TREE RINGS AND CLIMATE IN THE GREAT LAKES REGION PAST, PRESENT , AND FUTURE By Scott Matthew Warner A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Plant Biology Doctor of Philosophy Ecology, Evolution, and Behavior Program Dual Major 2021 ABSTRACT TREE RINGS AND CLIMATE IN THE GREAT LAKES REGION PAST, PRESENT , AND FUTURE By Scott Matthew Warner As climate change unfolds it is necessary to gain a better understanding of how tree growth is affected by contemporary climate. This can provide information about both prerecorded climate and future growth responses to climate change. Yet, tree growth and its relationship with climate remains understudied in one section of the northeastern United States, the Great Lakes Region. Here, I use dendrochronology to (1) reconstruct 500 years of moisture conditions on South Manitou Island, Lake Michigan, (2) compa re growth and growth - climate relationships within/among nine species along a latitudinal gradient from southern Indiana to Upper Michigan, and (3) forecast future growth under climate change. I find that (1) while drought was a regular occurrence since the mid - 1500s, the most severe droughts and the most variable conditions overall occurred in the 20 th century, (2) the predominant climatic factors associated with growth in the region are summer temperature (negative relationships) and summer precipitation ( positive relationships), and that the influence of these factors was strongest in the south, and (3) future growth was projected to decline over the rest of this century as climate change proceeds , independent of latitude . iii For Nola, beside me throughout iv ACKNOWLEDGMENTS I thank Frank Telewski , for his advice, support, and patience, and for being an excellent major advisor in general. Andrew M. Jarosz went far beyond the role of a typical guidance committee member providing ample amounts of support, feedback, and ideas. In particular, I thank D r. Jarosz for helping to conceive the study that comprises Chapter 3. I am also grateful to my other committee members, Jeffrey Andresen and Steve Chhin for their guidance. S pecific ally , thanks to Dr. Andresen for his advice on modeling and selecting clima te projections. Thanks to Dr. Chhin for advice on statistics and dendrochronological methods. Alan Arbogast, Justin Maxwell, and William Lovis also provided substantial advice. Many other persons also provided advice , too many to list . Many persons provide d help in the lab and field, particularly Opal Jain. Thanks also to Alan Arbogast, Jill Grimes, Andrew Jarosz, William Lovis, Jaime Salisbury, Andrew Warner, and Nola Warner for help in the field. Thanks to Hazel Anderson, Gillian Chirillo, Matthew Hadden, and Sam antha Jeffries for help in the lab. For help in the field and lab, thanks to Ross Chinavare, Garrett Mullanix, Hunter Salisbury, and Frank Telewski. Thank you to the landowners/stewards who generously allowed me to sample on their land: The Huron M ountain Club, the Indiana Audubon Society, the Jacobson Family (Tustin, MI), Kalamazoo Nature Center, the Michigan Audubon Society, the Michigan Department of Natural Resources, the National Parks Service (NPS Permit No. 85839), the Peterson Family (Stonin gton Peninsula, MI), Saginaw County Parks, the United States Forest Service, and the University of Michigan Biological Station. v Several funding agencies provided support. Thank you to the Plant Science Fellowship (Michigan State University [MSU] ) , the Alfred H. and Jean Goldner Endowment, W.J. Beal Botanical Garden (MSU) , the Huron Mountain Wildlife Foundation, the James E. Rodman Endowment, the UAW Cal Rapson Endowment, donors to the William A. Lovis Endowment for Environmental Archaeology, the D epartment of Geography, Environment and Spatial Sciences (MSU), the Department of Plant Biology (MSU), the Ecology, Evolution, and Behavior Program (MSU), the Paul A. Taylor Endowment, and the Dissertation Completion Fellowship (MSU). For funded internship s, thank you to the MSU Herbarium (MSC) and the Natural Science Collections Internship (MSU). To Justin Maxwell, I am grateful for providing data for A. saccharum , C. ovata , L. tulipifera , Q. alba , and Q. rubra at Pioneer Mothers Memorial Forest. To Alex Dye, I am grateful for providing data for T. canadensis at Huron Mountain Club. Thank you to Edward Cook for providing an unpublished T. occidentalis chronology from South Manitou Island. For hospitality, thank you to the rangers of South Manitou Island, Carl Wilms of Mary Gray Bird Sanctuary, the Kalamazoo Nature Center, and the Huron Mountain Wildlife Foundation. Thank you to Inter - Research for permitting me to reprint with slight formatting modifications a paper from Climate Research which comprises chapter 2 of this document: S.M. Warner, S.J. Jeffries, W.A. Lovis, A.F. Arbogast, & F.W. Telewski. 2021. Tree ring - reconstructed late summer moisture conditions, 1546 to present, northern Lake Michigan, USA. Climate Research . 83:43 56. https://doi.org/10.3354/cr01637 . © Inter - Research 2021. Resale or republication not permitted without written consent of the publisher. vi Thank you to those who most inspired me in my undergraduate program, Todd Barkman and Steve Keto. To Ralph Reitz, thanks for being the first person to t each me about botany. I am thankful to my immediate, original, and in - law family members, particularly my wife, parents, and children for their support. vii TABLE OF CONTENTS LIST OF TABLES . ......................................................................................................................... . .. .. i x LIST OF FIGURES . .............................................................. . ................................................ ........ ..... . x CHAPTER ONE . ............................................................................................................................. ... 1 INTRODUCTION . .................................................................................... ...................................... .... 1 Prehistoric, Historic, and Future Climate Change ............................................................... 1 Plant Responses to Climate Change ........................... .. ....................................................... 2 Forests and Climate of the Great Lakes Region .................................................................. 9 Study Objectives . ..................................................................... ......................................... 11 APPENDIX . ......................................................................................................................... 12 CHAPTER TWO . .................................................................. .......................................................... .1 4 TREE - RING - RECONSTRUCTED LATE - SUMMER MOISTURE CONDITIONS, 1546 TO PRESENT, SOUTH MANITOU ISLAND, LAKE MICHIGAN, USA ...................................... ............ ..... ............. .1 4 5 Abstract..................................................................................................................... ........ 1 5 Introduction............................................................................................................ .......... 1 7 Methods...................................................................................................................... ...... 21 Results ...................................................................................................... ...................... ... 2 5 Discussion ......................................................................................................................... 3 0 Conclusion .................................................................................. ...................................... .3 6 APPENDIX . ....................... .. ................................................................................................ 38 CHAPTER THREE . ...................................................................................................................... .....4 7 IN GREAT LAKES R EGION MESIC FORESTS, CONTEMPORARY GROWTH - CLIMATE RELATIONSHIPS SUGGEST THAT SOUTHERLY TREE POPULATIONS ARE MOST VULNERABLE TO CLIMATE CHANGE ................. ............................................................................4 7 Abstract............................................... . ............................................................................. 4 7 Introduction . .................................................................................................................... .4 8 Method s ............................................................................................................................ 51 Th e Study Region ................................................................................................. . 51 The Study Species ...................................................... ........................................... 52 Field Methods . ............................................................... ....................................... 52 Core Processing, Cross - Dating, and Detrending ...................................... ............. 53 Quantifying and Comparing Growth - Climate Relationshi ps . ....................... .. .... ... 55 Results ............................................................................................................................ .. .5 8 General Chronology Characteristics ................................................... ...................5 8 Comparing Tree - Ring Chronologies Across Populations ....................................... 5 8 The Strongest Growth - Climate Relationships, 190 3 2004 Single - Interval ........... 60 viii Other G rowth - Climate Relationships, 190 3 2004 Single - Interval ............ .......... .. 61 Comparing Growth - Climate Relationships Across Populations, 190 3 2004 Single Interval .................................................................................................................. 62 Changes in Growth - Climate Relationships, 190 3 2004 Moving - Intervals ........... . 6 4 Discussion . .............................................. .................................. ........................................6 5 Comparing Tree - Ring Chronologies Across Populations ........................... ............6 5 The Strongest Growth - Climate Relationships, 190 3 2004 Single Interval ........... 68 Other Growth - Climate Relationships, 190 3 2004 Single - Interva l.. . ................... .. 7 0 Changes in Growth - Climate Relationships, 190 3 2004 Moving - Intervals . ........... 7 2 Conclusion ......................................................................................................................... 7 3 APPENDIX ..................................................................................................................... ..7 5 CHAPTER FOUR . ......................................................................................................................... . 1 0 1 CLIMATE CHANGE IS PROJECTED TO HINDER PRODUCTIVITY OF TREE SPECIES IN INDIANA AND MICHIGAN OVER THE 21 st CENTURY ............................................................ ............................... 1 0 1 Abstract . .......................................................................................................................... 1 0 1 Introduction . ................................................................................................................... 1 0 2 Methods............................................................................................................. .............1 0 4 Field Methods and Tree - Core Processing .................................. .........................1 0 4 Growth - Climate Model ing ................... ........................................................ ...... . 1 0 5 Future - Growth Projections ................................................................................ . 1 0 7 Results.................................................. ..................................................... ......................1 0 9 Model Calibration and Verification .. ................................................................... 1 0 9 Characteristics of Validated Models . .................................................................. 1 0 9 Comparison of Historic and Projected Growt h.... .............................................. . 1 1 0 Future - Growth Projection s.. .................... ....... ................................ .................... 1 1 1 Discussion . ...................................................................................................................... 1 1 2 Conclusion ...................................................................................................................... .1 1 8 A PPENDI .................................................................................................................. .. 1 1 9 CHAPTER FIVE ................. ............................................................................................................ 1 2 7 CONCLUDING REMARKS . ......................................................................................................... 1 2 7 LITERATURE CITED . ............................................................................................................... ...... 1 3 1 ix LIST OF TABLES Table 2.1. Statistics and coefficients for models relating year t year t+1 Thuja occidentalis ring width index ........ ........................................................ ... .. .. . 46 Table 3.1. Site characteristics........................................... ........................................ . . .................. . 86 Table 3.2. General chronology characteristics............................................................ .. .................8 7 Table 3.3 . Correlations (Pearson r) between loadings on each principal component (PC) vs. latitude. .. .................... ....................................................................................................... 90 Table 3.4 . Partial correlation m atrix (T mean only) for the growth - climate coefficient (Pearson r) for each of 34 climate variables for each population ... .................................................... 9 1 Table 3.5 . Partial correlation matrix (Ppt only) for the growth - climate coefficient (Pearson r) for each of 34 climate variables for each population . .................... .................................. 9 4 Table 3.6 . The monthly mean temperature (T mean ) and total p recipitation (Ppt) variables with the steepest slope in a regression of growth - climate coefficient (Pearson r) vs. time ... ..9 7 Table 3.7 . Correlation matrix (Pearson r) for 190 3 2004 mean annual temperature among study sites............................. ............................................................................ .............. ..9 9 Table 3.8 . Correlation matrix (Pearson r) for 190 3 2004 total annual precipitation among study sites ...... ......... .................................................................................................................. 1 0 0 Table 4.1 . Calibration and verification statistics and parameters for models relating climate and ring - width index ........................................................................... ............... 1 2 3 1953 calibration period were fit over the entire common 6 x LIST OF FIGURES Figure 1.1 . Global average annual temperature deviations from 1850 1900 reference period .. 1 3 Figure 2.1. Study site, South Manitou Island, is indicated by a red square off northwest lower Michigan 39 Figure 2.2. Rooted tree in paleosol in the dune field on South Manitou Island ................. ......... . 40 Figure 2.3. Ring width index (purple), sample size (green), expressed population signal (EPS; dark blue) over time, and year at which our chronology was truncated due to insufficient sample size (<1 tree [<3 cores]; black vertical line) ... . ................ .. ... ............................. ...... 4 0 Figure 2.4. Evolution of the relationship between observed year t July September Palmer Z Index and year t+1 Thuja occidentalis ring - width index........................................ . ... ........... 41 Figure 2.5. Standardized observed vs. reconstructed Palmer Z i ndex (a) and standardized observed, reconstructed Palmer Z Index vs. time (b)......... ..................................... .. ......... 42 Figure 2.6. Reconstructed annual (gray) and 9 yr moving average of (black) standardized July September Palmer Z Index, 1546 2014............................................... .......... . ........... 43 Figure 2.7. The most extreme 10 yr non - overlapping plu vial and non - overlapping drought events, the entire reconstruction ......................... ........................... . .................... .................. ....... 43 F igure 2.8. Reconstructed extreme 5 yr non - period of the reconstruction and observational record (a) and the return interval of ............... ............. . ................... 44 Figure 2.9. Mean and SD of standardized reconstructed Palmer Z index per century (a) and number of extreme years and their frequency throughout each century and the entire reconstructed period (b) .. . ...................................................................... ....................... ... 45 Figure 3.1. The study sites and their latitude. ............................................................... ............... .. 76 Figure 3.2. Detrended, standardized residual ring - width - index chronologies............................. .. 77 Figure 3.3 . Loadings of each stand on principal components (PCs ) 1 & 2 (a) and on PCs 3 & 4 (b) ............................................ .......................................................................................... 7 8 xi Figure 3.4 . Latitude vs. (a) growth - Jun Tmean and (b) growth - Jun Ppt correlation coefficients (Pearson r) . ....................... ................................. ................................................................ 80 Figure 3.5 . Growth vs. climate principal component analysis biplots with colors selected to highlight (a) clustering among sites and (b) species ........................................... .............. .. 8 2 Figure 3.6 . Growth - climate coefficients (Pearson r) vs. time for (a) prior - Jul Tmean , (b) prior - Sep Ppt , (c) Jun Tmean , and (d) Jun Ppt ................................................................. ................................ . 8 4 Figure 4.1. Comparison of historic and projected June August mean temperature (a) and June July total precipitation (b) . .............. ................................................................................1 2 0 Figure 4.2. Mean forecasted ring - width indices (RWI) over years 2022 - 2099 under two representative concentration pathways (RCPs 4.5 and 8.5), under each of two GCMs, MRI - CGCM3 (Yukimoto et al. 2012) and GFDL - CM3 (Griffies et al. 2011), relative to simulated RWI over common interval 1903 - 2004 .. .............................. ...........................1 2 2 1 CHAPTER ONE INTRODUCTION Prehistoric, Historic, and Future Climate Change Although climate is ever - changing, it has changed at an unprecedented rate for more than a century. Global mean surface temperature increased by 0.85 °C from 1880 2012 (Hartmann et al. 2013). Further, in the mid - latitudes of the northern - hemisphere (NH) increases in precipitation, near - surface specific humidity, and heavy precipitation events were observed (Hartmann et al. 2013). By 2035, mean surface air temperature is projected to increase by 0.3 0.7 °C und er any of the four Representative Concentration Pathways (RCPs), i.e., greenhouse - gas emissions scenarios identified by the Intergovernmental Panel on Climate Change (Kirtman et al. 2013 ) . By the end of this century, temperature is expected to increase by another 0.3 to 1.7 °C under the lowest - emissions RCP and, under the highest - emissions RCP, by 2.6 to 4.8 °C (Collins et al. 2013). However, the earth has been even warmer in the past than what is predicted. The most extreme example of the current geologic era is the Early Eocene Climatic Optimum (52 48 mya), in which temperature was 9 14 °C higher than present, and atmospheric CO 2 concentration was around 1000 ppm (Masson - Delmotte et al. 2013), compared to 391 in 2011 (Hartmann et al. 2013). Indeed, change s to the climate are the rule. Over the last 2.6 million years, the NH was subjected to about 20 cycles of glaciation and deglaciation (Davis 1983). Considering that living things persisted through warm, CO 2 - rich periods of the distant past and 2 through mor e recent ice ages, life may be resilient to climatic fluctuations including contemporary climate change. A lthough warming is nothing new, the rate of contemporary warming exceeds anything known. For example, during the most recent deglaciation, the fastest warming rate was 0.010 0.015 °C/decade, but from 1880 2012 it was 0.064 °C/decade (Hartmann et al. 2013, Masson - Delmotte et al. 2013). The extent to which plant life will be resilient to ongoing climate change is uncertain. Due to the importance of plants for shelter, food, fiber, medicine, recreation, and ecosystem services, it is necessary to predict how climate change will affect plants. This is particularly salient for trees, many species of which are long - lived and slow to evolve. In in Thuja occidentalis (white cedar; Kelly & Larson 2007). Such long - lived organisms must have som e degree of resilience to environmental change. A 2,000 - year - old white cedar alive today would be one that has experienced both prolonged cool and prolonged warm periods (Fig. 1.1). Most notably, it would have endured the Little Ice Age (ca. 1400 1900) and contemporary warming (ca. 1900 present). However, will such trees be able to cope with the anticipated unprecedented rapidity of further climate change? Plant Responses to Climate Change Many tree species were able to cope with climate change of the past . Tree responses to the most recent glacial cycle included continental - scale migration, adaptation, and extinction, and responses were characterized by differences among species. Davis (1983) summarized tree er colonization of deglaciated land. Some 3 species migrated faster than others, resulting in continual novel community formation and dissolution. For example, Tsuga canadensis (eastern hemlock) and Pinus strobus (eastern white pine) commonly occur together today along with many hardwood tree species in the mesic northern forest community type common in the northern Great Lakes Region (GLR) (Cohen et al. 2014), but in their glacial refugia, T. canadensis and P. strobus grew o n the Atlantic central coastal pla in and on the continental shelf, largely absent from the hardwoods, which are believed to have survived in small isolated refugia far to the south. Some species may not have been able to migrate with sufficient speed, such as the formerly widespread Picea critchfieldii ies underwent substantial adaptation, too. For example, Cheddadi et al. (2016) reconstructed median January temperature throughout the distribution of three European tree species between 10 and 3 kya and compared the reconstruction to modern temperatures. Fagus sylvatica (European beech) formerly occurred in areas cooler than its modern range and Picea abies (Norway spruce) in areas warmer. The ancient climate of the range of Abies alba (European silver fir) was comparable to its modern climate. Tree - ring - b ased growth projections suggest that tree growth, in addition to tree ranges, will also change heterogeneously in response to climate. This is true within species (e.g., Huang et al. 2013, Chhin 2015, Rimkus et al. 2018), and it is also true among species (e.g., Huang et al. 2013, Su et al. 2015, Rahman et al. 2018). Not only will tree growth be affected by climate change itself but also likely by increasing atmospheric CO 2 . Higher CO 2 levels may ameliorate the additional 4 evapotranspirative demand that wil l come with higher summer temperatures, allowing plants to keep their stomata closed more often, reducing moisture loss. This effect, CO 2 fertilization, has been found in greenhouse trials (Bazzaz et al. 1990), and in forest trees growing in nature, but ev idence is mixed. In observational studies, no support was found in the tropics (Van der Sleen et al. 2015) or boreal Canada (Girardin et al. 2016). However, support was found in the temperate western and eastern U.S. ( Wang et al. 2006 and McMahon et al. 20 10 , respectively ) and the GLR (Cole et al. 2010). This was corroborated in experimental studies (Telewski et al. 1999, Walker et al. 2019). Lamarche et al. (1984) attributed unprecedented recent growth of ancient Pinus longaeva (bristlecone pine) in the Am erican southwest to rising CO 2 , but Salzer et al. (2009) claimed this was due to rising temperatures. It may be that CO 2 fertilization is sustainable only until nitrogen becomes depleted to the point of being a greater limiting factor (Norby et al. 2010). Though CO 2 fertilization is equivocal, other effects of nascent climate change are clearer. In some cases, one need not turn to models or theory to understand the effects of climate change but can look at effects already observed. In semi - arid forests of China, climate change has led to increases in drought, fire, plant - pathogen outbreaks, tree mortality and tree - growth reductions (Liu et al. 2013). Increased tree mortality has also been found on the Tibetan Plateau (Liang et al. 2015) . In the United St ates, recent severe droughts led to widespread tree mortality in Texas (Crouchet et al. 2019) and reduced radial growth in Indiana (Kannenberg et al. 2019) . Further, climate change has led to phenological changes. Willis et al. (2008) compared contempo rary phenological and abundance data to mid - 19 th - century data. Those species which 5 adjusted their phenology in response to climate change over the period increased in abundance, while those that failed to adjust decreased. Hufkens et al. (2012) examined ho w an early, intense spring warm - up followed by a late - spring frost affected tree canopy development and forest - wide productivity. All three study species leafed out early in response to unseasonably warm weather. Betula alleghaniensis (yellow birch) and Fagus grandifolia (American beech) were able to endure the subsequent frost with little damage. However, canopy development of Acer saccharum (sugar maple) was suppressed, leading to declines in forest productivity. Phenological changes can result in mi smatch between interacting organisms. Well - known are the disruptions to pollination that occur when plants and their animal pollinators differentially respond to climate change (Pyke 2016, Hutchings 2018). Additionally, wildflowers, particularly spring eph emerals, those early emerging forest - understory herbs whose aboveground parts soon senesce after the forest canopy develops, respond differently than do overstorey trees. Heberling et al. (2019) compared 19 th century and present - day data and found that, in response to climate change in eastern North America, the leaf - out date of both wildflowers and trees had advanced, but in trees it had advanced at a more rapid rate, shortening the crucial window during which spring wildflowers capture most of their sunli ght. Climate plays a large role in determining the range limit of tree species (Siefert et al. 2015), and thus it is no surprise that the distributions of trees and other plants have changed as climate change has proceeded. Generally, species are moving upward in elevation (Kelly & Goulden 2008, Harsch et al. 2009, Kharuk VI et al. 2010, Kopp & Cleland 2014, Bruening et al. 6 the Santa Catalina Mountains, Arizona revealed that the mean lower range limit across species had moved upward, though some individual species were exceptions (Brusca et al. 2013). A wider exception to the trend was found in a large swath of California, where a comparison of mean 1930s and pre sent - day elevation of 64 plant species revealed a significant downhill shift, attributable to a regional decrease in climatic water deficit ( Crimmins et al. 2011). Just as species are moving upward in elevation, so too are they thriving in previously margi nal habitats (Sturm et al. 2001) and moving upward in latitude (Boisvert - Marsh et al. 2014). Clearly, plant range s, including those of trees, are shifting . The question is whether tree migration can keep pace with climate change in the long - term. Recent r eports suggest that some species will succeed, and others fail (Chen et al. 2011). Emigration and immigration can have large, cascading effects. For example, Rodriguez - Cabal et al. (2013) found that the local in Pata gonia led to the extinction of the mistletoe (a keystone species), a hummingbird (important pollinator), and both a marsupial and a bird (important seed dispersers). st ructure, storage of biochemicals, competing for light with other plants, and transporting water, food, and other substances. In trees, stem growth has already been affected by climate change. Gamache and Payette (2004) studied height growth of Picea marian a (black spruce) at and near treeline in northern Quebec. Historically, growth at treeline was lower than growth at more southerly locations, but from the 1970s through the end of the study period there was no difference between the two. Latte et al. (2016 ) studied the radial growth of F. sylvatica that occurred from 1930 2008 in Belgium. They found a marked decrease in overall growth and an 7 increase variability beginning in the 1950s 1960s and 1970s 1980s, respectively, and persisting through the study per iod, which they attributed to climate change, soil compaction, and nitrogen deposition. At the southern range limit of F. sylvatica , radial growth decline over the most recent three decades examined was observed. The decline was greatest at low elevations (Jump et al. 2006). By contrast, climate - change - attributable radial growth increase was observed at an alpine tree line on the Tibetan plateau. Abies faxoniana increased its growth rate 16 - fold over 1900 2012 relative to 1764 1899. This was attributed t o increases in atmospheric CO 2 and soil moisture (Silva et al. 2016). Pretzsch et al. (2014) found accelerated tree growth and forest development in Europe as climate change proceeded. Along a latitudinal gradient in the central Siberian Taiga, Larix