doi:10.1016/j.gca.2004.01.006 A stable isotope-based approach to tropical dendroclimatology MICHAEL N. EVANS 1, * and DANIEL P. SCHRAG 2 1 Laboratory of Tree-Ring Research, The University of Arizona, 105 W. Stadium, Tucson, AZ 85721, USA 2 Department of Earth and Planetary Sciences, Harvard University, 20 Oxford St., Cambridge, MA 02138, USA (Received June 6, 2003; accepted in revised form January 6, 2004) Abstract—We describe a strategy for development of chronological control in tropical trees lacking demon- strably annual ring formation, using high resolution  18 O measurements in tropical wood. The approach applies existing models of the oxygen isotopic composition of alpha-cellulose (Roden et al., 2000), a rapid method for cellulose extraction from raw wood (Brendel et al., 2000), and continuous flow isotope ratio mass spectrometry (Brenna et al., 1998) to develop proxy chronological, rainfall and growth rate estimates from tropical trees lacking visible annual ring structure. Consistent with model predictions, pilot datasets from the temperate US and Costa Rica having independent chronological control suggest that observed cyclic isotopic signatures of several permil (SMOW) represent the annual cycle of local rainfall and relative humidity. Additional data from a plantation tree of known age from ENSO-sensitive northwestern coastal Peru suggests that the 1997-8 ENSO warm phase event was recorded as an 8‰ anomaly in the  18 O of -cellulose. The results demonstrate reproducibility of the stable isotopic chronometer over decades, two different climatic zones, and three tropical tree genera, and point to future applications in paleoclimatology. Copyright © 2004 Elsevier Ltd 1. INTRODUCTION A particular gap in the network of high resolution proxy observations used for paleoclimate and paleoenvironmental studies is in the terrestrial tropics (World Data Center-A for Paleoclimatology, 2003; Fig. 1). One potential source of annu- al-resolution paleoclimate data lies in dendroclimatology, the science linking tree-rings to environmental conditions (Fritts, 1976). This approach requires large numbers of individuals of a single species to be sampled, statistically processed, and assembled into a site average. Yet, although exciting successes have recently been reported (Stahle, 1999; Stahle et al., 1998; D’Arrigo et al., 2001a; Biondi, 2001; Morales et al., 2001; Worbes, 2002), dendrochronology remains unestablished in much of the tropics, for several reasons. Although there are tropical tree species which exhibit annual ring structure (Worbes, 2002; Jacoby and D’Arrigo, 1990), other species or the environment do not appear to encourage annual ring for- mation (Fig. 2). Once target species are identified, finding the necessarily large sample population of annual ring-forming trees within the diversity of tropical forests (Wilson, 1988) makes species characterization and dataset development diffi- cult. For these reasons and others, attempts to develop reliable paleoclimatic time serie from tropical trees using classical ring dating methods (Stokes and Smiley, 1968) have proceeded slowly. Here we present a strategy to develop chronometric esti- mates in tropical trees lacking demonstrably annual ring struc- ture, using high resolution stable isotopic measurements in tropical woods. We call this approach “tropical isotope den- drochronology.” The approach exploits recent advances in mechanistic modeling of the oxygen isotopic composition of the -cellulose component of wood, and a rapid protocol for extracting the -cellulose component of wood from very small- samples in combination with online, continuous flow mass spectrometric techniques for necessary high-throughput analy- ses. The approach has the potential advantage of generality across species, and hypothetically permits development of proxy paleo-rainfall/humidity and growth rate data as an addi- tional output. We describe the approach and lay out the primary methodological assumptions in Section 2. We describe methods and data in Section 3 and results in Section 4. A synthesis of the results and the prospects for dendroclimatology from isotopic data developed from tropical trees is in Section 5. 2. HYPOTHESIS The basic premise of this study is that most tropical locales experience precipitation seasonality, even if they do not expe- rience the temperature seasonality which is primarily respon- sible for the reliable formation of tree-rings in the extratropics (Fritts, 1976, Fig. 2). Hence, we seek to detect seasonality in the amount of tropical convective rainfall, as mirrored in the stable isotopic composition ( 18 O) of the annual cycle in trop- ical convective rainfall (Fig. 3). Following process modeling studies of the isotopic composition of tree-ring cellulose in extratropical trees, we expect to find that the seasonality of the isotopic composition of tropical rainfall, primarily modified by evaporative effects at the leaf, is resolved by intraseasonal- resolution measurements of the  18 O of the -cellulose com- ponent of contemporaneously-formed wood. In this manner we seek to establish chronology within tropical wood samples we wish to employ for paleoclimatic study. Once chronology is established, we can proceed to the calibration and development of proxy rainfall/humidity and growth estimates from tropical trees. In a recent set of careful greenhouse, field and modeling studies, Roden and colleagues showed that the oxygen isotope composition of the -cellulose component of wood depends * Author to whom correspondence should be addressed (mevans@ltrr.arizona.edu). Pergamon Geochimica et Cosmochimica Acta, Vol. 68, No. 16, pp. 3295–3305, 2004 Copyright © 2004 Elsevier Ltd Printed in the USA. All rights reserved 0016-7037/04 $30.00 + .00 3295