Figure 6. Changes in pollen percentages for different plant taxa during the last 3700 years. Black and gray silhouettes (5x exaggeration) represent changes in the proportion of pollen for individual taxa over time as a percentage of all terrestrial pollen. The arboreal pollen (AP) and non-arboreal pollen (NAP) ratio includes all upland tree pollen divided by all non-aquatic shrubs and herbs. Fire frequency is recorded as number of episodes per 1000 years. Examples of vegetation response to climate change include a distinct change from steppe vegetation to forest accompanied by increased fire frequency ca. 2125 cal yr BP and a return to more open vegetation ca. 750 cal yr BP. Source: Power et al. 2006; reprinted with permission.
These shifts in vegetation were accompanied by pronounced changes in fire patterns as evident in the charcoal record (fig. 6). Intervals dominated by forests coincide with high magnitude and frequent fires (e.g., stand replacing fires), periods dominated by steppe-parkland vegetation are associated with smaller and less frequent fires (surface fires), and a decline in charcoal deposition in the last century likely reflects the impact of fire suppression (fig. 6). The Foy Lake record demonstrates the impacts of centennial-scale climate variations and their associated ecosystem response during the last two millennia. While relatively modest changes in vegetation cover occurred after the conifer forests were established ca. 2700 cal yr BP, multidecadal shifts in climate are evident in the fire reconstruction for the last several millennia.
Central U.S. Rocky Mountains and the Greater Yellowstone Area
Pollen records from Yellowstone and Grand Teton national parks show the nature of the biotic change that occurred in conjunction with broad climatic changes at different elevations (fig. 7, Whitlock 1993). Deglaciation (ca. 17,000–14,000 ka BP) was followed by colonization of tundra vegetation which, during the late Glacial and into the mid-Holocene, was replaced by subalpine communities of spruce, fir, and pine in many regions, first as an open parkland and then as a closed forest. With the warmer growing season conditions of the early Holocene, pine, juniper (Juniperus), and birch (Betula) were present at low elevations, whereas lodgepole pine, Douglas fir, and aspen (Populus) established and characterized mid-elevation forests. Subalpine forests (Picea and Abies) expanded their ranges to higher elevations. , and the upper treeline was at an elevation similar to that of the present. Decreased summer insolation in the late Holocene (ca. 3000–4000 ka BP) led to cooler, wetter conditions. Sagebrush-steppe became present at low elevations and forests of limber pine (Pinus flexilus), Douglas fir, and lodgepole pine dominated mid-elevations (Whitlock et al. 1993). Subalpine communities were comprised of mixed spruce, fir, and pine forests, and the increasingly cooler conditions resulted in a lowering of the upper treeline to an elevation close to its present position.
Figure 7. Ecological response to changing climatic conditions following glacial retreat in the Central Rockies and the summer-dry region of the Greater Yellowstone Area (derived from Whitlock 1993). The figure illustrates that tundra occupied most of the area >14 ka BP, but is only present at high elevations in the late Glacial and late Holocene.
Figure 8. Ecological response to changing climatic conditions following glacial retreat in Yellowstone National Park. Pollen and charcoal diagram from Cygnet Lake, central Yellowstone (modified from Millspaugh et al. 2000).
An examination of charcoal, pollen, and climate conditions from central Yellowstone provides an example of the linkage between fire and climate even in the absence of vegetation change (fig. 8; Whitlock 1993). Pollen from Cygnet Lake indicates the area was dominated by a tundra community of sagebrush and grass (Poaceae) when cool and wet conditions prevailed prior to 12,000 ka BP, after which the vegetation was dominated by lodgepole pine, which is favored in infertile rhyolite soils. The last 12,000 years reveal little change in the vegetation at Cygnet Lake compared to sites on non-rhyolite substrates that show strong responses to early, middle, and late Holocene climate forcings (Whitlock 1993). Despite the complacency of the vegetation, the Cygnet Lake fire record shows the effect of early Holocene drought on fire regimes; the charcoal data suggests rising fire activity in response to increasing summer insolation. Fire return intervals were 75–100 years between 11,000 and 7000 cal yr BP, and cooler wetter conditions after 7000 cal yr BP coincide with decreasing fire frequency. The present fire return interval of 200 to 400 years was reached in the late Holocene.
