Clim Dyn DOI 10.1007/s00382-015-2694-y
Patterns of moisture source and climate variability in the southeastern United States: a four‑century seasonally resolved tree‑ring oxygen‑isotope record D. M. Labotka1 · H. D. Grissino‑Mayer2 · C. I. Mora3 · E. J. Johnson4
Received: 13 November 2014 / Accepted: 29 May 2015 © Springer-Verlag Berlin Heidelberg 2015
Abstract This study presents a climate reconstruction utilizing a seasonally resolved 417-year oxygen-isotope record of tree rings from southern Georgia, United States (1580–1997 CE). Oxygen isotopes within the cellulose predominately reflect moisture source observed on a seasonal scale between earlywood and latewood growth. Signatures of large climate oscillations were captured in modern and subfossil wood. Spectral and wavelet transform analyses of seasonally resolved oxygen isotopes showed distinct periodicities coinciding with the Atlantic multidecadal oscillation and other major climate oscillation phenomena. Oxygen-isotope values in latewood growth revealed a significant correlation with North Atlantic sea surface temperature anomalies. This correlation suggests that the precipitation source was strongly influenced by fluctuations in the Atlantic multidecadal oscillation and teleconnections with other major climate phenomena such as the North Atlantic subtropical high-pressure system, El Niño Southern Oscillation, and Pacific Decadal Electronic supplementary material The online version of this article (doi:10.1007/s00382-015-2694-y) contains supplementary material, which is available to authorized users. * D. M. Labotka
[email protected] 1
Illinois State Geological Survey, Prairie Research Institute, University of Illinois, 615 E. Peabody Drive, Champaign, IL 61820, USA
2
Department of Geography, University of Tennessee, Knoxville, TN 37996‑0925, USA
3
Earth and Environmental Sciences Division, Los Alamos National Laboratory, Los Alamos, NM 87545, USA
4
Department of Earth and Planetary Sciences, University of Tennessee, Knoxville, TN 37996, USA
Oscillation. These results emphasize the utility of oxygen isotopes in tree rings for revealing seasonal influences associated with major climate drivers over centuries and enhance our understanding of long-term climate behavior on a detailed scale. Keywords Tree rings · Oxygen isotopes · Climate oscillations · Paleoclimate
1 Introduction Numerous studies have demonstrated the importance of understanding climate dynamics associated with ocean– atmosphere interactions (i.e., Enfield et al. 2001; Nogueira et al. 2013). Long-term climate records stored in paleoproxies elucidate the connection between modern climate and paleoclimate thus facilitating the understanding of mechanisms driving climate change. Trees are high-resolution in situ recorders of climate and capture interactions among the atmosphere, ocean, and biosphere. Stable isotope analysis of tree-ring cellulose is a fundamental tool for understanding past trends in climate. Oxygen isotopes, in particular, offer detailed insights into source water and precipitation, allowing reconstructions of climate phenomena on a seasonal scale. Oxygen isotopes in tree rings reflect the isotope value of source water. Three main isotopic influences are reflected in the oxygen isotope value of plant cellulose: (1) the δ18O value of the source water, (2) the effects of evaporative enrichment of leaf water due to transpiration and exchange, (3) and the biochemical fractionation (27 ‰) that occurs during photosynthesis (Barbour et al. 2001). The δ18O of precipitation and humidity is the dominant signal recorded in tree-ring cellulose (Epstein et al. 1977;
