Abstract

This paper presents a multicentury reconstruction of May precipitation (1200–1997) for the mid-Atlantic region of the United States. The reconstruction is based on the first principal component (PC1) of two millennial-length Juniperus virginiana L. (eastern red cedar) tree-ring chronologies collected from rocky, limestone sites in the Ridge and Valley province of West Virginia. A split-calibration linear regression model accounted for 27% of the adjusted variance in the instrumental record and was stable through time. The model was verified by the reduction of error (RE = 0.21) and coefficient of efficiency (CE = 0.20) statistics. Multidecadal changes in precipitation were common throughout the reconstruction, and wetter than median conditions and drier than median conditions occurred during the medieval climate anomaly (1200–1300) and the Little Ice Age (1550–1650), respectively. The full reconstruction contained evidence of interannual and decadal variability; however, the twentieth century recorded the greatest number of decadal extreme wet and dry periods. A comparison of the May precipitation reconstruction to other regional reconstructions [Potomac River, Maryland, streamflow (Cook and Jacoby); Virginia/North Carolina July Palmer hydrologic drought index (PHDI; Stahle et al.); Missouri July PHDI (Cleaveland and Stahle); and White River, Arkansas, streamflow (Cleaveland)] showed that the eastern U.S. decadal drought and pluvial events extended into the mid-Atlantic region. A positive correlation between PC1 and the winter North Atlantic Oscillation (NAO) index and comparisons of smoothed May precipitation and the NAO (Luterbacher et al.) indicated that J. virginiana’s response to May precipitation was mediated by winter temperature.

1. Introduction

The mid-Atlantic region (MAR) of the United States is home to ~15% (35.2 million) of the country’s population, with about two-thirds of that population living in urban areas (e.g., Philadelphia; Washington, D.C.) near the coast or large rivers (Polsky et al. 2000). While water resources in the region are generally abundant, periodic drought and pluvial events require careful management of water resources and the surrounding watersheds to minimize the negative impact of changes in water quantity and quality (Neff et al. 2000; Najjar et al. 2000; Cook et al. 2007). Water managers face additional challenges because the MAR is predicted to become warmer and wetter in the coming decades (Polsky et al. 2000; Solomon et al. 2007; Cook et al. 2010). The expected changes in climate and hydrology, in combination with a projected increase in population, will place greater demand on the region’s water resources and impact ecosystem services from the region’s forests, wetlands, and freshwater and coastal ecosystems (McKenney-Easterling et al. 2000; Najjar et al. 2000; Rogers and McCarty 2000; Lenihan et al. 2008). Increases in precipitation, seasonal streamflow, and a shift to earlier streamflow peaks in the winter and spring are projected (Neff et al. 2000); however, it is not known if these changes are within the natural range of variability for the region. Currently, water resource managers in the MAR rely on instrumental records of precipitation and streamflow (<120 yr) that are not adequate to assess the variability in water quantity within the context of past centuries (Stockton and Jacoby 1976; Meko et al. 1995; Woodhouse and Lukas 2006). Therefore, any information on past variability in spring precipitation may be used to better prepare for future changes in water quantity and quality.

The development of tree-ring reconstructions of precipitation, drought, snowfall, and streamflow is well established, and reconstructions are commonly used to evaluate changes in water resources over the centuries (Stockton and Jacoby 1976; Graumlich 1987; Stahle and Cleaveland 1992; Pederson et al. 2001; Meko et al. 2001; Woodhouse 2003). However, annually resolved multicentury proxy records of hydroclimate are rare in eastern North America because suitable sources of high-resolution natural archives (i.e., old-growth tree species) have been destroyed by human land alteration or are limited by tree species biology. An emerging source of climate history for the MAR is Juniperus virginiana L. (eastern red cedar), which grows on moisture-limited limestone sites near the headwaters of the Potomac River. Hawley (1937) was the first to publish on the utility of long-lived (500+ yr), climate-sensitive J. virginiana trees growing in eastern Tennessee. In her work, Hawley found a positive correlation (r = 0.69) between annual growth and water year precipitation. Guyette et al. (1980) further investigated the potential of using J. virginiana annual growth for reconstructing climate in Missouri. The authors found positive correlations between ring-width index and growing season precipitation (r = 0.48) and a negative correlation with early growing season temperature (r = −0.54). In southwest Virginia, Larson (1997) found a positive correlation between J. virginiana annual increment and total May–July precipitation (r = 0.56), stating that the species increases in growth during cool and wet springs. Juniperus virginiana trees sampled for the studies described above were all sampled on similar cliff, bluff, knob, or barren sites throughout the eastern United States that are 1) primarily limestone based with poor soil development; 2) south–southwest facing; 3) steep sloped; 4) sparsely populated by other trees and understory vegetation; and 5) mostly undisturbed by fire, logging, and grazing. The previous research suggests that J. virginiana is well suited for reconstruction of climate variables because it is long-lived and sensitive to variations in temperature and precipitation. Because J. virginiana is the most widely distributed conifer in the eastern United States (Lawson 1990), additional dendroclimatic sites likely exist where limestone is the primary component of the soil and trees remain undisturbed.

In this paper, we present two multicentury chronologies developed from subfossil wood and long-lived J. virginiana trees growing on moisture-limited sites in the Ridge and Valley physiographic province of West Virginia. The first chronology was developed explicitly for this study. The second chronology (Cedar Knob) was used previously as a predictor chronology in Cook et al.’s (1999) reconstruction of the Palmer drought severity index (PDSI) for the continental United States and Cook et al.’s (2002) multiproxy reconstruction of the winter North Atlantic Oscillation (NAO) index, but it has not been investigated as a predictor in a single-species reconstruction for the MAR. We examine both chronologies’ dendroclimatic response to regional temperature and precipitation, reconstruct May precipitation from 1200–1997, and evaluate the temporal variation in the reconstruction for decadal to half-centennial drought and pluvial events. Then, we compare the reconstruction to four regional proxy records of moisture, including 1) Cook and Jacoby’s (1983) Potomac River, Maryland, streamflow (1983); 2) Stahle et al.’s (1998) Virginia/North Carolina July Palmer hydrologic drought index (PHDI); 3) Cleaveland and Stahle’s (1994) Missouri July PHDI; and 4) Cleaveland’s (2000) White River, Arkansas, streamflow. Finally, we discuss the winter North Atlantic Oscillation teleconnection in the MAR and factors mitigating J. virginiana’s dendroclimatic response.

2. Data

a. Tree-ring data

Increment cores from live J. virginiana trees and cross sections from subfossil wood were collected from two moisture-limited locations (i.e., Smoke Hole Canyon and Cedar Knob) in the Ridge and Valley province of West Virginia. Sample locations were previously identified as cedar or limestone glades by Bartgis (1993) and are characterized by shallow, rocky soils with sparse annual and perennial understory vegetation. Sampling in Smoke Hole Canyon (38°53′02″N, 79°14′12″W; Fig. 1) occurred in 2006 and 2007 at three locations with similar site conditions. All locations are ~650 m MSL, southwest facing, and moderate to steep in slope (>25°). Tree-ring samples were returned to the laboratory and prepared and cross-dated using standard dendrochronological techniques (Speer 2010). While several hundred increment cores and cross sections were collected in Smoke Hole, we eliminated live samples <250 yr old and cross sections <150 yr old from the chronology to retain low-frequency signals (Cook et al. 1995). In total, 48 increment cores from 36 live trees and 80 radii from 45 cross sections were used. Samples cross-dated well between Smoke Hole locations with a 0.536 series intercorrelation and 0.326 average mean sensitivity (Grissino-Mayer 2001). Because tree growth between sites was well correlated, a Smoke Hole chronology (517–2007) was created by merging tree-ring samples from the three collection sites.

