic activities and leaf cover (i.e. NDVI) declines with increase in LST. These relationships show the possibility of land surface temperature being a major inducing element of leaf fall and changes in canopy structure in the Miombo woodland.