Southern U.S. Rocky Mountains
A network of vegetation reconstructions from pollen and macrofossil data provides a history of climate and vegetation change in southwestern Colorado during the late Glacial and Holocene (Fall 1997). Prior to 14,000 cal yr BP, cooler and wetter conditions (2–5ºC [4–9°F] cooler and 7–16 cm [3–6"] wetter than at present) supported tundra vegetation at high elevations (~300–700 m [984–2,297'] below the present treeline) and spruce parkland at low elevations (fig. 9, Fall 1997). During the late Glacial, fir (Abies) increased in abundance and subalpine forest was established at low and middle elevations (Fall 1997). Summer insolation increased during the early Holocene, and warmer temperatures allowed subalpine forests to expand above their present elevation. While the central and northern U.S. Rockies generally experienced warm dry conditions during the early Holocene, the southern Rockies experienced warmer, wetter growing seasons driven by a more intense North American monsoon (Thompson et al. 1993). Markgraf and Scott (1981) recorded an upslope advance of subalpine forests due to warmer conditions and an expansion of pine forests at both the lower and upper treeline facilitated by warmer and still wet conditions. Similarly, Fall (1997) found that from 9–4 ka BP, warm summers (mean temperatures 1.9ºC [3.4°FC] above present) facilitated the expansion of forests of spruce and fir (Picea engelmannii and Abies lasiocarpa) upward to an elevation of almost 4,000 meters (13,123'), at least 300 meters (948') higher than today (Fall 1997), and downward to elevations below 3,000 meters (9.483').
Figure 9. Ecological response to changing climatic conditions following glacial retreat in the southern Rockies. Derived from Fall 1997.
In the late Holocene, low-elevation montane forests mixed with steppe vegetation and low-elevation subalpine forests were defined by an increasingly open stand structure. Summer temperatures declined to pre-industrial levels ca. 1850, and spruce and fir dominated subalpine forests and krummholz vegetation. Montane taxa retreated upslope, sagebrush steppe expanded at lower elevations, and alpine tundra dominated a larger range of high elevation areas, suggesting that drier conditions increased for parts of the southern Rockies (Markgraf and Scott 1981; Fall 1997). Conditions similar to the present were established approximately two millennia ago, with modest treeline elevation fluctuations during the Medieval Climate Anomaly.
Figure 10. Ecological response to changing climatic conditions in the western portion of the Upper Columbia Basin following deglaciation. Derived from Carp Lake, Washington (Whitlock et al. 2000).
Upper Columbia Basin
Paleoenvironmental data are available from the Upper Columbia Basin (Barnosky 1985; Whitlock et al. 2000; Mack et al. 1978, 1976; Mehringer 1996; Blinnikov et al. 2002), the Snake River Plain (Davis 1986; Beiswenger 1991; Bright and Davis 1982), and the mountains of southern Idaho and Montana (Doerner and Carrara 2001; Whitlock et al. in review; Mumma 2010).
During glacial times, tundra-steppe communities dominated by Artemisia and Poaceae were widespread in the basins, reflecting cold dry conditions (figs. 10–11). As the climate warmed in the late Glacial period, pine, spruce, and fir parkland developed. The early Holocene period in the western Columbia Basin and Snake River Plain featured steppe vegetation, and records from adjacent mountains record an expansion of juniper, sagebrush, and Chenopodiaceae. Summer drought was more pronounced throughout much of the region during the early Holocene as the amplification of the seasonal insolation cycle resulted in warmer and effectively drier conditions at low and mid-elevations (Whitlock et al. 2000). Cool dry conditions in the Okanogan Highlands of northern Washington continued to support grasses and sagebrush, and the forest-steppe ecotone was north of its present location by as much as 100 kilometers (Mack et al. 1978). Pine, spruce, and fir were present in areas with greater precipitation (e.g., Waits Lake, eastern Washington) (Whitlock and Brunelle 2007, fig. 11). In the mid-Holocene, increased effective moisture allowed the establishment and expansion of pine woodland at middle and higher elevations, and the upper treeline was higher than at present. In the western Columbia Basin, the expansion of pine woodland (primarily Pinus ponderosa) was followed by an invasion of mixed forest in the late Holocene (Douglas fir, larch, fir, western hemlock, and oak) (Whitlock and Brunelle 2007). Some sites at higher elevations signal an interval of cooling during the late Holocene (ca. 1.7–3.5 ka BP) when the abundance of spruce and fir pollen increased (Whitlock and Brunelle 2007) and led to an expansion of mixed forests throughout the Okanogan highlands. Modern assemblages of Douglas fir, fir, western hemlock and spruce were established during this period (Whitlock and Brunelle 2007).
Figure 11. Late Glacial and early Holocene vegetation history along the southern margin of the Cordilleran ice sheet, based on a transect of pollen records from western Washington to western Montana. Abies, fir; Alnus, alder; Artemisia, sagebrush; Picea, spruce; Pinus contorta, lodgepole pine; Poaceae, grass; Pseudotsuga, Douglas fir; Shepherdia, buffaloberry; Tsuga heterophylla, western hemlock. Source: Whitlock and Brunelle 2007; reprinted with permission.