Fig. 1.

Map of the MAR, including the Smoke Hole and Cedar Knob J. virginiana tree-ring sampling locations.

Fig. 1.

Map of the MAR, including the Smoke Hole and Cedar Knob J. virginiana tree-ring sampling locations.

Tree-ring data from a second location, Cedar Knob, 20 km south of Smoke Hole were collected in 1998 (38°39′30″N, 79°22′45″W; Fig. 1) and have been used in previous dendroclimatic research reconstructing a nationwide PDSI grid (Cook et al. 1999) and winter NAO index (Cook et al. 2002). The Cedar Knob chronology (481–1998) is a composite of three sites located within 3 km of each other. All the Cedar Knob collection sites share similar slope, aspect, and soil conditions to the Smoke Hole sites but are higher in elevation (~850 m MSL). Similar to Smoke Hole, hundreds of samples were collected at Cedar Knob but only longer segments were retained (see above). In total, 15 increment cores from 11 live trees and 137 radii from 98 cross sections were used. Samples from the Cedar Knob sites cross-date well across sites with a 0.55 series intercorrelation and 0.31 average mean sensitivity.

Individual raw ring-width series from the two sample locations were standardized in the program ARSTAN with a 300-yr smoothing spline (50% frequency response cutoff) to remove biological growth trends and preserve annual- to centennial-scale climate signals (Cook 1985). Low-order autocorrelation (i.e., lag 1 and lag 2) was removed from each series with an autoregressive model. Then, individual series were averaged together to create ARSTAN chronologies for each location (Fig. 2). The ARSTAN chronologies were calculated by adding the pooled autoregression back into the residual chronologies and thus, low-frequency variability was reintroduced. The variance of the ARSTAN time series was stabilized using the Briffa RBAR-weighted method to account for changes in variance due to the reduction in sample size backward in time (Osborn et al. 1997). The expressed population signal (EPS) was calculated for each chronology using 50-yr segments overlapped 25 yr. The EPS is a measure of the common variance in a chronology and weakens as the sample size decreases (Briffa 1984; Wigley et al. 1984). While no rule exists for use of the EPS, Wigley et al. (1984) suggested that chronologies could be truncated when the EPS drops below 85%. We chose to restrict our analyses when the EPS for one of the chronologies (i.e., Smoke Hole) dropped below 85% at year 1200.

Fig. 2.

ARSTAN chronologies for the (a) Smoke Hole and (b) Cedar Knob locations with sample size (dashed line). A 10-yr fourth-order smoothing spline is used to highlight decadal trends (thick black line) in each time series.

Fig. 2.

ARSTAN chronologies for the (a) Smoke Hole and (b) Cedar Knob locations with sample size (dashed line). A 10-yr fourth-order smoothing spline is used to highlight decadal trends (thick black line) in each time series.

The Smoke Hole and Cedar Knob Arstan chronologies are correlated with each other (r = 0.83) during the period of the instrumental record (1895–1997) and an initial correlation function analysis between individual chronologies and regional climate data showed similar responses to monthly temperature and precipitation (data not shown). Therefore, we elected to use a principal components analysis (PCA) to combine the chronologies into a single predictor variable. PCA is widely used in dendroclimatic analysis to reduce the number of predictor chronologies by extracting a common pattern based on similar variance characteristics between variables that are collinear (e.g., Briffa 1995; Hidalgo 2004; Timilsena et al. 2007). Our primary concern was to reduce the collinearity between predictor chronologies and create a linear combination of the original variables that maintained the greatest possible variance in the tree-ring data. The PCA of the ARSTAN chronologies resulted in one principal component (PC) with an eigenvalue > 1 (Kaiser–Guttman rule) that explained 81% of the variance in the predictors (Guttman 1954; Kaiser 1960). For subsequent dendroclimatic analyses, we used the first principal component (PC1) and PC1t+1 (lagged) as the primary predictor variables.

b. Climate data

Monthly precipitation and temperature data (1895–1998) were obtained from the National Climatic Data Center for West Virginia climate region 6 (National Climatic Data Center 2009). Region 6 includes the portion of West Virginia in the Ridge and Valley physiographic province and drains into the Chesapeake Bay. The majority of the region lies within the rain shadow of the Appalachian Mountains, receiving <90 cm of precipitation annually. While the region does not have defined wet and dry seasons, precipitation peaks in summer months (mean = 92.35 mm) with a corresponding peak in temperature (mean = 21.8°C). Both precipitation and temperature reach their seasonal minima during the winter months, averaging 59.4 mm and 2.2°C, respectively.

The climate of the MAR is influenced by three large-scale circulation patterns emanating from the continental landmass to the west, the polar region to the north, and the Atlantic Ocean to the east (Barry and Chorley 2003). In the spring, temperature in the eastern portion of West Virginia is controlled largely by the northward movement of the polar jet stream rather than the maritime circulation patterns coming off the Atlantic Ocean. Spring precipitation, however, is influenced by moisture from the Gulf of Mexico and the Atlantic Ocean, and less so by continental systems. Because of the orographic effect of the Appalachian Mountains, much of the Ridge and Valley province lies within a rain shadow and receives less precipitation than coastal areas (National Climatic Data Center 2009). The NAO is the most prominent and recurrent pattern of atmospheric variability in the Northern Hemisphere (Hurrell et al. 2003), affecting temperature and precipitation in the MAR. The NAO index is defined by the seesaw of atmospheric mass between the Icelandic low (northern pole) and the Azores high (southern pole) and is climatologically strongest during the cold-season months, from November to April (Jones et al. 1997; Cook et al. 2002; Hurrell et al. 2003). In the eastern United States, the negative winter NAO results in more frequent outbreaks of cold-air temperatures and frozen precipitation in winter, while the positive winter NAO contributes to mild, wet weather conditions in winter (Hartley and Keables 1998; Hartley 1999; Hurrell et al. 2003). Additionally, increased precipitation and runoff in eastern North America may have some relationship to the NAO but the association has not been well defined (Perreault et al. 1999).

Correlation function analysis between PC1 and monthly climate data was performed in the program DendroClim 2002 to identify a suitable period for a precipitation reconstruction (Biondi 1997; Biondi and Waikul 2004). DendroClim 2002 implements bootstrapped (1000×) correlation and response functions to identify significant climatic signals in tree-ring data using static, evolutionary, and moving intervals. Evolutionary and moving intervals allow the user to test for temporal changes of dendroclimatic relationships. To examine the lag effect of climate on tree growth, the analysis window was extended from May of the previous year to October of the current year.

