Precipitation signal in pollen rain from tropical forests, South India

We have analyzed the pollen content of 51 surface soil samples collected in tropical evergreen and deciduous forests from the Western Ghats of South I ndia sampled along a west to east gradient of decreasing rainfall (between 11 degrees 30-13 degrees 20 'N and 75 degrees 30-76 degrees 30 'E). Values of mean annual precipitation (Pann, mm/yr) have been calculated at each of the 51 sampling sites from a great number of meteorological stations in Sou th India, using a method of data interpolation based on artificial neural n etwork. Interpolated values at the pollen sites of Pann range from 1200 to 5555 mm/yr, while mean temperature of the coldest month (MTCO) remains > 15 degreesC and humidity factor (AET/PET, the actual evapotranspiration to po tential evapotranspiration ratio) remains also included between 65 and 72%. Results are presented in the form of percentage pollen diagrams where sampl es are arranged according to increasing values of annual precipitation. The y indicate that the climatic signal of rainfall is clearly evidenced by dis tinct pollen associations. Numerical analyses show that annual precipitatio n is an important parameter explaining the modern distribution of pollen ta xa in this region. Pollen taxa markers of high rainfall (Pann > 2500 mm/yr) are Mallotus type, Elaeocarpus, Syzygium type, Olea dioica, Gnetum ula, an d Hopea type, associated with Ixora type and Caryota. Pollen taxa markers o f low rainfall (Pann < 2500 mm/yr) are Melastomataceae/Combretaceae, Mayten us type, Lagerstroemia and Grewia. The proportions of evergreen taxa and of arboreal taxa vary according to rainfall values. Indeed, when rainfall is < 2500 mm/yr, percentage of arboreal pollen (AP) is < 50% and proportion of evergreen taxa is < 20%. When rainfall exceeds 2500 mm/yr, AP values avera ge 70%, and proportion of evergreen taxa increases from 60 to 90%. Moreover , a,good correlation between precipitation and proportion of evergreen taxa (0.85) presumes that precipitation can be estimated from pollen data. (C) 2001 Elsevier Science B.V. All rights reserved.