cajanderi growth did not change over the growth period studied, but growth of Picea obovata (Siberian spruce) and P. sylvestris accelerated, especially in P. obovata. Consistent with among - species differences in post - glacial recolonization rates, such among - species differences in contemporary climate change responses are the norm. Granda et al. (2013) studied growth of four co - occurring species in a Mediterranean forest. Growth of one species had accelerated, that of another declined, and that of two more did not change. Not only do climate - change responses differ among species but wit hin species, too. Along gradients of altitude and latitude, growth of populations at low elevation or latitude is generally hindered and that of high - altitude/latitude populations helped, consistent with wider trends in the field of global change biology ( Parmesan 2006). Dulamsuren et al. (2017) studied F. sylvatica in Germany in the middle of its range. At low elevations, growth had declined since the 1980s, while at high elevations it increased. Similar results were found in P. abies in east - 8 central Europ e and in three conifer species in British Columbia (Lo et al. 2010, Ponocna et al. 2016). However, exceptions have been found. Housset et al. (2015) studied Thuja occidentalis (white cedar) in boreal Canada over three degrees of latitude. Northerly populat ions were unable to take advantage of recent warming, and in fact their growth recently declined due to concomitant moisture stress. Though incipient climate change has been substantial in some regions, it generally is still relatively minor and within th e range of natural historic variability . Thus, growth trends of many tree populations have not yet been affected. In these cases, one can still use contemporary relationships between growth and climate to anticipate the effects of ongoing climate change. F indings are generally consistent with the expectation that low - altitude/latitude populations will suffer and high - altitude/latitude populations benefit. For example, in a meta - analysis of 378 P. mariana tree - ring - O rangeville et al. (2 016) found that in the southern part of the chronology network, growth - temperature relationships were negative, but they gradually became more positive moving north, until by around 49° N the majority were positive. Similar results were found for Pinus pin ea in Iberia (Natalini et al. 2016) and for Sequoia sempervirens (coast redwood) in California (Carroll et al. 2014), but exceptions are found. No latitudinal trend in growth - climate relationships was found for Sorbus torminalis (wild service tree) in Euro pe (Rasmussen 2007). As for altitudinal gradients in growth - climate relationships, Picea schrenkiana spruce) in northwest China had negative growth - temperature and positive growth - precipitation relationships at low elevation and positive grow th - temperature relationships at high elevation 9 (Huo et al. 2017). However, no such trend was found in a different part of northwest China in Juniperus przewalskii Clearly, growth - climate relationships differ withi n and among species, unsurprising given that each species is physiologically adapted to its environment, and, within species, populations are locally adapted. Robakowski et al. (2011) found, among four broad - leaved tree species of eastern North America, di fferences in photosynthetic temperature optima and differences among different provenances within species, even when grown in common gardens for more than a decade. Further, tree species differ in how they adjust stomatal conductance as soil and air moistu re change (Hinckley et al. 1979, Pataki & Oren 2004). Thus, species and populations differ in their response to climate, and so too climate change differs from region to region. For example, high latitudes are warming more than low latitudes. Precipitation has increased in eastern North America, and many other places around the globe, but it has decreased in the Sahel Region of Africa and scattered pockets elsewhere (Hartmann et al. 2013). Because each region has its unique climate change prognosis and each tree species its own response, it is necessary to study a variety of species across different regions to understand how each species will respond to climate change in different parts of its range. Forests and Climate of the Great Lakes Region The GLR is an area with ample, variable forests which play important ecological and economic roles. In Michigan, the annual state and regional economic impact of the timber industry is $14 billion (Leefers 2017). Further, there are more than 20 million acr es of forest in 10 that state, including 56% of the land area and amounting to 1,400 trees for every person (Dickmann & Leefers 2016). The region has a high concentration of distribution limits of tree species, both northern and southern limits, because of th e tension zone a diffuse region in which the hardwood forests dominant in the south gradually give way to the mixed hardwood - conifer forests of the north (Anders e n 2005) allowing the comparison of southern - and northern - range - margin populations. Most of t he region experienced a cycle of glacial and nonglacial conditions over at least the last 800,000 years, except the southern extreme, which did not experience the two most recent glaciations, although it may have been affected by more ancient ones. The mos t recent for most of the current soil types, topography, and lake levels in the study region (Larson & Kincare 2009). The variation in soils affects modern veget ation, with mesic plants growing in fertile, coarse - textured soils, and xeric plants in dry, sandy soils (Harman 2009). The contemporary climate of the region is unique due to the temperature - moderating, moisture - influencing Great Lakes. For example, due to the predominantly west - flowing wind, both major peninsulas of Michigan are downwind of a Great Lake, resulting in an o verall more wet, cloudy, snowy, and moderate climate than areas upwind (Andresen & Winkler 2009). Recent climate change has resulted in a generally warmer, wetter regime (Andresen 2012), and long - term Great Lakes Region forecasts predict these trends to co ntinue (Hayhoe et al. 2010, Christensen et al. 2013, Byun & Hamlet 2018). 11 Study Objectives With the unique climate and diversity of forest types in the Great Lakes Region, it is important to examine relationships between tree growth and climate in the region. This will provide information about both future responses to climate and prerecorded climate (climate preceding the era of widespread instrumentally derived records kept by humans ). Yet, the relationship between tree growth and climate has been in sufficiently examined in the region with relatively few published tree - ring chronologies coming from there (Zhao et al. 2019). In this dissertation, I use 47 tree - ring chronologies from Indiana and Michigan to accomplish four major objectives. First, I rec onstruct ed 500 years of summer moisture conditions on South Manitou Island in northern Lake Michigan to put current and predicted future climate into a historical perspective (Chapter 2) . Second, I quantified relationships between tree growth and climate in nine species along a latitudinal gradient to establish baseline data for this understudied region and facilitate comparisons within and among species (Chapter 3) . Third, I examine d the temporal stability of growth - climate relationships to determine whet her tree rings were an appropriate proxy to use for climate reconstruction in the region (Chapter 3) . Finally, I project ed growth over the rest of this century in the context of climate change via parsimonious growth - climate models to identify vulnerable s pecies/regions and to identify potentially climate - change - resilient populations (Chapter 4). 12 APPENDIX 13 APPENDIX Figures Figure 1.1 . Global average annual temperature deviations from 1850 1900 reference period . Note that the medieval warm period (Middle Holocene Climatic Optimum) was a predominantly North Atlantic phenomenon and hence does not register on this global scale. Thanks to Ed Hawkins for sharing the figure, which he made from PAGES2k (Neukom et al. 2 019) and HadCRUT 4.6 (updated version of Morice et al. 2012). 14 CHAPTER TWO TREE - RING - RECONSTRUCTED LATE - SUMMER MOISTURE CONDITIONS, 1546 TO PRESENT, SOUTH MANITOU ISLAND, LAKE MICHIGAN, USA SCOTT M. WARNER 1,* , SAMANTHA J. JEFFRIES 2 , WILLIAM A. LOVIS 2,3 , ALAN F. ARBOGAST 4 , FRANK W. TELEWSKI 1,5 1 ECOLOGY, EVOLUTION, AND BEHAVIOR PROGRAM, DEPARTMENT OF PLANT BIOLOGY, MICHIGAN STATE UNIVERSITY, EAST LANSING, MICHIGAN 48824, USA 2 DEPARTMENT OF ANTHROPOLOGY, MICHIGAN STATE UNIVERSITY, EA ST LANSING, MICHIGAN 48824, USA 3 MICHIGAN STATE UNIVERSITY MUSEUM, MICHIGAN STATE UNIVERSITY, EAST LANSING, MICHIGAN 48824, USA 4 DEPARTMENT OF GEOGRAPHY, ENVIRONMENT AND SPATIAL SCIENCES, MICHIGAN STATE UNIVERSITY, EAST LANSING, MICHIGAN 48824, USA 5 W. J. BEAL BOTANICAL GARDEN, MICHIGAN STATE UNIVERSITY, EAST LANSING, MICHIGAN 48824, USA * Corresponding author: warner91@msu.ed u 15 A cknowledgments Thank you to Inter - Research for permitting us to reprint with slight formatting modifications a paper from Climate Research which comprises this chapter: S.M. Warner, S.J. Jeffries, W.A. Lovis, A.F. Arbogast, & F.W. Telewski. 2021. Tree ring - reconstructed late summer moisture conditions, 1546 to present, northern Lake Michigan, USA. Climate Research . 83:43 56. https://doi.org/10.3354/cr01637 . © Inter - Research 2021. Resale or republication not permitted without written consent of the publisher. We thank Jill Grimes for assistance with field work, Matthew J. Hadden for assistance with tree ring measuring and dating, and Edward R. Cook for providing an unpublished tree - ring chronology and for advice on crossdating and detrending. We thank the National Park Service for allowing us to sample trees (NPS Permit No. 85839) and the South Manitou Island rangers for their hospitality: providing transportation to and from the island and providing lodging and access to a vehicle while on the island. For accepting our cores into their collection, we thank the Michigan State Universit y Herbarium (MSC). For funding we are grateful to the Alfred H. and Jean Goldner Endowment, donors to the William A. Lovis Endowment for Environmental Archaeology, and the Department of Geography, Environment and Spatial Sciences. To three anonymous review ers and a journal editor we are grateful for insightful suggestions which h elped us to improve our presentation. Abstract Drought can affect even humid regions like northeastern North America, which experienced significant, well - documented dry spells in the 1930s, 50s, 60s, and 80s, and proxies 16 tell us that in the years before instrumentally recorded climate, droughts could be even more severe. To get a more complete picture of pre - recorded climate, the spatial coverage of proxy - based climate reconstructi ons must be extended. This can better put in context past, current, and future climate, and it can lend anthropological and historical insights. With regard to tree rings as climate proxies, however, there is increasing evidence that relationships between tree growth and climate can be inconsistent over time, in some cases decreasing the utility of tree rings in the representation of climate. We developed a chronology from white cedar Thuja occidentalis ich we modeled the Z index). The relationship was consistent across the period of instrumentally recorded climate, e, we used the model to century to be the 20th, the least the 18th. The severest decade - mean) occurred in the 1560s, 1600s/10s, 1630s, 1 770s/80s, 1840s, and 1910s/20s, the severest of severe droughts throughout the reconstruction, increasing variability in the 20th century, and expected climate ch ange - enhanced late summer drought, portend a future punctuated with severe droughts. KEY WORDS: Great Lakes climate Lake Michigan climate Thuja occidentalis dendrochronology White cedar dendrochronology Dendroclimatology Drought reconstruction Moisture reconstruction Tree ring reconstruction 17 Introduction In the North American Great Lakes Region (GLR) several episodes of unusual weather have recently occurred, at least from a modern perspective. Anomalously high tempe ratures have been observed: the warmest March on record across the US Midwest in 2012 (National Oceanic and Atmospheric Administration [NOAA] National Centers for Environmental Information 2012a), the second warmest February in the US Midwest in 2017 (NOAA National Centers for Environmental Information 2017a), and 5 consecutive record - high daily temperatures in Chicago, Illinois, September 2017 (NOAA National Centers for Environmental Information 2017b). August 2012 brought severe to exceptional drought to 40% of the US Midwest (NOAA National Centers for Environmental Information 2012b). Conversely, 2013 was the wettest in the upper GLR since 1900 (Knutson et al. 2014). Recent anomalous cold temperatures have also orth American Midwest was the third coldest since 1920 (Wolter et al. 2015), and February 2015 was the third coldest February on record in Michigan (NOAA National Centers for Environmental Information 2015). However, the widespread reliable instrumental climate record extends only to 1895 in the region (Andresen 2012), so it is difficult to say how unusual recent events were. Further, climate has been dynamic was in Michigan an increasing trend per decade (linear correlation, r 2 = 0.263, 2 - tailed p < 0.0001; http://www.ncdc.noaa.gov/cag/statewi de/time - series/20/tavg/ann/9/1895 - 2019 ). Throughout 18 5.94°C, with was a warmer period, on average 6.51°C; it st arted very warm and had a marginally significant cooling trend overall (p = on average 7.19 °C, with a marginally significant warming trend (p = 0.077). Precipitation has also been on the rise in Michigan. , annual values increased by 1.09 cm decade 1 (linear correlation, r 2 = 0.224, 2 - tailed p < 0.0001; http://www.ncdc.noaa.gov/cag/stat ewide/time - series/20/pcp/ann/9/1895 - 2019 at 766 mm yr 1 , and it was without wetter, mean of 801 mm yr 1 . The most recent period, a trend of 7.81 mm yr 1 (p = 0.012) and a mean of 878 mm yr 1 . These moistening trends are consistent with other regions of eastern North America, including Indiana/ Ill inois (Mishra & Cherkauer 2010), southeastern New York (Pederson et al. 2013), and Iowa (Ford 2014). Trends in snowfall are mixed in the GLR, being modulated by location relative to Great Lakes shorelines. Areas near the leeward shore, such as western lowe r Michigan, receive greater snowfall than inland and windward coastal areas due to the lake effect (Scott & Huff 1996). In lake effect areas alone, snowfall has been increasing since the 1940s. In non - lake effect areas, however, snowfall did not significan tly change over the same period (Burnett 2003, Andresen 2012). Consistent with this trend, Suriano et al. (2019) found that from but snowfall had increased in many areas along the lakes leeward shores. Further, since 1960 19 snowmelt is increasing in lake effect areas such as the east coast of the Georgian Bay and decreasing outside these areas such as the north coast of Lake Superior (Suriano & Leathers 2017). Throughout the Great Lakes Basin, a shift toward earlier snowmelt has progressed since 1960 (Suriano & Leathers 2017). As expected by the precipitation trend, cloudiness has increased since at least the 1960s (Andresen 2012). Thus, the G LR is getting warmer, wetter, cloudier, and in some areas snowier. In the future, additional changes in the regional climatology are anticipated. To account for uncertainty of future anthropogenic contributions to climate change, the Intergovernmental Pane l on Climate Change (IPCC) has established 4 potential scenarios called representative concentration pathways (RCPs) that correspond to a range of greenhouse gas concentrations. From lowest to highest concentration, the pathways are RCP 2.6, 4.5, 6.0, and 8.5. Mid - century projections for eastern North America under, for example, RCP 4.5 predict precipitation increases in the northern part of the continent, particularly in winter, but also summer, and for temperature to increase beyond the range of natural v ariability (Christensen et al. 2013). Projections for the GLR specifically suggest that temperature will increase during all seasons throughout the rest of this century, and precipitation will increase in winter and spring while remaining the same or decre asing in summer and fall (Hayhoe et al. 2010, Byun & Hamlet 2018). Byun & Hamlet (2018) predict a 6.5°C increase by 2100 under RCP 8.5 and 3.3°C under RCP 4.5. Their predictions for precipitation are more equivocal. Under RCP 8.5, there may be a ease in annual precipitation by the 2080s, with large increases in winter/spring and perhaps little change in summer/fall. Thus, climate has changed and is expected to continue changing in the GLR. To put this climatic change in a historical perspective, however, 20 the climate record must be extended into the period preceding reliable records with proxies such as fossilized pollen in lakes/bogs, elemental isotope concentrations, and tree rings. Proxy the cool, wet Little Ice Age trees and logs. Temporally, information about paleoclimates becomes more limited moving backward in time. Here in northeast Nort 2007), but even a quarter of that age is rare, and the mesic climate often results in decay of the most ancient wood. Thus, any tree ring chronology dating back further than perhaps 250 yr about t wice the length of recorded data is a valuable contribution to the understanding of ancient climate. Tree ring chronologies that remain consistently correlated with climate over time are particularly valuable. The field of dendrochronology has traditionall y stood on the principle of uniformitarianism, meaning present - day relationships between natural phenomena are like those of the past (Fritts 1976). However, it is increasingly evident that the factors affecting growth in some trees are inconsistent. For e xample, numerous tree ring chronologies recent decades (reviewed in D relationships have been found occ asionally at lower latitudes, including the northeastern US. Maxwell et al. (2016) found across several species in southern Indiana/Illinois a weakening relationship between growth and Palmer drought severity index (PDSI) and attributed it to a 21 scarcity of droughts in recent decades. Saladyga & Maxwell (2015) found in the central Appalachians weakening relationships between Tsuga canadensis growth and both summer precipitation and late winter temperature, and they attributed these changes to infestation by hemlock woody adelgid and a shift in the temperature regime. Toward a better understanding of both prerecorded climate in the GLR and the temporal stability of using tree rings as climate proxies there, we present a tree ring chronology for the northeaster n part of Lake Michigan extending to 1469 C.E. We use it to reconstruct Z index, a drought metric, to 1546 C.E. Our specific objectives were to (1) quantify moisture over the last several centuries, (2) identify significant droughts a nd pluvials at the 5 to 10 yr scale, (3) put recent droughts and pluvials into historical context by quantifying their return intervals, and (4) examine how the strength of the relationship between tree growth and moisture changes over time. This complemen ts other studies in the region (Cook et al. 1999, Buckley et al. 2004, Pederson et al. 2013, Ford 2014) to present a more complete picture of ancient climate in the greater GLR. Methods In May 2016, we used increment borers to take 101 cores from 43 white cedars Thuja occidentalis on South Manitou Island, Michigan, located in northern Lake Michigan 11 km west of mainland lower Michigan (Fig. 2.1). The area has warm, wet summers and cold, dry winters, with annual precipitation at 791 mm and annual temperature at 6.0 °C. February, the coldest, among the wetter months, averages 19.4 °C and 73 mm (189 22 ftp://ftp.ncdc.noaa.gov/pub/data/cirs/climdiv/ ). We focused our sampling on two subsites: first, the Valley of the Giants - gr owth mesic northern forest and, second, among the dunes just to Valley of the Giants, we selected a variety of trees representing what appeared to be large old - growth specimens as well as some younger t rees in order to create a cross - dated chronology from the living grove, for a total of 57 cores from 27 trees and logs. The dunes are peppered with ghost forests stands of dead trees and logs; these trees and logs were presumably covered and uncovered by t he movement of sand over the decades and centuries, with their preservation aided by sand cover (Fig. 2. 2). We took 44 cores from 16 trees and logs scattered about the dunes. We glued the cores to wooden mounts (Frame and Trim Molding) and surfaced them with sandpaper up to a grit of 400. We skeleton - plotted each core, that is, made graphical representations of the relative narrowness of each tree ring (Stokes & Smiley 1968), focusing first on the cores from the Valley of the Giants. We cross - dated a cross skeleton plots, first on cores within the same tree to build tree - level composite plots, and then across composites to build a stand - level master composite. For individual - core plots which did not match well against the master, we checked the wood fo r potential false rings, and identified potential missing rings, years for which a tree does not lay down xylem along part of its cam bial surface due to stress. From this, we had tentative dates for each tree ring from the Valley of the Giants, and then used the same strategy for the trees from the dunes. As an additional check, we cross - dated our skeleton plots against an independently derived, unpublished T. occidentalis chronology from South Manitou Island (E. R. Cook pers. comm.). 23 We used a scanner to generate images of the cores for computerized ring width program CooRecorder (version 9.0.1, Cybis Elektronik & DATA AB), we measured the width of each tree ring to the nearest hundredth of a millimeter. To test our tentative dates, we ran the ring measurements through the program COFECHA (Holmes 1983), which compares each individual series of ring measurements to the mean of all measurements and identifies a ny cores which do not correlate well against the rest. These problems were addressed through an examination of the wood and a review of the measurements. If there were no ambiguous cracks in the wood, ring boundaries that looked suspiciously false, or meas urement errors, potential false and missing rings were identified through skeleton plot cross - dating and then re - tested in COFECHA until all measurement series were accurately dated. At the conclusion of the dating process, cores were prepared for permanen t storage in a natural history collection. These are being accessioned into the Michigan State University Herbarium (MSC). Ring width series for individual cores were standardized and detrended using program ARSTAN (Cook 1985). We focused on extracting a c limatic signal from the ring - width series. To do so, we fit to each ring - width series a cubic smoothing spline with a length equal to 67% of the total length of the ring width series. Year by year ring width indices (RWI) were calculated by dividing the ob served ring width by that expected according to the fitted spline. This process attenuates long - term confounding trends, and accentuates interannual and interdecadal variation caused by climate fluctuation. To ameliorate unstable variance due to inconsiste nt sample depth over time, we used Briffa r - bar - weighted stabilization in ARSTAN (Osborne et al. 1997, Pederson et al. 2012). The bi - 24 in ARSTAN, which yielded three composite chronologies, standard , residual and ARSTAN . We selected the ARSTAN chronology for further analysis. It retains the pooled variance at the site level, which is hypothesized to be due to climate, and therefore is skilled at examining long - term climate variability (Cook 1985). The quality of the chronology was assessed with the expressed population signal (EPS) statistic, which expresses how well a finite sample represents a theoretical infinite population (Wigley et al. 1984), and has been used widely i n dendrochronological climate reconstructions ( e.g., Buckley et al. 2004, Pederson et al. 2012, Maxwell et al. 2016; but see critique from Buras 2017; see Fig. 2. 3). We calculated EPS using 50 yr segments overlapping 49 yr, i.e. , each 50 yr window represen ted by at least two trees was assessed. Our predetermined EPS minimum for chronology quality was 0.80 (D Orangeville et al. 2018). To quantify the influence of climate on our RWI chronology, we took the common period the interval shared by both the RWI and climate - observation datasets and used ordinary least - squares (OLS) regressio n on one - period. To find the climate variable best correlated with growth and that could thus provide the most information on prer ecorded climate, we tested several variables related to moisture balance, including precipitation, mean temperature, PDSI, Palmer hydrological drought index (PHDI), and Palmer Z index. Pilot analyses revealed the best source for climate data to be US Clima te Division data from NOAA ( ftp://ftp.ncdc.noaa.gov/pub/data/cirs/climdiv/ ). Our site falls within Michigan Division 3 (Fig. 2. 1). With each tested variable, we determined the influence of the most influential month(s). Because tree growth can be affected by both current - and prior - year conditions, the window of months considered ran from the May preceding the year of ring 25 formation to the September of the year of ring formation (Fritts 1976). W e examined - , 2 - , 3 - and 4 - mo nth scales. To clarify, this included for example prior - November through current - February, current - - August alone. We selected the final variable based on r 2 and the consistency of r 2 when comparing the same model across the calibration period and the verification period procedure, we further verified the final model with the reduction of error (RE) and co efficient of efficiency statistics (CE), which both compare the ability of the model to predict observed values within the verification period. RE compares the model's predictive ability relative to the calibration period mean's predictive ability. CE does the same relative to the verification period mean (Fritts 1976, Cook et al. 1999). To confirm that the model was consistent throughout the period, not in only the two specific 60 yr windows chosen for the calibration and verification periods, we quantifie d the r 2 value in each 60 yr window of the common period, i.e. , the correlation inspected visually (Harvey et al. 2020) . After finding similar results across each window, the calibration and verification periods were combined into a single period across which the relationship between the selected climate variable and RWI was quantified. To reconstruct climate, we applied this regression equation on the RWI chronolog y preceding the observed climate record. Results Our final ring width chronology cross - dated well, with an intercorrelation among all cores of 0.547 (Grissino - Mayer 2001). The chronology is based off 101 cores and spans 26 the sample is limited to a single tree, one from which the interval in which the sample included at least two trees (four cores). After 1547, the sample size in creased rapidly, from 9 cores in 1560 to 14 in 1580 and 17 in 1600. It remained robust thereafter (Fig. 2. 3). Despite the low sample size early in the chronology, EPS was high throughout, with a mean of 0.944 and no 50 yr segments falling below our predete rmined The detrended, standardized ring width chronology the RWI chronology was modeled against several moisture - balance - related climate variables. The climate variable most associated with RWI was prior - Z index. It was consistent acros s the 2 of 0.239 and 0.301, respectively (Table 2.1). Additionally, it was relatively stable in each of the 60 yr windows overlapping the calibration and verification periods (Fig. 2. 4). As an additional model - quality check, RE and CE were calculated. These statistics assess the pred ictive skill of a model and can 1976, Cook et al. 1999), a threshold which our model met (Table 2. 1). Prior - September Palmer Z index was directly correlated wit h RWI, that is, a wetter summer resulted in a wider ring the following year. For every unit increase in Palmer Z index there was a 9.6 and 8.2% increase in RWI during the calibration and verification periods, respectively (Table 2.1). Because we obtained similar results across the calibration and verification periods, we combined them into a full model (Pederson et al. 2012; Table 2.1) which was used to Z index. Because of the lagged effect, we used RWI at year t +1 27 to predict Pa lmer Z index at year t . Additionally, we tried a model in which both year t and year t +1 RWI were used as predictors, but this improved r 2 only mildly (0.266 for the single predictor vs. 0.271 for the double), and the Akaike information criterion (AIC) suffer ed (373 for the single vs. 374 for the double). AIC is a model - quality metric that assesses the goodness of fit while penalizing for having too many parameters; a lower value is better (Akaike 1974). We checked Z against obse rved values; the datasets were significantly correlated (r 2 = 0.266, p < 0.0001), but reconstructed Palmer Z was less variable than observed values (standard deviation [SD] = 0.58 vs. 1.31). Therefore, both reconstructed and observed Palmer Z indices were standardized to a mean of zero and SD of 1. This was done by Ford 2014). The standardized reconstructed data correlated well with the standardized observed data, with a few notable exceptions: early 1900s decades, late 1910s/early 1920s, and late 1960s/early 1970s (Fig. 2. 5). Additionally, our reconstructed summer drought index agreed with that of a previous study from a nearby site. Cook et al. (1999) conducted gridded reconstructions of summer moisture throughout the continental USA. Although none of their grid points were on South Manitou Island or any of the nearby islands, there were reconstructions in adjacent mainland Michigan. The nearest reconstruction was on mainland northwest lower Michigan, Grid Point 225 in their study. Despite their use of a different drought index (PDSI) over a different time window throughout common period 154 their data in the same manner as ours (r = 0.498, p < 0.0001). 28 Over the entire reconstruction there is no trend in moisture over time (p = 0.82), but there are multidecadal, even centennial, trends (Fig. 2 . 