Annual tree growth of J. virginiana depends on cool, moist May conditions; mild winter and early spring temperatures; and partly on previous growing season precipitation (Fig. 3a). The relationship between J. virginiana tree growth and May precipitation and temperature has been reported previously by Guyette et al. (1980) and Larson (1997). A positive relationship (r = 0.41, p < 0.05) between growth and May PDSI signaled that the annual increment of J. virginiana was integrating the effects of the previous month’s temperature and precipitation. A preliminary reconstruction of May PDSI, using similar methods to those described in the next section, revealed a weak calibration (r2adj = 0.16; 1895–1997; data not shown) with the instrumental record, and PDSI was excluded from further analysis.

Fig. 3.

(a) Correlation analysis results of PC1 and WV climate region 6 temperature, precipitation, and PDSI. Capitalized months are from the previous year. Values exceeding the dashed lines are significant at p < 0.05. (b) Correlation analysis between PC1 and May precipitation using a 45-yr moving window (black line) to assess time stability of the relationship and the instrumental winter NAO (Jones et al. 1997) index smoothed with a 10-yr fourth-order spline (gray line). Dashed line is the mean correlation coefficient between May precipitation and PC1. Correlation was significant (p < 0.05) for the entire instrumental period (1895–1997).

Fig. 3.

(a) Correlation analysis results of PC1 and WV climate region 6 temperature, precipitation, and PDSI. Capitalized months are from the previous year. Values exceeding the dashed lines are significant at p < 0.05. (b) Correlation analysis between PC1 and May precipitation using a 45-yr moving window (black line) to assess time stability of the relationship and the instrumental winter NAO (Jones et al. 1997) index smoothed with a 10-yr fourth-order spline (gray line). Dashed line is the mean correlation coefficient between May precipitation and PC1. Correlation was significant (p < 0.05) for the entire instrumental period (1895–1997).

The response of J. virginiana to May climate on our sites is likely a function of microclimate and site conditions. In a physiological study of J. virginiana growing in old fields in southern Illinois, Ormsbee et al. (1976) found that photosynthesis decreased from the maximum rate when the air temperature rose above 20°C. During clear, dry days, the rocky soils of cedar glades and barrens absorb additional radiation, causing an increase in evapotranspiration and a decrease in photosynthesis. Cool, moist May days allow J. virginiana trees to take advantage of the available moisture prior to runoff or evaporation. Also, warmer winter and early spring temperatures appear beneficial to the annual increment of J. virginiana on our sites. Juniperus virginiana has been shown to photosynthesize in winter and spring months when the temperature rises above freezing (Ormsbee et al. 1976). The ability to photosynthesize when mild winter–spring temperatures prevail may allow J. virginiana to build carbohydrates and take advantage of favorable growing conditions in May. The correlation between PC1 and May precipitation (r = 0.44) was further investigated in DendroClim 2002 using a 45-yr moving window to assess the strength of the correlation through time (Fig. 3b). The correlation coefficient represented the leading year of the window, and bootstrapped confidence intervals were used to test significance (Biondi 1997). For example, the correlation coefficient for year 1965 represents the correlation over the window from 1921 to 1965. While the correlation weakens through the late 1950s into the early 1960s, the relationship remains significant (p < 0.05) throughout the entire instrumental period, and the May precipitation was deemed suitable for reconstruction.

3. Calibration and verification of the reconstruction model

May precipitation was examined for normality prior to the calibration of the reconstruction model. Visual inspection of the normal quantile plot and a test of goodness of fit using the Shapiro–Wilk W test (W = 0.97, p < 0.01) revealed that the May precipitation is not adequately modeled by the normal distribution. May precipitation was log transformed to meet the assumptions of linear regression, and predicted values were back-transformed following verification of the model. Additionally, least squares regression was used to assess an apparent increasing trend in the instrumental record of precipitation (Fig. 4). May precipitation in region 6 has increased 0.23 mm yr−1 from 1895 to 1997. The increase in precipitation is consistent with regional trends in observational data and predictive modeling (Polsky et al. 2000; Solomon et al. 2007).

Fig. 4.

Calibration and verification of modeled May precipitation (black line) compared to instrumental record (gray line). Reconstructed values were back-transformed from log values and then scaled to the mean and variance of the instrumental period. Trend line shows the 0.23 mm yr−1 increase in instrumental May precipitation from 1895 to 1997; significant at p < 0.001.

Fig. 4.

Calibration and verification of modeled May precipitation (black line) compared to instrumental record (gray line). Reconstructed values were back-transformed from log values and then scaled to the mean and variance of the instrumental period. Trend line shows the 0.23 mm yr−1 increase in instrumental May precipitation from 1895 to 1997; significant at p < 0.001.

The instrumental dataset was split into two periods (1895–1945 and 1946–97) to test the quality and stability of the calibration models (Fritts 1976; Cook and Kairiukstis 1990). A stepwise multilinear regression (F to enter = 0.25; F to remove = 0.10) was used to predict log May precipitation with PC1 and PC1t+1 as the predictors. The term PC1t+1 (p > 0.87) was removed from the calibration model from 1895 to 1945 but was significant when calibrating from 1946 to 1997 (p < 0.05) and from 1895 to 1997 (p < 0.01). The reconstruction model based on the entire instrumental period (1895–1997) for year t is

 
formula

where xt and xt+1 are PC1 and PC1t+1 for a given year, respectively. The model explained 27% of the adjusted variance for the period from 1895 to 1997. The calibration models were verified with two rigorous tests of fit, the reduction of error (RE) statistic and the coefficient of efficiency (CE; Fritts 1976; Cook et al. 1999). The RE ranges from −∞ to +1. When the RE exceeds zero, the calibration model shows greater skill than the mean of the instrumental data from the calibration period. The CE has the same range and calculation, except the CE relies on the verification period mean as a baseline of predictive skill. The RE and CE statistics verified the models (RE = 0.20–0.29; CE = 0.15–0.25) across the two periods, indicating that the models were not affected by the increasing trend in May precipitation over the last century (Table 1; Fig. 4). Back-transforming and scaling the reconstruction forced the proxy record to have the mean and variance of the instrumental record. A closer examination of the calibration showed that the reconstruction replicated some multiyear events, such as the 1960s drought. Individual drought years (e.g., 1911, 1964, and 1977) were replicated; however, large pluvial years were not consistently replicated in the reconstruction. The lack of pluvial replication is a common occurrence in tree-ring reconstructions because of a physiological limitation in water uptake during brief, heavy rain events (Fritts 1976).

Table 1.

Calibration and verification statistics for the May precipitation models. RMSE = root-mean-square error. All statistics calculated on log-transformed data.

Calibration and verification statistics for the May precipitation models. RMSE = root-mean-square error. All statistics calculated on log-transformed data.
Calibration and verification statistics for the May precipitation models. RMSE = root-mean-square error. All statistics calculated on log-transformed data.

