Page  193



Alex Dye1 Department of Forest Ecosystems and Society Oregon State University Corvallis, OR 97333

Kerry Woods

Natural Sciences, Bennington College One College Drive Bennington, VT 05201


Availability of high-quality primary ecological datasets like tree-ring growth is critical to the progress of ecological sciences. The International Tree-Ring Databank (ITRDB) is the premier public archive for interannual records of tree-ring growth (i.e., “chronologiesâ€); however, there is currently a dearth of available chronologies that extend into the 21st century. In Michigan, none of the available records for eastern hemlock extends past 1983. Unfortunately, this reduces the availability of these chronologies for study of recent ecological and climatological change or for their integration with new environmental measuring technologies. In this paper, we fill part of this gap for northern Michigan and examine how our datasets can inform long-term studies of ecology-climate interactions in the Midwest region. We present multi-century Tsuga canadensis (L.) Carrière (eastern hemlock) tree-ring growth records for four sites in the Upper Peninsula of Michigan, U.S.A., covering the periods 1708–2015, 1754–2015, 1794–2015, and 1857–1995. We explore potential applications of these datasets by examining basic correlations between interannual growth on the one hand and regional temperature and Palmer Drought Severity Index (PDSI, an estimate of dryness) values on the other. At all four sites, growth is negatively correlated with previous summer and current spring temperature, while growth is positively correlated with previous summer PDSI at three of the four sites.

KEYWORDS: eastern hemlock, tree rings, Michigan, Huron Mountains


Insights into past environmental conditions, including climate, are primarily gained through the study of natural proxies. In forested areas, tree rings are one of the most accessible and informative natural proxies available, providing annually resolved information about past environmental variability, such as temperature and precipitation. Public archives that store these proxies for public use, such as the International Tree-Ring Databank (ITRDB), maintained by the US National Oceanic and Atmospheric Administration, provide critical benefits to

1 Author for correspondence (

Page  194 194 THE GREAT LAKES BOTANIST Vol. 58

the progress of environmental research (Babst et al. 2017). However, temporal coverage of available tree-ring records (“chronologiesâ€) on the ITRDB is not homogeneous; for instance, nearly half of all chronologies archived have a final ring year preceding 1990 (Larson et al. 2013, Babst et al. 2017). This means that most chronologies do not extend into the 21st century, a period that has witnessed unprecedented shifts in climate (IPCC 2014). This temporal limitation discourages potential integration with newer methods for recording environmental change, such as eddy covariance towers, remote sensing, or phenocams (Babst et al. 2017).

Furthermore, in the Great Lakes region, like many regions in eastern North America, the combined effects of logging, agriculture, and human settlement have removed much of the presettlement forest cover (Gleason 1923, Whitney 1987, White and Mladenoff 1994). Old trees facilitate the construction of multi- century growth records and add intrinsic value to tree-ring studies of the few old- growth forests that still exist in these anthropogenically altered regions. Without information on long-term growth of trees, it is difficult for researchers to assess tree growth patterns over time.

The humid climate of eastern North America prevents extended preservation of dead trees, and old living trees are often the only source of multi-century tree growth information (Fritts 1976). One of the longest-lived tree species in eastern North America is eastern hemlock, Tsuga canadensis (L.) Carrière (Cook and Cole 1991). For this reason, it is a preferred species in eastern North America for developing multi-century tree-ring chronologies, and it has been used successfully to assess climate–growth relationships. Often, annual eastern hemlock growth will exhibit a seasonally lagged negative correlation with summer or fall temperature (D’Arrigo et al. 2001, Tardiff et al. 2001, Hart et al. 2010, Saladyga and Maxwell 2015).

However, few new or updated tree-ring chronologies have been made for eastern hemlock in recent years. Of the 76 eastern hemlock chronologies available on the ITRDB, 54 end prior to 1990. Most subsequent published work has focused primarily on the southeastern United States (Hart et al. 2010, Saladyga and Maxwell 2015) or northeastern North America (D’Arrigo et al. 2001, Tardiff et al. 2001, Black and Abrams 2005). In Michigan, all publicly available hemlock chronologies end in 1983 or earlier, thereby creating a critical temporal informational gap that can be filled with strategic data additions and updates to existing data (Larson et al. 2013).