6). For example, there was a rapid moistening trend 1 ) and more modest trends 1 1 ). 1 (p = 0.0008, slope = 0.054 yr 1 ). The driest periods in our reconstruction, either especially intense or long - ualization, some of the above Toward putting climate of the modern era into a historical perspective, we more closely examined the most extreme non - overlapping half - reconstruction (Fig. 2.8a), quantifying the return intervals of these extreme events relative to .8 a). If one were to look only at the la st 120 yr, 29 one would think these intervals were more common than they were from a historical perspective. Extreme 5 yr pluvials such as these four occurred every 116, 29, 23, and 58 yr , and 93 yr from - decades occurred once every 15, 465, 233, and 19 yr. This tendency, for extremes to be more frequent in the most recent 120 yr than in the rest of the reconstruction, held for the less extreme 5 yr pluvials and droughts too (Fig. 2.8b). From the perspective of human - delineated centuries, the driest were the 17th (mean = 16th (mean = 0.032 ± 0.152 SE) and the 20th (mean = 0.031 ± 0.122 SE). However, these differences were not significant; there were no among - century differences in the mean (1 - way ANOVA; F = 0.091, df = 5463, p = 0.994). The most variable centuries were the 16 th and 20 th (SD = 1.12, 1.22), the least the 18 th (SD = 0.79). These variability differences are corroborated by the number of extreme events per century (Fig. 2.9b). The 16 th century was one in which extreme (1 SD) about every fifth year (frequency = 0.222). In the 20 th centur y, extreme drought years occurred about once in 5 yr (0.220 frequency), extreme pluvial years about every fourth year (0.260 frequency), whereas over the reconstruction at large the frequency of extreme droughts and pluvials was about every 6 yr (0.168 and 0.164, respectively). Additionally, the 20 th century contained the variable century, the 18th, had frequencies 30 for drought and pluvial years of 0.110 and 0.080, respectively. For pluvial years, there was a significant deviation from the null hypothesis of equivalent frequencies per century (chi - squared goodness - of - 2 =12.62, d f = 5, p = 0.027). However, there was no significant difference in frequency of extreme drought years a mong centuries (chi - squared goodness - of - fit 2 = 8.94, df = 5, p = 0.111). Discussion Our reconstruction model explained 26.6% of the variability in prior - Palmer Z index. Doubtless, this relatively low number is due in part to a paucity of weather station data near our island site ( http:// www.ncdc.noaa.gov/cdo - web/datatools/findstation ). Although we relied on a regionally coarse estimate for all northwest lower Michigan (Fig. 2.1), pilot analysis revealed it worked better than any single station in isolation. The low variance explained may also reflect the relatively benign, cool and moist climate of northeastern North America, especially since our island site i s buffered from extreme summer heat by Lake Michigan. Benign climates lead to low correlations between tree ring chronologies and climate (Fritts 1976). Cook et al. (1999) reconstructed summer drought throughout the continental USA, and their tree ring - cli mate verification statistics were lowest in the upper Midwest and New England. Finally, it could be specific to white cedar Thuja occidentalis . In using that species to reconstruct precipitation in adjacent Ontario, Canada, Buckley et al. (2004) found r 2 = 0.36 in their calibration period and lower values in each of their verification periods, with the growth - climate relationship in the more recent of the two verifications largely deteriorating. 31 The phenomenon of growth - climate relationships deteriorating i n recent decades has Maxwell et al. 2016), but it was not observed in this dataset. Our growth - climate relationship was consistent across each half of the dataset (Ta ble 2.1), as well as in each 60 yr window overlapping with those halves (Fig. 2.4). There was variability, but overall there was no great difference between the minimum r 2 = 0.239 and maximum r 2 = 0.415. Numerous hypotheses have been put forth to explain t he phenomenon in which tree growth - climate relationships have deteriorated, especially for high northern latitudes, where it is posited that global warming has rendered tree growth less limited by insufficient heat, thereby weakening the growth - temperature variable, Palmer Z this was true of the reconstruction (linear correlation , p = 0.52) and observational reco rd (linear correlation , p = 0.16). The lack of change in the study variable may explain why tree growth relationships with it did not deteriorate. Additionally, Maxwell et al (2016) found that the deteriorating growth - precipitation relationships he observe d were due to a lack of droughts in the calibration period, but our calibration period did contain significant droughts (Figs. 2.6, 2.7, & 2.8a). Though our r 2 was relatively low, we are confident in the model and reconstruction because the RE and CE value s exceeded zero, a rigorous quality check. Other tree ring - based reconstructions have reported similar r 2 , RE, and CE values for all or part of their study period for all or some of their tree ring chronologies, mostly in humid eastern North America (Cook et al. 1999, Maxwell et al. 2011, 2012, Ford 2014), but sometimes even in the drier North 32 American Great Plains (Cook et al. 1999). Our finding that summer moisture is most important to growth is consistent with other similar studies in temperate Nor th America. Palmer Z index incorporates both incoming and outgoing moisture, i.e. , precipitation and evapotranspiration. This is consistent with the wealth of literature that has found temperate tree growth consistently correlated negatively with temperatu re, and positively with precipitation ( e.g., Martin - al. 2020). In agreement with our results, T. occidentalis in adjacent Ontario was most strongly correlated (inversely) with prior - July/August maximum temperature (Kelly et al. 1994). Conversely, inverse correlations with current June temperature were the most consistent along a 3 - 45°) (Housset et al. 2015). It may be counterintuitive that prior - year, not current - year, conditions can most influence radial growth, but this is not anomalous. Favorable weather at year t 1 leads to strong root growth, starch storage, budset, and leaf pr oduction, which can boost radial growth at year t , particularly in evergreen trees, which will still have leaves from year t 1 , the boon year, at year t . Palmer Z index was better correlated with growth than were other moisture indices. Unlike PDSI and PHDI, the Z index has no memory. It reflects conditions in the current month(s) only (Heim 2002). Our Z index - based model performed better than the PDSI - and PHDI - bas ed models because T. occidentalis growth is correlated primarily with mid - to late summer conditions of either the prior or current year (Kelly et al. 1994, Housset et al. 2015). In early summer months, the magnitude is weaker, and in spring correlations a re either weaker or the inverse of the summer. Growth relationships with spring temperature are often positive (Kelly et al. 1994, Housset et al. 2015) and with precipitation negative (Kelly et al. 1994, 33 Housset et al. 2015). This would potentially confoun d the reconstruction of an index which incorporates conditions from several preceding months. To repel confusion, it should be reiterated that while the Palmer Z index has no memory, tree growth does have memory and can thus be affected by prior - year Palme r Z . Our reconstructed moisture conditions did not reveal the most recent several decades to be remarkably wet, either in the reconstruction or the observed record, and this contrasts with a recent moistening trend found in the instrumental record for Indi ana/ Illinois (Mishra & Cherkauer 2010) and Michigan (Andresen 2012, http:// www.ncdc.noaa.gov/cag/statewide/time - series/20/pcp/ann/9/1895 - 2019 ) and with some other de ndroclimatic reconstructions in northeastern North America. Ford (2014) found a moistening trend in Iowa from about 1940 to present, with the most recent 20 yr of recorded ee ring - reconstructed, back to 1640. However, the results for Michigan and Iowa were based on precipitation alone, not a drought index like Palmer Z , which incorporates both incoming and outgoing moisture. Pederson et al. (2013) did reconstruct a moisture in southeastern New York back to 1531 and found the most recent 43 yr period to be uniquely moist and an overall moistening trend extending back to about 1800. Cook et al. (1999), too, reconstructed a moisture trend. At their studie d geographic point nearest to our site, in , r = 0.107, p = 0.026). We found neit her the 20th nor 21st century to be particularly moist but did find a 5 yr and two other 10 yr pluvials in the second half of the 1900s (Fig. 2. 7). In agreement 34 with our study, Buckley et al. (2004) did not find a moistening trend in adjacent Ontario, with none of their most moist decades occurring after the 1830s. Differences among studies may reflect different variables and time windows examined, as well as local - scale climatic differences. We found three severe 6 yr droughts in the late 1500s: 2. 6). Signatures of drought during this time the Late 16th - Century Megadrought have been found widely in North America, including in the mid - Atlantic ( Stahle et al. 1998, Maxwell et al. 2011), Southwest (Grissino - Mayer 1996), throughout the West (Fritts 1965), Arkansas (Stahle et al. 1985), and elsewhere (Woodhouse & Overpeck 1998). This dry episode has been implicated for being potentially causal to sev eral important events of human history, including the abandonment of pueblos in New Mexico by indigenous people (Douglass 1935, Schroeder 1968, Burns 1983), the disappearance of the Lost Colony of Roanoke Island, Virginia (Stahle et al. 1998), conflict bet ween American Indians and the Spanish, and abandonment of the Santa Elena colony on Parris Island, South Carolina (Anderson et al. 1995), and perhaps even the Chichimeca War in Central Mexico (Stahle et al. 2000). Future research in our group will focus on the effects of prerecorded climate on the indigenous peoples of the upper GLR, particularly their degree of reliance on maize horticulture. We also found correlation with the nearby reconstruction of Cook et al. (1999) and with Buckley et al. (2004), the latter working just across Lake Huron from northern lower Michigan. They and we found droughts in the 1600s decade, the 1770s, the 1840s, and the 1910s, and a 35 pluvial in the 1830s, though there were also periods both dry and wet which did not coincide. Unlike us, they did not find the Late 16th - Century Megadrought. Although drought conditions dominated the late 16th century, we also found that century to include pluvials in the 1550s, 1580s, and 1590s (Fig. 2. 6). The late 16th century was a variable ti me (SD = 1.12), though not to the same extent as the 20th century (SD = 1.22) (Fig. 2. 9b). This is consistent with the observation and expectation that climate change has made and will continue to make climate more variable (Medvigy & Beaulieu 2012, Christ ensen et al. 2013). Additionally, Maxwell et al. (2012) found the 20 th century to contain both the most extreme wet and dry decades in an 800 yr May precipitation reconstruction in the mid - Atlantic Region. Indeed, some of the events in the 20th century wer e unusual in their extremity. The year 1921 was the only year of the entire reconstruction 3 SD from the drought of the entire reconstruction (Fig. 2. 9). There was a general tendency for the extreme 20th centur y events to be more common in that century than in the drought had a 20th - century return interval of 58 yr and for the entire reconstruction a return interval of 233 yr. It was the same for - century return intervals of 116 and 58 yr, respectively, and entire r econstruction return intervals of 155 and 93 yr. The high prevalence of extreme droughts in the 20th century, coupled with predic tions of increased temperature without significant increases in summer/ f all precipitation (Hayhoe et al. 2010, Byun & Hamlet 2018), suggest that extreme droughts will be more common in the coming century than in the last several. 36 In future research, we wil l use our climate reconstruction to better understand the history of human habitation and use of South Manitou Island, proximal islands, and the nearby mainland (Lovis et al. 1976, 2017, 2020). Further, additional climate reconstructions must also be condu cted in the GLR. The region has lost the great majority of its old - growth forest, but significant remaining pockets present a trove of uncollected data. There are numerous long - lived species in the region (Rocky Mountain Tree - Ring Research, Inc. & the Tree Ring Laboratory of Lamont - Doherty Earth Observatory and Columbia University 2013), with tree rings known to reflect climate in the region (Graumlich 1993), including Tsuga canadensis , Pinus strobus , P. resinosa , Acer saccharum , Quercus alba , and Betula alleghaniensis . When it is desirable to avoid coring living trees, alternatives should be sought such as dead trees, fallen logs, stumps, and structures with beams of a known geographical source (Larson & Rawling 2016). The unfortunate imminent decl ine of the long - lived T. canadensis to hemlock woody adelgid could provide an opportunity to obtain more data. To preserve data sources, park policies banning backcountry campfires, and wood collection even in the front - country, are needed, such as the pol icy of Bruce Peninsula National Park, Ontario, in the heart of ancient T. occidentalis country. Conclusion We derived a composite ring width index chronology from a living old - growth stand and from ghost forest stands of white cedar Thuja occidentalis in northern Lake Michigan and modeled it against observations of late summer moisture conditions (Palmer Z inde x). The model was consistent throughout the observational record and was used to reconstruct moisture conditions back to 1546 C.E. The first half - century of the reconstruction included 37 This corroborates a body of literature documenting a widespread severe drought in late 1500s North America. in the 1600s/10s, 1630s,1770s 80s, 1840s, and 1910s/20s. Ten - found in the 1610s/20s, 1660s/70s, and the 1970s/80s. The 20th century was the most variable century of the reconstruction and was a period during which both pluvials and droughts occurred with a greater frequency than in the reconstruction at large. As climate variability and late summer heat continue to increase, society should prepare to cope with extreme drought even in humid regions. 38 APPENDIX 39 APPENDIX Figures Figure 2.1. Study site, South Manitou Island, is indicated by a red square off northwest lower Michigan . The subsites Dunes and Valley of the Giants are indicated, respectively, with blue and red pins. Boundaries within Michigan are counties; counties with an asterisk repre sent US Climate Michigan Division 3. 40 Figure 2.2. Rooted tree in paleosol in the dune field on South Manitou Island (photo credit: Alan Arbogast). Figure 2.3. Ring width index (purple), sample size (green), expressed population signal (EPS; dark blue) over time, and year at which our chronology was truncated due to insufficient sample size (<1 tree [<3 cores]; black vertical line) . EPS measures how well a finite sample represents a theoretical infinite population (Wigley et al. 1 984) and was calculated in 50 yr windows, overlapping by 49 yr. The first plotted value in 1565 represents the window Before 1516, sample size is 1; thus, there are no EPS values prior to the window beginning at that year. Our predetermined EPS threshold which each of our 50 yr windows exceeded. Ring width index and EPS, both unitless, are plotted on the same axis. 41 Figure 2.4. Evolution of the relationship between observed year t Z index and year t +1 Thuja occidentalis ring width index . The relationship was modeled over 60 yr windows overlapping 59 yr, i.e. , 42 Figure 2.5. Standardized observed vs. reconstructed Palmer Z index (a) and standardized observed, reconstructed Palmer Z Index vs. time (b) . Dashed black line in (a) is the best fit estimated by linear regression. 43 Figure 2.6. Reconstructed annual (gray) and 9 yr moving average (black) of standardized Z The year at which the moving average is plotted is the middle year of the 9 yr window. Figure 2.7. The most extreme 10 yr non - overlapping pluvial and non - overlapping drought events, defined here as having a mean standardized reconstructed Palmer Z ind the entire reconstruction . Year on the x - axis is the last year of the decade, for example, 1559 44 Figure 2.8. Reconstructed extreme 5 yr non - period of the reconstruction and observational record (a) and the return interval of extreme 5 (b) . Extreme events are here defined as having a mean standardized Pal mer Z x - axes is the last year of the 5 yr period, corresponded to extreme reconstructed events. In cases where an extreme event was reconstructed but not reflected in the observational record, the corresponding mean over the observational record is still displayed. Numbers above the bars are the return intervals in years. 45 Figure 2.9. Mean and SD of standardized reconstructed Palmer Z index per century (a) and number of extreme years and their frequency throughout each century and the entire reconstructed period (b) . In (a) the error bars on the gray columns are the standard error of the mean; there were no significant differences in the mean among centuries (1 - way ANOVA, p - value >0.05). In (b) the number above each bar is the frequency of extreme years within each period. 46 Tables Table 2.1 . Statistics and coefficients for models relating year t and year t+1 Thuja occidentalis ring width index . RE: reduction of error; CE: coefficient of efficiency see Methods section for more details (Fritts 1976, Cook et al. 1999). NA: not applicable. Period July Sep Palmer Z Coefficient r 2 RE CE Calibration (1896 1955) 0.0960 0.239 NA NA Verification (1956 2015) 0.0818 0.301 0.259 0.202 Full Model (1896 2015) 0.0855 0.266 NA NA 47 CHAPTER THREE IN GREAT LAKES REGION MESIC FORESTS, CONTEMPORARY GROWTH - CLIMATE RELATIONSHIPS SUGGEST THAT SOUTHERLY TREE POPULATIONS ARE MOST VULNERABLE TO CLIMATE CHANGE Abstract The North American Great Lakes Region is forest - rich and ecologically important, yet relationships between tree growth and climate there have been insufficiently quantified and the temporal consistency of those relationships insufficiently assessed. A network of 46 tree - ring - width chronologies, 35 previously unpublished was established. The network included nine species from 12 sites along an eight - degree latitudinal gradient in Indiana and Michigan . I sought to compare growth and growth - climate relationships within and among tree species and to examine how temporally consistent those relationships were. Relationships between tree - ring widths and monthly temperature and precipita tion were quantified and compared among species and sites. Principal components analysis was conducted to identify patterns in ring - width anomalies and growth - climate relationships within and among species. The temporal consistency of growth - climate relati onships was examined along moving intervals with bootstrapped correlation analysis . The most consistent growth - temperature relationship was a negative association with current - June mean temperature. The most consistent growth - precipitation relationship wa s also in the current - June month; in this case, the relationships were positive . This was true throughout the gradient but especially in the south, and this held when species were pooled or 48 considered individually . Relationships with some climate variables were inconsistent. For example, the relationship between growth and prior - September precipitation went from r = 0.226 in the first interval examined, 1904 1937, to r = - 0.030 in the final interval, 1971 2004. However, the strongest growth - climate relation ships were stable over time. Strongly and temporally consistent negative growth - June - temperature and positive growth - June - precipitation relationships portend a future of reduced radial growth as climate change proceeds. Further, the temporal consistency s uggests that growth - climate modeling is a reliable method for reconstructing past climate and predicting future growth in the region, when the most growth - influencing variables are used in the model. Introduction The Great Lakes Region (GLR) of North Ame rica contains ample, diverse forests which play important ecological and economic roles. In Michigan, the annual state and regional economic impact of the timber industry is $14 billion (Leefers 2017). T here are more than 20 million acres of forest, includ ing 56% of the land area and amounting to 1,400 trees for every distribution limits, both northern and southern, because of the tension zone a diffuse region in which th e hardwood forests dominant in the south gradually give way to the mixed hardwood - conifer forests of the north (Anders e n 2005) allowing the comparison of southern - and northern - range - margin populations. Most of the region experienced a lternating cycle s of glacial and interglacial conditions over the last 800,000 years, the most recent being the Wisconsin glaciation , the peak of which 49 was . This glaciation is responsible for most of the current soil types, topography, and lake levels (Larson and Kinc are 2009). The variation in soils affects modern vegetation, with mesic plants growing in fertile, coarse - textured soils and xeric plants in dry, sandy soils (Harman 2009). The contemporary climate of the region is unique due to the temperature - moderatin g, moisture - influencing Great Lakes. Due to the predominantly west - flowing wind, both major peninsulas of Michigan are downwind of a Great Lake, resulting in an overall more wet, cloudy, snowy, and moderate climate than areas upwind (Andresen & Winkler 200 9). Recent climate change has resulted in a generally warmer, wetter regime. Widespread reliable weather data collection began in 1895 in Michigan. From then through 2019, mean annual temperature increased by 0.127 °C/decade and total annual precipitation by 1.09 cm/decade ( http:// www.ncdc.noaa.gov/cag/statewide/time - series/20/tavg/ann/9/1895 - 2019 , http:// www.ncdc.noaa.gov/cag/statewide/time - series/20/pcp/ann/9/1895 - 2019 ). Long - term forecasts for North America predict further increasing precipitation and temperatu re. Mid - century projections under Representative Concentration Pathway (RCP) 4.5 predict precipitation increases in the northern part of the continent, particularly in winter, but also summer, and for temperature to increase beyond the range of natural var iability for all the continent (Christensen et al. 2013). For the GLR, Byun & Hamlet (2018) predict a 6.5 °C increase by 2100 under RCP 8.5 and 3.3 °C under RCP 4.5. Their predictions for precipitation are more equivocal. Under RCP 8.5, the GLR may experie nce a 5 25% increase in annual precipitation by the 2080s, with large increases in winter/spring and perhaps little change in summer/fall. This is consistent with a previous prediction, as well (Hayhoe et al. 2010). 50 With the unique climate and diversity of forest types, climate change responses among tree species may be unique in the GLR. In any case, it is important to anticipate how the major tree species will be affected to inform conservation and forest management. For example, will tree gro wth responses to climate change in the region be more species - specific or site - specific? In some studies of the tree growth - climate relationships in Eastern North America, growth - climate relationships are more similar within species across sites than withi n sites across species. In the Hudson River Valley, Pederson et al. (2004), found that species - specific differences in growth - climate relationships were bigger than site - specific differences , suggesting that the species at a given site will not respond uni formly to climate change . Similar results were found by Gruamlich (1993) in the Western Great Lakes Region and by Cook et al. (2001) in the U.S. West Gulf Coast, at least at the more environmentally benign sites. By contrast, growth - climate relationships w ere more similar across sites than across species along a latitudinal gradient in the eastern United States (Martin - Benito & Pederson 2015). In this regard, the GLR remains understudied. There does exist a latitudinal comparison of growth - climate relati onships among several species in Wisconsin and the Western Upper Peninsula of Michigan (Graumlich 1993), a model predicting how different trees will migrate under ongoing climate change (Walker et al. 2002), a model of climate change vulnerability for 11 M ichigan tree species (Penskar & Derosier 2013), and work on the growth - climate relationships of trees in the Lower Great Lakes Region (Maxwell et al. 2016, Au et al. 2020), but a better understanding of how the major tree species are affected by climate is required. Further, there is evidence that growth - precipitation relationships are weakening over time in Indiana (Maxwell et al. 2015). The potential temporal inconsistency in growth - climate 51 relationships has implications both for tree - ring - based climate r econstruction and forecasts of future growth and productivity. Is the deteriorating growth - precipitation relationship a wider trend in the GLR? Below I present a study on the growth - climate relationships of nine major tree species in Indiana and Michigan. My objectives were to (1) quantify growth - climate relationships, (2) compare relationships within species , (3) among species, and (4) among groups based on ( 5 ) test the temporal stability of those relat ionships. Methods The Study Region The study region includes 12 sites along a latitudinal gradient from southern Indiana at 38.54 °N to the Upper Peninsula of Michigan at 46.88 °N (Fig. 3.1, Table 3.1). According to one estimate, this 8.34° gradient wou ld be equivalent, in terms of temperature lapse, to an altitudinal gradient of 1668 m (Montgomery 2006). In selecting sites, mature forests were sought with the aim of finding trees old enough that their growth could be compared against the regional climat e record which generally extends from 1895 to the present . To control somewhat for site conditions, forests of the mesic type were selected, a common, important type in the region (Cohen et al. 2014). Overall, site selection was a balance of including a wi de latitudinal gradient while efficiently using limited resources. 