The current reconstruction of May precipitation explained less of the variability in the instrumental record compared to Cook et al.’s (1999) reconstruction of summer PDSI for a grid point representing the mid-Atlantic region. The Cook et al. drought reconstruction explained 56% of the variance in the instrumental record and was more strongly verified (RE = 0.53; CE = 0.49). However, the Cook et al. reconstruction utilized not only the J. virginiana chronologies from Cedar Knob but also included 29 additional chronologies in the common period (1698–1978) from multiple species growing in the region. When using just J. virginiana, the drought reconstruction explained 18% of the variance in the instrumental record, which is weaker than the current May precipitation reconstruction and on par with the preliminary May PDSI reconstruction (r2adj = 0.16) described in the previous section. In comparison to other single-species reconstructions in the eastern United States, our calibration model was comparable. Buckley et al. (2004) used ring widths from Thuja occidentalis L. to reconstruct summer precipitation for southern Ontario back to a.d. 610. Principal components were extracted from seven T. occidentalis chronologies to explain 21% of the variance in the twentieth century (33% from 1900 to 1960). Stahle et al. (1988) reconstructed June PDSI (372–1985) for North Carolina from a single Taxodium distichum chronology. The calibration model explained 47% of the adjusted variance in the instrumental record. A second reconstruction of July PHDI in southeastern Virginia, using two T. distichum chronologies, explained 44% of the adjusted variance in the calibration period (Stahle et al. 1998). An expanded network of moisture-sensitive J. virginiana chronologies could be used to improve the reconstruction model for the MAR. Given the widespread distribution of J. virginiana, additional sampling sites likely exist throughout the eastern United States.

4. Reconstructed May precipitation

The full multicentury reconstruction (1200–1997) of May precipitation contained interannual, decadal, and multidecadal variability (Fig. 5). Ranking of the top 10 driest and wettest nonoverlapping decadal events in the reconstruction showed that the twentieth century had some of the most extreme events in the proxy record, with three droughts (1957–66, 1911–20, and 1921–30) and two pluvials (1988–97 and 1931–40) ranking in the top 10 extreme events (Table 2). The most extreme drought (1957–66) and the second-most extreme pluvial decadal (1988–97) periods in the instrumental record were also the most extreme in the reconstructed record, though there was not a complete overlap between the instrumental and proxy extreme periods. Abrupt decadal changes from wet to dry or dry to wet appear common throughout the record. One of the most severe transitions from wet to dry conditions happened in the mid-thirteenth century, when the third wettest decade (1238–47) transitioned to the seventh driest decade (1251–60) in the reconstruction (Table 2). The 1251–60 period is also significant because it was the longest period of below-median precipitation.

Fig. 5.

Full May precipitation reconstruction from 1200 to 1997. Gray line represents the annual predicted values. A 50-yr fourth-order smoothing spline is used to highlight multidecadal trends (thick black line). Thin black line is the median of the reconstruction.

Fig. 5.

Full May precipitation reconstruction from 1200 to 1997. Gray line represents the annual predicted values. A 50-yr fourth-order smoothing spline is used to highlight multidecadal trends (thick black line). Thin black line is the median of the reconstruction.

Table 2.

Top 10 driest and wettest nonoverlapping 10-yr events for the scaled May precipitation reconstruction (1200–1997). Decadal events were calculated with a 10-yr moving average with each year representing the leading edge of the 10-yr window (mm in parentheses). Average May precipitation for the instrumental record and the entire scaled reconstruction is ~86.0 mm.

Top 10 driest and wettest nonoverlapping 10-yr events for the scaled May precipitation reconstruction (1200–1997). Decadal events were calculated with a 10-yr moving average with each year representing the leading edge of the 10-yr window (mm in parentheses). Average May precipitation for the instrumental record and the entire scaled reconstruction is ~86.0 mm.
Top 10 driest and wettest nonoverlapping 10-yr events for the scaled May precipitation reconstruction (1200–1997). Decadal events were calculated with a 10-yr moving average with each year representing the leading edge of the 10-yr window (mm in parentheses). Average May precipitation for the instrumental record and the entire scaled reconstruction is ~86.0 mm.

We constructed box-and-whisker plots to investigate the distribution of reconstructed precipitation by half-century periods and compared the half-centennial distributions to the distribution of the entire reconstruction (Fig. 6). Additionally, we have smoothed the reconstruction with a 50-yr fourth-order smoothing spline to emphasize multidecadal variability (Fig. 5). We have made a few observations about low-frequency trends. First, shifts in median precipitation were common on half-centennial to centennial scales and the smoothed reconstructed record confirmed this low-frequency variability through time. Greater-than-median precipitation occurred in the early thirteenth century, coincident with the Medieval Climate Anomaly (MCA, ~1000–1300). The early fourteenth century shifted below the long-term median level of precipitation and had the smallest variance of the entire reconstruction. Second, decreased variability in precipitation from the late fifteenth to the late seventeenth centuries and a century of below-median precipitation from 1550 to 1650 coincided with the period of the Little Ice Age (LIA, ~1450–1850). Also, the presence of a multidecadal drought in the late sixteenth century provided further evidence of the sixteenth-century megadrought reported by Stahle et al. (2000, 2007) and Woodhouse and Overpeck (1998). While the drought was not as severe as droughts occurring during the reconstructed early twentieth century, our record showed that the megadrought extended into the mid-Atlantic region. Third, variability in precipitation has increased in the past 300 yr and was greater than any time since 1200. The nineteenth century was not the most variable in the entire reconstruction but had the largest change in median precipitation, increasing from 65.7 to 103.0 mm (Fig. 6; Table 2). Changes in variance in the recent centuries is not attributed to sample size, as evidenced by the high sample size during this period (Fig. 2), nor errors in dating, as shown by the high series intercorrelations (>0.536) and EPS > 0.85. Finally, despite scaling the reconstruction to the mean and variance of the instrumental record, our record has underestimated and overestimated the medians of the early twentieth and late twentieth centuries, respectively. To determine if the overestimation was partly an effect of our method of standardization, we examined individual samples from each site to determine if an increasing growth trend was present during the past half-century. Indeed, 68.5% of all samples showed an increasing growth trend in the raw measurement data. The interpretation of the increasing growth trend is somewhat confounded by the tree growth response to temperature as well as precipitation and the increase in atmospheric CO2 and nitrogen deposition over the past century, and will require further investigation into its cause. The primary strength of our reconstruction of May precipitation is its length and representation of low-frequency variability in a region with a dearth of multicentury, high-resolution proxy records of precipitation. However, the moderate explanatory value (r2adj = 0.27) precludes the record from precisely estimating precipitation for individual years.

Fig. 6.

Box-and-whisker plots (10th, 25th, 50th, 75th, and 90th percentiles) were calculated for each half-century of the May precipitation reconstruction and the twentieth-century instrumental period. Dashed line represents the median of the reconstruction.

Fig. 6.

Box-and-whisker plots (10th, 25th, 50th, 75th, and 90th percentiles) were calculated for each half-century of the May precipitation reconstruction and the twentieth-century instrumental period. Dashed line represents the median of the reconstruction.

5. Regional comparison

We compared our reconstruction of May precipitation for West Virginia climate region 6 against four regional climate and streamflow reconstructions, including 1) Cook and Jacoby’s (1983) Potomac River, Maryland, streamflow (July–September); 2) Stahle et al.’s (1998) July PHDI for southern Virginia–northern North Carolina; 3) Cleaveland and Stahle’s (1994) July PHDI for Missouri; and 4) Cleaveland’s (2000) White River, Arkansas, streamflow (June–August). We chose not to compare our reconstruction to the gridded PDSI reconstruction network (Cook et al. 1999) because the Cedar Knob J. virginiana chronology and Stahle et al.’s (1998) Blackwater River and Nottoway River Taxodium distichum chronologies were included as predictors in the PDSI reconstruction model. Prior to 1500, the J. virginiana and T. distichum chronologies represent three of the four chronologies closest to the mid-Atlantic PDSI grid points (i.e., 246, 247, 255, 256, 262, and 263) and any comparisons to these points would be collinear and redundant.