For eastern hemlock, the need to develop high-quality chronologies is even more urgent, since the spread of the invasive Adelges tsugae (Annand) (hemlock woolly adelgid) continues to cause hemlock mortality throughout the eastern United States (Hessl and Pederson 2012, Orwig et al. 2012). The northward spread of hemlock woolly adelgid has been slow because it is suspected to be limited by the cooler northern climate (Trotter et al. 2009). However, the range of hemlock woolly adelgid has been expanding further northward as temperatures in North America rise, and it is likely only a matter of time before all hemlocks are affected, including those in northern Michigan (Paradis et al. 2008, Fitzpatrick et al. 2012).

We recognize that there is a need for ecologists to continue to update, de


velop, and publish quality tree-ring chronologies in order to provide users with the most current and beneficial information available. In regions lacking significant old-growth forest cover, such as eastern North America, multi-century records are essential to extending the temporal scope of analysis available. In this paper, we describe new multi-century tree-ring chronologies at four old-growth eastern hemlock forests in northern Michigan. With each of these datasets, we evaluate their suitability for recording regional climate parameters by comparing interannual ring growth with monthly temperature and Palmer Drought Severity Index (PDSI, a measure of dryness (Palmer 1965)).


Study sites and sampling procedure

The Huron Mountain Club (HMC) is a privately managed area covering 8,100 ha along the southern coast of Lake Superior in Michigan’s Upper Peninsula, approximately 50 km northwest of the city of Marquette. Founded in 1889 as a private retreat, HMC encompasses forests that have largely escaped major logging operations, resulting in the preservation of some of the largest tracts of old- growth forest in the eastern United States. The growing season is typically less than 120 days per year (Sommers 1977).

We sampled three sites at HMC, which we have designated HMC-West, HMC-Rush Lake, and HMC-Mountain Lake (Figure 1). In summer 2016, we collected increment cores at HMC-West and HMC-Rush Lake as part of a project reconstructing biomass dynamics over the last four decades (Dye 2018). However, we have not yet published long-term chronology development and climate response. At both sites, we cored all trees within two 16-m radius plots with diameter at breast height (DBH, approximately 1.3 m above ground) greater than 10 cm. We also incorporated unpublished tree-ring data sampled at HMC-Mountain Lake in 1995. Because HMC-Mountain Lake was sampled 23 years ago, we cannot infer information on recent tree-ring growth as we can at the other sites. Despite this shortcoming, we include it primarily for its contribution to eastern hemlock chronology construction in the region.

Dukes Research Natural Area (Dukes RNA) lies within the Munising Ranger District of the Hiawatha National Forest approximately 40 km inland from Lake Superior (Figure 1). Established in 1979 as a protected research area by the US Forest Service, Dukes RNA harbors stands of old growth hemlock-northern hardwood forests that have escaped significant logging activity. Trees at Dukes RNA were cored in 2016 and have not yet been published. At Dukes RNA, we cored two radii from 30 eastern hemlock trees > 20 cm DBH. All Dukes RNA and HMC trees were synchronized with long-term remeasurement plots to maximize the applicability of the dataset (Woods 2007; Woods 2014).

Tree-ring chronology development

We mounted and sanded each increment core using progressively finer grit sandpaper and confirmed annual dating of each ring through crossdating (Stokes and Smiley 1968). We measured all annual ring widths at 0.001 mm precision with a sliding stage micrometer (Velmex Inc., Bloomfield, NY, USA). Measurements were then verified statistically for crossdating accuracy using the software COFECHA (Holmes 1983). To compile our final site chronologies, we omitted ring width series with segment lengths less than 75 years and correlations with the master chronology of less than 0.32 to avoid bias in the climate growth relationships (Fritts 1976). For each ring-width chronology, we are primarily interested in two statistical characteristics: interseries correlation and mean sensitivity. The interseries correlation is the average of all correlations calculated between each individual series and the remaining series upon removal of the series being tested (Holmes 1983). Average mean sensitivity quantifies relative differences between consecutive growth rings, serving as an indicator of growth sensitivity to environmental conditions (Fritts 1976). Both of these measures are useful for comparing characteristics of the chronologies presented here with other chronologies.