52 The Study Species Ecologically and economically important species that occur commonly as canopy dominants in the mesic forests were selected, including the southerly beech - maple type and northerly northern - hardwoods type (Table 3.1). Additionally, species with long ranges span the entire study region: Acer saccharum Marshall (sugar maple) , Fagus grandifolia Ehrh. (American beech) , and Quercus rubra L . (red oak). Others find a northern limit in the region, Carya ovata (Mill.) K. Koch (shagbark hickory) , Liriodendron tulipifera L . (tulip poplar) , and Quercus alba L. (white oak) , and others a southern limit, Betula alleghaniensis Britton (yellow birch) , Pinus strobus L. (eastern white pine) , and Tsuga canadensis (L.) Carrière (eastern hemlock). Chronologies derived for A. saccharum and Q. alba from Price Nature Center (PNC) and from Voorhees (Vrh) and Warner (War) Audubon Sanctuaries were published previously (Au et al. 2020). Chronologies for A. saccharum (Au et al. 2020), L. tulipifera , Q . alba , and Q. rubra from Pioneer Mothers M emorial Forest (PMF) (Maxwell et al. 2015), C. ovata from PMF (Maxwell & Harley 2017), and T. canadensis from Huron Mountain Club at Rush Lake (HMC) (Dye & Woods 2019) also were published previously. Field Methods Of the 46 populations that ultimately went into this study, 40 were selected a priori . These included all but T. canadensis at HMC and A. saccharum , C. ovata , L. tulipifera , Q. alba , and Q. rubra at PMF. For those 40 populations, field work was conducted f rom 2016 2018 at 12 sites, sampling between one and five species per site, from nine species. Upon arrival at a site, a 53 general reconnaissance was performed to identify suitable stands . Suitable stands were defined as ones with at least ten trees per focal species which appeared to be mature. Trees then were selected from throughout these suitable stands to minimize within - stand differences, targeting mature trees. Making the most of limited resources, between 10 and 27 trees were sampled per species per si te, obtaining generally two cores per tree but occasionally only one or up to four. According to stipulations from landowners/managers, for some populations only living trees were cored and from others both living and dead trees. In the case of standing tr ees, all cores were taken at breast height (1.37 ± 0.5 m) or the equivalent position along a downed tree. For the other six populations, data were obtained from other researchers. For A. saccharum , C . ovata , Q. alba , L. tulipifera , and Q. rubra at PMF, field work was done in 2012 2014, using methods consistent with those described above, except the sample size for C. ovata was six trees (Maxwell & Harley 2017, Au et al. 2020). For T. canadensis at HMC, field work was done in 2016; all trees with diameter at breast height greater than 10 cm were sampled within two 16 - m radius plots, for a total of 52 T. canadensis trees which were suitable to my study based on temporal coverage (Dye & Woods 2019). Core Processing, Cross - Dating, and Detrending Co res were air dried and glued to wooden mounts, then sanded with a palm sander using progressively finer grades of sandpaper beginning with 120 - or 150 - grit and ending with 400 - or 600 - grit, depending on species. The 120 - grit grade was used only for hardwoo ds. With the conifers, 150 - grit was the first grade used. For some A. saccharum populations, 600 - grit was 54 the final grade used because that species often had quite narrow rings that were difficult to see. For all other species, 400 - grit was the final grade used. To ensure the calendar date of each tree ring was known, cross - dating was performed across the cores of each population. Cross - dating was accomplished according to either the list (Yamaguchi 1991) or skeleton plot method (Stokes & Smiley 1996), depe nding on species. The list method is a simpler, less conservative approach that was used only for some populations of the most easy - to - date species (i.e., L. tulipifera , Quercus spp. , and C. ovata . Methods applied to cores in previous studies are reported in their respective publications (Maxwell & Harley 2017, Dye & Woods 2019, Au et al. 2020). Depending on the chronology, tree - ring widths were measured either to a resolution of .001 millimeters using a Velmex measuring system (Velmex, Bloomfield, NY, US A) or to .01 millimeters in the computer program C OO R ECORDER (version 9.0.1, Cybis Elektronik & DATA AB, - Foy, Quebec, CN). In the case of computerized ring - width measu ring, digital images were generated for each core by scanning them with a flatbed computer scanner at a resolution of 1200 3200 dpi, depending on core length. Initial cross - dating was confirmed by running for each population all the ring - width measurement series through program C OFECHA (Holmes 1983). Potential cross - dating issues were flagged by the program, and these issues were troubleshot by checking for measurement errors or misplaced missing or false rings. Revised ring - width series were re - run through C OFECHA until successfully cross - dated or eliminated due to failure in the cross - dating process. In total, fewer than 5% of cores were discarded, either due to severe rot or failure to cross - date. 55 Non - climatic trends due to endogenous stand dynamics or tree age/size were dampened through standardization and detrending in program ARSTAN version 44 (Cook 1985). First, all pre - 1895 years were excised from ring - width series because they were not relevant to this dendroclimatological study due to the absence of meteorological data prior to 1895. Further, sample sizes tended to be low that far in the past, rendering cross - dating less reliable. In ARSTAN, indices were calculated by dividing observed ring widths by expected widths based on a fitted curve. The fit ted curves were cubic smoothing splines, with a rigidity typically 40 70% the length of the cores in years, but occasionally as low as 20% or 30% to cope with isolated extremely exaggerated ring - width indices. Among the three detrended chronologies ARSTAN provides, I selected the Residual Chronology for further analysis because it was stripped of autocorrelation and is therefore suitable for regression analysis (Speer 2009), and it is frequently used in similar studies (e.g., Huang et al. 2010, Harvey et al. 2020, LeBlanc et al. 2020) . Quantifying and Comparing Growth - C limate Relationships To quantify growth - climate relationships (Objective 1), the residual chronologies were run in program D ENDROCLIM 2002 (Franco & Biondi 2004) against monthly mean temperature and total precipitation data. For meteorological data, gridd ed 4 km × 4 km interpolations were obtained from the PRISM Climate Group (Daly et al. 2008). In a bootstrapping procedure, DENDROCLIM2002 randomly selects one rin g - width index from the dataset, replaces it, randomly selects another, and so on, until 1000 r ing - width indices are randomly selected and linearly regressed against the corresponding meteorological data. I used the interval common 56 to all 46 ring - width - index chronologies, 1903 2004. The climatological window ran from the May preceding the year of ri ng formation through the September of the year of ring formation. To compare growth - climate relationships within and among species (Objectives 2 & 3), a combination of principal components analysis (PCA) and linear regression was employed. A correlation matrix was constructed consisting of 46 rows, one for each population, and 34 columns, one for each climate variable (17 months × 2 types of climate variables). Each cell - width indice s and a climate variable, as calculated in D ENDROCLIM 2002 . Rather than focusing intensely on so many variables, the two most influential variables, one for temperature, one for precipitation, were retained for in - depth analysis and a handful of others for brief interpretation. The most influential variables were determined by calculating the absolute value of the mean across all populations. For each of the two retained climate variables, all species were pooled and linear regression of the 46 growth - climat e coefficients vs. latitude was conducted. To compar e growth - climate relationships within species (Objective 2 ), a subset of species, A. saccharum , F. grandifolia , and Q. rubra w as selected , the three species for which at least seven populations were sampl ed in this study, enough to make a meaningful comparison along the latitudinal gradient. A similar approach was then taken, in that linear regression of the growth - climate coefficients for the two most influential climate variables vs. latitude was conduct ed (Martin - Benito & Pederson 2015, Harvey et al. 2020) . To determine whether trends in growth - climate (Objective 4) 57 - employed within these categories as above. Further, to identify general patte rns within and among species in overall growth - climate relationships (Objectives 2 and 3), a wider subset of the 34 climate variables was retained. Because all growth - climate coefficients with a p - value < 0.05 were considered significant, 2.3 of 46 correla tion coefficients for a given climate variable would be expected to turn out significant just by chance. I therefore retained only those climate variables for which 5 of 46 populations turned out significant, about double the number expected by chance (mod ified from Huang et al. 2010). Those climate variables were retained in a separate correlation matrix, and PCA was conducted on that matrix. Two biplots, one for principal components (PCs) 1 and 2 and one for PCs 3 and 4, were generated for visualization o f patterns in growth - climate relationships. To test the temporal stability of growth - climate relationships (Objective 5 ), D ENDROCLIM 2002 was again used, but instead of the 1903 2004 single - interval analysis, moving - interval analysis was conducted. The int ervals were equal to 34 years, twice the number of predictors (17 months) for both mean temperature and total precipitation. Thus, the first interval was 1904 1937, the second 1905 1938, and so on up to 1971 2004. Rather than focusing on such a large numbe r of variables (17 months × 2 classes of climate variables), the two climate variables which among all 46 populations changed the most over time, both the most dynamic temperature and precipitation variable, were retained for further analysis and interpret ation (Harvey et al. 2020) . 58 Results General Chronology Characteristics For 46 tree populations across nine species and 12 sites, I collected tree cores, established calendar dates for each tree ring, measured tree - ring widths, and detrended and standardized tree - ring - width series resulting in 46 ring - width - index chronologies ( Table 3.2 , Fig. 3.2 ). All chronologies extended back to 1895 except B. alleghaniensis at Beaver Island which extended to 1903. All chronologies extended up to at least 2011 except F. grandifolia at Colonial Point. That chronology was truncated after 2004, because the populat ion developed symptoms of beech - bark disease, a potentially confounding factor, around 2005 (Adam Schubel pers. comm.). Sample size for each population ranged from 6 trees/12 cores to 52 trees/84 cores. Sample quality was assessed with the expressed popula tion signal (EPS), which measures how well a finite sample represents the total population (Wigley et al. 1984 but see Buras 2017). My predetermined minimum EPS value of 0.80 was met by all populations ( Table 3.2 ; Comparing Tr ee - Ring Chronologies Across Populations To identify general patterns in overall ring - width - index chronologies, PCA was conducted on a correlation matrix in which each row was a year and each column a tree population. Of the 46 PCs, one for each dependent v ariable (population), four were retained for interpretation, those which accounted for at least 5% of the variation (Legendre & Legendre 1998). The first four PCs together explained 55.6% of the variation in ring - width indices, the first 32.8%, the second 10.3%, the third 7.0%, and the fourth 5.5% ( Fig. 3.3 ). 59 On the first PC, all populations loaded positively, with a general trend that the hardwoods loaded higher than the conifers and the southerly populations higher than the northerly ones ( Fig. 3.3 a). Lo adings on this PC were correlated significantly and negatively with latitude (r = - 0.552; Table 3.3 ). The second PC separated even more clearly the northerly and southerly populations. Loadings were significantly and positively correlated with latitude (r = 0.912; Table 3.3 ), with all populations north of the site at 43.3° N loading positively and 23/24 to the south of it, negatively ( Fig. 3.3 explain 42.8% of the variability, both species and site a re important for determining ring - width anomalies. Generally, populations cluster together within sites, i.e., they have similar loadings on both PCs. This is particularly evident at PMF (38.5 °N), Mary Gray Bird Sanctuary (MGB, 39.6°), and Vrh (42.4°), al l of which were sites where hardwoods alone were sampled. At sites where both conifers and hardwoods were sampled, there is still within - site clustering, but only when hardwoods and conifers are separated, particularly for the proximal sites of Maywood His tory Trail/Peterson Hemlock Grove (Mwd/Pet, 45.8°). Within subregions, i.e., latitudinal bands spanning 1 3°, there was within - species clustering, too. This was particularly evident with northerly populations of T. canadensis , P. strobus , B. alleghaniensis , and A. saccharum . The third PC explained 7.0% of the variation, and it was not associated with latitude according to linear regression (r 2 = - 0.242, p = 0.106). The mid - latitude populations tended to load negatively, and the high - latitude populations, o ther than the hemlocks, to load around zero ( Fig. 3.3 b). The high - loading populations included two classes: populations from the two most southerly sites and the hemlocks (all northerly). The fourth PC explained 5.5% of the variability. Along this component, T. canadensis was well separated from the other sp ecies, 60 loading quite positively. Taking these PCs together, the hemlocks clustered together, and the white pines also clustered together. Other than that, site was more important for determining growth anomalies. Within - site clustering was particularly evi dent at PMF (38.5° N), MGB (39.6° N), Vrh (42.4° N), and PNC (43.3° N). The Strongest Growth - Climate Relationships, 190 3 2004 Single - Interval Relationships between radial growth and climate were estimated using bootstrapped correlation analysis in DENCROC LIM 2002 over the common interval 1903 2004. Among the 17 months over which mean temperature (T mean ) was considered, the most important month was June of the current year, the year the ring was formed. The mean of the growth - JunT mean correlation coefficien t across all 46 populations was r = - 0.268. It was also the most consistent T mean variable, significant in 31 populations and virtually always negative (42/46 populations). Pooling all species, I regressed all 46 correlation coefficients against latitude; the lower the latitude, the more that June heat dampened growth (r 2 = 0.588, p < 0.00001; Table 3.4 , Fig. 3.4 a). To illustrate, this variable was significant in just one of ten populations at the three most northerly sites and fifteen of fifteen populations at the three most southerly ( Table 3.4 , Fig. 3.4 a). This phenomenon held true within species, too. Considering the well - replicated species in isolation, the amount of variability in growth - Jun Tmean relationships explained by latitude was r 2 = 0 .908 for A. saccharum , r 2 = 0.652 for F. grandifolia , and r 2 = 0.827 for Q. rubra ( Table 3.4 , Fig. 3.4 a). The phenomenon also held true within the pooled trans - gradient species and the pooled southerly species (r 2 = 0.343 for southerly and r 2 = 0.815 for t rans - gradient species, Fig. 3.4 a). However, northerly species did not follow this pattern (r 2 = 0.016, Fig. 3.4 a). 61 Among all months considered, the most important for precipitation - growth relationships was again current - June ( Fig. 3.4 b, Table 3.5 ). It was the most consistently significant across populations (33/46) and had the highest magnitude (mean growth - Jun Ppt correlation coefficient = 0.283). For all populations, the relationship was positive, significantly for the 24 most southerly populations. Latitu de and growth - Jun Ppt correlation coefficients were inversely correlated: the lower the latitude, the more precipitation boosted growth (r 2 = 0.521, p < 0.0001). As with T mean , species - specific results for the three well - replicated species were similar to t he pooled - species results. The amount of variability in growth - Jun Ppt relationships explained by latitude was r 2 = 0.467 for A. saccharum , r 2 = 0.372 for F. grandifolia , and r 2 = 0.643 for Q. rubra ( Table 3.5 , Fig. 3.4 b). This was also true within the pooled trans - gradient species and the pooled southerly species (r 2 = 0.451 for southerly and r 2 = 0.468 for trans - gradient species, Fig. 3.4 b). However, northerly species showed a slight trend in the opposite direction, i.e. , growth - June Ppt relationships were slightly stronger at the most northern site (r 2 = 0.161, Fig. 3.4 b). Other Growth - Climate Relationships, 190 3 2004 Single - Interval Among all climate variables considered, 18/34 were significant in at least five populatio ns ( Tables 3.4 & 3.5 ). Five is about double what is expected by chance alone with a significance threshold of p = 0.05. Briefly, some of the variables other than Jun Tmean and Ppt which were often important were Jul Ppt (significant in 32/46 populations), Jul Tmean (28/46), Jan Ppt (18/46), May Ppt (16/46), prior - Aug Tmean (15/46), and prior - Aug Ppt (12/46) ( Tables 3.4 & 3.5 ) . For each of these variables, except prior - Aug Ppt , there was a significant association between 62 latitud e and the growth - climate relationship, when species were pooled. Relationships with Jan, May, Jun, and Jul Ppt were generally positive, and they became more positive moving south. Relationships with Jun and Jul Tmean were generally negative and became more n egative moving south. Relationships with prior - Aug Tmean , too, were generally negative, but here they became less negative moving south. Overall, 18/34 growth - climate relationships were significantly associated with latitude ( Tables 3.4 & 3.5 ) . Comparing Growth - Climate Relationships Across Populations, 1904 2004 Single Interval PCA was conducted on a correlation matrix of growth - climate coefficients. Only coefficients significant in at least 5 of 46 tree populations were retained in the matrix. E ach column was a climate variable, each row a population, and each cell a correlation coefficient. A biplot of the two most important PCs was generated ( Fig. 3.5 ), this for visualization of patterns in growth - climate relationships, to interpret on a single figure a prodigious amount of information. The biplot was duplicated to highlight both within - site clustering ( Fig. 3.5 a) and within - species clustering ( Fig. 3.5 b). The first two PCs explained 68.4% and 11.9% of the variation, respectively. There was muc h overlap in how populations clustered, both by species and site, but it was clear that, within latitudinal bands, populations clustered according to species more than site ( Fig. 3.5 ). For example, at each of the most northerly sites at which at least thre e species were sampled, latitudes 44.1, 45.6, 45.8, and 46.9 °N, there was not close clustering among species. There was tighter within - site clustering at southerly sites, but also much overlap among sites. Within species, there was tight clustering within most of the narrowly sampled 63 taxa: B. alleghaniensis , T. canadensis , L. tulipifera , Q. alba , and P. strobus . However, for the widely sampled species, A. saccharum , F. grandifolia , and Q. rubra , there was wide separation within species. For Q. rubra and A. saccharum , a latitudinal gradient was apparent. Northerly populations tended to load more negatively on PC 1. However, within subregions, both A. saccharum and Q. rubra , did cluster well withi n species. Together, these results suggest that both species and site are important for determining growth - climate relationships, but that species is more important than site at small spatial scales. Additionally, the biplot allows approximation of the im portance of each climate variable and the relationship of each population with each climate variable. The longer the vector arrow, the more important the climate variable. For example, Jun and Jul Ppt and Jun and Jul Tmean are quite important. May and prior - Dec Tmean are less important. The relationship between angle indicates a perfect inverse relationship (r = - 1), 0° a perfect direct relationship (r = 1), and 90° no relationship (r = 0). In this way one can, for example, determine what sets apart T. canadensis from the other species. Those populations had relatively weak relationships with current - summer conditions and relatively strong relationships with prior - July c onditions. Additionally, each of the four T. canadensis populations were among the few to have a significant affinity for current - Mar Tmean . Similar comparisons could be made by studying the growth - climate correlation matrix ( Tables 3.4 & 3.5 ), but it is ea sier to study the biplot ( Fig. 3.5 ). That was the purpose of conducting this PCA. 64 Changes in Growth - Climate Relationships, 190 3 2004 Moving - Intervals The moving - interval correlation analysis showed that the monthly T mean variable that changed the most was prior - July ( Table 3.6 ). For 8/46 populations, the slope was steepest for this month, with relationships moving from positive to negative among all eig ht. This was most prevalent in the south but also occurred at two more - northern sites. All eight of the populations for which prior - July was the largest - changing month were A. saccharum or F. grandifolia , but the positive to negative trend was also observe d among most populations of other species, too ( Fig. 3.6 a). The monthly precipitation variable that was most variable across the timeline was prior - September ( Table 3. 6 ). For 13/46 populations, the slope was steepest for this month, with relationships mov ing from positive to negative in more recent years, among all 13. This was most prevalent at the sites Vrh and Kalamazoo Nature Center (KNC), in the middle of the latitudinal gradient, in southern Michigan, but it was found also at more northerly sites. It was not clustered among a few species but was found in seven species, six hardwoods and P. strobus . Even for most of the other populations, those for which prior - September was not the largest - changing month, a similar trend was still observed ( Fig. 3.6 b). By contrast, growth relationships with the strongest growth - related variables, Jun Tmean and Jun Ppt changed little over the interval ( Fig. 3.6 c,d). 65 Discussion Comparing Tree - Ring Chronologies Across Populations In a PCA of all 46 ring - width - index chronologies, the first PC, explaining 32.8% of the variation, was one on which all populations loaded positively ( Fig. 3.3 a). The shared sign indicates a direct relationship among populations regarding growth anomalies (variations from th e mean). Thus, along the first PC, a good growth year in southern Indiana also meant a good year as far away as the northern Upper Peninsula (Brubaker 1980). Other studies, also examining many tree - ring chronologies over large areas, have found all (or nea rly all) populations to load in the same direction on the first principal component: in the Great Lakes Region (Graumlich 1993), eastern boreal Canada (Huang et al. 2010), the southeastern U.S. (Pederson et al. 2012), and the U.S. Pacific Northwest (PNW) ( Brubaker 1980, Peterson & Peterson 1994). It is believed that this represents a shared response to a coherent regional climate signal. Though my study area covered eight degrees latitude (940 km), a 6.7 - °C difference between the warmest and coolest site, a nd a 473 - mm/year difference between the driest and wettest (Table 3.1), annual weather was correlated among sites. For common period 1903 2004, mean annual temperature of the most southerly sites was well correlated even with that of most northerly sites ( lowest Pearson r among all pairwise comparisons = 0.555; Table 3.7 ). Correlations for total annual precipitation were not as strong, but there was still a clear relationship (lowest r = 0.160; Table 3.8 ). Although all populations loaded positively on PC 1, there were differences in magnitude. The PCA does not examine growth - climate relationships directly, but for two reasons it is likely that different loadings among populations are due largely to climate ( Cook et al. 2001). First is 66 because of the prodigious influence of climate on tree growth (Fritts 1976). Second, much of the non - climatic influence on growth was removed in the detrending process (Cook et al. 2001). The conifers, in this study sampled only at northern sites, all loaded on PC 1 only slightly negatively, whereas the hardwoods loaded with a higher magnitude, even in the north. This - summer conditions and with Jan Ppt , as all those factors were strongly correlated with latitude, along with PC 1 loadings. That could also explain the lower loadings of northerly populations, as the northerly hardwoods tended to have weaker relationships with those variables than did southerly hard woods. PC 2 revealed an even more clear - cut distinction between northerly and southerly populations, with each population north of the site at 43.3 °N loading negatively and virtually all to the south of south of it, positively. Indeed, PC 2 loadings were highly correlated with latitude (r = 0.928). However, this PC accounted for only 10.3% of the variation. Thus, in a small number of years, a good growth year in the north is opposed by a poor year in the south and vice versa. Geographical - gradient threshol ds across which PC loadings flip sign have been found in other studies. Brubaker (1980) studied conifer tree rings at 38 sites in the PNW. On the second PC, all sites west of the crest of the Cascade Mountains loaded positively and all but two east of the crest negatively. Littell et al. (2008) studied Pseudotsuga menziesii (Douglas fir) tree rings at four mountain ranges in the Pacific Northwest. In a PCA of the tree - ring chronologies, along the second PC , nearly all populations from the two eastern ranges loaded positively and all from the western ranges, negatively. In contrast, no threshold was found across wide swaths of the West Gulf Coast, USA (Cook et al. 2001) eastern boreal Canada (Huang et al. 2010), or the GLR (Graumlich 1993). 67 PC 3 explained 7.0 % of the variation, and on that PC T. canadensis loaded strongly positively, together with the hardwoods at the two most southerly sites. These two classes exhibited quite different suites of growth - climate relationships ; thus , it is unclear why they would share this common loading along PC 3. PC 4 explained just 5.5% of the variation, but that was higher than the 5% threshold selected (Legendre & Legendre 1998). Along this PC, T. canadensis was well separated from the other species, loading quite positivel y. That species had many atypical relationships with climate variables. All four of its populations had significantly positive relationships with Mar Tmean , a phenomenon exhibited by another species in just one case, and relationships usually being negative among all other populations. Beyond just March, T. canadensis tended to be positively correlated with temperatures outside of the growing season with atypically high correlations November through March ( Table 3.4 ). This may be due to the ability of evergr een trees to photosynthesize over the winter given sufficiently high temperatures (Schaberg et al. 1998), however the other conifer, P. strobus , did not at all show the same phenomenon. Additionally, T. canadensis showed an unusually strong aversion to pri or - summer heat (Jun Sep), which again may be attributable to its evergreen habit; the effects of poor leaf production at year t will carry over to year t+1 (Fritts 1976). However, again P. strobus did not share this signature. Radial growth of T. canadensis was also found to be unique by Pederson et al. (2012), separating from the growth of several hardwood species along one PC. Though the two most important PCs were associated with latitude, species was also important. The narrowly sampled species especiall y tended to load similarly to their conspecifics, especially on the first PC, but on that PC the widely sampled species also loaded 68 similarly with their conspecifics at least among nearby sites. Overall, it was concluded that both species and site were imp ortant for separating ring - width - index chronologies. By contrast, Pederson et al. (2012) found species to be more important than site, as did Cook et al. (2001) and Graumlich (1993). However, though Graumlich was also working in the GLR, the latitudinal gr adient studied was four degrees, narrower than that of the present study. Huang et al. (2010) found both site and species to be important, with the first PC separating populations by site and the second, third, and fourth by species. The Strongest Growth - Climate Relationships, 190 3 2004 Single - Interval By far, the most important growth - climate relationships were in the current - summer months, especially June and July. Growth - temperature relationships were quite negative, - precip i tation rela tionships, quite positive. This was also true of August, but to a lesser extent. Negative growth - temperature and positive - precipitation relationships are a signature of moisture stress (e.g., Martin - Benito & Pederson 2015). Summer moisture stress leads t o leaf senescence (Marchin et al. 2010), fine - root mortality (Gaul et al. 2008), and stomatal closure (Panek & Goldstein 2001). The dominant signature of summer moisture stress is nearly ubiquitous in the wood of temperate trees, and it bodes ominously wit h the ongoing acceleration of climate change. To list a few examples, this signature was found across several hardwood species along a Georgia - to - Vermont latitudinal gradient in the eastern U.S. (Martin - Benito & Pederson 2015), Pseudotsuga menziesii in the montane PNW (Littell et al. 2008), several hardwood and Pinus species in Louisiana and east Texas (Cook et al. 2001), across a suite of hardwood and conifer species along a latitudinal 69 gradient from southern Wisconsin to northern Michigan (Graumlich 1993) , and across several oak species and in L. tulipifera throughout the eastern United States (LeBlanc & Berland 2019 , LeBlanc et al. 2020 ). It is generally only at the far northern end of the temperate world, and further to the north, that there is little si gnature of summer moisture stress in tree rings. Across a 46 54° N latitudinal gradient, there were few stands exhibiting negative summer temperature or positive summer precipitation relationships, especially in the current summer, and in fact positive gro wth - summer - temperature relationships were more common than negative in the north (Huang et al. 2010). Babst et al. (2013) studied 1000 tree - ring chronologies from the major European tree species in north Africa and throughout Europe (30 70° N) and found pr edominantly positive growth - precipitation and negative growth - temperature relationships at low - latitude/altitude sites and the inverse at high sites. Harvey et al. (2020) worked on Fagus sylvatica (European beech), Quercus robur (English oak), and Pinus sy lvestris (Scots pine) at European sites between the 51 st and 59 th parallels. Positive current - June precipitation relationships were quite important in the hardwoods, but few other summer variables were consequential, and in the conifers current - March tempe rature was most important. In Picea mariana (black spruce) chronologies throughout northeastern North America above the 60 th parallel, current - summer - temperature relationships were mostly positive and - precipitation relationships negative ( In the present study, the signature of summer moisture stress was nearly ubiquitous, but it was strongest in the south ( Fig. 3.4 ). This is consistent with other studies in temperate eastern North America (e.g., Martin - Benito & P ederson 2015, LeBlanc & Berland 2019, Au et al. 70 2020). By contrast, LeBlanc et al. ( 2020) found no such latitudinal trend in L. tulipifera across the eastern U.S. In the present study, the latitudinal trend in the strength of the summer - moisture stress sig nal held for both trans - gradient and southerly species. Thus , a s the southerly species, C. ovata , Q. alba , and L. tulipifera , approached their northern limit , the strength of the summer - moisture stress weakened, but it was still present, and it was not low er than that of co - occurring species that were not near their northern limit. Thus, the southerl y species were not relieved of the negative effects of high temperatures and insufficient precipitation near their northern limit ( Fig. 3.4 ). This is consistent with other dendrochronological work at and near northern range limits in the temperate world (e.g., LeBlanc & Berland 2019, LeBlanc et al. 2020) and with the hypothesis that in benign environments, such as temperate eastern North America, range limits are determined by abiotic not biotic factors ( e.g., Darwin 1859, MacArth u r 1972). No clear latitudinal trend in the summer moisture stress signal was found in the present study for the northerly species ( Fig. 3.4 ). However, caution should be exercised in the interpretation because the sample size was low (n = 10) and the latitudinal gradient narrow (2.76°). Other Growth - Climate Relationships, 1904 2004 Single - Interval Aside from the signature of current - summer moisture stress, relatively few variables were consistently important. Jan Ppt was one of these few, being significantly positive in 18 populations. This precipitation would most often fall as snow in the GLR. Winter snow insulates and protects soil from freezing temperature (Zhang 2005). Without this protection, fine - root mortality increases, and trees must devote more resources to fine - root production the 71 following growing season (Tierney et al. 2001). Further, lack of snow leads to roots taking up less nitrogen and to more nitrogen leac hing out of the soil (Campbell et al. 2014). Though winter snow accumulation can also boost tree growth by melting and providing moisture in the spring (Wu et al. 2019), that is unlikely a factor here because growth in this network was not limited by insuf ficient moisture in early spring ( Table 3.5 ). The next - most - important additional variable was May Ppt , significant in 16 populations. Relationships with May Ppt were generally positive, more so in the south. Interpretation is like that of Jun and Jul Ppt . Rel ationships with prior - Aug Tmean were significant in 15 populations. These relationships were generally negative, but unlike relationships with current - summer temperature, these relationships became less negative moving south. Rather than directly affecting radial growth as with current - summer temperatures, this is a lagged effect caused by decreased root production, bud set, carbon assimilation, and in the conifers leaf production in the previous year (Fritts 1976). Some of these relationships and others co uld offset the enhanced summer moisture stress that ongoing climate change will bring. Winter/spring precipitation and temperature are likely to increase (Byun & Hamlet 2018). Significantly positive relationships with prior - Nov Ppt , Jan/Feb Ppt , and prior - De c Tmean ( Fig. 3.5 , Tables 3.4 & 3.5 ) could ameliorate the effects of summer drought, but this is complicated by the conversion of winter precipitation from snow to rain, which would not necessarily have the same effect. Further, warmer winter temperatures w ill increase winter respiration (Wibbe et al. 1994). Modeling the combined effects of a variety of climate variables on future growth, and comparing this within and among species, will be the subject of future research the next chapter . 