PATTERNS AND CONTROLS OF PRIMARY PRODUCTION IN THE PATAGONIAN STEPPE: A REMOTE SENSING APPROACH*

Abstract

We took advantage of regional gradients to study the spatial relationships between aboveground net primary production (ANPP) and climate in the Patagonian steppe of South America. We explored the same relationships through time, considering the natural variations of ANPP and climate for 11 yr. Based on NOAA/AVHRR satellite normalized difference vegetation index (NDVI) data, we evaluated the effects of climate on annual and seasonal ANPP across regional gradients of precipitation (100-500 mm/yr) and tem- perature (218-98C of annual mean). We studied ANPP climatic controls through time at four sites using NDVI and meteorological data. We used annual NDVI integral as a surrogate of annual ANPP. Annual NDVI integral increased linearly along regional gradients of precipitation, and its annual variability de- creased exponentially. Annual NDVI integral was, in most cases, unrelated to precipitation through time. We described the seasonality of ANPP using four variables derived from seasonal NDVI curves: the dates of growing season start and end, the date of maximum NDVI, and the length of the growing season. The growing season started later toward the cold extreme of the regional temperature gradients and, within a given site through time, during the coldest years. The dates of maximum NDVI and end of the growing season occurred later toward the humid or cold extremes of the regional gradients, whereas the length of the growing season was positively affected by precipitation and temperature along these gradients. These variables were not associated with climate through time. The response of the start of the growing season to temperature was greater in time, following the natural climatic fluctuations, than in space, accompanying regional temperature gradients. This difference probably resulted because the time required for shifts in community composition and plant adaptation is longer than one year. Climatic determinants of ANPP shifted from precipitation alone to precipitation plus temperature when the temporal scale of analysis changed from annual to seasonal. Our results indicate the feasibility of forecasting forage availability a few months prior to the beginning of the growing season, but not during the whole year. Longer term data sets and manipulative experiments are required to forecast annual ANPP and predict its response to climate change.