To remove non-climatic growth responses such as release events and age-related growth trends (Cook and Peters 1981), we detrended and standardized each raw ring width series using a 30-year

Page  196 196 THE GREAT LAKES BOTANIST Vol. 58

FIGURE 1. Map showing the location of the study sites. The Huron Mountain Club area is expanded in an inset for detail. Map layout was created with QGIS Version 2.8.2 (QGIS, 2015). Basemap layers consist of U.S. state border lines (U.S. Census Bureau, 2017a), local lake boundaries (U.S. Census Bureau 2017b), and Canadian national borders (Statistics Canada, 2016).

smoothing spline in the R package dplR (Bunn 2008). The 30-year spline has been used as an effective detrending technique in eastern hemlocks (Hart et al. 2010, Saladyga and Maxwell 2015). Although tree-ring widths are often autocorrelated with past growth, we opted not to remove this autocorrelation to enable comparisons with previous year climate. Each year of all detrended series per site were then averaged to create a single mean chronology time series for each site. We constructed chronologies only for years when there were at least five individual samples.

Climate-growth relationships

We compare each of our four detrended tree-ring chronologies with monthly mean temperature and monthly Palmer Drought Severity Index (PDSI). Instrumental records for each of these variables were downloaded for the years 1895–2015 from the National Climate Data Center Michigan climate division 1 (National Oceanic and Atmospheric Administration 2016). Divisional climate data tends to be more strongly related to tree growth than do single stations (Blasing et al. 1981). In particular, NCDC divisional data has successfully been used in several dendroclimatological studies in the eastern U.S. (Hart et al. 2010, Maxwell et al. 2015, Saladyga and Maxwell 2015).

We conducted correlation analyses between each tree-ring chronology and monthly temperature and PDSI, from previous-year June to current-year September. Because we correlate growth with previous year climate, we begin climate-growth comparisons in 1896. All correlation analyses were conducted in the R package treeclim (Zang and Biondi 2016). Package treeclim conducts Pearson product moment correlation coefficients between the chronology and monthly climate, constructing confidence intervals and statistical significance based on partial regression coefficients from 1000 bootstrapped estimates (a < 0.05).


For each monthly climate variable that returned a significant growth-climate relationship, we additionally produced separate correlation coefficients for the pre-1983 period (1895–1983), the post1983 period (1983–2015) period, and the full period (1895–2015) at Dukes RNA, HMC-Rush Lake, and HMC-West. We did not produce separate correlations for HMC-Mountain Lake because this chronology ends in 1995. Because there are no eastern hemlock chronologies publicly available on the ITRDB for Michigan that extend past 1983, this shows whether there has been an observable change in the climate-growth relationships for Michigan hemlocks since the period when most hemlock chronologies end.


Tree-ring chronologies

Each chronology strongly crossdates internally and has interseries correlations confirming confident crossdating that are consistent with other publicly available eastern hemlock tree-ring chronologies (Table 1; Figure 2). Mean sensitivity, an indicator of interannual variation in tree growth, is high for each chronology. The average mean sensitivity of eastern hemlock chronologies archived on the ITRDB is 0.24 (maximum = 0.34 and minimum = 0.19; Hart et al. 2012). Consequently, mean sensitivity for each of our chronologies is towards the upper limit of what is typical for eastern hemlock.