72 Changes in Growth - Climate Relationships, 190 3 2004 Moving - Intervals Moving - interval growth - climate analysis was conducted for 34 - year windows overlapping by one year, and changes in growth - climate relationships were quantified. Dendrochronology has stood traditionall y on the premise that growth - climate relationships are stable over time, allowing the use of observed relationships to reconstruct centuries or millennia of unrecorded climate. This premise will need to be reworked, as more studies have revealed the dynami c nature of growth - Changes in growth - climate relationships have been found chiefly at high latitudes where the historically well - linked variables of tree - growth and heat have diverged from each othe r as , Briffa et al. 1998 , 2004 ). Further, the phenomenon is not limited to recent decades, as it has also occurred during a period of early - 20 th - century warming in Fennoscandia (Sch neider et al. 2014), nor is it limited to high latitudes as Maxwell et al. (2015) found a recent weakening between tree growth and moisture stress in southern Indiana, and Marquardt et al. (2019) found a shift in the seasonality of growth - climate sensitivi ty in Pinus of sky islands in Arizona. However, growth - climate divergence is not universal, even at high latitudes. For example, among 64 Larix and Picea tree - ring chronologies from the Alps, no divergence was found (Buntgen et al. 2008). The present study found evolving growth - climate relationships, most dramatically for prior - Jul Tmean and prior - Sep Ppt , but relationships evolved in other months, too ( Table 3.6 ). However, it is unlikely that recent shifts in growth - climate relationships in the GLR will impede climate reconstruction because reconstruction is usually done with the most - growth - correlated climate variables. The most frequently evolving growth - climate relationships ( Table 3.6 ) were 73 not th e ones that most influenced growth ( Fig. 3.6 , Tables 3.4, 3.5, & 3.6 ). Similar results were found in L. tulipifera in the eastern U.S. (LeBlanc et al. 2020). However, caution should still be exercised because even with the two most influential climate vari ables, JuneT mean and June Ppt , with which growth was relatively consistently correlated across time, there were stray populations for which the relationship was unstable ( Fig. 3.6 c,d). Several hypotheses to explain high - latitude divergence exist, including global dimming (e.g., Stine & Huybers 2014), changes in non - focal variables (Vaganov et al. 1999), non - linear growth - 2000). In the GLR, the fading drought signal w as attributed to a lack of droughts in recent decades (Maxwell et al. 2015). This is consistent with the finding that in the eastern U.S., growth - precipitation relationships are weak when there is ample precipitation (LeBlanc and Berland 2019). However, th e changes to growth - p rior - Jul Tmean and - p rior - Sep Ppt observed in this study were not due to changes to the climate variables over time, as they changed little according to the PRISM - derived data (data for individual months not shown). It may be that a diff erent climate variable, one that influenced the growth - climate relationships considered, changed (Vaganov et al. 1999). This topic deserves further research. Conclusion Among 46 tree populations from nine species across an eight - degree latitudinal gradient in the Great Lakes Region, annual ring - width anomalies are determined by latitude more than by species. Within narrow latitudinal bands, species becomes more important than site, however. Across all populations except T. canadensis in the north, current - Jun/Jul Tmean and Ppt 74 are the most growth - limiting variables. The negative growth - temperature and positive growth - precipitation relationships suggest sensitivity to moist ure stress, and this portends a future of reduced productivity as climate change continues. The growth - summer - climate relationships were strongest in the south, and thus the north may be temporarily buffered from the negative effects of climate change. Acr oss most sites and species, some of the minor growth - climate relationships changed over time considerably, in a consistent direction, however the most limiting factors were stable over time. Projected changes to the most limiting variables can thus likely be used to predict future growth changes . 75 APPENDIX 76 APPENDIX Figures Figure 3.1. The study sites and their latitude . Sites sharing a latitude are further differentiated with the first letter of their name (Table 3.1). 77 Figure 3.2 . Detrended, standardized residual ring - width - index chronologies . Two - letter species codes: are in Table 3.1. Species codes: As = sugar maple Acer saccharum , Ba = yellow birch Betula alleghaniensis , Co = shagbark hickory Carya ovata , Fg = American beech Fagus grandifolia , Lt = tulip poplar Liriodendron tulipifera , Ps = white pine Pinus strobus , Qa = white oak Quercus alba , Qr = red oak Quercus rubra , Tc = eastern hemlock Tsuga canadensis . Sites are depicted by their latitude; when two sites share a latitude, they are disambiguated with a one - letter code: K = Kalamazoo Nature Center, V = Voorhees Audubon Sanctuary. 78 Figure 3.3 . Loadings of each stand on principal components (PCs) 1 & 2 (a) and on PCs 3 & 4 (b) . Loadings were calculated with a PC analysis on a correlation matrix in which each stand was a - width index in a given year. Species codes: As = sugar maple Acer saccharum , Ba = yellow birch Betula alleghaniensis , Co = shagbark hickory Carya ovata , Fg = American beech Fagus grandifolia , Lt = tulip poplar Liriodendron tulipifera , Ps = white pine Pinus strobus , Qa = white oak Quercus alba , Qr = red oak Quercus rubra , Tc = eastern hemlock Tsuga canadensis . 79 Figure 3.3 80 Figure 3.4 . Latitude vs. (a) growth - Jun Tmean and (b) growth - Jun Ppt correlation coefficients (Pearson r) . Species are grouped according to distribution within the study gradient: circles represent trans - gradient, triangles northerly, and squares southerly species. -0.60 -0.50 -0.40 -0.30 -0.20 -0.10 0.00 0.10 38 39 40 41 42 43 44 45 46 47 Correlation Coefficient (r) Latitude (a) A. saccharum F. grandifolia Q. rubra B. alleghaniensis P. strobus T. canadensis C. ovata L. tulipifera Q. alba Linear (All) Linear (Northerly) Linear (Southerly) Linear (Trans-Gradient) 81 Figure 3.4 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 38 39 40 41 42 43 44 45 46 47 Correlation Coefficient (Pearson r) Latitude (b) A. saccharum F. grandifolia Q. rubra B. alleghaniensis P. strobus T. canadensis C. ovata L. tulipifera Q. alba Linear (All) Linear (Northerly) Linear (Southerly) Linear (Trans-Gradient) 82 Figure 3.5 . Growth vs. climate principal component analysis biplots with colors selected to highlight (a) clustering among sites and (b) species . Note that the subfigures are identical other than the color scheme. Each colored label represents a population, each vector a climate variable. Climate variables are labeled with numbers corresponding to months (mont h 1 = Jan., 12 = Dec.). The letters preceding the months indicate whether the month occurred in the year codes: As = sugar maple Acer saccharum , Ba = yellow bir ch Betula alleghaniensis , Co = shagbark hickory Carya ovata , Fg = American beech Fagus grandifolia , Lt = tulip poplar Liriodendron tulipifera , Ps = white pine Pinus strobus , Qa = white oak Quercus alba , Qr = red oak Quercus rubra , Tc = eastern hemlock Tsuga canadensis . Sites are denoted by their latitude; when two sites share a latitude, they are disambiguated with a one - letter code: K = Kalamazoo Nature Center, V = Voorhees Audubon Sanctuary. 83 Figure 3.5 84 Figure 3.6 . Growth - climate coefficients (Pearson r) vs. time for (a) prior - Jul Tmean , (b) prior - Sep Ppt , (c) Jun Tmean , and (d) Jun Ppt . The year is the final year of a 34 - year window, for example, the final year, 2004, corresponds to the 1971 2004 window. Aside from the mean, each curve represents an individual population. -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1930 1940 1950 1960 1970 1980 1990 2000 2010 Correlation Coefficient (r) (a) A. saccharum B. alleghaniensis C. ovata F. grandifolia L. tulipifera P. strobus Q. alba Q. rubra T. canadensis Mean -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1930 1940 1950 1960 1970 1980 1990 2000 2010 Correlation Coefficient (r) (b) A. saccharum B. alleghaniensis C. ovata F. grandifolia L. tulipifera P. strobus Q. alba Q. rubra T. canadensis Mean 85 Figure 3.6 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 1930 1940 1950 1960 1970 1980 1990 2000 2010 Correlation Coefficient (r) (c) A. saccharum B. alleghaniensis C. ovata F. grandifolia L. tulipifera P. strobus Q. alba Q. rubra T. canadensis Mean -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 1930 1940 1950 1960 1970 1980 1990 2000 2010 Correlation Coefficient (r) (d) A. saccharum B. alleghaniensis C. ovata F. grandifolia L. tulipifera P. strobus Q. alba Q. rubra T. canadensis Mean 86 Tables Table 3.1. Site characteristics . Meteorological data are 1981 2010 normals ( https://prism.oregonstate.edu/ ). Species codes: As = sugar maple Acer saccharum , Ba = yellow birch Betula alleghaniensis , Co = shagbark hickory Carya ovata , Fg = American beech Fagus grandifolia , Lt = tulip poplar Liriodendron tulipifera , Ps = white pine P inus strobus , Qa = white oak Quercus alba , Qr = red oak Quercus rubra , Tc = eastern hemlock Tsuga canadensis. Site Species Latitude (decimal degrees) Longitude (decimal degrees) Mean Annual Temperature (°C) Mean Annual Ppt. (mm) Huron Mountain Club (HMC) As, Ba, Ps Qr, Tc 46.88070 - 87.91703 5.7 766.2 Maywood History Trail (Mwd) As, Ba, Qr, Ps, WP 45.83690 - 86.98482 5.8 738.5 Petersons' Hemlock Grove (Pet) Tc 45.78080 - 86.96793 5.7 732.3 Beaver Island (BvI) As, Ba Tc, 45.63522 - 85.51808 6.8 800.2 Colonial Point (CoP) Fg, Ps 45.48848 - 84.68538 6.4 779.6 Jacobson Sugar Bush (JSB) Fg, Tc, As 44.11998 - 85.56775 6.6 857.2 Price Nature Center (PNC) As, Fg, Qa Qr 43.32607 - 83.92752 8.8 828.5 Warner Audubon Sanctuary (War) As, Fg, Lt, Qa, Qr 42.61938 - 85.38603 9.0 938.9 Voorhees Audubon Sanctuary (Vrh) As, Co, Qa, Qr 42.35675 - 84.83425 9.0 907.9 Kalamazoo Nature Center (KNC) As, Fg, Qr, Lt, 42.35583 - 85.59538 9.5 982.5 Mary Gray Bird Sanctuary (MGB) As, Co, Fg, Lt, Qr 39.59205 - 85.22300 10.9 1085.1 Pioneer Mothers Memorial Forest (PMF) As, Co, Fg, Lt, Qa, Qr 38.54017 - 86.45593 12.4 1204.2 87 Table 3.2 . General chronology characteristics . EPS = expressed population signal (Wigley et al. 1984). Site codes: BvI = Beaver Island, CoP = Colonial Point, HMC = Huron Mountain Club, JSB = Mwd = Maywood History Tr Memorial Forest, PNC = Price Nature Center, Vrh = Voorhees Audubon Sanctuary, Warner = Warner Audubon Sanctuary. Species codes: As = sugar maple Acer saccharum , Ba = yellow birch Betula alleghanien sis , Co = shagbark hickory Carya ovata , Fg = American beech Fagus grandifolia , Lt = tulip poplar Liriodendron tulipifera , Ps = white pine Pinus strobus , Qa = white oak Quercus alba , Qr = red oak Quercus rubra , Tc = eastern hemlock Tsuga canadensis . Popu lation (three - letter site and two - letter species codes) Latitude Longitude Timespan Number of trees (cores) Series intercorrelation Mean ring width (SD) in mm Mean EPS (lowest 50 - y segment) Original publication (if different from the present) HMC Ac 45.83690 - 86.98482 1895 2017 18 (35) 0.502 0.86 (0.539) .927 (.918) HMC Ba 45.83690 - 86.98482 1895 2017 21 (49) 0.574 0.92 (0.553) 0.954 (0.952) HMC Qr 45.83690 - 86.98482 1895 2017 12 (24) 0.611 2.35 (0.710) .939 (.924) HMC Ps 45.83690 - 86.98482 1895 2017 14 (26) 0.526 2.89 (1.052) 0.912 (0.898) HMC Tc 46.88070 - 87.91703 1895 2015 52 (84) 0.672 0.81 (0.361) .986 (.985) Dye & Woods 2019 (Rush Lake) Mwd As 45.83690 - 86.98482 1895 2016 10 (18) 0.571 2.03 (1.082) 0.894 (0.843) Mwd Ba 45.83690 - 86.98482 1895 2015 9 (20) 0.579 1.25 (0.757) .906 (.901) Mwd Ps 45.83690 - 86.98482 1895 2015 10 (22) 0.575 1.82 (0.949) 0.916 (0.887) 88 Table 3.2 Mwd Qr 45.83690 - 86.98482 1895 2015 10 (20) 0.527 2.27 (0.820) 0.902 (0.901) Pet Tc 45.78080 - 86.96793 1895 2015 10 (20) 0.685 1.87 (0.772) 0.916 (0.906) BvI As 45.63522 - 85.51808 1895 2017 12 (22) 0.622 1.59 (0.842) 0.917 (0.865) BvI Ba 45.63522 - 85.51808 1903 2017 10 (20) 0.608 1.57 (0.803) 0.920 (0.903) BvI Tc 45.63522 - 85.51808 1895 2016 12 (24) 0.634 1.64 (0.882) 0.958 (0.956) CoP Fg 45.48848 - 84.68538 1895 2004 17 (34) 0.615 1.62 (0.651) 0.948 (0.935) CoP Ps 45.48848 - 84.68538 1895 2017 14 (27) 0.562 2.16 (0.940) 0.906 (0.892) JSB As 44.11998 - 85.56775 1895 2015 13 (26) 0.574 2.02 (0.996) 0.927 (0.917) JSB Fg 44.11998 - 85.56775 1895 2015 10 (20) 0.603 1.69 (0.900) 0.905 (0.856) JSB Tc 44.11998 - 85.56775 1895 2016 11 (22) 0.598 1.71 (1.170) 0.924 (0.909) PNC As 43.32607 - 83.92752 1895 2015 11 (21) 0.558 1.57 (0.878) 0.902 (0.889) Au et al. 2020 PNC Fg 43.32607 - 83.92752 1895 2015 11 (22) 0.633 1.66 (0.747) 0.925 (0.916) PNC Qa 43.32607 - 83.92752 1895 2015 9 (18) 0.523 2.22 (0.931) 0.879 (0.827) Au et al. 2020 PNC Qr 43.32607 - 83.92752 1895 2015 10 (20) 0.566 2.66 (0.932) 0.928 (0.908) War As 42.61938 - 85.38603 1895 2015 10 (20) 0.509 1.85 (0.420) 0.8715 (0.8344) Au et al. 2020 War Fg 42.61938 - 85.38603 1895 2015 12 (23) 0.537 1.61 (0.440) 0.9096 (0.8828) War Lt 42.61938 - 85.38603 1895 2015 10 (19) 0.647 2.71 (0.460) 0.9297 (0.9079) War Qa 42.61938 - 85.38603 1895 2015 10 (20) 0.64 1.93 (0.460) 0.9453 (0.9397) Au et al. 2020 War Qr 42.61938 - 85.38603 1895 2015 11 (22) 0.6 2.70 (0.446) 0.9478 (0.9367) Vrh As 42.35675 - 84.83425 1895 2015 10 (20) 0.598 1.81 (0.429) 0.9253 (0.9164) Au et al. 2020 89 Vrh Co 42.35675 - 84.83425 1895 2015 10 (20) 0.578 1.59 (0.504) 0.9209 (0.9022) Vrh Qa 42.35675 - 84.83425 1895 2015 10 (20) 0.64 2.03 (0.462) 0.9468 (0.9402) Au et al. 2020 Vrh Qr 42.35675 - 84.83425 1895 2015 14 (26) 0.699 2.91 (0.488) 0.9711 (0.9694) KNC As 42.35583 - 85.59538 1895 2015 10 (20) 0.552 2.23 (0.430) 0.9279 (0.8934) KNC Fg 42.35583 - 85.59538 1895 2015 10 (19) 0.604 2.26 (0.482) 0.9220 (0.9118) KNC Lt 42.35583 - 85.59538 1895 2015 10 (20) 0.663 2.89 (0.436) 0.9318 (0.8963) KNC Qr 42.35583 - 85.59538 1895 2015 10 (20) 0.604 3.40 (0.474) 0.9440 (0.9242) MGB As 39.59205 - 85.22300 1895 2016 14 (27) 0.526 2.54 (1.212) 0.918 (0.914) MGB Co 39.59205 - 85.22300 1895 2016 14 (28) 0.603 1.98 (0.821) 0.931 (0.915) MGB Lt 39.59205 - 85.22300 1895 2016 15 (30) 0.630 3.61 (1.676) 0.936 (0.898) MGB Fg 39.59205 - 85.22300 1895 2016 13 (27) 0.517 1.93 (0.455) 0.8838 (0.8474) MGB Qr 39.59205 - 85.22300 1895 2016 14 (28) 0.549 2.83 (1.042) 0.930 (0.928) PMF As 38.54017 - 86.45593 1895 2013 10 (19) 0.501 1.62 (0.853) .881 (.831) Au et al. 2020 PMF Co 38.54017 - 86.45593 1895 2012 6 (12) 0.580 1.54 (.702) .891 (.877) Maxwell & Harley 2017 PMF Fg 38.54017 - 86.45593 1895 2016 16 (31) 0.575 2.13 (1.062) 0.940 (0.926) PMF Lt 38.54017 - 86.45593 1895 2012 20 (22) 0.643 2.12 (1.081) .937 (.931) Maxwell et al. 2015 PMF Qa 38.54017 - 86.45593 1895 2011 22 (38) 0.605 2.39 (0.869) .942 (.935) Maxwell et al. 2015 PMF Qr 38.54017 - 86.45593 1895 2012 25 (44) 0.598 3.34 (1.173) 0.954 (.947) Maxwell et al. 2015 90 Table 3.3 Correlations (Pearson r) between loadings on each principal component (PC) vs. latitude . The PC loadings were derived from a PC analysis of the ring - width - index chronologies ( Fig. 3.3 PC 1 PC 2 PC 3 PC 4 Variance explained 32.8% 10.3% 7.2% 5.5% Correlation (Pearson r) with latitude - 0.552 * 0.928 * - 0.242 - 0.002 91 Table 3.4 . Partial correlation matrix (T mean only) for the growth - climate coefficient (Pearson r) for each of 34 climate variables for each population year p rior to the year of ring format ion. Cells in bold are significant (p < 0.05). Sites are arranged from north (top) to south (Table 3.1). Kalamazoo Nature Center, MGB = Mar Pioneer Mothers Memorial Forest, PNC = Price Nature Center, Vrh = Voorhees Audubon Sanctuary, Warner = Warner Audubon Sanctuary. Species codes: As = sugar maple Acer saccharum , Ba = yellow birch Betula alleghaniensis , Co = shagbark hickory Carya ovata , Fg = American beech Fagus grandifolia , Lt = tulip poplar Liriodendron tulipifera , Ps = white pine Pinus strobus , Qa = white oak Quercus alba , Qr = red oak Quercus rubra , Tc = eastern hemlock Tsuga canadensis . Population pMay Tmean pJun Tmean pJul Tmean pAug Tmean pSep Tmean pOct Tmean pNov Tmean pDec Tmean Jan Tmean Feb Tmean Mar Tmean Apr Tmean May Tmean Jun Tmean Jul Tmean Aug Tmean Sep Tmean HMC As . 15 . 07 - . 14 - . 22 . 07 . 03 . 04 . 04 - . 03 - . 04 - . 11 - . 06 . 19 - . 01 - . 08 . 00 - . 21 HMC Ba - . 01 - . 09 - . 17 - . 20 - . 05 - . 07 - . 17 - . 04 - . 04 - . 09 - . 09 - . 04 . 11 - . 19 - . 16 - . 06 - . 07 HMC Qr - . 03 . 14 . 00 - . 11 . 20 . 07 . 14 . 05 . 14 . 12 . 12 . 13 . 20 - . 03 . 05 . 18 - . 13 HMC Ps . 05 . 13 . 21 - . 02 . 24 - . 06 - . 07 - . 06 . 08 . 15 . 08 . 19 . 18 - . 15 - . 08 . 10 - . 04 HMC Tc . 14 - . 22 - . 32 - . 33 - . 26 - . 19 . 13 . 08 - . 07 . 12 . 34 . 29 . 03 . 06 . 17 . 14 . 11 Mwd As - . 02 - . 15 - . 29 - . 37 . 09 - . 05 - . 05 - . 04 . 05 . 02 . 19 . 07 . 23 - . 04 - . 04 . 04 . 04 Mwd Ba - . 02 - . 11 - . 13 - . 25 - . 05 - . 03 - . 18 - . 08 - . 14 - . 20 - . 06 . 04 . 20 - . 33 - . 22 - . 09 - . 18 Mwd Ps - . 01 . 23 . 16 - . 09 . 12 . 02 . 07 - . 01 - . 09 - . 06 - . 17 . 04 . 21 - . 19 - . 19 - . 05 - . 13 Mwd Qr . 01 . 09 - . 11 - . 20 . 05 . 06 - . 02 . 07 - . 05 . 06 - . 07 . 16 . 11 - . 18 - . 14 - . 02 . 08 Pet Tc . 08 - . 30 - . 42 - . 30 - . 23 - . 08 . 23 . 06 . 15 . 12 . 34 . 19 - . 04 . 02 . 22 . 20 - . 01 BvI As . 06 - . 06 - . 20 - . 37 . 04 . 00 . 10 . 07 . 03 - . 03 . 10 - . 06 . 20 - . 16 - . 03 . 07 - . 06 92 Table 3.4 BvI Ba . 02 . 03 - . 04 - . 21 - . 04 . 11 - . 03 - . 13 - . 14 - . 09 . 09 . 04 . 05 - . 36 - . 27 - . 07 - . 03 BvI Tc . 09 - . 17 - . 34 - . 41 - . 10 - . 06 . 23 . 03 . 06 . 12 . 39 . 16 - . 13 . 01 . 16 . 18 . 12 CoP Fg - . 07 . 00 - . 10 - . 22 - . 03 . 10 - . 04 - . 03 . 10 - . 01 . 05 . 14 . 27 - . 16 - . 12 - . 08 . 01 CoP Ps - . 12 . 11 . 18 - . 16 . 00 . 04 - . 05 - . 04 - . 12 - . 07 . 03 . 15 . 15 - . 24 - . 30 - . 06 - . 06 JSB As . 04 . 00 - . 05 - . 15 . 12 . 05 . 07 . 25 . 05 - . 05 - . 14 - . 09 . 06 - . 19 - . 28 . 01 - . 14 JSB Fg . 14 . 06 . 14 - . 01 . 11 . 01 . 06 . 20 . 13 . 04 - . 11 - . 20 . 14 - . 17 - . 26 - . 11 - . 21 JSB Tc . 10 - . 14 - . 30 - . 26 - . 12 - . 05 . 19 . 24 . 21 . 19 . 35 . 13 - . 31 . 06 . 08 - . 08 . 04 PNC As - . 23 - . 13 - . 23 - . 26 . 07 . 00 . 04 . 13 . 09 - . 05 - . 09 . 00 . 12 - . 23 - . 13 - . 01 - . 08 PNC Fg - . 28 - . 06 - . 10 - . 17 . 07 - . 01 - . 04 . 08 . 03 - . 05 - . 08 . 00 . 09 - . 28 - . 28 - . 14 - . 11 PNC Qa . 04 . 22 . 03 - . 08 . 20 . 18 . 18 . 36 . 16 . 05 - . 12 - . 09 . 02 - . 04 - . 19 . 04 - . 21 PNC Qr . 06 . 07 - . 01 - . 14 . 05 . 10 . 05 . 23 . 11 - . 13 - . 19 - . 07 - . 02 - . 24 - . 19 - . 16 - . 21 War Fg . 01 . 08 - . 01 - . 08 . 12 . 00 - . 11 . 10 - . 04 - . 16 - . 17 - . 03 . 22 - . 38 - . 22 - . 16 - . 05 War Lt . 12 - . 12 - . 11 - . 07 . 12 . 04 - . 07 . 25 . 12 . 12 - . 05 . 11 . 02 - . 26 - . 11 - . 04 - . 18 War Qa . 15 . 11 . 02 - . 09 . 12 . 04 - . 01 . 18 . 04 - . 11 - . 19 - . 06 - . 04 - . 45 - . 36 - . 18 - . 10 War Qr . 19 . 04 . 05 - . 13 . 07 - . 02 - . 05 . 09 - . 04 - . 13 - . 22 - . 02 . 13 - . 29 - . 22 - . 16 - . 13 Vrh As - . 01 . 14 . 03 - . 10 - . 05 . 08 - . 07 . 26 . 00 - . 07 - . 22 . 01 . 17 - . 31 - . 26 - . 25 - . 16 Vrh Co . 08 . 19 . 24 - . 06 . 15 . 14 - . 05 . 09 . 11 . 02 - . 14 . 03 . 16 - . 14 - . 18 - . 13 - . 01 Vrh Qa . 11 . 07 - . 03 - . 15 . 03 . 09 . 07 . 18 . 09 . 01 - . 06 . 02 - . 02 - . 35 - . 33 - . 19 - . 12 Vrh Qr . 05 . 04 . 00 - . 10 . 01 - . 01 - . 01 . 08 . 01 - . 11 - . 15 . 07 . 04 - . 39 - . 27 - . 15 - . 15 KNC As . 11 . 06 . 02 - . 12 . 09 . 03 - . 06 . 14 - . 06 - . 10 - . 09 - . 07 . 06 - . 38 - . 29 - . 21 - . 27 93 Table 3.4 KNC Fg . 16 . 12 . 12 - . 06 . 23 . 03 . 06 . 09 - . 04 - . 03 - . 10 - . 06 - . 05 - . 39 - . 38 - . 24 - . 25 KNC Qr . 10 . 02 . 10 - . 08 . 18 . 03 - . 09 . 16 - . 01 . 01 - . 22 - . 02 . 04 - . 44 - . 33 - . 06 - . 16 KNC Lt . 06 - . 10 - . 13 - . 22 . 07 - . 03 - . 13 . 09 - . 01 - . 11 - . 15 . 00 - . 04 - . 41 - . 27 - . 11 - . 19 MGB As . 00 . 00 - . 02 - . 07 . 08 . 17 . 02 . 09 . 02 . 06 - . 04 - . 02 - . 02 - . 41 - . 22 - . 18 . 02 MGB Co . 09 . 02 . 00 . 03 . 00 . 18 . 12 - . 06 . 03 - . 03 . 04 - . 07 - . 11 - . 52 - . 35 - . 26 - . 13 MGB Fg - . 08 - . 01 - . 09 - . 14 . 03 . 14 . 12 . 11 . 08 - . 02 - . 06 . 15 - . 01 - . 48 - . 19 - . 17 - . 12 MGB Lt - . 17 - . 08 - . 20 - . 10 - . 01 . 10 - . 04 - . 03 . 09 . 09 . 09 . 02 - . 02 - . 40 - . 30 - . 20 - . 02 MGB Qr . 05 . 02 - . 01 - . 02 . 05 . 08 . 00 . 08 - . 01 - . 08 - . 07 . 04 - . 01 - . 47 - . 32 - . 16 - . 14 PMF As . 16 . 13 . 14 . 06 . 16 - . 03 . 06 . 14 . 00 . 00 - . 09 . 08 - . 11 - . 42 - . 22 - . 20 - . 11 PMF Co . 15 - . 08 - . 17 - . 07 . 00 . 03 . 16 - . 04 . 04 . 03 . 01 - . 06 - . 21 - . 47 - . 23 - . 26 - . 13 PMF Fg - . 01 . 10 . 03 - . 03 . 06 . 16 . 17 . 09 - . 09 . 03 - . 02 . 07 - . 04 - . 37 - . 19 - . 14 - . 19 PMF Lt - . 06 - . 15 - . 26 - . 15 - . 03 . 08 . 05 - . 01 - . 04 - . 12 - . 04 - . 06 . 01 - . 33 - . 20 - . 20 - . 11 PMF Qa . 03 . 01 - . 07 - . 07 - . 02 . 01 . 13 - . 02 . 12 . 05 - . 01 - . 15 - . 13 - . 50 - . 16 - . 21 - . 21 PMF Qr . 01 . 14 - . 01 . 02 . 04 . 00 . 10 . 03 . 01 . 01 . 00 . 14 - . 90 - . 45 - . 24 - . 15 - . 21 Total significant 2 4 1 15 6 2 3 8 1 1 11 3 11 31 28 11 8 Mean . 03 . 01 - . 06 - . 14 . 05 . 03 . 03 . 08 . 03 - . 01 - . 03 . 03 . 05 - . 27 - . 18 - . 09 - . 10 Pearson r, Growth - Climate Correlation v. Latitude . 00 - . 12 - . 17 - . 62 - . 11 - . 44 - . 16 - . 11 - . 05 . 10 . 29 . 30 . 56 . 77 . 50 . 72 . 36 94 Table 3.5 . Partial correlation matrix (Ppt only) for the growth - climate coefficient (Pearson r) for each of 34 climate variables for each population p rior to the year of ring formation. Cells in bold are significant (p < 0.05). Sites are arranged from north (top) to south (Table 3.1) . Site c odes: BvI ial Forest, PNC = Price Nature Center, Vrh = Voorhees Audubon Sanctuary, Warner = Warner Audubon Sanctuary. Species codes: As = sugar maple Acer saccharum , Ba = yellow birch Betula alleghaniensis , Co = shagbark hickory Carya ovata , Fg = American beech Fagu s grandifolia , Lt = tulip poplar Liriodendron tulipifera , Ps = white pine Pinus strobus , Qa = white oak Quercus alba , Qr = red oak Quercus rubra , Tc = eastern hemlock Tsuga canadensis . Population pMay Ppt pJun Ppt pJul Ppt pAug Ppt pSep Ppt pOct Ppt pNov Ppt pDec Ppt Jan Ppt Feb Ppt Mar Ppt Apr Ppt May Ppt Jun Ppt Jul Ppt Aug Ppt Sep Ppt HMC As - .05 - .07 .12 - .01 .11 .03 - .11 .04 .11 - .12 .04 .06 .06 .20 .30 - .01 - .02 HMC Ba - .12 - .10 .09 .12 .09 .02 - .05 .14 .20 .05 - .02 - .04 .08 .33 .24 - .09 .05 HMC Ps - .26 - .10 - .17 .03 .01 .04 - .03 - .12 - .03 .07 - .01 - .05 - .05 .20 .16 .12 .04 HMC Qr - .13 - .05 - .01 .01 .05 - .02 - .10 - .09 - .04 - .08 .08 - .01 .09 .18 .20 .13 .09 HMC Tc - .07 .18 .09 - .05 - .15 .00 .05 - .08 - .05 .02 - .07 .08 .15 - .06 - .01 - .03 - .07 Mwd As - .04 - .04 .20 .09 .13 - .08 .12 - .03 - .07 .07 .13 - .01 .05 .04 .13 - .01 .07 Mwd Ba - .01 - .03 .01 .20 .05 .04 .00 - .06 .13 .08 - .15 - .12 .04 .16 .30 - .10 .04 Mwd Ps - .05 - .16 .09 .24 .16 .13 .03 - .04 .05 .19 - .11 .03 - .02 .08 .32 .02 .07 Mwd Qr .00 - .10 .11 .28 .10 .10 - .03 .09 .04 .00 - .05 .01 .01 .24 .35 - .01 .11 Pet Tc .07 .01 .26 .08 .15 - .26 .13 - .16 - .07 - .17 .00 .01 .09 .10 - .02 - .09 - .02 BvI As - .08 .02 .25 .21 .08 - .11 - .01 - .10 .15 - .03 .07 .03 .09 .21 .23 .05 .09 BvI Ba - .06 - .05 .10 .28 - .07 .02 - .05 .07 .08 .05 - .04 .03 .21 .24 .39 .05 .11 95 Table 3.5 BvI Tc - .01 - .09 .19 .15 .05 - .27 .13 .06 .05 .01 - .06 - .14 .12 .11 .09 - .05 - .02 CoP Fg - .07 .02 .21 .26 - .02 - .09 .07 .01 .21 .13 - .07 .00 .10 .25 .34 - .04 .00 CoP Ps .05 - .12 .04 .14 .06 .15 .07 - .08 .06 .08 - .12 .04 .12 .11 .22 .16 .08 JSB As .04 - .05 .00 - .02 - .06 .24 .05 - .01 .15 .14 - .04 - .04 .30 .21 .28 - .07 - .02 JSB Fg - .05 - .19 .03 .16 .04 .00 .16 .03 .28 .21 - .10 .10 .20 .26 .11 .07 .04 JSB Tc .15 .10 .10 .05 .04 - .05 .22 .13 .08 .02 - .05 - .06 .13 .16 .08 - .05 .02 PNC As .01 - .03 .26 - .12 .06 .07 .09 - .07 .18 .20 - .05 - .10 - .01 .02 .23 .15 - .06 PNC Fg .05 - .03 .31 - .02 .04 .03 .17 .00 .15 .16 - .15 - .13 .11 .05 .18 .18 .09 PNC Qa - .22 - .15 .17 .02 - .17 - .07 .21 .08 .20 - .03 - .03 .19 .12 .02 .47 .09 - .12 PNC Qr - .09 - .17 .09 .04 - .14 .00 .11 .00 .26 - .01 - .02 .08 .16 .17 .29 .08 .00 War As - .03 - .10 - .04 - .14 .09 .12 - .02 .07 .28 .11 .05 .12 .00 .23 .17 .23 - .13 War Fg - .11 .03 - .01 - .01 .12 .12 .09 .05 .12 .17 - .06 - .05 .21 .26 .19 .33 - .07 War Lt - .16 .12 .00 .01 .06 .16 .17 .15 .24 .05 .03 .08 .24 .28 .30 .15 - .16 War Qa - .14 - .04 .04 .06 .08 - .03 .19 .13 .18 .00 .07 .05 .23 .28 .28 .16 .04 War Qr - .08 - .06 .05 .11 .01 .09 .18 .01 .04 - .10 - .15 .03 .19 .33 .22 .17 - .02 Vrh As - .12 .03 - .15 .03 .00 .21 .09 .03 .27 .10 - .01 - .01 - .02 .35 .29 .16 - .13 Vrh Co .00 - .13 - .20 - .03 .03 .05 - .06 .13 .28 .05 .06 .06 .15 .33 .08 - .01 - .03 Vrh Qa - .08 .01 - .11 .12 .01 .02 .19 .13 .27 - .04 .01 .10 .26 .40 .37 .03 - .08 Vrh Qr - .01 .00 - .03 .15 - .07 .12 .24 .06 .09 - .10 - .11 - .01 .22 .50 .33 .06 - .07 KNC As - .17 .09 - .10 - .01 .18 .15 .12 .10 .27 .21 - .05 - .07 .02 .37 .30 .27 - .10 96 Table 3.5 KNC Fg - .20 .10 - .16 .03 .15 .05 .22 .02 .18 .15 - .12 - .04 .19 .49 .32 .24 - .08 KNC Lt - .20 .10 - .02 .05 .13 .12 .17 .05 .21 .14 - .06 - .04 .20 .40 .31 .26 - .01 KNC Qr - .11 .00 - .10 - .10 .12 .14 .21 .01 .17 .13 - .23 .06 .12 .41 .32 .22 .00 MGB As - .19 .08 - .12 - .02 .07 .13 .19 - .05 .13 .06 - .01 - .13 .18 .44 .32 .09 .02 MGB Co - .14 .07 - .02 .18 .08 .12 .12 .07 .15 - .01 .09 - .09 .14 .40 .33 .12 .03 MGB Fg - .07 - .03 .04 .18 .15 .12 .14 .14 .17 .03 - .04 - .26 .20 .37 .35 .11 - .03 MGB Lt .03 .12 .01 .15 .12 - .08 .17 .08 .08 .04 .07 - .18 .26 .57 .40 .21 - .03 MGB Qr - .14 .02 - .15 .19 .03 .05 .10 .08 .18 - .01 - .03 - .07 .23 .45 .45 .15 .06 PMF As - .06 - .09 - .07 .12 .36 .28 .00 - .09 - .06 - .09 - .07 .12 .36 .28 .00 - .09 - .06 PMF Co - .07 .13 - .02 .21 - .02 - .08 .10 .11 .13 .03 - .01 - .13 .23 .32 .20 .09 - .02 PMF Fg - .06 .00 - .07 .29 - .02 .11 .21 .13 .09 - .02 - .06 - .07 .23 .48 .22 .12 - .21 PMF Lt - .15 .05 - .11 .20 .58 .26 .13 - .21 - .15 .05 - .11 .20 .58 .26 .13 - .21 - .15 PMF Qa .04 .05 .03 .21 .46 .27 .05 - .06 .04 .05 .03 .21 .46 .27 .05 - .06 .04 PMF Qr - .08 - .05 - .11 .12 - .07 - .02 .13 .05 .13 .03 .03 - .08 .19 .49 .37 .20 - .11 Total significant 3 2 8 12 2 3 10 1 18 4 1 1 16 33 32 7 2 Mean - .07 - .02 .03 .10 .04 .03 .10 .04 .14 .04 - .03 - .02 .14 .28 .26 .09 - .02 Pearson r, Growth - Climate Correlation v. Latitude - .05 - .07 .12 - .01 .11 .03 - .11 .04 .11 - .12 .04 .06 .06 .20 .30 - .01 - .02 97 Table 3.6 . The monthly mean temperature (T mean ) and total precipitation (Ppt) variables with the steepest slope in a regression of growth - climate coefficient (Pearson r) vs. time . Sites are arranged from north (top) to south. Months in bold indicate the variable which changed most in a column. Site codes: BvI = Beaver Island, CoP = Colonial Point, HMC = Huron Mountain Club, Mothers Memorial Forest, PNC = Price Nature Center, Vrh = Voorhees Audubon Sanctuar y, Warner = Warner Audubon Sanctuary. Species codes: As = sugar maple Acer saccharum , Ba = yellow birch Betula alleghaniensis , Co = shagbark hickory Carya ovata , Fg = American beech Fagus grandifolia , Lt = tulip poplar Liriodendron tulipifera , Ps = white p ine Pinus strobus , Qa = white oak Quercus alba , Qr = red oak Quercus rubra , Tc = eastern hemlock Tsuga canadensis . Population T mean , biggest changer (direction) Ppt, biggest changer (direction) HMC As prior - Jun ( - ) current - Mar ( - ) HMC Ba current - Feb ( - ) current - Mar ( - ) HMC Ps current - Feb ( - ) current - Apr (+) HMC Qr prior - May ( - ) current - Mar ( - ) HMC Tc current - Jul (+) current - Aug (+) Mwd As prior - Jun ( - ) prior - May ( - ) Mwd Ba current - Feb ( - ) prior - Sep ( - ) Mwd Ps prior - May ( - ) current - Jan ( - ) Mwd Qr current - Jun ( - ) prior - Jun (+) Pet Tc current - Jul (+) current - Jan ( - ) MaB As prior - Jun ( - ) prior - Oct ( - ) MaB Ba current - Mar ( - ) current - Sep (+) MaB Tc prior - Jun ( - ) current - Jan ( - ) CoP Fg prior - Jul ( - ) prior - Jun (+) CoP Ps current - Jan (+) prior - Sep ( - ) JSB As prior - Jul ( - ) prior - Sep ( - ) JSB Fg prior - May ( - ) current - Feb ( - ) JSB Tc current - Jan (+) prior - May ( - ) PNC As current - Jan (+) prior - Dec ( - ) PNC Fg current - Jan (+) prior - Aug ( - ) PNC Qa prior - May ( - ) prior - Sep ( - ) PNC Qr current - Sep (+) prior - Sep ( - ) War As prior - Jul ( - ) prior - Oct ( - ) War Fg prior - May ( - ) prior - Aug ( - ) War Lt current - Jan (+) prior - May (+) War Qa current - Feb ( - ) prior - Sep ( - ) War Qr prior - Dec ( - ) current - Mar ( - ) Vrh As prior - Jul ( - ) prior - Sep ( - ) 98 Table 3.6 Vrh Co prior - Sep ( - ) current - Feb (+) Vrh Qa current - Jan (+) prior - Sep ( - ) Vrh Qr current - Sep (+) prior - Sep ( - ) KNC As prior - Jul ( - ) prior - Sep ( - ) KNC Fg prior - Jul ( - ) prior - Sep ( - ) KNC Lt prior - June (+) prior - Sep ( - ) KNC Qr current - Sep (+) prior - Sep ( - ) MGB As prior - Jul ( - ) current - Jul ( - ) MGB Co current - Jan (+) current - Apr ( - ) MGB Fg prior - Jul ( - ) prior - May (+) MGB Lt current - Jul (+) prior - May (+) MGB Qr current - May (+) prior - May (+) PMF As current - Jul (+) current - Sep (+) PMF Co current - Jan (+) prior - Nov (+) PMF Fg current - Jul (+) current - Apr ( - ) PMF Lt prior - May ( - ) prior - Dec ( - ) PMF Qa prior - May ( - ) prior - May (+) PMF Qr prior - May ( - ) prior - Dec ( - ) 99 Table 3.7 . Correlation matrix (Pearson r) for 190 3 2004 mean annual temperature among study sites . Site codes: BvI = Beaver t, PNC = Price Nature Center, Vrh = Voorhees Audubon Sanctuary, Warner = Warner Audubon Sanctuary. PMF MGB KNC Vrh War PNC JSB CoP BvI Pet Mwd HMC PMF 1.000 0.949 0.743 0.804 0.801 0.663 0.745 0.642 0.634 0.564 0.555 0.589 MGB 1.000 0.820 0.865 0.876 0.722 0.829 0.717 0.723 0.658 0.650 0.687 KNC 1.000 0.965 0.972 0.928 0.924 0.875 0.887 0.892 0.889 0.869 Vrh 1.000 0.978 0.922 0.925 0.842 0.873 0.859 0.856 0.850 War 1.000 0.910 0.934 0.860 0.890 0.860 0.856 0.850 PNC 1.000 0.902 0.894 0.903 0.895 0.893 0.870 JSB 1.000 0.937 0.933 0.897 0.890 0.903 CoP 1.000 0.959 0.928 0.922 0.913 BvI 1.000 0.942 0.939 0.922 Pet 1.000 0.997 0.969 Mwd 1.000 0.969 HMC 1.000 100 Table 3.8 . Correlation matrix (Pearson r) for 190 3 2004 total annual precipitation among study sites . Site codes: BvI = Beaver Island, Bird Price Nature Center, Vrh = Voorhees Audubon Sanctuary, War = Warner Audubon Sanctuary. PMF MGB KNC Vrh War PNC JSB CoP MaB Pet Mwd HMC PMF 1.000 0.810 0.378 0.399 0.334 0.357 0.192 0.280 0.215 0.220 0.229 0.317 MGB 1.000 0.441 0.494 0.417 0.395 0.201 0.219 0.160 0.185 0.180 0.213 KNC 1.000 0.891 0.922 0.676 0.663 0.310 0.269 0.278 0.277 0.244 Vrh 1.000 0.912 0.742 0.604 0.249 0.191 0.211 0.218 0.188 War 1.000 0.732 0.630 0.232 0.174 0.230 0.232 0.171 PNC 1.000 0.662 0.257 0.220 0.232 0.223 0.165 JSB 1.000 0.589 0.519 0.495 0.484 0.335 CoP 1.000 0.885 0.701 0.701 0.614 MaB 1.000 0.684 0.677 0.638 Pet 1.000 0.995 0.722 Mwd 1.000 0.747 HMC 1.000 101 CHAPTER FOUR CLIMATE CHANGE IS PROJECTED TO HINDER PRODUCTIVITY OF TREE SPECIES IN INDIANA AND MICHIGAN OVER THE 21 st CENTURY Abstract Global c limate is projected to continue changing over the rest of this century and beyond. Projections of local monthly temperature and precipitation are available through the end of this century based on the models of various meteorological research groups (Taylor et al. 2012) . Such projections can be used along with modeled relationships between tree growth and climate to project future growth changes under climate change. Growth - climate models were calibrated over one half of available data for each of 46 tree populations from nine species across Indiana and Michigan. Model verification was then attempted on the other half of each dataset. Successfully verified models were used to project future growth under four climate - change scenarios. Model verification was successful in 14 of 46 populations. In these populations, the most influential variables were related to summer - moisture stress. A negative association with current - June maximum temperatur e was the most influential variable in five of these models, followed by a negative association with current - July maximum temperature in three, and a positive association with current - July precipitation in two. Growth was projected to decline significantly in 12/14 populations under the two highest - warming climate - change scenarios and to increase in one population. Under the mildest climate - change scenario, growth was projected to decline in four populations, to increase in one, and to have no significant c hange in 102 nine. A preponderance of projected growth decline in all but the mildest scenario suggest that significant management options should be considered including assisted colonization and gene flow. Introduction Climate change is expected to have a n egative impact on the trees of the eastern United States. Using forest inventories across the eastern U.S. and Canada, Prasad et al. (2020) modeled future habitat suitability and migration potential of 25 tree species. For the U.S., they found that under R epresentative Concentration Pathway (RCP) 8.5 , 500,000 square km of habitat may be lost, depending on species, and much of this will be high - quality habitat . This will be partially offset by gains in potential habitat, however most of the gained po tential habitat will be lower quality, and little of it will be realized over this century due to dispersal limitations. Further, tree - ring - based studies of growth - climate relationships suggest that strong limitations placed on growth by summer - moisture st ress are likely to be exacerbated as climate change proceeds. These studies were cited in greater detail in the previous chapter. To recapitulate, this pattern was found for both Liriodendron tulipifera (tulip poplar) and Quercus spp. (oaks) across their r ange in eastern North America (LeBlanc & Berland 2019, LeBlanc et al. 2020), for eight hardwood species spanning Georgia to Vermont (Martin - Benito & Pederson 2015), for eleven species in the w estern Great Lakes Region (Graumlich 1993), and for Acer saccharum (sugar maple) and Q. alba (white oak) across much of the eastern U.S. (Au et al. 2020). 103 This pattern was corroborated for Indiana and Michigan in the previous chapter . I n a network of tree - ring chronologies from nine species spanning an eight - degree latitudinal gradient, the dominant factor limiting radial growth was summer - moisture stress, i.e., the strongest relationships between growth and climate were negative relationships with June average temperature and positive relat ionships with June precipitation. This suggest s that ongoing climate change in Indiana and Michigan, which is expected to bring increased summer temperature and more erratic precipitation patterns (Hayhoe et al. 2010, Christensen et al. 2013, Byun & Hamlet 2018), will lead to reduced tree growth in the region. However, the previous chapter also showed that the summer - moisture - stress signal was weaker at the more northerly sites, corroborating other studies (Martin - Benito & Pederson 2015, LeBlanc & Berland 2019, LeBlanc et al. 2020). The weaker signal was particularly apparent in Tsuga canadensis (eastern hemlock). Further, other growth - climate relationships found in the previous chapter and elsewhere in the literature suggest that certain aspects of climate change could ameliorate the growth reductions due to anticipated increases in summer moisture stress. Winter temperature, winter precipitation and spring temperature are expected to increase (Hayhoe et al. 2010, Christensen et al. 2013, Byun & Hamlet 2018 ). Growth of many of the 46 tree populations studied in Chapter 3 was positively correlated with those variables. For example, current - year growth of 18/46 populations was positively correlated with January precipitation ( Jan Ppt ). For the remaining populat ions, the relationship was nonsignificant. Growth of eight populations was positively correlated with prior - December mean temperature (prior - Dec Tmean ). Similar results were found elsewhere in the northeastern U.S. (Pederson et al. 2004; Martin - Benito and P ederson 2015). Additionally, growth of all four T. canadensis 104 populations studied in Chapter 3 was positively correlated with current - Mar Tmean , which has also been found elsewhere (Cook and Cole 1991, Dye & Woods 2019). Disentangling the potentially positi ve effects of warmer springs and warmer, wetter winters from the negative effects of warmer, more drought - prone summers requires mathematical modeling. This has been done to some extent in trees around the world, including in Mexico (Brienen et al. 2010, Pompa - Garcia et al. 2017), Australia (Nitschke et al. 2017), Bangladesh (Rahman et al. 2018), China (Su et al. 2015), western North America (Chhin et al. 2008; Chen et al. 2010; Charney et al. 2016), Spain (Sanchez - Salguero et al. 2017), Lithuania (Rimkus et al. 2018), and eastern North America (Huang et al. 2010, Charney et al. 2016, Chhin 2015, Chhin 2016, Tei et al. 2016). However, little work has been done on the native species of temperate eastern North America, outside of conifers and oaks (Charney et al. 2016). Here, I select the most parsimonious growth - climate models for each of the 46 tree populations discussed above. I then attempt to verify the statistical skill of those models. Finally, I employ all models which have sufficient predictive skill , together with projections of future climate, to project future growth over the rest of this century. I hypothesized that growth will generally decline over this century, but that the decline will be milder at more northerly sites especially for T. canade nsis . Methods Field Methods and Tree - Core Processing A network of tree - ring chronologies was generated at 12 sites along a latitudinal gradient from southern Indiana to northern Michigan (Fig. 3.1, Table 3.1). Between one and six species are represented at each site for a total of 46 populations (stands) fro m nine species. 105 Details about site/species selection and field methods are described in Chapter 3. Cores were later sanded, the calendar dates of tree rings were established, ring widths were measured, and ring - width measurement series were standardized an d detrended according to disciplinary standards (Stokes & Smiley 1996, Speer 2009). Again, further detail is provided in Chapter 3. Growth - Climate Modeling Modeling the relationship between tree growth and climate was conducted in the R statistical envir onment (R Core Team 2017). The climate variables considered in the modeling process included total monthly precipitation (Ppt) and minimum, mean, and maximum temperature (T min , T mean , and T max ), each averaged over the month. Variables were considered at th e single - month scale throughout a window from May of the year preceding tree growth through September of the year of tree growth (Fritts 1976). The single - month scale was selected because pilot analysis revealed that single - month variables explained more v ariability than did variables spanning, two, three, and/or four months. Gridded 4 km × 4 km interpolations of these variables were obtained from the PRISM Climate Group (Daly et al. 2008). The interval held in common by all tree - ring chronologies was spli t into two periods: the calibration period, 1903 1953, and the verification period, 1954 2004. Models were selected (Venables & Ripley 2002) was used to conduct forward - stepwise multiple regression (Chhin et al. 2008). This regression was conducted over three steps, with each step adding the single variable which lowered the Akaike Information Criterion (AIC) by the largest amount (Akaike 106 1974). I stopped at thre e steps, ensuring after each step that AIC went down, because it was found in pilot modeling that more than three variables in the final model resulted in over - fitting, i.e., though AIC typically continued to go down and adjusted r 2 to go up, the success o f model verification over the 1954 2004 verification period tended to go down. To be deemed reliable for projection of future growth, models had to pass three criteria - value had to be less than 0.05. Secon d, the reduction of error (RE) had to be greater than zero. This statistic was first used in meteorology for the verification of weather forecasting (Lorenz 1956), and it has been adopted extensively in the dendroclimatological literature (Fritts 1976, Coo k et al. 1999). It involves comparing the predictive power of the selected model with the predictive power of the calibration - period skill. It can be found acco rding to where is the observed ring - width index (RWI) at year i , is the predicted RWI at that year, and is the mean RWI of the calibration period. The final criterion was that the coefficient of efficiency (CE) also had to be greater than zero. CE was first used in hydrology (Nash & Sutcliffe 1971) and has also been adopted by dendroclimatologists (Cook et al. 1999). It has the same theoretical range and predictive - skill threshold as RE. It differs in that it compares the predictive power of the selected model with that of the verification - period mean according to 107 Where is the mean RWI of the verification peri od, and all other abbreviations are as stated previously. As soon as a model failed any of these criteria, the corresponding population was discarded from further analysis. Future - Growth Projections If a model passed the above three criteria, then that model was refit over the entire common interval, i.e., the calibration and verification periods were combined (Pederson et al. 2012). Corresponding climate data were then entered into the entire - interva l model to simulate growth over the observational record (Brienen et al. 2010, Nitzsche et al. 2017). Gridded, downscaled Coupled Model Intercomparison Project Phase 5 (CMIP5) ite from https://gdo - dcp.ucllnl.org/downscaled_cmip_projections/ for two climate - change scenarios, RCPs 4.5 and 8.5 (Taylor et al. 2012). RCP 4.5 represents the second - most benign scen ario out of four, wherein atmospheric greenhouse gas concentrations stabilize by 2100. RCP 8.5 is the most severe scenario, wherein concentrations continue to increase eve n beyond 2100. The appropriate 1/8° × 1/8° grid was selected for each study site. All sites for which at least one growth - climate model was successfully verified fell within a downscaled grid except for the projections from many CMIP5 global circulatio n models (GCMs) were available. I selected from a pool of 13 models which include the influence of the Great Lakes in their simulations (J. A. Andresen pers. comm. ). I chose to forecast a wide range of potential future growth. O f the 13 potential GCMs, tha t which predicted the greatest amount of 21 st - century warming and that 108 which predicted the least were selected (Moser et al. 2020). This was determined based on the median latitude and longitude of the tree - ring network. For the grid within which the media n latitude /longitude fell, RCP 4.5 and 8.5 projections were obtained from each of the 13 potential among the GCMs under each RCP. For each RCP, the GCM MRI - CGCM3 predicted the least warming (Yukimoto et al. 2012) and the GCM GFDL - CM3 predicted the most (Griffies et al. 2011). Projected climate from these two GCMs was compared across the different modeled scenarios and compared to the historic common interval. Rathe r than focusing on all examined variables, the variables which were consistently most important in the selected growth - climate models were examined. Differences in projected temperature and precipitation were visually compared with bar and box plots, respe ctively. To project future growth, relevant projected variables from MRI - CGCM3 and GFDL - CM3 were entered into the models selected for each retained tree - ring chronology. The function Core Team 2017). Projected RWI from each scenario (2 GCM s × 2 RCPs) were averaged over this interval and 95% confidence intervals of the mean were calculated. Comparisons of the mean of these four scenarios and of simulated RWI over the observational reco rd were made for each population with ANOVA. 109 Results Model Calibration and Verification Of the 46 growth - climate models fit over the 1903 1953 calibration period, 27 were significant (p < 0.05) when fit to the 1954 2004 verification period ( Table 4.1 ). The next two validation criteria were that RE and CE had to be > 0 over the verification period. Of the 27 models which made it to these steps, 14 passed each of these criteria ( Table 4.1 ). Model verification was generally more successful in the south of the gradient than in the north. Among 22 populations in the northern half of the gradient, there were only five successful verifications, three of which were T. canadensis . Validations were successful for 9/24 populations in the southern half of the gradie nt, from a mix of all six species that were sampled in the south . Characteristics of Validated Models For 9 of the 14 successfully validated models, the first variable to enter the model in stepwise regression over the calibration period was a current - summer maximum - temperature variable, whether in the month of June (5 populations), July (3), or August (1); these were all populations in the southern half of the study gradient; the correlation was always negative ( Table 4.2 ). For the remaining 5 successfully validated populations, each in the northern half of the gradient, current - Jul Ppt was the first variable in 2 populations (positive relationship), prior - JulT Tmax and prior - Aug Tmean were each first in 1 (negative relationship), and prior - Nov Ppt was first in 1 (positive relationship). Among the second and third variables to go in the model, frequent relationships were found with current - Jun or - Jul Tmax (3 populations, relationships all negative), 110 with current - Jun or - Jul Ppt (5 populations, all pos itive), and with T min , T mean , or T max in the prior - October, November, or December (7 populations, all positive) ( Table 4.2 ). When the verified model was applied to the entire common interval, the three retained variables in combination explained between 14.7 and 48.3% of the variability of each Table 4.2 ). Partial regression coefficients reveal the effect of retained variables in isolation. For example, the strongest growth - temperature relationship was between current - Jul Tmax and A. saccharum RWI at Voorhees Audubon Sanctuary (Vrh). If all other variables were held constant, a 1 °C increase in Vrh Jul Tmax would result in a .0571 decrease in A. saccharum RWI ( Table 4.2 ). The strongest positive growth - temperature relationship was between current - Mar Tmin and T. canadensis There, a 1 C increase in current - Mar Tmin led to a 0.0315 increase in T. canadensis RWI. The effects of retained precipitation variables were always positive ( Table 4.2 ). The maximum w as a 0.00229 increase in RWI per mm increase of current - Jun Ppt for L. tulipifera at Pioneer Mothers Memorial Forest (PMF). Comparison of Historic and Projected Growth The most consistently important variables to enter the retained models were summer tempe rature (June, July, or August) and precipitation (June or July). At all sites, June August mean temperature was projected to increase significantly under all scenarios relative to the historic common interval. Projected increases in the mean range from 1.1 °C at PNC under MRI - CGCM3 for RCP 4.5 to 7.1 °C under GFDL - 4.1a). Under the GFDL - CM3 model, increases were more dramatic, even for RCP 4.5, than under 111 MRI - CGCM3 for RCP 8.5 (Fig. 4.1a). Projected temperatu re increases were significant throughout the gradient, but generally greater in the north. Regarding June July total precipitation, it was generally projected to increase (Fig. 4.1b). This was true under both models, MRI - CGCM3 and GFDL - CM3. In the south of the gradient, projected increases were generally more dramatic under GFDL - CM3. Future - Growth Projections Growth over the rest of this century under both RCP 4.5 and 8.5 under both GCMs MRI - CGCM3 and GFDL - CM3 was generally projected to decline in relation to the historic common interval (Fig. 4.2). However, for some populations there would be no significant gr owth changes under the milder GCM, MRI - CGCM3. That is true for all three of the T. canadensis populations and for A. saccharum at Vrh (single - population one - way ANOVA p - values > 0.05). All ten other populations, except for one, would experience growth decl ine under MRI - CGCM3, at least for RCP 8.5. In some cases, this would be dramatic. A. saccharum at Maywood History Trail (Mwd) would decline by 13.1 and 16.4% under RCPs 4.5 and 8.5, respectively. Carya ovata at Mary Gray Bird Sanctuary (MGB) would decline by 8.13 and 9.91% (Fig. 4.2). The sole population projected to increase in growth under MRI - CGCM3 scenarios is Q. alba at Price Nature Center (PNC), where its growth would increase by 5.18 and 8.20% under RCPs 4.5 and 8.5, respectively. Under GFDL - CM3 scen arios, growth would significantly decline for all but two populations, and the decline would generally be more dramatic than under the MRI - CGCM3 scenarios. Even for RCP 4.5, the decline under GFDL - CM3 would generally be more dramatic than either MRI - CGCM3 4.5 or MRI - CGCM3 8.5 (Fig. 4.2). For GFDL - CM3 under RCP 4.5, growth 112 declines would range from a non - significant 3.98% for L. tulipifera at PMF to 28.0% for A. saccharum at Mwd (Fig. 4. 2). For GFDL - CM3 under RCP 8.5, declines would range from 4.47% for L. t ulipifera at PMF (p > 0.05) to 34.2% for A. saccharum at Mwd. As under MRI - CGCM3, Q. alba growth at PNC would significantly increase under GFDL - CM3, by 11.8 and 12.1% under RCPs 4.5 and 8.5, respectively. There was no latitudinal trend in future - growth pro jections. For example, the three populations projected to decline the most under GFDL - CM3 came from both the most northerly site (Mwd) and the third - most - southerly site (Vrh); the sole population projected to accelerate came from a mid - gradient site, PNC; the populations projected to face the mildest decline in growth came from the most southerly site, PMF. However, the lack of a clear trend may be confounded by differences among species. A different suite of species was sampled at each end of the gradient. Further, t he maximum representation of any species was three populations, a number too small for the examination of within - species latitudinal trends. Discussion Model validation was relatively unsuccessful throughout the gradient, with 14/46 populations passing all three validation criteria. For the remaining 32 populations, it is not possible to project changes in growth with the present dataset. By contrast, calibrated growth - climate models were successfully verified for 3/5 populations in Australia (N itzschke et al. 2017), 4/4 sites in Lithuania (Rimkus et al. 2019), 3/3 populations in Bangladesh (Rahman et al. 2018), 3/3 populations in northern Mexico (Pompa - Garcia et al. 2017), 33/33 populations in eastern boreal Canada, and 3/3 populations in subtro pical China (Su et al. 2015). The low 113 verification rate in the present study may be due to the relatively benign climate of temperate eastern North America. Its humid climate leads to complacent (low - variability) tree - ring chronologies (Phipps 1982). In a gridded reconstruction of summer moisture based on composites of tree - ring chronologies available near each of 154 points throughout the continental United States, the lowest verification statistics were found for the upper Midwest and northern New England (Cook et al. 1999). The present finding that northerly populations had a lower model - verification success rate than did southerly ones supports that. Further, relatively low variability was also explained by growth - climate models derived from tree - ring chronologies in nearby Ontario: on the Bruce Peninsula and proximal islands (Buckley et al. 2004). The finding in this study that summer precipitation and temperature variables were most important, being the first to enter the calibrated models in 13/14 successfully val idated models, was consistent with the findings in Chapter 3 and with other published information on temperate tree growth in eastern North America. For example, LeBlanc et al. (2020) performed stepwise multiple regression to model ring - width - climate relat ionships for 45 L. tulipifera populations across eastern North America. For 35 populations, current - summer precipitation was the first or only variable to enter the model, and for 8 of 10 remaining populations a different summer precipitation or temperatur e variable was first to enter. Similar results were found for non - native Pinus hybrids growing in Michigan (Chhin 2015). The only other classes of variables consistently retained in validated models were with current - Mar Temp , and prior - Nov/Dec Temp , and pr ior - Nov/Dec Ppt . Relationships with these variables were positive, and it was hypothesized a priori that projected increases in these 114 variables could ameliorate projected exacerbation of summer moisture stress. However, even when these variables were select ed in models, they tended to be selected second or third, and their influence tended to be relatively low ( Table 4.2 ). Again, this was consistent with the modeling of LeBlanc et al. (2020). For 45 calibrated L. tulipifera growth - climate models, prior - Winte r Tmin was included in 17, however it was the first variable to enter the model only twice. - climate models for hybrid Pinus and hybrid Populus (aspen) growing in Michigan plantations, winter and spring temperature and precipitation tended to have more significant contributions to the models. Moving northward, Huang et al. (2013) found in eastern boreal Canada that climate variables outside of the growing season were frequently first to enter their calibrated models. Further, at these higher latitudes, growth was often positively related with temperature even during the growing season. It was hypothesized a priori that projected reductions in 21 st century tree growth would be greatest in the south of the gradient because the most southerly populations exhibited the strongest indications of summer moisture stress, i.e., the most strongly negative growth - temperature and strongly positive growth - precipitation relationships (Chapter 3). However, growth reductions were not projected to be greater in the south (Fig. 4.2). It should be reiterated that this question is confounded by the differing assemblage of species across the gradient. Another factor is that the projected warming of the GCMs selected for this study was greater in the north than in the south (Fig. 4.1a), as has been more widely predicted (Christensen et al. 2013). Further, projected summer moistening was greater in the south than in the north (Fig. 4.1b). Thus, climate change impacts on ra dial growth will not necessarily be 115 consistent with the expected trend of low - latitude/altitude populations suffering more than their conspecifics at higher latitudes/altitudes (Parmesan 2006). The opposite trend may even be found for some species, as pred icted for Pseudotsuga menziesii (Douglas fir) in western North America (Chen et al. 2010). A bright spot of the present study was that growth of one population, Quercus alba at Price Nature Center, was projected to increase under all climate change scenarios considered. This was because this population was relatively insensitive to summer temperature ( Fig. 3.4 a, Tables 3.4 and 4. 2 ). It was more sensitive to summer precipitation and prior - fall temperature, two variables with which its growth was posit ively correlated and which were projected to increase under the climate projections selected. This unique set of growth - climate relationships could be due to site - level factor s, however the other species sampled at this site, including the congeneric Q. ru bra , did not share the insensitivity to summer temperature. It could also be due to adaptation to high temperatures . I t is recommended that this be tested experimentally. If this population proved insensitive to experimental warming, it would be a good can didate for assisted gene flow (Aitken & Whitlock 2013). By contrast, the other 13 populations for which future growth was projected are likely to experience lower growth for the rest of this century relative to historic observations under GCM GFDL - CM3 and 9/14 for MRI - CGCM3. This concerning result has been found for several other species/regions: P. menziesii in western North America (Chen et al. 2010), Pinus contorta (lodgepole pine) in Alberta (Chhin et al. 2008) , Mimosa acantholoba in Mexico (Brienen et al. 2017), several lineages of both Pinus nigra × P. densiflora (hybrid pine) and Populus × smithii (hybrid aspen) in Michigan (Chhin 2015, 2016), three urban landscaping trees in Australia 116 (Nitzsche et al. 2017), and three species in Bangladesh (Rahman et al. 2018). By contrast, only two of three endangered conifer species studied in northern Mexico were projected to face growth declines, and the other to experience increased growth (Pompa - Garcia et al. 2017). Similarly, two conifers in Spain were projected to lose growth, the other to gain growth (Sanchez - Salguero et al. 2017). In subtropical China, one species was projected to lose growth, one to experience no change, and one to gain growth (Su et al. 2015). In Lithuania, several pop ulations of Pinus sylvestris (Scots pine) were projected to experience radial growth increases (Rimkus et al. 2018). In four species of eastern boreal Canada growth increases were largely projected in the north of the region and little change in the south (Huang, et al. 2013). Tei et al. (2017) projected circumboreal future growth and found that growth was likely to increase across wide areas, such as in northern Eurasia and the west coast of Canada but likely to decrease over large areas elsewhere, such as interior Alaska and Canada. Charney et al. (2016) projected growth across North America and found it was likely to decrease in the interior west and Midwest and to increase in the far west, the southeast, and the far northeast. It should be noted that the future growth projections presented here should be interpreted cautiously because the climate projections used in modeling future growth fall outside the range of climate variability used to calibrate the models (Chen et al. 2010). Further, there are a variety of other factors not considered in the models which will be influenced by global change, such as increasing erraticism of precipitation (Byun & Hamlet 2018), increased fire prevalence (Tang et al. 2015), conversion of snow to rain (Byun & Hamlet 2 018), more rapid snowmelt (Suriano & Leathers 2017), successional maturation of eastern forests (Moser et al. 2020), nonnative insects and pathogens (Lovett et al. 2016), and a growing human population 117 (Moser et al. 2020). Further, modeling future growth i s sensitive to the GCM selected for obtaining climate projections. By selecting two GCMs, one which is unusually mild in its warming projection and another which is unusually extreme, the present study attempts to model the full range of potential growth c hange in the absence of the effects of other changing variables. Finally, CO 2 fertilization is a factor which could mitigate the negative consequences of summer moisture stress. As temperature increases or as precipitation decreases, trees must either clo se their stomata and cut off their CO 2 supply or leave their stomata open at the cost of heavy water loss. However, as atmospheric CO 2 concentrations continue to rise , the cost of keeping stomata closed more often could be offset. As discussed in Chapter 1, some evidence of growth enhancement due to CO 2 fertilization has been found (Lamarche et al. 1984, Bazzaz et al. 1990, Telewski et al. 1999, Wang et al. 2006, Cole et al. 2010, McMahon et al. 2010, Walker et al. 2019 ) . In other cases, either no enhancement or only a temporary enhancement due to CO 2 fertilization was found (Salzer et al. 2009, Norby et al. 2010, Van der Sleen et al . 