An initial correlation analysis between the paleorecords showed weak but significant positive relationships between reconstructed May precipitation and the eastern U.S. proxy records ranging from 0.16 to 0.24 (Table 3). The weak results of the correlation analysis were not surprising because the tree-ring reconstructions represent different species, regions, and months of the growing season; however, the correlations showed that some regional synchrony existed. Also, low correlations between the proxy records were likely an effect of the amount of variance in the instrumental records explained by each proxy. Next, we visually compared the records of moisture to determine if the extreme periods of wet and dry in the May precipitation reconstruction (Table 2) were replicated across the eastern United States. Cook and Jacoby’s Potomac River reconstruction recorded three of five decadal pluvials and only two decadal droughts from the May precipitation reconstruction (Fig. 7). The small correspondence between the July and September Potomac River and May precipitation reconstructions was likely an effect of the season of moisture reconstructed and the use of a single species in our reconstruction versus four species in the Potomac River reconstruction. The use of multiple species has been shown to improve the strength of the relationship between tree growth and climate (Cook and Pederson 2011).

Table 3.

A correlation matrix of regional reconstructions of moisture, including 1) Cook and Jacoby’s (1983) Potomac River, MD, streamflow (July–September); 2) Stahle et al.’s (1998) July PHDI for southern VA–northern NC; 3) Cleaveland and Stahle’s (1994) July PHDI for MO; and 4) Cleaveland’s (2000) White River, AK, streamflow (June–August). All correlations are significant at p < 0.01.

A correlation matrix of regional reconstructions of moisture, including 1) Cook and Jacoby’s (1983) Potomac River, MD, streamflow (July–September); 2) Stahle et al.’s (1998) July PHDI for southern VA–northern NC; 3) Cleaveland and Stahle’s (1994) July PHDI for MO; and 4) Cleaveland’s (2000) White River, AK, streamflow (June–August). All correlations are significant at p < 0.01.
A correlation matrix of regional reconstructions of moisture, including 1) Cook and Jacoby’s (1983) Potomac River, MD, streamflow (July–September); 2) Stahle et al.’s (1998) July PHDI for southern VA–northern NC; 3) Cleaveland and Stahle’s (1994) July PHDI for MO; and 4) Cleaveland’s (2000) White River, AK, streamflow (June–August). All correlations are significant at p < 0.01.
Fig. 7.

Regional comparison of the (a) May precipitation reconstruction to each (b) Cook and Jacoby’s (1983) Potomac River, MD, streamflow (July–September); (c) Stahle et al.’s (1998) July PHDI for southern VA–northern NC; (d) Cleaveland and Stahle’s (1994) July PHDI for MO; (e) Cleaveland’s (2000) White River, AK, streamflow (June–August); and (f) Luterbacher’s (1999) NAO. All records were smoothed with a 10-yr fourth-order smoothing spline to enhance decadal trends. Stippled and solid bars represent extreme decadal droughts and pluvials, respectively, recorded in the May precipitation reconstruction (Table 2). Vertical dashed lines represent the Lost Colony (1587–89) and Jamestown (1606–12) droughts from Stahle et al. (1998).

Fig. 7.

Regional comparison of the (a) May precipitation reconstruction to each (b) Cook and Jacoby’s (1983) Potomac River, MD, streamflow (July–September); (c) Stahle et al.’s (1998) July PHDI for southern VA–northern NC; (d) Cleaveland and Stahle’s (1994) July PHDI for MO; (e) Cleaveland’s (2000) White River, AK, streamflow (June–August); and (f) Luterbacher’s (1999) NAO. All records were smoothed with a 10-yr fourth-order smoothing spline to enhance decadal trends. Stippled and solid bars represent extreme decadal droughts and pluvials, respectively, recorded in the May precipitation reconstruction (Table 2). Vertical dashed lines represent the Lost Colony (1587–89) and Jamestown (1606–12) droughts from Stahle et al. (1998).

We found that 12 of 19 extreme dry and wet decadal periods in the May precipitation reconstruction were replicated in Stahle et al.’s (1998) July PHDI proxy record from 1200 to 1984. Additional annual and decadal periods of similarity are apparent throughout the overlapping period. Stahle et al.’s PHDI reconstruction was used to link the disappearance of the Roanoke Island Colony (1587–89) and near abandonment of the Jamestown Colony (1606–12) to extreme drought events around the turn of the sixteenth century. Both events were present in the May precipitation reconstruction, indicating that the droughts were regionwide. On the half-centennial time scale, Stahle et al.’s PHDI reconstruction showed a similar decrease in variance during the early sixteenth century and below-median moisture conditions in both the late sixteenth and early seventeenth centuries that are coincident with the LIA and the sixteenth-century megadrought (Stahle et al. 2007). Moving farther to the southwest, Cleaveland and Stahle’s (1994) reconstruction of Missouri PHDI showed matching with 6 of 12 extreme decadal events, though the duration and magnitude of these events were not always similar. Cleaveland’s (2000) Missouri PHDI reconstruction showed 10 of 20 matching extreme events and also replicated the Lost Colony and Jamestown drought events. Despite some disagreement between the regional records, the comparison was valuable because it demonstrated that 1) decadal to multidecadal trends in moisture were persistent across the eastern United States for some periods, 2) historic droughts (e.g., Lost Colony and Jamestown) were regional in occurrence, and 3) tree-ring proxy records in the MAR have adequate correlation to warrant the collection of additional multicentury J. virginiana tree-ring chronologies for future modeling efforts.

6. Mid-Atlantic region climate patterns

To better understand the potential influence of the NAO on our May precipitation reconstruction and climate of the MAR, we calculated correlations between the winter NAO (1826–2007; Jones et al. 1997) and West Virginia climate region 6 winter precipitation and temperature, reconstructed May precipitation, Cook and Jacoby’s (1983) Potomac River, Maryland, streamflow; Stahle et al.’s (1998) Virginia–North Carolina July PHDI; Cleaveland and Stahle’s (1994) Missouri July PHDI; and Cleaveland’s (2000) White River, Arkansas, streamflow. Correlations were calculated for the common period of the reconstructed records (1825–1977) and only significant correlations will be discussed. Confirming the results of previous research (Hartley and Keables 1998; Hartley 1999; Hurrell et al. 2003), winter NAO (December–February) is positively correlated with winter temperature (r = 0.32, p < 0.05) and winter precipitation (r = 0.35, p < 0.05). The winter NAO was positively correlated with Cook and Jacoby’s Potomac River streamflow (r = 0.18, p < 0.05), Cleaveland and Stahle’s Missouri PDHI (r = 0.21, p < 0.05), and Cleaveland’s White River streamflow (r = 0.21, p < 0.05). The NAO was not significantly (p > 0.05) correlated with Stahle’s Virginia–North Carolina PHDI reconstruction.