Climate-tree growth relationships

Tree growth at all four sites is significantly negatively correlated with temperature for one or more months in the previous summer and significantly positively correlated with one or more months in the current spring (Figure 3). Growth at HMC-Rush Lake is correlated with temperature for each of previous June-October, HMC-West and Dukes RNA are correlated with previous June- September, and HMC-Mountain Lake is correlated only with previous August. Current-year March and/or April temperature is positively correlated with growth at all sites.

PDSI is positively correlated with growth for all months except current-year August and September at HMC-West. Previous summer PDSI is positively correlated with growth at Dukes RNA (previous June-November) and HMC-Rush

TABLE 1. Statistics for each of the four tree-ring chronologies.

Number Average Average of trees Dated Time interseries mean Site sampled radii span correlation sensitivity Dukes RNA 25 39 1693–2015 0.692 0.257 HMC-Rush Lake 53 86 1747–2015 0.690 0.336 HMC-West 77 140 1777–2015 0.687 0.326 HMC-Mountain Lake 37 43 1836–1995 0.544 0.386

Page  198 198 THE GREAT LAKES BOTANIST Vol. 58

FIGURE 2. Detrended, standardized ring width indices (left axis) with sample depth (right axis) for

(A) Dukes RNA; (B) HMC-Rush Lake; (C) HMC-West; and (D) HMC-Mountain Lake. The solid black line denotes the average chronology, and light gray shaded regions indicate 95% confidence intervals around the mean. Lake (previous August and November). In contrast, growth at HMC Mountain Lake is negatively correlated with PDSI in previous year June.

At some sites, the climate-growth relationship differs between the pre-and post-1983 periods (Table 2). At Dukes RNA, monthly temperature and PDSI relationships are consistently much less correlated with growth in the pre-1983 period; for temperature, the sign changes from negative (pre-1983) to positive (post-1983) during all months. Changes in correlations are less consistent at HMC-Rush Lake and HMC-West. At these sites, correlations become stronger with some monthly climate variables but become weaker with others (Table 2).


Climate-growth relationships of hemlock chronologies

In the eastern U.S., studies relating tree growth and climate have a robust history for eastern hemlock (D’Arrigo et al. 2001, Black and Abrams 2005, Hart et al. 2010, Saladyga and Maxwell 2015). Typically, hemlock is one of the oldest- lived trees in the eastern U.S. and can exceed 500 years old (Cook and Cole 1991), making it a primary target for development of long term tree-ring archives. In other dendroclimatological analyses, eastern hemlock growth is often negatively correlated with previous summer temperatures (D’Arrigo et al. 2001, Tardiff et al. 2001, Hart et al. 2010, Saladyga and Maxwell 2015). Likewise, this is the case with each of our four chronologies (Figure 3). Potentially,


TABLE 2. Monthly correlation coefficients for temperature and for Palmer Drought Severity Index (PDSI) for the pre-1983 period (1895–1983), the post-1983 period (1983–2015) period, and the full period (1895–2015) at Dukes RNA, HMC-Rush Lake, and HMC-West. Correlations for HMC- Mountain Lake are not shown because this chronology ends in 1995. Only those monthly climate variables that were deemed significant are included. Previous-year months are listed with a lowercase “p†preceding the name of the month; for example, previous-year June is listed as “pJun.â€

Dukes RNA HMC-Rush Lake HMC-West Month Pre1983 Post1983 Full Pre1983 Post1983 Full Pre1983 Post1983 Full Temperature pJun pJul pAug pSept pOct pNov pDec Mar –0.28 –0.36 –0.28 –0.30 0.24 0.04 0.16 0.19 0.16 0.15 –0.32 –0.38 –0.32 –0.29 0.24 –0.15 –0.37 –0.34 –0.21 0.31 –0.28 –0.40 –0.25 –0.12 0.21 –0.21 –0.36 –0.31 –0.20 0.27 –0.33 –0.27 –0.30 –0.33 –0.33 0.30 –0.52 –0.43 –0.44 –0.10 –0.52 0.10 –0.38 –0.35 –0.34 –0.26 –0.38 0.20 PDSI pJun pJul pAug pSept pOct pNov pDec Jan Feb Mar Apr May Jun Jul 0.24 0.35 0.32 0.29 0.25 0.27 –0.08 0.05 0.13 0.15 0.02 0.01 0.16 0.26 0.27 0.25 0.19 0.19 –0.08 –0.01 0.20 0.07 0.19 0.18 0.20 0.37 0.38 0.36 0.39 0.38 0.34 0.31 0.32 0.31 0.24 0.28 0.40 0.34 0.21 0.33 0.38 0.37 0.24 0.22 0.23 0.26 0.22 0.15 0.12 0.20 0.19 0.13 0.21 0.35 0.37 0.36 0.33 0.32 0.29 0.28 0.28 0.25 0.18 0.24 0.32 0.26