2015, Girardin et al. 2016). In a projection of future growth across North America. Charney et al. (2016) found that under RCP 8.5 it would take a 72% water - use efficiency increase, due to CO 2 fertilization, to offset growth declines. Despite the variabl es for which the present models did not account , the tree - ring based projections of the present study and other studies from the region (Chhin 2015, 2016; Charney et al. 2016) are consistent with projected range retraction (Walker et al. 2002, Prasad et al . 2020). Together, projections of reduced growth and range retraction suggest that land managers need to consider dramatic management strategies such as assisted colonization (Albrecht et al. 2012) and assisted gene flow (Aitken & Whitlock 2013). 118 To identi systematic dendrochronological sampling is recommended for the region, followed by growth - climate modeling and projection of future growth under the climate of various locat ions (Chen et al. 2010). Conclusion The relationship between growth and climate was modeled for each of 46 tree populations from nine species across Indiana and Michigan. The models were calibrated on one half of available data. Verification was attempt ed on the other half of the data by comparing model - predicted growth with observations. Verification was successful for 14/46 populations. Growth of these 14 populations was forecast under ongoing climate change for the rest of this century according to pr ojections for four climate - change scenarios. Growth of one population was projected to increase by 5.18 12.1% depending on the scenario. Growth of the other 13 populations was generally projected to decline. Under the mildest scenario, only 4 of these 1 3 would decline significantly. However, under the second mildest 9 would decline significantly. Under the two scenarios with the most warming, declines would be significant in 12 populations, ranging from 4.79% in one A. saccharum population to 34.2% in a different population of that species. With large and widespread growth declines under all but the mildest warming scenario considered, significant management strategies should be considered. 119 APPENDIX 120 APPENDIX Figures Figure 4.1 . Comparison of historic and projected June August mean temperature (a) and June July total precipitation (b) . The number after each three - letter site code represents decimal degrees north latitude (Table 3.1 ). Climate projections are from global circulation models (GCMs) MRI - CGCM3 (Yukimoto et al. 2012) and GFDL - CM3 (Griffies et al. 2011) under representative co ncentration pathways (RCPs) 4.5 and 8.5. 121 122 Figure 4.2 . Mean forecasted ring - width indices (RWI) over years 2022 - 2099 under two representative concentration pathways (RCPs 4.5 and 8.5), under each of two GCMs, MRI - CGCM3 (Yukimoto et al. 2012) and GFDL - CM3 (Griffies et al. 2011 ), relative to simulated RWI over common interval 1903 - 2004 . Historic and future RWI were simulated according to models fit over the 1903 - 1953 calibration period (Table 1) . Note that RWI of all populations was standardized to a mean of 1 in the pre - modeling detrending process. Error bars represent 95% confidence intervals about the mean. Tree populations are arranged from north (left) to south (right); site abbreviations ar e in Table 3. 1. Within each population, scenarios sharing a letter did not significantly differ in an ANOVA test (p > 0.05). Species codes: As = sugar maple Acer saccharum , Co = shagbark hickory Carya ovata , Fg = American beech Fagus grandifolia , Lt = tuli p poplar Liriodendron tulipifera , Qa = white oak Quercus alba , Qr = red oak Quercus rubra , Tc = eastern hemlock Tsuga canadensis . a a a a a a a a a a a a a a b a a a b a,b a,b a b b a b a,b a,b b a a a c b b,c a,b c c b b b b,c c b b b d c c b,c d d b,c c a,b c,d d b b b d d d c e e c c a,b d 60% 70% 80% 90% 100% 110% 120% Mwd As Pet Tc MaB Tc JSB Tc PNC Qa War Fg War Qa Vrh As Vrh Qa Vrh Qr KNC Fg MGB Co PMF Lt PMF Qr RWI Relative to Historic Simulations Simulated Historic MRI-CGCM3, RCP 4.5 MRI-CGCM3, RCP 8.5 GFDL-CM3, RCP 4.5 GFDL-CM3, RCP 8.5 123 Tables Table 4.1 . Calibration and verification statistics and parameters for models relating climate and ring - width index . Data were split into calibration (1903 - 1953) and verification periods (1954 - 2004). Forward stepwise multiple regression established the three most explanatory variables over the calibration period. Models were then applied to the verifica tion period and judged sequentially according to three criteria: p - value, reduction of error (RE), and coefficient of efficiency (CE). If a model failed any criterion it was then discarded and not judged by subsequent criteria. Parameters are listed in the order in which they entered the model. Climate variables are abbreviated with numbers corresponding to months (month 1 = Jan., 12 = Dec.), letters preceding months indicate whether the month occurred in the year prior to ring e minimum, and maximum temperature C), each averaged over the month. RE and CE are explai ned in the Methods section above. Sites are arranged north to south (Table 3.1). Species codes: As = sugar maple Acer saccharum , Ba = yellow birch Betula alleghaniensis , Co = shagbark hickory Carya ovata , Fg = American beech Fagus grandifolia , Lt = tulip p oplar Liriodendron tulipifera , Ps = white pine Pinus strobus , Qa = white oak Quercus alba , Qr = red oak Quercus rubra , Tc = eastern hemlock Tsuga canadensis. Population Model Parameters (Climate Variables) Adj. r 2 , 1903 - 1953 p - value, 1903 - 1953 Adj. r 2 , 19 54 - 2004 p - value, 1954 - 2004 RE CE HMC As c7Ppt + p8Tmin + c3Ppt 0.389 8.12E - 06 5.12E - 04 0.397 HMC Ba c6Ppt + c7Ppt + p6Ppt 0.326 7.75E - 05 0.0262 0.241 HMC Ps c2Tmax + p7Tmin + p8Tmin 0.325 7.80E - 05 - 0.0263 0.635 HMC Qr c3Ppt + p11Tmin + p6Tmax 0.236 1.29E - 03 - 0.0449 0.0177 HMC Tc p9Tmax + c4Tmax + p10Tmin 0.264 5.68E - 04 0.0731 0.0879 Mwd As c7Ppt + p11Ppt + p8Tmin 0.256 7.22E - 04 0.125 0.0260 0.0478 0.0200 Mwd Ba c7Ppt + p8Ppt + c6Tmin 0.388 8.52E - 06 0.128 0.0241 - 0.174 124 Table Mwd Ps c7Ppt + p9Ppt + p8Ppt 0.394 6.65E - 06 0.0402 0.180 Mwd Qr c7Ppt + p8Ppt + p9Ppt 0.380 1.15E - 05 .0305 0.220 Pet Tc p7Tmax + c6Ppt + c3Tmean 0.358 2.57E - 05 0.316 1.06E - 04 0.156 0.156 MAB As c7Ppt + c6Ppt + p10Ppt 0.432 1.51E - 06 0.140 0.0177 - 0.438 MaB Ba c7Ppt + c6Tmin + p8Ppt 0.393 7.01E - 06 0.158 0.0113 - 0.281 MaB Tc p8Tmean + c3Tmin + p7Ppt 0.360 2.37E - 05 0.325 7.92E - 05 0.224 0.211 CoP Fg c7Ppt + c2Ppt + p10Ppt 0.355 2.79E - 05 0.118 0.031 - 0.494 CoP Ps c7Tmax + p9Ppt + c5Tmin 0.353 2.98E - 05 0.0550 0.131 JSB As c7Tmax + p12Tmax + c4Ppt 0.459 5.04E - 07 0.0701 0.0942 JSB Fg c2Ppt + p12Tmax + c9Tmean 0.409 3.76E - 06 - 0.0430 0.816 JSB Tc p11Ppt + p7Tmin + p12Tmax 0.170 8.09E - 03 0.118 0.0309 0.0891 0.0879 PNC As c7Ppt + c2Ppt + p8Tmax 0.432 1.55E - 06 - 0.0131 0.508 PNC Fg c7Ppt + p7Ppt + c6Tmin 0.370 1.67E - 05 0.0369 0.193 PNC Qa c7Ppt + p7Ppt + p11Tmin 0.446 8.59E - 07 0.148 0.0144 0.146 0.136 PNC Qr c7Ppt + c9Tmin + p10Tmin 0.378 1.21E - 05 - 0.0261 0.633 War As p12Tmax + c7Tmax + c5Tmean 0.444 9.55E - 07 0.0971 0.0505 War Fg c8Tmax + c5Tmin + c6Tmax 0.460 4.78E - 07 0.294 2.17E - 04 0.0904 0.0890 War Lt c6Tmax + p10Ppt + c4Tmin 0.249 8.74E - 04 0.0960 0.0519 War Qr c6Tmax + p12Tmean + c8Tmax 0.408 4.31E - 06 0.101 0.0457 - 0.0821 125 Table 4.1 Vrh As c7Tmax + p12Tmean + c5Tmean 0.442 1.02E - 06 0.188 4.98E - 03 0.122 0.121 Vrh Co p7Tmax + c7Tmax + c1Tmax 0.388 8.62E - 06 - 0.0396 0.779 Vrh Qa c7Tmax + p12Tmin + c6Tmax 0.458 5.25E - 07 0.206 3.02E - 03 0.111 0.108 Vrh Qr c6Tmax + p12Tmin + c7Tmax 0.397 6.02E - 06 0.268 5.01E - 04 0.110 0.0975 KNC As c7Tmax + p9Ppt + c8Ppt 0.453 6.53E - 07 .0687 0.0972 KNC Fg c7Tmax + p5Tmax + c6Ppt 0.521 3.06E - 08 0.343 4.27E - 05 0.0643 0.0622 KNC Lt c6Tmax + c7Ppt + c8Ppt 0.481 1.90E - 07 0.219 2.13E - 03 - 0.0551 KNC Qr c6Tmax + p7Tmax + c8Ppt 0.420 2.50E - 06 0.194 4.31E - 03 - 0.0890 MGB As c6Tmax + c7Ppt + p5Ppt 0.523 2.75E - 08 0.127 0.0246 - 0.436 MGB Co c6Tmax + c3Ppt + p10Tmin 0.516 3.78E - 08 0.267 5.16E - 04 0.191 0.190 MGB Fg c6Tmax + c1Tmin + c4Tmin 0.572 2.27E - 09 0.178 6.57E - 03 - 0.257 MGB Lt c7Tmax + c8Tmax + c2Ppt 0.481 1.93E - 07 .0408 0.180 MGB Qr c6Tmax + c7Ppt + p12Tmax 0.592 7.45E - 10 0.275 3.96E - 04 - 0.0332 PMF As c6Tmax + p8Tmax + p9Tmax 0.571 2.32E - 09 0.107 0.0403 - 0.643 PMF Co c6Tmax + c8Tmean + c7Ppt 0.590 8.13E - 10 7.09E - 04 0.396 PMF Fg c6Tmax + c6Ppt + c7Ppt 0.536 1.46E - 08 0.122 0.0279 - 0.327 PMF Lt c6Tmax + c6Ppt + p12Ppt 0.605 3.38E - 10 0.262 6.00E - 04 0.134 0.128 PMF Qa c6Tmax + c7Ppt + c1Ppt 0.581 1.38E - 09 0.242 1.08E - 03 - 0.130 PMF Qr c6Tmax + p6Tmax + c7Ppt 0.590 8.22E - 10 0.277 3.80E - 04 0.134 0.0951 126 Table 4.2 . Model p arameters and their coefficients when models established over the . Variables are listed in the order in which they entered the calibration - period model according to forward stepwise multiple regression. Climate variables are abbreviated with numbers corresponding to months (month 1 = Jan . , 12 = Dec.), letters preceding months indicate whether the month month. Sites are arranged north (top) to south (Table 3.1). Species codes: As = sugar maple Acer saccharum , Co = shag bark hickory Carya ovata , Fg = American beech Fagus grandifolia , Lt = tulip poplar Liriodendron tulipifera , Qa = white oak Quercus alba , Qr = red oak Quercus rubra , Tc = eastern hemlock Tsuga canadensis . Population Model Parameters (Climate Variables) Cons tant Parameter 1 Coefficient Parameter 2 Coefficient Parameter 3 Coefficient Adj. r 2 Mwd As c7Ppt + p11Ppt + p8Tmin 1.43 8.26E - 4 8.77E - 4 - 0.0466 0.147 Pet Tc p7Tmax + c6Ppt + c3Tmean 2.18 - 0.0462 7.75E - 4 0.0277 0.279 MaB Tc p8Tmean + c3Tmin + p7Ppt 2.06 - 0.0473 0.0315 7.53E - 4 0.324 JSB Tc p11Ppt + p7Tmin + p12Tmax 1.52 1.49E - 3 - 0.0496 0.0235 0.159 PNC Qa c7Ppt + p7Ppt + p11Tmin 0.816 1.83E - 3 9.10E - 4 0.0211 0.307 War Fg c8Tmax + c5Tmin + c6Tmax 2.23 - 0.0211 0.0270 - 0.0324 0.338 War Qa c6Tmax + p5Tmax + c7Ppt 1.72 - 0.0402 0.0136 6.87E - 4 0.359 Vrh As c7Tmax + p12Tmean + c5Tmean 2.28 - 0.0571 0.0239 0.0272 0.300 Vrh Qa c7Tmax + p12Tmin + c6Tmax 2.86 - 0.0338 0.0168 - 0.0309 0.336 Vrh Qr c6Tmax + p12Tmin + c7Tmax 2.45 - 0.0312 9.16E - 3 - 0.0210 0.268 KNC Fg c7Tmax + p5Tmax + c6Ppt 1.99 - 0.0508 0.0150 1.86E - 3 0.415 MGB Co c6Tmax + c3Ppt + p10Tmin 2.42 - 0.0569 5.53E - 4 0.0184 0.384 PMF Lt c6Tmax + c6Ppt + p12Ppt 1.73 - 0.0359 2.29E - 3 6.93E - 4 0.400 PMF Qr c6Tmax + p6Tmax + c7Ppt 1.77 - 0.0401 0.0110 6.93E - 4 0.483 127 CHAPTER FIVE CONCLUDING REMARKS In this dissertation, I used tree rings in Indiana and Michigan to better understand the climate of the era preceding instrumentally derived weather records, to compare relationships between growth and climate within and among tree species, and to project future tree growth under ongoing climate change. It was found that even on humid South M anitou I sland , Lake Michigan, in the humid east ern United States, severe drought was a common occurrence. Colleagues and I reconstructed 469 years of the moisture index Palmer Z Index over late - summer and found that 17% of years were at least one standard deviation (SD) drier than the mean; 2% were at least two SD s drier (Fig. 2.9b) . Intense pluvials, which can have their own negative effects such as erosion and flooding, were also common , with 16% of years at least one SD wetter than the mean and 1% at lea st two SD s wetter. Extreme years occurred throughout the interval, but they were most common in the 20 th century , with 22% of its years being one SD drier and 26% wetter than the mean. Further, all four of the driest years were in that century. Thus, climate change has already resulted in a more variable environment in northern Lake Michigan relative to historic conditions and led to droughts unmatched by anything from the preceding four centuries. Climate change is expected to bring warmer temperature s to the region, more erraticism in rainfall, and more variability in general. With many trees being ecological foundation species and with society relying on the products and services they provide, it is 128 important to understand how climate affects tree gr owth and whether this information can be valuable in projecting future growth. A 12 - site latitudinal gradient was established spanning southern Indiana to the Upper Peninsula of Michigan. Tree - core samples from 46 populations were taken at these sites fr om a mix of nine species. Relationships between ring widths and temperature/precipitation were quantified and compared within and among species. The dominant factors associated with tree growth were conditions in the summer , with negative growth - temperatur e and positive - precipitation relationships ( Fig. 3.4 ). The strength of this relationship linearly decreased moving to the north. This was found across all species except Tsuga canadensis , however that species was still limited by summer conditions, but i t was the conditions of the summer preceding the year of growth, unlike the other species ( Tables 3.4 & 3.5 ). It has been found in some tree - ring studies that the strength of growth - climate relationships has weakened in recent decades, rendering climate reconstructions and future - growth projections less reliable. I tested whether this had occurred in my study gradient by analyzing the strength of monthly growth - temperature and - precipitation relationships in overlapping 34 - year windows. It was found that some growth - climate relationships did evolve but only in the months which were not significantly associated with growth ( Fig. 3.6 a & b). For the most influential months, the relationships were stable ( Fig. 3.6 c & d ). Thus, tree rings are still recommended for reconstructing climate and projecting future growth in the Great Lakes Region . In this study, g rowth was also commonly associated with spring temperatures (positively), winter temperature (positively), and winter precipitation (positively ). Each of these 129 variables is expected to increase as climate change proceeds. Thus, even as climate change exacerbates summer moisture stress, which is so detrimental to tree growth, it may bring some benefit to tree growth through warmer, wetter winters and springs. To project future growth, the same network of 46 tree - ring chronologies was used to establish parsimonious models between growth and climate. Forward stepwise multiple regression was used to select the most variability - explaining set of thr ee variables over one half of the dataset. It was then attempted to verify the calibrated model on the other half of the dataset. This process tended to be unsuccessful (14 successes/46 attempts). This is likely due to the remote location of some of my sit es, away from weather stations, and due to the relatively benign climate of the temperate east which leads to complacent tree growth. Nonetheless, future growth could be projected for the 14 successfully validated populations. This was done for the period 2022 2099 under two climate - change models, one which predicted relatively mild warming and one which predicted severe warming. Under the severe - warming model, growth was projected to decline significantly in 12/14 populations and to increase in one populat ion. Under the mild scenario, growth was projected to decline in four populations, to increase in one, and to have no significant change in nine. This suggests that land managers need to prepare for severe imminent tree growth declines. In future researc h, it is recommended that tree cores continue to be sampled from the major species in the Great Lakes Region. Researchers should carefully select sites that are nearby weather stations with long and consistent records. Using methods similar to those in thi s dissertation, models should be constructed that relate growth and climate, and these should be used to project future growth. When potentially climate - change - resilient populations 130 are identified, such as the Quercus alba population studied here at Price Nature Center, experimental warming under controlled conditions should be applied to seedlings from the site to test whether the growth - climate relationships observed were due to genetics or microsite. Populations for which climate - change resiliency is sup ported both through tree - ring - climate analysis and experimental warming should be considered candidates for use in assisted gene flow. Another avenue of future research is sampling ancient trees, stumps, and wood in human - made structures to reconstruct climate. This will paint a clearer picture of what climate was l i ke in the era before the keeping of instrumentally derived records. This is important to do soon because such wood is continuously lost to decay and fire. Thank you for reading this disser tation. Now go out and hug a tree! 131 LITERATURE CITED 132 LITERATURE CITED Aitken, S.N., Whitlock, M.C., 2013. Assisted Gene Flow to Facilitate Local Adaptation to Climate Change. Annual Review of Ecology, Evolution, and Systematics 44, 367 388. Akaike, H., 1974. A new look at the statistical model identification. IEEE Transactions on Automatic Control 19, 716 723. Albrecht, G.A., Brooke, C., Bennett, D.H., Garnett, S.T., 2013. The Ethi cs of Assisted Colonization in the Age of Anthropogenic Climate Change. Journal of Agricultural and Environmental Ethics 26, 827 845. Andersen, B.J., 2005. The Historical Development of the Tension Zone Concept in the Great Lakes Region of North America. The Michigan Botanist 44, 127 138. Anderson, D.G., Stahle, D.W., Cleaveland, M.K., 1995. Paleoclimate and the Potential Food Reserves of Mississippian Societies: A Case Study from the Savannah River Valley. American Antiquity 60, 258 286. Andresen, J.A ., 2012. Historical climate trends in Michigan and the Great Lakes region, in: Climate Change in the Great Lakes Region. Michigan State University Press, East Lansing, Michigan, USA 17 34. Andresen, J.A., Winkler, J.A, 2009. Weather and Climate, in: Michi gan Geography and Geology. Pearson Custom Publishing, Boston, Massachusetts, USA 288 311. Au, T.F., Maxwell, J.T., Novick, K.A., Robeson, S.M., Warner, S.M., Lockwood, B.R., Phillips, R.P., Harley, G.L., Telewski, F.W., Therrell, M.D., Pederson, N., 2020. Demographic shifts in eastern US forests increase the impact of late - season drought on forest growth. Ecography 43, 1475 1486. Babst, F., Poulter, B., Trouet, V., Tan, K., Neuwirth, B., Wilson, R., Carrer, M., Grabner, M., Tegel, W., Levanic, T., Panayo tov, M., Urbinati, C., Bouriaud, O., Ciais, P., Frank, D., 2013. Site - and species - specific responses of forest growth to climate across the European continent. Global Ecology and Biogeography 22, 706 717. Barber, V.A., Juday, G.P., Finney, B.P., 2000. R educed growth of Alaskan white spruce in the twentieth century from temperature - induced drought stress. Nature 405, 668 673. Bazzaz, F.A., Coleman, J.S., Morse, S.R., 1990. Growth responses of seven major co - occurring tree species of the northeastern Uni ted States to elevated CO 2 . Canadian Journal of Forest Research 20, 1479 1484. 133 Biondi, F., Waikul, K., 2004. DENDROCLIM2002: A C++ program for statistical calibration of climate signals in tree - ring chronologies. Computers & Geosciences 30, 303 311. Bo isvert - Marsh, L., Périé, C., de Blois, S., 2014. Shifting with climate? Evidence for recent changes in tree species distribution at high latitudes. Ecosphere 5, 1 33. - growth analysis for a Mexican dry forest tree shows strong impact of sea surface temperatures and predicts future growth declines. Global Change Biology 16, 2001 2012. Briffa, K.R., Schweingruber, F.H., Jones, P.D., Osborn, T.J., Shiyatov, S.G., Vaganov, E.A., 1998. Reduced sensitivity of recent tree - growth to temperature at high northern latitudes. Nature 391, 678 682. Brubaker, L.B., 1980. Spatial Patterns of Tree Gr owth Anomalies in the Pacific Northwest. Ecology 61, 798 807. Bruening, J.M., Tran, T.J., Bunn, A.G., Weiss, S.B., Salzer, M.W., 2017. Fine - scale modeling of bristlecone pine treeline position in the Great Basin, USA. Environmental Research Letters 12, 0 14008. Brusca, R.C., Wiens, J.F., Meyer, W.M., Eble, J., Franklin, K., Overpeck, J.T., Moore, W., 2013. Mountain plant transect revisited. Ecology and Evolution 3, 3307 3319. Buckley, B.M., Wilson, R.J.S., Kelly, P.E., Larson, D.W., Cook, E.R., 2004. Inferred summer precipitation for southern Ontario back to AD 610, as reconstructed from ring widths of Thuja occidentalis 34, 2541 2553. Büntgen, U., Frank, D., Wilson, R., Carrer, M., Urbinati, C., Esper, J., 2008. Testing for tree - ring divergence in the European Alps. Global Change Biology 14, 2443 2453. Buras, A., 2017. A comment on the expressed population signal. Dendrochronologia 44, 130 132. Burnett, A.W., Kir by, M.E., Mullins, H.T., Patterson, W.P., 2003. Increasing Great Lake Effect Snowfall during the Twentieth Century: A Regional Response to Global Warming? Journal of Climate 16, 3535 3542. Burns, B.T., 1983. Simulated Anasazi Storage Behavior Using Crop Yields Reconstructed from Tree Rings, A.D. 652 - 1968. Dissertation, University of Arizona, Tucson, Arizona, USA. 134 Byun, K., Hamlet, A.F., 2018. Projected changes in future climate over the Midwest and Great Lakes region using downscaled CMIP5 ensembles. I nternational Journal of Climatology 38, e531 e553. Campbell, J.L., Socci, A.M., Templer, P.H., 2014. Increased nitrogen leaching following soil freezing is due to decreased root uptake in a northern hardwood forest. Global Change Biology 20, 2663 2673. Carroll, A.L., Sillett, S.C., Kramer, R.D., 2014. Millennium - Scale Crossdating and Inter - Annual Climate Sensitivities of Standing California Redwoods. PLoS ONE 9, e102545. Cedro, A., Cedro, B., 2015. Growth - climate relationships at yew and wild service t rees on the eastern edge of their range in Europe. Forest Systems 24, e044. Charney, N.D., Babst, F., Poulter, B., Record, S., Trouet, V.M., Frank, D., Enquist, B.J., Evans, M.E.K., 2016. Observed forest sensitivity to climate implies large changes in 21 st century North American forest growth. Ecology Letters 19, 1119 1128. Cheddadi, R., Araújo, M.B., Maiorano, L., Edwards, M., Guisan, A., Carré, M., Chevalier, M., Pearman, P.B., 2016. Temperature Range Shifts for Three European Tree Species over the La st 10,000 Years. Frontiers in Plant Science. 7, 1581. Chen, I. - C., Hill, J.K., Ohlemüller, R., Roy, D.B., Thomas, C.D., 2011. Rapid Range Shifts of Species Associated with High Levels of Climate Warming. Science 333, 1024 1026. Chen, P. - Y., Welsh, C., H amann, A., 2010. Geographic variation in growth response of Douglas - fir to interannual climate variability and projected climate change. Global Change Biology 16, 3374 3385. Chhin, S., 2016. Screening the Resilience of Short - Rotation Woody Crops to Clima te Change. Geosciences 6, 7. Chhin, S., 2015. Impact of future climate change on a genetic plantation of hybrid pine Botany 93, 397 404. Chhin, S., Hogg, E.H. (Ted), Lieffers, V.J., Huang, S., 2008. Potential effects of climate change on the growth of l odgepole pine across diameter size classes and ecological regions. Forest Ecology and Management 256, 1692 1703. 135 Christensen, J.H., Kanikicharla, K.K., Aldrian, E., An, S. - I., de Castro, M., Dong, W., Goswami, P., Hall, A., Kanyanga, J.K., Kitoh, A. , Kossin, J., Lau, N. - C., Renwick, J., Zhou, T., Abraham, L., Ambrizzi, T., Anderson, B., Arakawa, O., Arritt, R., Baldwin, M., Barlow, M., Barriopedro, D., Biasutti, M., Biner, S., Bromwich, D., Brown, J., Cai, W., Carvalho, L.V., Chang, P., Choi, J., Ema nuel, K., Endo, H., Enfield, D.B., Evan, A., Giannini, A., Gillett, N., Hariharasubramanian, A., Huang, P., Jones, J., Karumuri, A., Katzfey, J., Kjellström, E., Knight, J., Knutson, T., Kulkarni, A., Kundeti, K.R., Lau, W.K., Lenderink, G., Mackellar, N.C ., Meehl, G., Menéndez, C., Murakami, H., Nath, M.J., Neelin, J.D., van Oldenborgh, J., Solman, S., Stendel, M., Takayabu, I., Turner, J., Ummenhofer, C., Wa ng, B., Wang, C., Watterson, I., Wittenberg, A., Woollings, T., Yeh, S. - W., Zhang, C., Zhang, L., Zheng, X., Zou, L., Fyfe, J., Kwon, W. - T., 2013. Climate Phenomena and their Relevance for Future Regional Climate Change, in: Climate Change 2013: the Physic al Science Basis: Working Group I Contribution to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA 1217 1308. Cohen, J.G., Kost, M.A., Slaughter, B.S., Albert, D.A., 2014. A Field Guide to the Natural Communities of Michigan. Michigan State University Press, East Lansing, Michigan, USA, 196 199. Cole, C.T., Anderson, J.E., Lindroth, R.L., Waller, D.M., 2010. Rising concentrations of atmo spheric CO2 have increased growth in natural stands of quaking aspen ( Populus tremuloides ). Global Change Biology 16, 2186 2197. Collins, M., Knutti, R., Arblaster, J., Dufresne, J. - L., Fichefet, T., Gao, X., Jr, W.J.G., Johns, T., Krinner, G., Shongwe, M., Weaver, A.J., Wehner, M., Allen, M.R., Andrews, T., Beyerle, U., Bitz, C.M., Bony, S., Booth, B.B.B., Brooks, H.E., Brovkin, V., Browne, O., Brutel - Vuilmet, C., Cane, M., Chadwick, R., Cook, E., Cook, K.H., Eby, M., Fasullo, J., Forest, C.E., Forster, P., Good, P., Goosse, H., Gregory, J.M., Hegerl, G.C., Hezel, P.J., Hodges, K.I., Holland, M.M., Huber, M., Joshi, M., Kharin, V., Kushnir, Y., Lawrence, D.M., Lee, R.W., Liddicoat, S., Lucas, C., Lucht, W., Marotzke, J., Massonnet, F., Matthews, H.D., Mei nshausen, M., Morice, C., Otto, A., Patricola, C.M., Philippon, G., Rahmstorf, S., Riley, B., Stott, P.A., Webb, R., Zappa, G., Zickfeld, K., Joussaume, S., Mo kssit, A., Taylor, K., Tett, S., 2013. Long - term Climate Change: Projections, Commitments and Irreversibility, in: Climate Change 2013: the Physical Science Basis: Working Group I Contribution to the Fifth Assessment Report of the Intergovernmental Panel o n Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 1029 1136. Cook, E.R., 1985. A Time Series Analysis Approach to Tree Ring Standardization. Dissertation, University of Arizona, Tucson, Arizona, USA. 136 Cook, E. R., Cole, J., 1991. On predicting the response of forests in eastern North America to future climatic change. Climatic Change 19, 271 282. Cook, E.R., Glitzenstein, J.S., Krusic, P.J., Harcombe, P.A., 2001. Identifying Functional Groups of Trees in West Gulf Coast Forests (USA): A Tree - Ring Approach. Ecological Applications 11, 883 903. Cook, E.R., Meko, D.M., Stahle, D.W., Cleaveland, M.K., 1999. Drought Reconstructions for the Continental United States. Journal of Climate 12, 1145 1162. Crimmins, S. M., Dobrowski, S.Z., Greenberg, J.A., Abatzoglou, J.T., Mynsberge, A.R., 2011. Elevations. Science 331, 324 327. Crouchet, S.E., Jensen, J., Schwartz, B.F . Schwinning, S., 2019. Tree mortality after a hot drought: distinguishing density - dependent and - independent drivers and why it matters. Frontiers in Forests and Global Change 2, 21. Daly, C., Halbleib, M., Smith, J.I., Gibson, W.P., Doggett, M.K., Taylor, G.H., Curtis, J., Pasteris, P.P., 2008. Physiographically sensitive mapping of climatological temperature and precipitation across the conterminous United States. International Journal of Climatology 28, 2031 2064. Northern Forests: A review of the tree - ring evidence and possible causes. Global and Planetary Change 60, 289 305. , R.D., Kaufmann, R.K., Davi, N., Jacoby, G.C., Laskowski, C., Myneni, R.B., Cherubini, P., 2004. Thresholds for warming - induced growth decline at elevational tree line in the Yukon Territory, Canada. Global Biogeochemical Cycles 18. Darwin, C., 1859. On the Origin of the Species by Means of Natural Selection, John Murray, London, United Kingdom. Davis, M.B., 1983. Quaternary History of Deciduous Forests of Eastern North America and Europe. Annals of the Missouri Botanical Garden 70, 550 563. Davis, M. B., Shaw, R.G., 2001. Range Shifts and Adaptive Responses to Quaternary Climate Change. Science 292, 673 679. Dickmann, D.I., Leefers, L.A., 2016. The Forests of Michigan. University of Michigan Press, Ann Arbor, Michigan, USA, 1 30. 137 Duchesne, L., Houle, D., Kneeshaw, D., Côté, B., Pederson, N., 2016. Northeastern North America as a potential refugium for boreal forests in a warming climate. Science 352, 1452 1455. , L., Logan, T., Houle, D., Arseneault, D., Beier, C.M., Bishop, D.A., Druckenbrod, D., Fraver, S., Girard, F., Halman, J., Hansen, C., Hart, J.L., Hartmann, H., Kaye, M., Leblanc, D., Manzoni, S., Ouimet, R., Rayback, S., Rollinson, C.R., Phillips, R.P., 2018. Drought timing and local climate determine the sensitivity of eastern temperate forests to drought. Global Change Biology 24, 2339 2351. Douglass, A.E., 1935. Dating Pueblo Bonito and Other Ruins of the Southwest, National Geographic Society, Contr ibuted Technical Papers, Pueblo Bonito Series 1, 1 74. Washington, D.C., USA. Dulamsuren, C., Hauck, M., Kopp, G., Ruff, M., Leuschner, C., 2017. European beech responds to climate change with growth decline at lower, and growth increase at higher elevati ons in the center of its distribution range (SW Germany). Trees 31, 673 686. Dye, A., 2019. Growth and climate response of four new Tsuga canadensis (L.) CARRIÈRE (eastern hemlock) tree - Lakes B otanist 58, 193 204. Ford, T.W., 2014. Precipitation anomalies in Eastern - Central Iowa from 1640 Present. Journal of Hydrology 519, 918 924. Fritts, H.C, 1976. Tree Rings and Climate, Academic Press, London, UK, New York, New York, USA, and San Francisc o, California, USA. Fritts, H.C., 1965. Tree - ring evidence for climatic changes in western North America. Monthly Weather Review 93, 421 443. Gamache, I., Payette, S., 2004. Height growth response of tree line black spruce to recent climate warming across the forest - tundra of eastern Canada. Journal of Ecology 92, 835 845. Gao, L., Gou, X., Deng, Y., Liu, W., Yang, M., Zhao, Z., 2013. Climate growth analysis of Qilian juniper across an altitudinal gradient in the central Qilian Mountains, northwest China. Trees 27, 379 388 . Gaul, D., Hertel, D., Borken, W., Matzner, E., Leuschner, C., 2008. Effects of experimental drought on the fine root system of mature Norway spruce. Forest Ecology and Management 256, 1151 1159. 