INTRODUCTION The New Madrid, Missouri, earthquakes of A.D. 1811–1812 were the largest earthquakes in the conterminous United States in recorded history. Three main shocks, having estimated moment magnitudes of 8.1, 7.9, and 8.0, occurred on 16 December 1811, 23 January 1812, and 7 February 1812, respectively (Fig. 1) (Johnston, 1996; Johnston and Schweig, 1996). Damage was sustained within a radius of 600 to 700 km and was felt over an area of 5 000 000 km 2 during each of the three earthquakes (Nuttli, 1982). There were 203 damaging aftershocks through March 15, 1812, and there have been 20 damaging earth- quakes along the New Madrid seismic zone since 1812 (Nuttli, 1982). The New Madrid seismic zone remains the most seismically active area east of the Rocky Mountains (Fig. 1, inset) and poses a serious threat to the population and infra- structure of the central United States. The only known surface faulting in the New Madrid seismic zone is minor extensional and reverse faulting along the Reelfoot fault scarp (Fig. 1) (Russ, 1982; Kelson et al., 1992, 1996), although the Bootheel lineament is a possible coseismic fault (Schweig and Marple, 1991). Paleoseismic trench studies along the Reelfoot scarp have revealed evidence for two surface faulting events in the 2400 yr prior to 1811–1812 (Russ, 1982; Kelson et al., 1992, 1996). These two prehistoric events are dated as between A.D. 780 and 1000 and A.D. 1260 and 1650. On the basis of the trench studies, Kelson et al. (1996) proposed that the Reelfoot fault has a recurrence interval of 400–500 yr. Paleolique- faction studies within the southern portion of the New Madrid seismic zone reveal a minimum of three paleoliquefaction events in the 2000 yr prior to 1811 (Tuttle and Schweig, 1995; Tuttle et al., 1996; Collier et al., 1997). These events are believed to have occurred between A.D. 0 and 500, A.D. 800 and 1000, A.D. 1200 and 1400, and a possible earthquake between A.D. 1340 and 1640. At the northern end of the New Madrid seismic zone, paleoearthquakes have been dated at about A.D. 439, A.D. 539–911, and A.D. 770– 1020 (Tuttle et al., 1996). Because of this relatively short recurrence interval, we believe that tree-ring analyses may be useful in accurately dating prehistoric earth- quakes within the New Madrid seismic zone. Dendrochronology has been successfully applied to paleoseismology in Alaska (Beavan et al., 1979; Jacoby and Ulan, 1983), along the San Andreas fault in California (LaMarche and Wallace, 1972; Jacoby et al., 1988; Sheppard and Jacoby, 1989), and on the Cascadia subduction zone (Atwater and Yamaguchi, 1991; Jacoby et al., 1997). These authors documented death or detrimental effects on subsequent tree growth due to lost canopies, tree tilting, and root damage by fault rupture and marine inundation. The detri- mental effects resulted in the formation of narrow annual rings, or in some cases the complete cessation of radial growth, in the years immedi- ately following the earthquakes. We utilized dendrochronology of baldcypress (Taxodium distichum) as a paleoseismic tool for the following reasons. First, trees are sensitive to trauma caused by great earthquakes. Second, dendrochronology allows precise dating of the trauma. Finally, baldcypress are abundant in the New Madrid seismic zone, they provide a long record, and there are established regional chronologies already available (Stahle et al., 1985; Stahle and Cleaveland, 1992). We ana- lyzed tree-ring cores from baldcypress trees in Reelfoot Lake of northwestern Tennessee (Stahle et al., 1992) and the St. Francis sunkland of northeastern Arkansas (Fig. 1) to determine if the tree rings recorded effects of the great 1811–1812 earthquakes. Old portions of the cores were examined for similar tree-ring effects that might have resulted from prehistoric earthquakes. GEOLOGIC SETTING OF THE NEW MADRID SEISMIC ZONE The New Madrid seismic zone is within the upper portion of the Mississippi embayment, a broad and gentle south-southwest–plunging syn- cline of Cretaceous and Tertiary age (Stearns, 1957; Cox and Van Arsdale, 1997) that overlies the seismically active late Precambrian and early Geology; June 1998; v. 26; no. 6; p. 515–518; 2 figures. 515 Earthquake signals in tree-ring data from the New Madrid seismic zone and implications for paleoseismicity Roy B. Van Arsdale Department of Geological Sciences, University of Memphis, Memphis, Tennessee 38152 David W. Stahle Malcolm K. Cleaveland Tree-Ring Laboratory, Department of Geography, University of Arkansas, Fayetteville, Arkansas 72701 Margaret J. Guccione Department of Geology, University of Arkansas, Fayetteville, Arkansas 72701 ABSTRACT Severe ground shaking and the formation of Reelfoot Lake during the great New Madrid earthquakes of A.D. 1811–1812 had a profound effect on baldcypress trees that still survive in Reelfoot Lake of northwestern Tennessee. Inundation greatly increased baldcypress radial growth from 1812 to 1819 and permanently decreased wood density after 1811. Ground shaking fractured the baldcypress stems that were present during the 1811–1812 event, but fractures are absent in the post-1811 growth. In contrast, the growth of old baldcypress trees in the St. Francis sunkland of northeastern Arkansas was severely suppressed for almost 50 yr following the 1811–1812 New Madrid earthquakes. Thus, there are two opposite but profound growth responses to the same earthquake events preserved in baldcypress trees of the New Madrid seismic zone. The tree-ring chronology at Reelfoot Lake extends from A.D. 1682 to 1990, but the 1812–1819 growth surge was the only extreme growth anomaly in this 309 yr period. The St. Francis sunk- land chronology extends from A. D. 1321 to 1990, and the 1812–1857 growth suppression is the most severe and prolonged growth anomaly of this entire 670 year period. Thus, the tree-ring record indicates that there was not a great earthquake during the 129 yr prior to 1811 in the Reelfoot Lake basin, nor during the 490 yr prior to 1811 in the St. Francis sunkland.