this is related to a delayed negative effect of high temperatures—that is, warmer temperatures accelerate evapotranspiration, increasing water loss and environmental stress on the tree (Cook and Cole 1991). This effect can be particularly drastic for eastern hemlock, since the desired temperature range of the species is small and photosynthesis noticeably begins to decline after temperatures crest this threshold (Adams and Loucks 1971). Additionally, tree growth is positively correlated with current-year spring temperatures in at least one month (March and/or April) at all sites. Most likely, this is a response to longer growing season length. Warmer spring months induce an earlier start to the growing season, which facilitates enhanced tree growth over the course of that year (D’Arrigo et al. 2001).

Growth correlations with PDSI vary more strongly between sites (Figure 4). HMC West exhibits high positive correlations with PDSI in all months except

Page  200 200 THE GREAT LAKES BOTANIST Vol. 58

FIGURE 3. Correlations between average monthly temperatures and tree-ring chronologies at (A) Dukes RNA; (B) HMC-Rush Lake; (C) HMC-West; and (D) HMC-Mountain Lake. Correlation values are shown by the heights of the bar plots, with significantly correlated months shaded in gray. Vertical solid black lines indicate 95% confidence intervals for the monthly correlations.

current-year August and September. This likely reflects drier site characteristics. However, HMC-West also has many more samples (n = 75) than the other sites, potentially biasing the monthly PDSI correlations. In contrast, HMC-Rush Lake and Dukes RNA respond positively only to previous summer PDSI and weakly correlate with current-year PDSI. Growth is negatively correlated with previous June PDSI at HMC-Mountain Lake. Differences in climate response here may be related to either microsite climate differences due to its position on the northeast corner one of the larger lakes at HMC (Hinkel and Nelson 2012) or the lack of ring width measurements from 1995 to present. The Mountain Lake chronology also has both the highest mean sensitivity (0.391) and lowest interseries correlation (0.549), which could potentially be reflective of the anomalously negative June PDSI relationship.

Tree-ring chronologies of other species in Michigan and the surrounding region exhibit climate-growth relationships similar to those we observed for eastern hemlock. In the Lower Peninsula of Michigan, red pine (Pinus resinosa Aiton) biomass growth is negatively correlated with previous spring temperature and positively correlated with a current-summer drought index (Magruder et al. 2012). Sugar maple (Acer saccharum Marshall), balsam fir (Abies balsamea (L.) Mill.), and white spruce (Picea glauca (Moench) Voss) north of Lake Superior in southern Ontario are negatively correlated with previous spring temperature, although the current-year spring temperature relationship emerges only in balsam


FIGURE 4. Correlations between monthly PDSI and tree-ring chronologies at (A) Dukes RNA; (B) HMC-Rush Lake; (C) HMC-West; and (D) HMC-Mountain Lake. Correlation values are shown by the heights of the bar plots, with significantly correlated months shaded in gray. Vertical solid black lines indicate 95% confidence intervals for the monthly correlations.

fir (Goldblum and Rigg 2005). Across an Upper Midwest longitudinal gradient, sugar maple growth is negatively correlated with previous spring temperature (Lane et al. 1993); and, in Quebec, American beech (Fagus grandifolia Ehrh.) and sugar maple are both negatively correlated with previous summer temperature (Tardiff et al. 2001).