138 Girardin, M.P., Bouriaud, O., Hogg, E.H., Kurz, W., Zimmermann, N.E., Metsaranta, J.M., de Jong, R., Frank, D.C., Esper, J., Büntgen, U., Guo, X.J., Bhatti, J., 2016. No growth - century of combined war ming and CO 2 fertilization. Proceedings of the National Academy of Sciences USA 113, E8406 E8414. Granda, E., Camarero, J.J., Gimeno, T.E., Martínez - Fernández, J., Valladares, F., 2013. Intensity and timing of warming and drought differentially affect g rowth patterns of co - occurring Mediterranean tree species. European Journal of Forest Research 132, 469 480. Graumlich, L.J., 1993. Response of tree growth to climatic variation in the mixed conifer and deciduous forests of the upper Great Lakes region. Canadian Journal of Forest Research 23, 133 143. Griffies, S.M., Winton, M., Donner, L.J., Horowitz, L.W., Downes, S.M., Farneti, R., Gnanadesikan, A., Hurlin, W.J., Lee, H. - C., Liang, Z., Palter, J.B., Samuels, B.L., Wittenberg, A.T., Wyman, B.L., Yin, J., Zadeh, N., 2011. The GFDL CM3 Coupled Climate Model: Characteristics of the Ocean and Sea Ice Simulations. Journal of Climate 24, 3520 3544. Grissino - Mayer, H.D., 2001. Evaluating crossdating accuracy: a manual and tutorial for the computer program COFECHA. Tree - Ring Research, 57, 205 221. Grissino - Mayer, H.D., 1996. A 2129 - year reconstruction of precipitation for northwestern New Mexico, USA, in: Tree rings, Environment, and Humanity, Radioc arbon. Department of Geosciences , University of Arizona, Tucson, Arizona, USA, 191 204. Harsch, M.A., Hulme, P.E., McGlone, M.S., Duncan, R.P., 2009. Are treelines advancing? A global meta - analysis of treeline response to climate warming. Ecology Letters 12, 1040 1049. Hartmann, D.L., Tank, A.M.G.K., Rusticucci, M., Alexander, L.V., Brönnimann, S., Charabi, Y.A.R., Dentener, F.J., Dlugokencky, E.J., Easterling, D.R., Kaplan, A., Soden, B.J., Thorne, P.W., Wild, M., Zhai, P., 2013. Observations: Atmospher e and surface, in: Climate Change 2013: the Physical Science Basis: Working Group I Contribution to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA 15 9 254. ssen, M. van der, Vitas, A., Weigel, R., Kreyling, J., Wilmking, M., 2020. Tree growth influenced by warming winter climate and summer moisture availability in northern temperate forests. Global Change Biology 26, 2505 2518. 139 Hayhoe, K., VanDorn, J., Cro ley, T., Schlegal, N., Wuebbles, D., 2010. Regional climate change projections for Chicago and the US Great Lakes. Journal of Great Lakes Research 36, 7 21. Heberling, J.M., MacKenzie, C.M., Fridley, J.D., Kalisz, S., Primack, R.B., 2019. Phenological mi smatch with trees reduces wildflower carbon budgets. Ecology Letters 22, 616 623. Heim, R.R., 2002. A Review of Twentieth - Century Drought Indices Used in the United States. Bulletin of the American Meteorological Society 83, 1149 1166. Hinckley, T.M., Dougherty, P.M., Lassoie, J.P., Roberts, J.E., Teskey, R.O., 1979. A Severe Drought: Impact on Tree Growth, Phenology, Net Photosynthetic Rate and Water Relations. The American Midland Naturalist 102, 307 316. Holmes, R.L., 1983. Computer - Assisted Qualit y Control in Tree - Ring Dating and Measurement. Tree - Ring Bulletin 43, 69 78. Housset, J.M., Girardin, M.P., Baconnet, M., Carcaillet, C., Bergeron, Y., 2015. Unexpected warming - induced growth decline in Thuja occidentalis at its northern limits in North A merica. Journal of Biogeography 42, 1233 1245. Huang, J., Tardif, J.C., Bergeron, Y., Denneler, B., Berninger, F., Girardin, M.P., 2010. Radial growth response of four dominant boreal tree species to climate along a latitudinal gradient in the eastern Ca nadian boreal forest. Global Change Biology 16, 711 731. Huang, J. - G., Bergeron, Y., Berninger, F., Zhai, L., Tardif, J.C., Denneler, B., 2013. Impact of Future Climate on Radial Growth of Four Major Boreal Tree Species in the Eastern Canadian Boreal For est. PLOS ONE 8, e56758. 2012. Ecological impacts of a widespread frost event following early spring leaf - out. Global Change Biology 18, 2365 2377. Huo, Y., Gou, X., Liu, W., Li, J., Zhang, F., Fang, K., 2017. Climate growth relationships of Schrenk spruce (Picea schrenkiana) along an altitudinal gradient in the western Tianshan mountains, northwest China. Trees 31, 429 439. Hutchings, M.J., Robbirt, K.M ., Roberts, D.L., Davy, A.J., 2018. Vulnerability of a specialized pollination mechanism to climate change revealed by a 356 - year analysis. Botanical Journal of the Linnean Society 186, 498 509. Jackson, S.T., Weng, C., 1999. Late Quaternary extinction of a tree species in eastern North America. Proceedings of the National Academy of Sciences USA 96, 13847 13852. 140 forest change in Alaska. Global Biogeochemical Cycles 9, 227 234. Harman, J.R., 2009. Plant Geography, in: Michigan Geography and Geology. Pearson Custom Publishing, Boston, Massachusetts, USA 330 345. Jump, A.S., Hunt, J.M., Peñuelas, J., 2006. Rapid climate change - related growth dec line at the southern range edge of Fagus sylvatica . Global Change Biology 12, 2163 2174. Kannenberg, S.A., Novick, K.A., Alexander, M.R., Maxwell, J.T., Moore, D.J., Phillips, R.P. and Anderegg, W.R. L. , 2019. Linking drought legacy effects across scales: From leaves to tree rings to ecosystems. Global Change Biology 25 , 2978 2992. Kelly, A.E., Goulden, M.L., 2008. Rapid shifts in plant distribution with recent climate change. Proceedings of the National Academy of Sciences 105, 11823 11826. Kelly, P.E. , Cook, E.R., Larson, D.W., 1994. A 1397 - year tree - ring chronology of Thuja occidentalis from cliff faces of the Niagara Escarpment, southern Ontario, Canada. Canadian Journal of Forest Research 24, 1049 1057. Kelly, P.E., Larson, D.W., 2007. The Last Sta nd: A Journey Through the Ancient Cliff - face Forest of the Niagara Escarpment. Natural Heritage Books, Toronto, Ontario, Canada. Kharuk, V., Im, S., Dvinskaya, M., 2010. Forest tundra ecotone response to climate change in the Western Sayan Mountains, Sibe ria. Scandinavian Journal of Forest Research 25, 224 233. Kirtman, B., Power, S.B., Adedoyin, A.J., Boer, G.J., Bojariu, R., Doblas - Reyes, F., Fiore, A.M., Kimoto, M., Meehl, G., Prather, M., Sarr, A., Schär, C., Sutton, R., van Oldenborgh, G.J., Vecchi, G., Wang, H. - J., Bindoff, N.L., Cameron - Smith, P., Clifton, O., Corti, S., Durack, P.J., Fichefet, T., García - Serrano, J., Ginoux, P., Guemas, V., Hawkins, E., Holland, M., Infanti, J., Ishii, M., Jacob, D., John, J., Klimont, Z., Knutson, T., Krinner, G. , Lawrence, D., Stevenson, D.S., van Noije, T., Vavrus, S., Voulgarakis, A., Wild, O., Woollings, T., Young, P., Delecluse, P., Palmer, T., Shepherd, T., Zwiers, F., 2013. Near - Term Climate Change: Projections and Predictability, in: Climate Change 2013: the Physical Science Basis: Working Group I Contribution to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Pres s, Cambridge, United Kingdom and New York, NY, USA 953 1028. Knutson, T., Zeng, F., Wittenberg, A., 2014. Seasonal and annual mean precipitation extremes iety 95, S19 S23. 141 Kopp, C.W., Cleland, E.E., 2014. Shifts in plant species elevational range limits and abundances observed over nearly five decades in a western North America mountain range. Journal of Vegetation Science 25, 135 146. Lamarche, V.C., Gr aybill, D.A., Fritts, H.C., Rose, M.R., 1984. Increasing Atmospheric Carbon Dioxide: Tree Ring Evidence for Growth Enhancement in Natural Vegetation. Science 225, 1019 1021. Larson, E.R., Rawling, J.E., 2016. Developing new sources of proxy climate data from historical structures in the Lake Michigan - Huron Basin. Journal of Great Lakes Research 42, 328 335. Larson, G.J., Kincare, K., 2009. Late Quaternary history of the eastern Mid - Continent region, USA, in: Michigan Geography and Geology. Pearson Custo m Publishing, Boston, Massachusetts, 69 90. Latte, N., Perin, J., Kint, V., Lebourgeois, F., Claessens, H., 2016. Major changes in growth rate and growth variability of beech ( Fagus sylvatica L.) related to soil alteration and climate change in Belgium. Forests 7, 174. LeBlanc, D., Maxwell, J., Pederson, N., Berland, A., Mandra, T., 2020. Radial growth responses of tulip poplar (Liriodendron tulipifera) to climate in the eastern United States. Ecosphere 11, e03203. LeBlanc, D.C., Berland, A.M., 2019. Spatial variation in oak (Quercus spp.) radial growth responses to drought stress in eastern North America. Canadian Journal of Forest Research 49, 986 993. Michi - 2017 update. Michigan Department of Natural Resources Forest Resource Division. https://www.michigan.gov/documents/dnr/2016ForestProductsIndustriesContributions _535055_7.pdf Legendre, P., Legendre, L.F.J., 1998. Numerical Ecology. Elsevier Scientific, New York, NY, USA. Liang, E., Leuschner, C., Dulamsuren, C., Wagner, B., Hauck, M., 2016. Global warming - related tree growth decline and mortality on the north - eastern Tibetan plateau. Climatic Change 134, 163 176. Littell, J.S., Peterson, D.L., Tjoelker, M., 2008. Douglas - fir growth in mountain ecosystems: water limits tree growth from stand to region. Ecological Monographs 78, 349 368. 142 Liu, H., Park Williams, A., Allen, C.D., Guo, D., Wu, X., Anenkhonov, O.A., Liang, E., Sandanov, D.V., Yin, Y., Qi, Z., Badmaeva, N.K., 2013. Rapid warming accelerates tree growth decline in semi - arid forests of Inner Asia. Global Change Biology 19, 2500 2 510. Lorenz, E. N., 1956: Empirical orthogonal functions and statistical weather prediction. Statistical Forecasting Scientific Rep. 1, Department of Meteorology, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA. Lovett, G.M., Weiss, M., Liebhold, A.M., Holmes, T.P., Leung, B., Lambert, K.F., Orwig, D.A., Campbell, F.T., Rosenthal, J., McCullough, D.G., Wildova, R., Ayres, M.P., Canham, C.D., Foster, D.R., LaDeau, S.L., Weldy, T., 2016. Nonnative forest insects and pathogens in the Un ited States: Impacts and policy options. Ecological Applications 26, 1437 1455. Lovis, W.A., Mainfort, R.C., Noble, V.E., 1976. Archaeological inventory and evaluation of Sleeping Bear Dunes National Lakeshore. Archaeological Survey Report Number 5. Mich igan State University Museum, East Lansing, Michigan, USA. Lovis, W.A., Monaghan, G.W., Arbogast, A.F., 2017. Site20LU115, Sleeping Bear Dunes National Lakeshore: synthesis of archaeological and environmental data recovery. Report submitted to National Pa rk Service, Midwest Archaeological Center, Lincoln, Nebraska, USA (Contract P15AC01378). Lovis, W.A., Monaghan, G.W., Arbogast, A.F., 2020. Reconstructing the precontact Late Woodland archaeology of site 20LU115 and the landscape history of Sleeping Bear Point, Great Lakes, USA. The Journal of Island and Coastal Archaeology 15, 1 21. MacArthur, R.H., 1972. Geographical Ecology: Patterns in the Distribution of Species, Harper & Row, New York, New York, USA. Marchin, R., Zeng, H., Hoffmann, W., 2010. Drou ght - deciduous behavior reduces nutrient losses from temperate deciduous trees under severe drought. Oecologia 163, 845 854. Marquardt, P.E., Miranda, B.R., Telewski, F.W., 2019. Shifts in Climate Growth Relationships of Sky Island Pines. Forests 10, 1011 . climatic drivers across a latitudinal gradient in a temperate broadleaf forest. Journal of Biogeography 42, 925 937. 143 Masson - Delmotte , V., Schulz, M., Abe - Ouchi, A., Beer, J., Ganopolski, A., Fidel, J., Rouco, G., Jansen, E., Lambeck, K., Luterbacher, J., Naish, T., Ramesh, R., Rojas, M., Shao, X., Anchukaitis, K., Arblaster, J., Bartlein, P.J., Benito, G., Clark, P., Comiso, J.C., Crow ley, T., Deckker, P.D., de Vernal, A., Delmonte, B., DiNezio, P., Dowsett, H.J., Edwards, R.L., Fischer, H., Fleitmann, D., Foster, G., Fröhlich, C., Hall, A., Hargreaves, J., Haywood, A., Hollis, C., Krinner, G., Landais, A., Li, C., Lunt, D., Mahowald, N ., McGregor, S., Meehl, G., Mitrovica, J.X., Moberg, A., Mudelsee, M., Muhs, D.R., Mulitza, S., Müller, S., Overland, J., Shindell, D., Smerdon, J., Solomina, O., Tarasov, P., Vinther, B., Waelbroeck, C., Wolf, D., Yokoyama, Y., Yoshimori, M., Zachos, J., Zwartz, D., Gupta, A.K., Rahimzadeh, F., Raynaud, D., Wanner, H., 2013. Information from Paleoclimate Archives, in: Climate Change 2013: the Physical Science Basi s: Working Group I Contribution to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA 383 464. Maxwell, J.T., Harley, G.L., 2017. Increased tree - ring ne twork density reveals more precise estimations of sub - regional hydroclimate variability and climate dynamics in the Midwest, USA. Climate Dynamics 49, 1479 1493. Maxwell, J.T., Harley, G.L., Matheus, T.J., 2015. Dendroclimatic reconstructions from multip le co - occurring species: a case study from an old - growth deciduous forest in Indiana, USA. International Journal of Climatology 35, 860 870. Maxwell, J.T., Harley, G.L., Matheus, T.J., Strange, B.M., Van Aken, K., Au, T.F., Bregy, J.C., 2020. Sampling de nsity and date along with species selection influence spatial representation of tree - ring reconstructions. Climate of the Past 16, 1901 1916. Maxwell, J.T., Harley, G.L., Robeson, S.M., 2016. On the declining relationship between tree growth and climate in the Midwest United States: the fading drought signal. Climatic Change 138, 127 142. Maxwell, R.S., Hessl, A.E., Cook, E.R., Buckley, B.M., 2012. A Multicentury Reconstruction of May Precipitation for the Mid - Atlantic Region Using Juniperus virginiana Tree Rings. Journal of Climate 25, 1045 1056. Maxwell, R.S., Hessl, A.E., Cook, E.R., Pederson, N., 2011. A multispecies tree ring reconstruction of Potomac River streamflow (950 2001). Water Resources Research 47, W05512. McMahon, S.M., Parker, G.G., M iller, D.R., 2010. Evidence for a recent increase in forest growth. Proceedings of the National Academy of Sciences 107, 3611 3615. Medvigy, D., Beaulieu, C., 2012. Trends in Daily Solar Radiation and Precipitation Coefficients of 144 Variation since 1984. J ournal of Climate 25, 1330 1339. Mishra, V., Cherkauer, K.A., 2010. Retrospective droughts in the crop growing season: Implications to corn and soybean yield in the Midwestern United States. Agricultural and Forest Meteorology 150, 1030 1045. Montgomery, K., 2006. Variation in Temperature With Altitude and Latitude. Journal of Geography 105, 133 135. Morice, C.P., Kennedy, J.J., Rayner, N.A., Jones, P.D., 2012. Quantifying uncertainties in global and regional temperature change using an ensem ble of observational estimates: The HadCRUT4 data set. Journal of Geophysical Research: Atmospheres 117. Moser, W.K., Butler - Leopold, P., Hausman, C., Iverson, L., Ontl, T., Brand, L., Matthews, S., Peters, M., Prasad, A., 2020. The impact of climate chan ge on forest systems in the northern United States: Projections and implications for forest management, in: Achieving sustainable management of boreal and temperate forests. Burleigh and Dodds Science Publishing, Cambridge, UK, 239 290. Nash, J.E., Sutcl iffe, J.V., 1970. River flow forecasting through conceptual models part I A discussion of principles. Journal of Hydrology 10, 282 290. Natalini, F., Alejano, R., Vázquez - Piqué, J., Pardos, M., Calama, R., Büntgen, U., 2016. Spatiotemporal variability of stone pine ( Pinus pinea L.) growth response to climate across the Iberian Peninsula. Dendrochronologia 40, 72 84. Neukom, R., Barboza, L.A., Erb, M.P., Shi, F., Emile - Geay, J., Evans, M.N., Franke, J., Kaufman, D.S., Lücke, L., Rehfeld, K., Schurer, A ., Zhu, F., Brönnimann, S., Hakim, G.J., Henley, B.J., Ljungqvist, F.C., McKay, N., Valler, V., von Gunten, L., PAGES 2k Consortium, 2019. Consistent multidecadal variability in global temperature reconstructions and simulations over the Common Era. Nature Geoscience 12, 643 649. Nitschke, C.R., Nichols, S., Allen, K., Dobbs, C., Livesley, S.J., Baker, P.J., Lynch, Y., 2017. The influence of climate and drought on urban tree growth in southeast Australia and the implications for future growth under climate change. Landscape and Urban Planning 167, 275 287. NOAA National Centers for Environmental information (2012a) State of the climate: National overview for March 2012. www.ncdc.noaa.gov/sotc/national/201203 (accessed 2 Mar 2020) NOAA National Centers fo r Environmental information (2012b) State of the climate: National overview for August 2012. www.ncdc.noaa.gov/sotc/national/201208 (accessed 2 Mar 2020) 145 NOAA National Centers for Environmental information (2015) State of the climate: National overview f or February 2015. www.ncdc.noaa.gov/sotc/national/201502 (accessed 2 Mar 2020). NOAA National Centers for Environmental information (2017a) State of the climate: National overview for February 2017. www.ncdc.noaa.gov/sotc/national/201702 (accessed 2 Mar 2 020). NOAA National Centers for Environmental information (2017b) State of the climate: National overview for September 2017 www.ncdc.noaa.gov/sotc/national/201709 (accessed 2 Mar 2020). Norby, R.J., Warren, J.M., Iversen, C.M., Medlyn, B.E., McMurtrie, R.E., 2010. CO2 enhancement of forest productivity constrained by limited nitrogen availability. Proceedings of the National Academy of Sciences 107, 19368 19373. Osborn, T.J., Biffa, K.R ., Jones, P.D., 1997. Adjusting variance for sample - size in tree - ring chronologies and other regional - mean timeseries. Dendrochronologia 15, 89 99. Panek, J.A., Goldstein, A.H., 2001. Response of stomatal conductance to drought in ponderosa pine: implicat ions for carbon and ozone uptake. Tree Physiology 21, 337 344. Parmesan, C., 2006. Ecological and Evolutionary Responses to Recent Climate Change. Annual Review of Ecology, Evolution, and Systematics 37, 637 669. Pataki, D.E., Oren, R., 2003. Species d ifferences in stomatal control of water loss at the canopy scale in a mature bottomland deciduous forest. Advances in Water Resources 26, 1267 1278. Pederson, N., Bell, A.R., Cook, E.R., Lall, U., Devineni, N., Seager, R., Eggleston, K., Vranes, K.P., 20 13. Is an Epic Pluvial Masking the Water Insecurity of the Greater New York City Region? Journal of Climate 26, 1339 1354. Pederson, N., Bell, A.R., Knight, T.A., Leland, C., Malcomb, N., Anchukaitis, K.J., Tackett, K., Scheff, J., Brice, A., Catron, B., Blozan, W., Riddle, J., 2012. A long - term perspective on a modern drought in the American Southeast. Environ men tal Research Letters. 7, 014034. Pederson, N., Cook, E.R., Jacoby, G.C., Peteet, D.M., Griffin, K.L., 2004. The influence of winter temperatur es on the annual radial growth of six northern range margin tree species. Dendrochronologia 22, 7 29. Pederson, Neil, Tackett, K., McEwan, R.W., Clark, S., Cooper, A., Brosi, G., Eaton, R., Stockwell, R.D., 2012. Long - term drought sensitivity of trees in second - growth forests in a humid 146 region. Canadian Journal of Forest Research. 42:1837 1850. Penskar, M.R., Derosier, A.L. 2012. Assisting the Michigan wildlife action plan: Tools and information for incorporating plants. Final report to NatureServe. Mich igan Natural Features Inventory Report No. 2012 - , Lansing, Michigan, USA. Peterson, D.W., Peterson, D.L., 1994. Effects of climate on radial growth of subalpine conifers in the North Cascade Mountains. Canadian Journal Forest Research 24, 1921 1932. Phipps, R.L., 1982. Comments on Interpretation on Climatic Information from Tree Rings, Eastern North America. Tree - Ring Bulletin 42, 11 22. Pompa - García, M., Sanchez - Salguero, R., Camarero, J., 2017. Observed and projected impacts of climate on radial gr owth of three endangered conifers in northern Mexico indicate high vulnerability of drought - sensitive species from mesic habitats. Dendrochronologia 45, 145 155. Ponocná, T., Spyt, B., Kaczka, R., Büntgen, U., Treml, V., 2016. Growth trends and climate r esponses of Norway spruce along elevational gradients in East - Central Europe. Trees 30, 1633 1646. Prasad, A., Pedlar, J., Peters, M., McKenney, D., Iverson, L., Matthews, S., Adams, B., 2020. Combining US and Canadian forest inventories to assess habita t suitability and migration potential of 25 tree species under climate change. Diversity and Distributions 26, 1142 1159. Pretzsch, H., Biber, P., Schütze, G., Uhl, E., Rötzer, T., 2014. Forest stand growth dynamics in Central Europe have accelerated sin ce 1870. Nature Communications 5, 4967. Pyke, G.H., Thomson, J.D., Inouye, D.W., Miller, T.J., 2016. Effects of climate change on phenologies and distributions of bumble bees and the plants they visit. Ecosphere 7, e01267. Quiring, S.M., 2004. Growing - season moisture variability in the eastern USA during the last 800 years. Climate Research 27, 9 17. R Core Team, R.C., 2017. R: a language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Rahman, M., Is lam, M., Bräuning, A., 2018. Tree radial growth is projected to decline in South Asian moist forest trees under climate change. Global and Planetary Change 170, 106 119. Rasmussen, K.K., 2007. Dendroecological analysis of a rare sub - canopy tree: Effects of climate, latitude, habitat conditions and forest history. Dendrochronologia 25, 3 17. 147 2019. Scots pine radial growth response to climate and future pro jections at peat and mineral soils in the boreo - nemoral zone. Theoretical and Applied Climatology 136, 639 650. Robakowski, P., Li, Y., Reich, P.B., 2012. Local ecotypic and species range - related adaptation influence photosynthetic temperature optima in deciduous broadleaved trees. Plant Ecology 213, 113 125. Rocky Mountain Tree - Ring Research, Inc. & the Tree Ring Laboratory of Lamont - Doherty Earth Observatory and Columbia University, 2013. Easte rn OLDLIST: A database of maximum tree ages for Eastern North America. https://www.ldeo.columbia.edu/~adk/oldlisteast/ . Rodriguez - Cabal, M.A., Barrios - Garcia, M.N., Amico, G.C., Aizen, M.A., Sanders, N.J., 2013. Node - by - node disassembly of a mutualistic interaction web driven by species introductions. Proceedings of the National Academy of Sciences 110, 16503 16507. Salad yga, T., Maxwell, R.S., 2015. Temporal variability in climate response of eastern hemlock in the Central Appalachian Region. Southeastern Geographer 55, 143 163. Salzer, M.W., Hughes, M.K., Bunn, A.G., Kipfmueller, K.F., 2009. Recent unprecedented tree - ri ng growth in bristlecone pine at the highest elevations and possible causes. Proceedings of the National Academy of Sciences 106, 20348 20353. s, L., Linares, J.C., Seftigen, K., 2017. Assessing forest vulnerability to climate warming using a process - based model of tree growth: bad prospects for rear - edges. Global Change Biology 23, 2705 2719. Schaberg, P.G., Shane, J.B., Cali, P.F., Donnelly, J.R., Strimbeck, G.R., 1998. Photosynthetic capacity of red spruce during winter. Tree Physiology 18, 271 276. Schneider, L., Esper, J., Timonen, M., Büntgen, U., 2014. Detection and evaluation of an early divergence problem in northern Fennoscandian tre e - ring data. Oikos 123, 559 566. Schroeder, A.H., 1968. Shifting for survival in the Spanish Southwest. New Mexico Historical Review 43, 291 310. Scott, R.W., Huff, F.A., 1996. Impacts of the Great Lakes on regional climate conditions. Journal of Great Lakes Research 22, 845 863. Siefert, A., Lesser, M.R., Fridley, J.D., 2015. How do climate and dispersal traits limit ranges of tree species along latitudinal and elevational gradients? Global Ecology and 148 Biogeography 24, 581 593. Silva, L.C.R., Sun, G ., Zhu - Barker, X., Liang, Q., Wu, N., Horwath, W.R., 2016. Tree growth acceleration and expansion of alpine forests: The synergistic effect of atmospheric and edaphic change. Science Advances 2, e1501302. Speer, J.H., 2010. Fundamentals of Tree - ring Research. University of Arizona Press, Tucson, Arizona, USA. Stahle, D.W., Cleaveland, M.K., Blanton, D.B., Therrell, M.D., Gay, D.A., 1998. The lost colony and Jamestown droughts. Science 280, 564 567. Stahle, D.W., Cook, E.R., Cleaveland, M.K., Therrell, M.D., Meko, D.M., Grissino - Mayer, H.D., Watson, E., Luckman, B.H., 2000. Tree - ring data document 16th century megadrought over North America. Eos, Transactions American Geophysical Union 81, 121 125. Stine, A.R., Huybers, P., 2014. Arctic tree rings as recorders of variations in light availability. Nature Communications 5, 383 6. Stokes, M.A., 1996. An Introduction to Tree - ring Dating. University of Arizona Press, Tucson, Arizona, USA. Sturm, M., Racine, C., Tape, K., 2001. Increasing shrub abundance in the Arctic. Nature 411, 546 547. Su, H., Axmacher, J.C., Yang, B., Sang, W., 2015. Differential radial growth response of three coexisting dominant tree species to local and large - scale climate variability in a subtropical evergreen broad - leaved forest of China. Ecological Research 30, 745 754. Suriano, Z.J., Leathers, D.J., 2017. Spatio - temporal variability of Great Lakes basin snow cover ablation events. Hydrological Processes 31, 4229 4237. Suriano, Z.J., Robinson, D.A., Leathers, D.J., 2019. Changing snow depth in the Great Lakes basin (USA): Implications and trends. Ant hropocene 26, 100208. Tang, Y., Zhong, S., Luo, L., Bian, X., Heilman, W.E., Winkler, J., 2015. The Potential Impact of Regional Climate Change on Fire Weather in the United States. Annals of the Association of American Geographers 105, 1 21. Taylor, K.E ., Stouffer, R.J., Meehl, G.A., 2012. An Overview of CMIP5 and the Experiment Design. Bulletin of the American Meteorological Society 93, 485 498. Tei, S., Sugimoto, A., Yonenobu, H., Matsuura, Y., Osawa, A., Sato, H., Fujinuma, J., Maximov, T., 2017. Tree - ring analysis and modeling approaches yield contrary response of circumboreal forest productivity to climate change. Global Change Biology 23, 517 9 149 5188. Telewski, F.W., Swanson, R.T., Strain, B.R., Burns, J.M., 1999. Wood properties and ring width responses to long - term atmospheric CO2 enrichment in field - grown loblolly pine ( Pinus taeda L.). Plant, Cell & Environment 22, 213 219. Tierney, G.L., Fahey, T.J., Groffman, P.M., Hardy, J.P., Fitzhugh, R.D., Driscoll, C.T., 2001. Soil freezing alters fine root dynamics in a northern hardwood forest. Biogeochemistry 56, 175 190. Vaganov, E.A. , Hughes, M.K., Kirdyanov, A.V., Schweingruber, F.H., Silkin, P.P., 1999. Influence of snowfall and melt timing on tree growth in subarctic Eurasia. Nature 400, 149 151. van der Sleen, P., Groenendijk, P., Vlam, M., Anten, N.P.R., Boom, A., Bongers, F., Pons, T.L., Terburg, G., Zuidema, P.A., 2015. No growth stimulation of tropical trees by 150 years of CO 2 fertilization but water - use efficiency increased. Nature Geoscience 8, 24 28. Venables, W.N., Ripley, B.D., 2013. Modern Applied Statistics with S - PLUS. Springer Science & Business Media. Walker, A.P., De Kauwe, M.G., Medlyn, B.E., Zaehle, S., Iversen, C.M., Asao, S., Guenet, B., Harper, A., Hickler, T., Hungate, B.A., Jain, A.K., Luo, Y., Lu, X., Lu, M., Luus, K., Megonigal, J.P., Oren, R., Ryan, E. , Shu, S., Talhelm, A., Wang, Y. - P., Warren, J.M., Werner, C., Xia, J., Yang, B., Zak, D.R., Norby, R.J., 2019. Decadal biomass increment in early secondary succession woody ecosystems is increased by CO 2 enrichment. Nature Communications 10, 454. Walker , K.V., Davis, M.B., Sugita, S., 2002. Climate Change and Shifts in Potential Tree Species Range Limits in the Great Lakes Region. Journal of Great Lakes Research 28, 555 567. Wa n g, G.G., Chhin, S., Bauerle, W.L. 2006. Effect of natural atmospheric CO 2 f ertilization 601 610. Wang, T., Hamann, A., Yanchuk, A., , G.A., Aitken, S.N., 2006. Use of response functions in selecting lodgepole pine populations for future climates. Global Change Biology 12, 2404 2416. Warner, S.M., Jeffries, S.J., Lovis, W.A., Arbogast A.F., Telewski, F.W. 2021. Tree ring - reconstructed late summer moisture conditions, 1546 to present, northern Lake Michigan, USA. Climate Research 83, 43 56. Whittaker, R.H., Niering, W.A., 1964. Vegetation of the Santa Catalina Mountains, Arizona. I. Ecological Classification and Distribution of Species. Journal of the Arizona Academy of Science 3, 9 34. 150 Wibbe, M.L., Blanke, M.M., Lenz, F., 1994. Respiration of apple trees between leaf fall and leaf emergence. Environmental and Experimental Botany 34, 25 30. Wigley, T.M.L., B riffa, K.R., Jones, P.D., 1984. On the Average Value of Correlated Time Series, with Applications in Dendroclimatology and Hydrometeorology. Journal of Applied Meteorology and Climatology 23, 201 213. Williams, J.W., Jackson, S.T., 2007. Novel climates, n o - analog communities, and ecological surprises. Frontiers in Ecology and the Environment 5, 475 482. Willis, C.G., Ruhfel, B., Primack, R.B., Miller - Rushing, A.J., Davis, C.C., 2008. Phylogenetic climate change. Proceedings of the National Academy of Sciences 105, 17029 17033. Wolter, K., Hoerling, M., Eischeid, J.K., van Oldenborgh, G.J., Quan, X. - W., Walsh, J.E., Chase, T.N., Dole, R.M., 2015. How Unusual was the Cold Winter of 2013/14 in the U pper Midwest? Bulletin of the American Meteorological Society 96, S10 S14. Woodhouse, C.A., Overpeck, J.T., 1998. 2000 years of drought variability in the central United States. Bulletin of the American Meteorological Society 79, 2693 2714. Wu, X., Li, X ., Liu, H., Ciais, P., Li, Y., Xu, C., Babst, F., Guo, W., Hao, B., Wang, P., Huang, Y., Liu, S., Tian, Y., He, B., Zhang, C., 2019. Uneven winter snow influence on tree growth across temperate China. Global Change Biology 25, 144 154. Yamaguchi, D.K., 1 991. A simple method for cross - dating increment cores from living trees. Canadian Journal of Forest Research 21, 414 416. Yukimoto, S., Adachi, Y., Hosaka, M., Sakami, T., Yoshimura, H., Hirabara, M., Tanaka, T.Y., Shindo, E., Tsujino, H., Deushi, M., Mi zuta, R., Yabu, S., Obata, A., Nakano, H., Koshiro, T., Ose, T., Kitoh, A., 2012. A New Global Climate Model of the Meteorological Research Institute: MRI - CGCM3 Model Description and Basic Performance . Journal of the Meteorological Society of Japan. Ser. II 90A, 23 64. Zhang, T., 2005. Influence of the seasonal snow cover on the ground thermal regime: An overview. Reviews of Geophysics 43, RG4002.