13
D. M. Labotka et al.
McCarroll and Loader 2004; Treydte et al. 2014). If source water is related to precipitation, then oxygen isotopes record changes in the precipitation source. Recent studies have demonstrated this utility and advanced our understanding of climate oscillations that drive precipitation patterns, such as the El Niño Southern Oscillation (ENSO) and the North Atlantic Oscillation (NAO) (i.e., Brienen et al. 2012; Xu et al. 2014). Few studies, however, have extended tree-ring oxygen-isotope chronologies beyond 200 years, and even fewer have addressed precipitation changes associated with the Atlantic multidecadal oscillation (AMO). Many of these climate oscillations fluctuate on multidecadal time scales. Thus, temporally extensive chronologies are essential for understanding their longterm behavior. In this study, a 417-year, seasonally resolved tree-ring record is evaluated using stable oxygen-isotope variations in the earlywood and latewood growth of longleaf pine trees from southern Georgia. Statistical, spectral, and wavelet transform analyses of oxygen-isotope time series reveal significant periodicities and correlations on a seasonal basis. This study demonstrates the impact of climate oscillations and their teleconnections on precipitation over the last four centuries for the southeastern United States.
2 Materials and methods 2.1 Overview A continuous 417-year tree-ring oxygen isotope record (δ18O) of seasonally resolved growth from longleaf pine (Pinus palustris Mill.) modern and subfossil wood (1580‒1997 CE) from southern Georgia is presented. This data set is compared with Kaplan’s extended SST version 2 data set (1856–1997) of gridded sea surface temperature anomalies (SSTAs; http://www.esrl.noaa.gov/psd/ data/timeseries/AMO/; Kaplan et al. 1998). The monthly data set was averaged into growing seasons, for comparison with seasonal tree-growth components. The growing season for these longleaf pines is approximately February–June for earlywood growth and July–November for latewood growth (Meldahl et al. 1999). Earlywood growth for these longleaf pines is most strongly correlated with current spring precipitation, and latewood growth is correlated with current summer precipitation (Henderson and Grissino-Mayer 2009). The latewood oxygen isotopes are compared with tree-ring width reconstructed SSTAs of the AMO covering 1580–1990 from the data set published by Gray et al. (2004). The following methods describe the collection of tree rings and the stable isotope and statistical analyses.
13
2.2 Sampling and isotope analysis Cross-sectional slabs were collected from felled trees and remnant stumps of longleaf pine trees discovered in or near Lake Louise (30.43°N, 83.25°W). The shallow root system of these trees minimizes the potential for groundwater influences on the oxygen-isotope signatures (Waterhouse et al. 2002). The taproot depth for longleaf pines in deep sandy soils can extend to 4 m, whereas the fine root distribution generally extends to less than 2 m (Addington et al. 2006; Ford et al. 2008). The potentiometric surface for the Upper Floridan aquifer in the Valdosta area is 15–24 m above sea level, and the average elevation for Lowndes County is 40–70 m (Long and Kirk 1920; Burgoon 1991; Peck et al. 2011). Therefore, the source water for these trees was likely soil water that was directly related to precipitation and not groundwater (Anderson et al. 2002). These longleaf pines grew in moderate to well-drained loamy sand soils and produce consistent annual growth rings (Grissino-Mayer et al. 2001). According to Climate Division 8 data for Georgia, the seasonal precipitation average for earlywood growth (February–June 1901–1974) is 52 cm and that for latewood growth (July–November 1901–1974) is 49 cm. The seasonal temperature average for earlywood growth (February–June 1901–1974) is 19 °C, and that for latewood growth (July–November 1901–1974) is 22 °C (http://www. ncdc.noaa.gov/cdo-web/datasets). Subfossil and modern slabs of longleaf pine were dated using standard cross-dating techniques (Stokes and Smiley 1996; Grissino-Mayer et al. 2001). Earlywood and latewood segments of each annual ring were sampled with a scalpel and sliced into ~40-μm slivers. Seasonal growth segments were readily distinguished by cell wall thickness and color. Alpha-cellulose was extracted from each whole wood seasonal segment by using Soxhlet extraction methods (Green 1963; Leavitt and Danzer 1993; Loader et al. 1997). Juvenile tree effects do not significantly alter oxygen isotopes of the tree rings (Anderson et al. 1998); therefore, no statistical or methodological treatment is necessary. Oxygen-isotope compositions of 80–100 µg of α-cellulose were analyzed with a Thermo-Finnigan high-temperatureconversion elemental analyzer interfaced with a Thermo MAT Delta Plus continuous-flow isotope-ratio mass spectrometer. Analyses were reported in per mil (‰) relative to Vienna-Standard Mean Ocean Water (V-SMOW; see Online Resource TRO.xlsx). An in-house cellulose standard (Sigma, St. Louis, MO) calibrated to an NBS-19 carbonate standard was run routinely with samples. Some samples were run in duplicate (seasonal components for years 1650–1770 CE), and the majority of samples were run in triplicate. The 2σ standard deviation for all triplicate samples is 0.33 ‰, and the error for duplicates is the same.