Utility of the dataset

The primary goal of this paper is to construct four high-quality old-growth eastern hemlock tree-ring chronologies and make them available for use on a public archive, thereby accelerating environmental and earth science research (Babst et al. 2017). Although the spatial and temporal coverage of the ITRDB has grown to include over 4000 sites on six continents, many of the currently archived chronologies were submitted prior to 1990 and lack information on recent growth (Larson et al. 2013). For eastern hemlock alone, only 24 of the 76 available chronologies available for North America extend past the year 1990. In Upper Peninsula Michigan specifically, our chronologies are currently the only publicly available eastern hemlock chronologies that extend past 1983. We extend the temporal data availability for this region by 12 years at HMC-Mountain Lake and by 33 years at HMC-Rush Lake, HMC-West, and Dukes RNA. In a re

Page  202 202 THE GREAT LAKES BOTANIST Vol. 58

gion containing some of the largest tracts of old-growth hemlock forest in North America, extending and updating available growth records is essential (Hessl and Pederson 2012). Few other sources exist to study long-term tree growth in Michigan or in eastern deciduous forests in general besides utilizing the current old-growth trees (but see De Graauw (2017) and Rochner et al. (2017) for recent work on using logs from historic structures, although these lack information on current tree characteristics).

The location of our chronologies within two active research forests also amplifies their utility for a variety of environmental research. Current parallel datasets include dendroecology of red pine (Guyette et al. 2012) and white pine (Pinus strobus L.) (Fahey and Lorimer 2014), fine-scale climate monitoring (Hinkel and Nelson 2012), biomass dynamics (Woods 2014, Dye 2018) at HMC, and maintenance of long-term monitoring plots of growth and mortality (Woods 2007, Woods 2014) and LiDAR measurements (Fahey et al. 2015) at both HMC and Dukes RNA. Aggregation of ancillary co-located datasets such as these can be instrumental to improving scientific understanding of ecosystem processes and variability in the face of changing climate (Babst et al. 2017, Haase et al. 2018), but only if scientists take the time to publish, share, and archive their datasets.


This work was supported by funds from the Huron Mountain Wildlife Foundation, the National Science Foundation Macrosystems Biology grant number 1241930, the West Virginia University Department of Geology and Geography, and a Department of Energy ORISE fellowship through the US Forest Service Pacific Northwest Research Station. The authors thank the Munising Ranger District of the Hiawatha National Forest for sampling permission, and Jason Fridley, Ruth Stetler, Jonathon Suite, and Lynsey Blackburn for field and laboratory assistance


Adams, M. S., and O. L. Loucks. (1971). Summer air temperature as a factor affecting net photosynthesis and distribution of eastern hemlock (Tsuga canadensis L. (Carriere)) in southwestern Wisconsin. American Midland Naturalist 85: 1–10.

Babst, F., B. Poulter, P. Bodesheim, M. D. Mahecha, and D. C. Frank. (2017). Improved tree-ring archives will support earth-system science. Nature Ecology and Evolution 1,0008: 1–2.

Black, B. A., and M. D. Abrams. (2005). Disturbance history and climate response in an old-growth hemlock-white pine forest, central Pennsylvania. Journal of the Torrey Botanical Society 132: 103–114.

Blasing, T. J., Duvick, D. N., and West, D. C. (1981). Dendroclimatic calibration and verification using regionally averaged and single station and precipitation data. Tree-Ring Bulletin 41: 37–43. Bunn, A. G. (2008). A dendrochronology program library in R (dplR). Dendrochronologia 26: 115– 124. Cook, E. R., and Peters, K. (1981). The smoothing spline: A new approach to standardizing forest interior tree-ring width series for dendroclimatic studies. Tree-Ring Bulletin 41: 45–53. Cook, E. R., and J. Cole. (1991). On predicting the response of forests in eastern North America to future climatic change. Climatic Change 19: 271–282.