The positive correlation between winter NAO (December–March) and reconstructed May precipitation (r = 0.18, p < 0.05) is most pertinent to our reconstruction of May precipitation. The teleconnection between winter NAO and annual ring width of J. virginiana has been established previously (Cook et al. 2002). The positive correlation between winter NAO and reconstructed May precipitation is an effect of winter NAO on J. virginiana tree growth rather than a true association. This interpretation is confirmed by the lack of correlation between instrumental spring precipitation and winter NAO. The correlation does suggest that winter NAO can mediate the growth response to instrumental May precipitation via the species sensitivity to winter and early spring temperature. We further investigated the influence of winter NAO on J. virginiana’s growth response to instrumental May precipitation by plotting 10-yr smoothed winter NAO and the correlation coefficient between PC1 and instrumental May precipitation through time (Fig. 3b). The decrease in the correlation between PC1 and May precipitation coincides with a period of strong negative winter NAO (1950s–70s). A strong negative winter NAO may precondition J. virginiana to respond weakly to May precipitation, while a strong positive winter NAO may precondition a stronger growth response. We expanded the temporal comparison of the NAO to the record of May precipitation and the other eastern U.S. moisture reconstruction using Luterbacher’s (1999) reconstruction of monthly NAO back to 1658. We seasonalized Luterbacher’s record and found a positive (r = 0.28, p < 0.05) interannual correlation with May precipitation. Luterbacher’s winter NAO was also positively correlated with Potomac River streamflow (r = 0.19, p < 0.05), Missouri PHDI (r = 0.21, p < 0.05), and White River streamflow (r = 0.20, p < 0.05) but not Virginia–North Carolina PHDI. This suggests that the NAO broadly affects winter climate in the eastern United States. Then, we smoothed the reconstructed winter NAO with a 10-yr fourth-order smoothing spline to compare with the smoothed reconstruction of May precipitation and the other regional records (Fig. 7). There were periods of correspondence between three of five decadal droughts and five of eight decadal pluvials in the May precipitation reconstruction and Luterbacher’s winter NAO reconstruction. While greater-than-median precipitation was generally associated with positive NAO, there were some notable exceptions. For example, a 20-yr period of drought occurred from ~1910 to 1930, but the NAO was positive during this period instead of negative. The other regional reconstructions showed similar matching between drought and negative NAO and pluvial and positive NAO; however, the relationship was not consistent for all periods. The lack of correspondence suggests that another forcing mechanism is also driving tree growth and moisture in the eastern United States. The significant (p < 0.05) relationships between the regional records and the NAO was expected as many tree-ring records in the eastern United States show some teleconnection to instrumental NAO (Cook et al. 2002). Unfortunately, identifying and characterizing long-term teleconnections with Cook et al.’s (2002) multiproxy reconstruction of the winter NAO would be inappropriate because the model utilized the Cedar Knob and Taxodium distichum chronologies as well as many of the multicentury tree-ring records in the eastern United States. Further investigation into the teleconnection between the NAO and tree growth and the NAO and regional moisture may improve our understanding of past climate in the MAR.

7. Summary and conclusions

Over the past century, the MAR has received an increasing amount of precipitation (~0.23 mm yr−1). Climate models project that this trend likely will continue coincident with increasing regional temperatures (Polsky et al. 2000; Solomon et al. 2007). Expected regional changes in climate will affect water quantity and quality and impact urban populations and ecosystem services. The dearth of multicentury proxy records of climate in the eastern United States has made the evaluation of current and future changes in the MAR climate difficult. In our study, two multicentury J. virginiana chronologies were used to reconstruct May precipitation for the MAR of the United States. The calibration model explained 27% of the adjusted variance in the instrumental record from 1895 to 1997. The model was well verified (1946–97) by the reduction of error (RE = 0.29) and coefficient of efficiency (CE = 0.20) statistics. Calibration and verification statistics for our reconstruction model were comparable to other single-species reconstructions (Stahle et al. 1988, 1998; Buckley et al. 2004) for the eastern United States.

The 797-yr May precipitation reconstruction (1200–1997) showed interannual, decadal, and multidecadal variability throughout the entire proxy record. The nineteenth and twentieth centuries were the most variable and contained the greatest number of decadal wet and dry events since 1200. Half-centennial length changes in precipitation were common but wetter than median conditions and drier than median conditions occurred during the MCA (1200–1300) and the LIA (1450–1850), respectively. Comparison to other regional multicentury records of moisture for the eastern United States showed weak but significant correlations with the May precipitation reconstruction. While correlations were low, extreme wet and dry decadal events in May precipitation were recorded in Cook and Jacoby’s Potomac River, Maryland, streamflow; Stahle et al.’s (1998) July PHDI; Cleaveland and Stahle’s Missouri July PHDI; and Cleaveland’s White River, Arkansas, streamflow; this suggests that some drought and pluvial events were synchronous across the eastern United States. The correlation between the first principal component of the chronologies and instrumental May precipitation weakened but remained significant during the 1950s and 1960s, coincident with a period of strong negative winter NAO. Our results suggest that the dendroclimatic response of J. virginiana to May precipitation is mediated by winter NAO via its influence on winter temperature, and this weakening may have occurred in the past when NAO was strongly negative. When compared to a multicentury reconstruction of winter NAO (Luterbacher et al. 1999), decadal drought and pluvial periods in the May precipitation record coincided with negative and positive NAO, respectively, with some exceptions.

Water resources in the MAR are generally abundant but managers are still challenged to provide adequate and clean water during drought and pluvial events. General circulation models predict that the MAR will experience increased temperature and precipitation in the coming decades (Polsky et al. 2000; Solomon et al. 2007). Expected climate changes may impact the water managers’ ability to provide the water quantity and quality required to sustain the increasing population of the MAR. Climate changes are expected to impact forest and wetland ecosystems, putting additional stress on these habitats and their water holding and purification services on which we rely (McKenney-Easterling et al. 2000; Najjar et al. 2000; Rogers and McCarty 2000). Our May precipitation reconstruction is a conservative estimate of past climate that managers may use to assess the intensity and duration of future spring drought and pluvial events and to maintain a quality water supply for the MAR. Moisture-sensitive, multicentury J. virginiana chronologies can help to fill the gap in the climate record of the MAR. Juniperus virginiana is widespread in the MAR, and the development of a J. virginiana network across the region could improve estimates of past precipitation. However, future work should investigate the influence of the NAO on tree-ring growth response to strengthen reconstructions and better understand how the NAO affects the region. Additionally, the Smoke Hole chronology will be included in future updates of the North American Drought Atlas (Cook and Krusic 2010).

Acknowledgments

This research was funded in part by the NASA West Virginia Space Grant Consortium, West Virginia University Eberly College of Arts and Sciences, The Explorers Club Washington (D.C.) Group, and National Science Foundation Doctoral Dissertation Research (Grant 0925114). The U.S. Forest Service, the Nature Conservancy, and Tom and Eve Firor kindly provided access to sampling locations. Thank you to David Stahle, Malcolm Cleaveland, and Jürg Luterbacher for contributing their data to the NOAA Paleoclimatology Program. This manuscript was greatly improved by Anthony Broccoli and three anonymous reviewers.

