Transitions from grassland to savanna under drought through passive facilitation by grasses

Questions: Woody plant encroachment into former grasslands currently represents a major physiognomic shift globally. Seedling establishment is a critical demographic bottleneck and is considered to be alleviated by increases in water availability and negatively impacted by interactions with grasses, particularly when water stress increases. However, interactions with grasses that are not actively competing for resources ('passive interactions' when grasses are dead) has seldom been considered. Could the transition from a live to a dead grass (litter) canopy favour recruitment of woody seedlings in a semi-arid grassland of the American SW? How does the sign and intensity of grass-seedling interactions change across drastically different summer precipitation regimes with and without passive interactions? Location: Sonoran Desert shrub savanna at the Santa Rita Experimental Range, near Tucson, AZ, US. Methods: Four cohorts of Prosopis velutina seeds were planted annually (2002-2005) under rainout shelters that intercepted all incoming precipitation on a soil with sandy loam texture. Summer precipitation was manipulated to simulate either a 50% increase or decrease in the long-term mean, and cover was manipulated to simulate a grassland dominated by the C4 bunchgrass Heteropogon contortus or left unvegetated. Emergence and survival of P. velutina was monitored and compared across cover types, along with monitoring of soil water content and light interception. Results: Strong active competition was observed with live grasses, under both summer drought and also under ample summer water supply. However, the pattern was reversed and strong passive facilitation of P. velutina was observed over time when grass canopies died and remained in place. This passive facilitation under dry summers was so strong that recruitment under dead grass conditions was comparable to that observed when ample water supply removed the effects of competition on unvegetated plots. Conclusions: After significant mortality of standing grass canopies, which typically compete for limited soil moisture resources, rates of recruitment by shrubs may increase even with significant seasonal drought. This work extends our understanding of interactions among co-located organisms and their effects on plant community dynamics, and introduces a new hypothesis on how grass litter facilitates woody plant encroachment during seasonal droughts.