D’Arrigo, R. D., W. S. F. Schuster, D. M. Lawrence, E. R. Cook, M. Wiljanen, and R. D. Thetford. (2001). Climate-growth relationships of eastern hemlock and chestnut oak from Black Rock Forest in the highlands of southeastern New York. Tree-Ring Research 57: 183–190.

De Graauw, K. K. (2017). Historic log structures as ecological archives: A case study from eastern North America. Dendrochronologia 45: 23–34.


Dye, A., A. Barker Plotkin, D. Bishop, N. Pederson, B. Poulter, and A. Hessl. (2016). Comparing tree-ring and permanent plot estimates of aboveground net primary production in three eastern U.S. forests. Ecosphere 7(9): e01454.

Dye, A. (2018). Annual aboveground biomass growth of temperate forests in eastern North America. PhD Dissertation, West Virginia University.

Fahey, R. T., and C. Lorimer. (2014). Persistence of pine species in late-successional forests: Evidence from habitat-related variation in stand age structure. Journal of Vegetation Science 25: 584– 600.

Fahey, R. T., A. T. Fotis, and K. Woods. (2015). Quantifying canopy complexity and effects on productivity and resilience in late-successional hemlock-hardwood forests. Ecological Applications 25: 834–847.

Fitzpatrick, M. C., E. L. Preisser, A. Porter, J. Elkinton, and A. E. Ellison. (2012). Modeling range dynamics in heterogeneous landscapes: invasion of the hemlock woolly adelgid in eastern North America. Ecological Applications 22: 472–486.

Fritts, H. C. (1976). Tree rings and climate. The Blackburn Press, Caldwell, New Jersey.

Gleason, H. A. (1923). Botanical observations in northern Michigan. Journal of the NewYork Botanical Club 24: 273–283.

Goldblum, D., and L. S. Rigg. (2005). Tree growth response to climate change at the deciduous-boreal forest ecotone, Ontario, Canada. Canadian Journal of Forest Research 35: 2709–2718.

Guyette, R. P., M. C. Stambaugh, D. C. Dey, and R. Muzika. (2012). Predicting ecosystem fire using the physical chemistry of climate. Ecosystems 15: 322–335.

Haase, P., J. D. Tonklin, S. Stoll, B. Burkhard, M. Frenzel, I. R. Geijzendorffer, C. Häuser, et al. (2018). The next generation of site-based long-term ecological monitoring: Linking essential biodiversity variables and ecosystem integrity. Science of the Total Environment 614: 1376–1384.

Hart, J., S. L. van de Gevel, J. Sakulich, and H. D. Grissino-Mayer. (2010). Influence of climate and disturbance on the growth of Tsuga canadensis at its southern limit in eastern North America. Trees 24: 621–633.

Hessl, A., and N. Pederson. (2012). Hemlock Legacy Project (HeLP): a paleoecological requiem for eastern hemlock. Progress in Physical Geography 37: 114–129.

Hinkel, K. M., and F. E. Nelson. (2012). Spatial and temporal aspects of the lake effect on the southern shore of Lake Superior. Theoretical and Applied Climatology 109: 415–428. doi 10.1007/s00704-012-0585-2.

Holmes, R. L. (1983). Computer-assisted quality control in tree-ring dating and measurement. Tree- Ring Bulletin 43: 69–78.

IPCC. (2014). Climate change 2014: Synthesis report. Contributions of working groups I, II, and III to the fifth assessment report of the Intergovernmental Panel on Climate Change, Geneva, Switzerland.

Lane, C., D. D. Reed, G. D. Mroz, and H. O. Leichty. (1993). Width of sugar maple (Acer saccharum) tree rings as affected by climate. Canadian Journal of Forest Research 23: 2370–2375.

Larson, E. R., S. Allen, N. J. Flinner, S. G. Labarge, and T. C. Wilding. (2013). The need and means to update chronologies in a dynamic environment. Tree-Ring Research 69: 21–27.