REFERENCES

REFERENCES
Barry
,
R. G.
, and
R. J.
Chorley
,
2003
:
Atmosphere, Weather, and Climate
.
8th ed. Routledge, 421 pp
.
Bartgis
,
R. L.
,
1993
:
The limestone glades and barrens of West Virginia
.
Castanea
,
58
,
69
89
.
Biondi
,
F.
,
1997
:
Evolutionary and moving response functions in dendroclimatology
.
Dendrochronologia
,
15
,
139
150
.
Biondi
,
F.
, and
K.
Waikul
,
2004
:
DENDROCLIM 2002: A C++ program for statistical calibration of climate signals in tree-ring chronologies
.
Comput. Geosci.
,
30
,
303
311
.
Briffa
,
K. R.
,
1984
:
Tree-climate relationships and dendrochronological reconstruction in the British Isles
.
Ph.D. dissertation, University of East Anglia, 525 pp
.
Briffa
,
K. R.
,
1995
:
Interpreting high-resolution proxy climate data: The example of dendroclimatology
.
Analysis of Climate Variability: Applications of Statistical Techniques, H. Von Storch and A. Navarra, Eds., Springer-Verlag, 77–94
.
Buckley
,
B. M.
,
R. J. S.
Wilson
,
P. E.
Kelly
,
D. W.
Larson
, and
E. R.
Cook
,
2004
:
Inferred summer precipitation for southern Ontario back to AD 610, as reconstructed from ring widths of Thuja occidentalis
.
Can. J. For. Res.
,
34
,
2541
2553
.
Cleaveland
,
M. K.
,
2000
:
A 963-year reconstruction of summer (JJA) stream flow in the White River, Arkansas, USA, from tree-rings
.
Holocene
,
10
,
33
41
.
Cleaveland
,
M. K.
,
D. W.
Stahle
,
1994
:
Reconstruction of Ozark Highlands paleoclimate from tree rings: Applications to global change
.
Proc. Int. Conf. on Tree Rings, Environment and Humanity: Relationships and Processes, Tucson, AZ, University of Arizona
.
Cook
,
E. R.
,
1985
:
A time series analysis approach to tree-ring standardization
.
Ph.D. dissertation, University of Arizona, 171 pp
.
Cook
,
E. R.
, and
G. C.
Jacoby
,
1983
:
Potomac River streamflow since 1730 as reconstructed by tree rings
.
J. Climate Appl. Meteor.
,
22
,
1659
1672
.
Cook
,
E. R.
, and
L. A.
Kairiukstis
, Eds.,
1990
:
Methods of Dendrochronology: Applications in the Environmental Science
.
Kluwer Academic, 394 pp
.
Cook
,
E. R.
, and
P. J.
Krusic
, cited
2010
:
The North American drought atlas
.
Lamont-Doherty Earth Observatory and the National Science Foundation. [Available online at http://iridl.ldeo.columbia.edu/SOURCES/.LDEO/.TRL/.NADA2004/.pdsi-atlas.html.]
Cook
,
E. R.
, and
N.
Pederson
,
2011
:
Uncertainty, emergence, and statistics in dendrochronology
.
Dendroclimatology: Progress and Prospects, M. K. Hughes, T. W. Swetnam, and H. F. Diaz, Eds., Developments in Paleoecological Research Series, Vol. 11, Springer-Verlag, 77–112
.
Cook
,
E. R.
,
K. R.
Briffa
,
D. M.
Meko
,
D. A.
Graybill
, and
G.
Funkhouser
,
1995
:
The “segment length curse” in long tree-ring chronology development for paleoclimatic studies
.
Holocene
,
5
,
229
237
.
Cook
,
E. R.
,
D. M.
Meko
,
D. W.
Stahle
, and
M. K.
Cleaveland
,
1999
:
Drought reconstructions for the continental United States
.
J. Climate
,
12
,
1145
1162
.
Cook
,
E. R.
,
R. D.
D’Arrigo
, and
M. E.
Mann
,
2002
:
A well-verified, multiproxy reconstruction of the winter North Atlantic Oscillation index since a.d. 1400
.
J. Climate
,
15
,
1754
1764
.
Cook
,
E. R.
,
R.
Seager
,
M. A.
Cane
, and
D. W.
Stahle
,
2007
:
North American drought: Reconstructions, causes, and consequences
.
Earth Sci. Rev.
,
81
,
93
134
.
Cook
,
E. R.
,
R.
Seager
,
R. R.
Heim
Jr.
,
R. S.
Vose
,
C.
Herweijer
, and
C.
Woodhouse
,
2010
:
Megadroughts in North America: Placing IPCC projections of hydroclimatic change in a long-term paleoclimatic context
.
J. Quat. Sci.
,
25
,
48
61
.
Fritts
,
H. C.
,
1976
:
Tree Rings and Climate
.
Academic Press, 567 pp
.
Graumlich
,
L. J.
,
1987
:
Precipitation variation in the Pacific Northwest (1675–1975) as reconstructed from tree rings
.
Ann. Assoc. Amer. Geogr.
,
77
,
19
29
.
Grissino-Mayer
,
H. D.
,
2001
:
Evaluating crossdating accuracy: A manual and tutorial for the computer program COFECHA
.
Tree-Ring Res.
,
57
,
205
221
.
Guttman
,
L.
,
1954
:
Some necessary conditions for common-factor analysis
.
Psychometrika
,
19
,
149
161
.
Guyette
,
R.
,
E. A.
McGinnes
Jr.
,
G. E.
Probasco
, and
K. E.
Evans
,
1980
:
A climate history of Boone County, Missouri, from tree-ring analysis of eastern redcedar
.
Wood Fiber
,
12
,
17
28
.
Hartley
,
S.
,
1999
:
Winter Atlantic climate and snowfall in the south and central Appalachians
.
Phys. Geogr.
,
20
,
1
13
.
Hartley
,
S.
, and
M. J.
Keables
,
1998
:
Synoptic associations of winter climate and snowfall variability in New England, USA, 1950-1992
.
Int. J. Climatol.
,
18
,
281
298
.
Hawley
,
F. M.
,
1937
:
Relationship of southern cedar growth to precipitation and run off
.
Ecology
,
18
,
398
405
.
Hidalgo
,
H. G.
,
2004
:
Climate precursors of multidecadal drought variability in the western United States
.
Water Resour. Res.
,
40
,
W12504
,
doi:10.1029/2004WR003350
.
Hurrell
,
J. W.
,
Y.
Kushnir
,
G.
Otterson
, and
M. H.
Visbeck
,
2003
:
An overview of the North Atlantic Oscillation
.
The North Atlantic Oscillation: Climatic Significance and Environmental Impact, Geophys. Monogr., Vol. 134, Amer. Geophys. Union, 1–35
.
Jones
,
P. D.
,
T.
Jonsson
, and
D.
Wheeler
,
1997
:
Extension to the North Atlantic Oscillation using early instrumental pressure observations from Gibraltar and south-west Iceland
.
Int. J. Climatol.
,
17
,
1433
1450
.
Kaiser
,
H. F.
,
1960
:
The application of electronic computers to factor analysis
.
Educ. Psychol. Meas.
,
20
,
141
151
.
Larson
,
D. W.
,
1997
:
