Fire and flood in the bosque: a cottonwood population model for flow-restricted streams of the American Southwest

Background/Question/Methods

In the American Southwest, cottonwood (Populus spp.) forests cover a small percentage of the landscape but are highly valued for their aesthetic appeal and ecological services. Cottonwoods require specific hydrological conditions for survival and reproduction and are less fire-tolerant than other woody species. These characteristics make cottonwood populations sensitive to the changes in stream flow and fire regimes anticipated under climate change scenarios. Linkages between hydrology, fire, and riparian vegetation have been demonstrated, but there is a need to synthesize this information to predict the long-term persistence of cottonwoods in an increasingly arid Southwest. From 2003 to 2011, we measured cottonwood density, seedling establishment, and mortality in relation to fire, drought, and flood along the flow-restricted Middle Rio Grande in central New Mexico. We used this information to modify an existing cottonwood population model to one in which flood, drought, and fire are stochastic events that affect vital rates.

Results/Conclusions

Our cottonwood matrix model contains six stages: seedlings (N₁); two, three, and four year olds (N₂, N₃, and N₄, respectively); non-reproductive adults (N₅); and reproductive adults (N₆). We included two types of stage transitions: progression from one stage to the next (G_i), and retrogression from stages N₄, N₅, and N₆ to N₃ (R_i). The model also has a fecundity term (F) expressed as the amount of space suitable for seedling germination. Because germination space is typically occupied by litter and competing vegetation, F is > 0 only in years following a combination of flood and wildfire, as we observed along the Middle Rio Grande. G_i is a function of survival probability, (a multiple of self-thinning, drought mortality, and fire mortality) and transition probability. R_i is > 0 only after a wildfire and represents the probability of large, top-killed cottonwoods resprouting after a fire. This model can be used to compare population projections using current flood, fire, and drought conditions with those predicted by global circulation models under various carbon emission scenarios.