Magruder, M., S. Chhin, A. Monks, and J. O’Brien. (2012). Effects of initial stand density and climate on red pine productivity within Huron National Forest, Michigan, USA. Forests 3: 1084– 1103.

Maxwell, J. T., G. L. Harley, and T. J. Matheus. (2015). Dendroclimatic reconstructions from multiple co-occurring species: A case study from an old-growth deciduous forest in Indiana, USA. International Journal of Climatology 35: 860–870.

National Oceanic and Atmospheric Administration. (2016). National Climatic Data Center, Divisional Climate Data. Available at time-series/103/pdsi. (Accessed September 14, 2017).

Orwig, D. A., D. R. Foster, and D. L. Mausel. (2012). Landscape patterns of hemlock decline in New England due to the introduced hemlock woolly adelgid. Journal of Biogeography 29:1475–1487.

Palmer, W.C. (1965). Meteorological Drought. Research Paper, U.S. Weather Bureau, Washington, D.C., No. 45, 58 p.

Paradis, A., J. Elkinton, K. Hayhoe, and J. Buonaccorsi. (2008). Role of winter temperature and climate change on the survival and future range expansion of the hemlock woolly adelgid (Adelges

Page  204 204 THE GREAT LAKES BOTANIST Vol. 58

tsugae) in eastern North America. Mitigation and Adaptation Strategies for Global Change 13: 541–554.

QGIS Development Team. (2015). QGIS Geographic Information System. Open Source Geospatial Foundation Project:

Rochner, M. L., H. W. Kelley, C. S. Wilson, T. Bennett, and H. D. Grissino-Mayer. (2017). Dendrochronological dating of the Warner House and barn (20LV334), Livingston County, Michigan, USA. Dendrochronologia 43: 4–11.

Saladyga, T., and R. S. Maxwell. (2015). Temporal variability in climate response of eastern hemlock in the central Appalachian region. Southeastern Geographer 55: 143–163.

Sommers, L. M. (1977). Atlas of Michigan. Michigan State University Press, East Lansing.

Statistics Canada (2016). 2016 Census—Boundary files, Provinces and territories. www12.statcan.

Stokes, M. A., and T. L. Smiley. (1968). An introduction to tree-ring dating. University of Chicago Press, Chicago, Illinois.

Tardiff, J., J. Brisson, and Y. Bergeron. (2001). Dendroclimatic analysis of Acer saccharum, Fagus grandifolia, and Tsuga canadensis from an old-growth forest, southwestern Quebec. Canadian Journal of Forest Research 31: 1491–1501.

Trotter, R, and K.S. Shields. (2009). Variation in winter survival of the invasive hemlock woolly adelgid (Hemiptera: Adelgidae) across the eastern United States. Environmental Entomology 38: 557 – 587.

U.S. Census Bureau, Department of Commerce (2017a). TIGER/Line Shapefile, 2017, nation, U.S., Current State and Equivalent National. tiger-line-shapefile-2017-nationu- s-current-state-and-equivalent-national. U.S. Census Bureau, Department of Commerce (2017b). TIGER/Line Shapefile: Water, 2017, Area Hydrography -Marquette County, Michigan. 2017&layergroup=Water White, M. A., and D. J. Mladenoff. (1994). Old-growth forest landscape transitions from pre-European settlement to present. Landscape Ecology 9: 191–205. Whitney, G. G. (1987). An ecological history of the Great Lakes forest of Michigan. Journal of Ecology 75: 667–684. Woods, K. D. (2007). Predictability, contingency, and convergence in late succession: Slow systems and complex data-sets. Journal of Vegetation Science 18: 543–554. Woods, K. D. (2014). Multi-decade biomass dynamics in an old-growth hemlock-northern hardwood forest, Michigan, USA. Peer J 2:e598. Zang, C., and F. Biondi. (2016). treeclim: an R package for the numerical calibration of proxy-climate relationships. Ecography 38: 431–436.