Dendroecological potential of Juniperus virginiana L. growing on cliffs in western Virginia
.
Banisteria
,
10
,
13
18
.
Lawson
,
E. R.
,
1990
:
Juniperius virginiana L.: Eastern redcedar
.
Conifers, R. M. Burns and B. H. Honkala, Eds., Vol. 1, Silvics of North America, USDA Forest Service Agricultural Handbook 654, 131–140
.
Lenihan
,
J. M.
,
D.
Bachelet
,
R. P.
Neilson
, and
R.
Drapek
,
2008
:
Simulated response of conterminous United States ecosystems to climate change at different levels of fire suppression, CO2 emission rate, and growth response to CO2
.
Global Planet. Change
,
64
,
16
25
.
Luterbacher
,
J.
,
C.
Schmutz
,
D.
Gyalistras
,
E.
Xoplaki
, and
H.
Wanner
,
1999
:
Reconstruction of monthly NAO and EU indices back to AD 1675
.
Geophys. Res. Lett.
,
26
,
2745
2748
.
McKenney-Easterling
,
M.
,
D. R.
DeWalle
,
L. R.
Iverson
,
A. M.
Prasad
, and
A. R.
Buda
,
2000
:
The potential impacts of climate change and variability on forests and forestry in the mid-Atlantic region
.
Climate Res.
,
14
,
195
206
.
Meko
,
D. M.
,
C. W.
Stockton
, and
W. R.
Boggess
,
1995
:
The tree-ring record of severe sustained drought
.
Water Resour. Bull.
,
31
,
789
801
.
Meko
,
D. M.
,
M. D.
Therrell
,
C. H.
Baisan
, and
M. K.
Hughes
,
2001
:
Sacramento River flow reconstructed to A.D. 869 from tree rings
.
J. Amer. Water Resour. Assoc.
,
37
,
1029
1039
.
Najjar
,
R. G.
, and
Coauthors
,
2000
:
The potential impacts of climate change on the mid-Atlantic coastal region
.
Climate Res.
,
14
,
219
233
.
National Climatic Data Center
, cited
2009
:
West Virginia climate region 6 meteorological data
Neff
,
R.
,
H.
Chang
,
G.
Knight
,
R. G.
Najjar
,
B.
Yarnal
, and
H. A.
Walker
,
2000
:
Impact of climate variation and change on mid-Atlantic region hydrology and water resources
.
Climate Res.
,
14
,
207
218
.
Ormsbee
,
P.
,
F. A.
Bazzaz
, and
W. R.
Boggess
,
1976
:
Physiological ecology of Juniperus virginiana in oldfields
.
Oecologia
,
23
,
75
82
.
Osborn
,
T. J.
,
K. R.
Briffa
, and
P. D.
Jones
,
1997
:
Adjusting variance for sample-size in tree-ring chronologies and other regional mean time series
.
Dendrochronologia
,
15
,
89
99
.
Pederson
,
N.
,
G. C.
Jacoby
,
R. D.
D’Arrigo
,
E. R.
Cook
, and
B. M.
Buckley
,
2001
:
Hydrometeorological reconstructions from northeastern Mongolia derived from tree rings: 1651–1995
.
J. Climate
,
14
,
872
881
.
Perreault
,
T. C.
,
M.
Haché
,
M.
Slivitzky
, and
B.
Bobée
,
1999
:
Detection of changes in precipitation and runoff over eastern Canada and U.S. using a Bayesian approach
.
Stochastic Environ. Res. Risk Assess.
,
13
,
201
216
.
Polsky
,
C. D.
,
J.
Allard
,
N.
Currit
,
R.
Crane
, and
B.
Yarnal
,
2000
:
The mid-Atlantic region and its climate: Past, present, and future
.
Climate Res.
,
14
,
161
173
.
Rogers
,
C. E.
, and
J. P.
McCarty
,
2000
:
Climate change and ecosystems of the mid-Atlantic region
.
Climate Res.
,
14
,
235
244
.
Solomon
,
S.
,
D.
Qin
,
M.
Manning
,
M.
Marquis
,
K.
Averyt
,
M. M. B.
Tignor
,
H. L.
Miller
Jr.
, and
Z.
Chen
, Eds.,
2007
:
Climate Change 2007: The Physical Science Basis
.
Cambridge University Press, 996 pp
.
Speer
,
J. H.
,
2010
:
Fundamentals of Tree-Ring Research
.
University of Arizona Press, 368 pp
.
Stahle
,
D. W.
, and
M. K.
Cleaveland
,
1992
:
Reconstruction and analysis of spring rainfall over the southeastern U.S. for the past 1000 years
.
Bull. Amer. Meteor. Soc.
,
73
,
1947
1961
.
Stahle
,
D. W.
,
M. K.
Cleaveland
, and
J. G.
Hehr
,
1988
:
North Carolina climate changes reconstructed from tree rings: A.D. 372 to 1985
.
Science
,
240
,
1517
1519
.
Stahle
,
D. W.
,
M. K.
Cleaveland
,
D. B.
Blanton
,
M. D.
Therrell
, and
D. A.
Gay
,
1998
:
The Lost Colony and Jamestown droughts
.
Science
,
280
,
564
567
.
Stahle
,
D. W.
,
E. R.
Cook
,
M. K.
Cleaveland
,
M. D.
Therrell
,
D. M.
Meko
,
H. D.
Grissino-Mayer
,
E.
Watson
, and
B. H.
Luckman
,
2000
:
Tree-ring data document 16th century megadrought over North America
.
Eos, Trans. Amer. Geophys. Union
,
81
,
121
.
Stahle
,
D. W.
,
F. K.
Fye
,
E. R.
Cook
, and
R. D.
Griffin
,
2007
:
Tree-ring reconstructed megadroughts over North America since A.D. 1300
.
Climatic Change
,
83
,
133
149
.
Stockton
,
C. W.
, and
G. C.
Jacoby
Jr.
,
1976
:
Long-term surface-water supply and streamflow levels in the Upper Colorado River basin based on tree-ring analyses
.
Lake Powell Research Project Bull. 18, Institute of Geophysics and Planetary Physics, University of California, Los Angeles, 70 pp
.
Timilsena
,
J.
,
T.
Piechota
,
H.
Hidalgo
, and
G.
Tootle
,
2007
:
Five hundred years of hydrological drought in the Upper Colorado River basin
.
J. Amer. Water Resour. Assoc.
,
43
,
798
812
.
Wigley
,
T. M. L.
,
K. R.
Briffa
, and
P. D.
Jones
,
1984
:
On the average value of correlated time series, with applications in dendroclimatology and hydrometeorology
.
J. Climate Appl. Meteor.
,
23
,
201
213
.
Woodhouse
,
C. A.
,
2003
:
A 431-yr reconstruction of western Colorado snowpack from tree rings
.
J. Climate
,
16
,
1551
1561
.
Woodhouse
,
C. A.
, and
J. T.
Overpeck
,
1998
:
2000 years of drought variability in the central United States
.
Bull. Amer. Meteor. Soc.
,
79
,
2693
2714
.
Woodhouse
,
C. A.
, and
J. J.
Lukas
,
2006
:
Drought, tree rings and water resource management in Colorado
.
Can. Water Resour. J.
,
31
,
1
14
.

Footnotes

*

Lamont-Doherty Earth Observatory Contribution Number 7512.

+

Current affiliation: Department of Geography, The Pennsylvania State University, University Park, Pennsylvania.