Patterns of moisture source and climate variability in the southeastern United States: a…
2.3 Statistical analysis
3 Results
We calculated the seasonal AMO SSTAs for latewood (July–November) and earlywood (February–June) growing seasons by using Kaplan’s extended SST version 2 data set (Kaplan et al. 1998), averaging the seasonal indices for each year over the period 1856–1997. Correlations between the seasonally resolved oxygen isotopes of the tree rings and the seasonally resolved AMO SSTAs were calculated by applying regression analysis to 5-year averaged segments of each corresponding time series. Positive and negative phases of the AMO are based on the time periods defined in Alexander et al. (2014): positive AMO phases are 1875–1899, 1926–1965, and 1995–present, and negative AMO phases are 1900–1925 and 1965–1994. Tropical cyclones that occurred over the time period 1856–1997 and that passed within a 200-mile radius of the study area are documented (Landsea et al. 2004). Tropical cyclones for this study are limited to tropical storms (maximum sustained wind speeds of 63–118 km/h) and hurricanes (maximum sustained wind speeds of ≥119 km/h; http://www.nhc.noaa.gov/climo/). The time series data were analyzed by spectral analysis (REDFIT; Schulz and Mudelsee 2002) and continuous wavelet analysis (Torrence and Compo 1998). REDFIT, an open access FORTRAN 90 program available from the National Oceanographic and Atmospheric Administration (NOAA) at the National Climatic Data Center (NCDC), was used to analyze the seasonally resolved oxygen-isotope time-series data. REDFIT estimates red noise spectra without requiring interpolation. The univariate spectra were biased-corrected using 1,000 Monte Carlo simulations. A 95 % false-alarm level (~confidence level) was established using a χ2 distribution of automatically computed autoregression [1 year (AR1)] values. The 95 % confidence level (CL) line shows the spectral values that exceed the rednoise false-alarm level. Continuous wavelet analysis uses a Gaussian (m = 2, Mexican hat) waveform that uses real values separating the negative and positive oscillations of a time series into individual peaks in wavelet power. Therefore, the Gaussian wavelet reveals fine-scale features of the time series. Wavelet analysis was applied to the oxygen-isotope latewood (1580–1997 CE) time series, the AMO paleoproxy reconstruction data from Gray et al. (2004), and the AMO unsmoothed SST version 2 data set (1856–1997; Kaplan et al. 1998). Seasonal oxygen-isotope data were standardized to a zero mean and normalized by 1/σ. Gaussian wavelet analysis models were padded with zeroes to minimize wraparound effects attributable to a finite data set, and a 95 % significance level was calculated against a red-noise background. A cone of influence was applied to the wavelet spectra to emphasis areas susceptible to the effects of zero padding (Torrence and Compo 1998).
The earlywood oxygen-isotope time series in these analyses routinely exhibited 18O-enriched values compared with latewood (~2 ‰; Fig. 1). The seasonal oxygen-isotope average over the 417-year time period for earlywood is 34.7 ‰ and for latewood is 32.7 ‰. Regression analysis between the oxygen-isotope time series of earlywood and latewood growing seasons over the period 1580–1997 CE showed a significant correlation of r = 0.20 (p