What do paleoenvironmental records tell us about millennial scale climate variations?
The western United States has experienced large-scale changes in climate, vegetation, and disturbance regime since the last glaciation 20,000 years ago. With the initial recession of glacial ice more than 14,000 years ago, the climate was colder and generally drier than at present and most regions were colonized by tundra communities. As the climate warmed and precipitation increased from 14,000 to 11,000 years ago, these tundra communities were replaced by subalpine parkland and then closed subalpine forest. During the early Holocene (11,000–7,000 cal yr BP), the development of warmer and drier than present conditions led to more xerophytic vegetation and more fires in most areas. After 7000 cal yr BP, the climate became cooler and effectively wetter. In most regions, the modern vegetation and climate were established during the late Holocene (the last 4000 cal yrs). These changes highlight several important lessons for understanding the impacts of climate change on vegetation:
Millennial-scale variations in climate over the last 20,000 years were caused by changes in the latitudinal and seasonal distribution of incoming solar radiation, the size and extent of the continental ice sheets, and attendant shifts in atmospheric circulation (e.g., southward displacement of the jet stream, the strength of the northeastern Pacific subtropical high- pressure system, and the intensity of monsoonal circulation). These slowly varying changes determined the distribution and composition of plant communities.
Abrupt climate variation during the last 20,000 years led to rapid changes in the assemblages and distribution of vegetation across the U.S. Rocky Mountains and the Upper Columbia Basin and influenced ecosystem processes such as fire.
The Medieval Climate Anomaly (MCA, ca. 950–1250) and Little Ice Age (ca. 1400–1700) resulted in shifts in plant distributions and disturbance regimes in some locations, but were not uniformly manifested across the study area. For example, warm dry conditions in Yellowstone during the MCA led to increased fires in the summer-dry areas while the summer-wet areas of Yellowstone were effectively wetter (Whitlock and Bartlein 1993; Whitlock and et al. 2003).
The ecological impact of these shorter climatic events (lasting centuries) is variable either because the climate signal was regionally heterogeneous or because plant communities were not responsive to climate change on this relatively short time scale.
The western United States is influenced by two precipitation regimes—a summer-dry area under the influence of the northeastern Pacific subtropical high-pressure system and a summer-wet region strongly influenced by summer monsoon circulation. In the northern U.S. Rocky Mountains, the location of these regimes is sharply delimited and constrained by topography. These two regimes were enhanced during the early Holocene, when summer solar radiation was higher than at present. As a result, summer-wet areas became wetter and experienced fewer fires than at present, and summer-dry areas became drier with more fires and xerophytic vegetation than at present (Whitlock and Bartlein 1993; Huerta et al. 2009; Millspaugh et al. 2004). The contrast between summer-wet and summer-dry regions was greatest in the early Holocene. This past response suggests that future changes in precipitation regimes will also likely be spatially heterogeneous, and that the boundary between precipitation regimes will likely be quite sharp in mountainous regions.
Although climate exerts strong controls on the distribution of vegetation at large spatial and long temporal scales, edaphic factors can amplify or minimize the response at smaller scales, as illustrated by the persistence of lodgepole pine forests on rhyolitic soils in Yellowstone’s Central Plateau throughout the Holocene (Whitlock et al. 1993; Millspaugh et al. 2000) and the influence of ultramafic soils on Holocene vegetation and fire regimes (Briles et al. in review).
Climate change can influence the distribution of vegetation via direct climate constraints (e.g., temperature and precipitation) or indirectly by influencing key ecosystem processes such as fire and nutrient cycling. Feedbacks related to vegetation changes can also influence fire by changing fuel availability.
The superposition of climate changes occurring on multiple time scales means that no period in the last 20,000 years is an exact analogue for the future. Nonetheless, the paleoenvironmental records show the resilience of vegetation to periods of extreme drought, changes in disturbance regimes, and rapid climate change. These examples provide insights about the sensitivity and pathways by which ecosystems respond to climate changes of different duration and intensity.
Natural variations in climate and the accompanying ecological responses occur at multiple temporal and spatial scales, all of which must be understood to explain modern plant communities and their distributions. Paleoecological data provide evidence of a range of responses that are not adequately represented in the last two centuries. A baseline of natural variability for restoration efforts must therefore consider a longer time scale.
Many terrestrial ecosystems in the study region were established during the last 3000 to 4000 years in response to gradual cooling and increased effective moisture in the late Holocene. More subtle changes occurred during the Medieval Climate Anomaly and Little Ice Age (as discussed in the next section). The sequence of events that led to present vegetation is different from that projected for the future, which argues against strategies to restore to a reference condition. Instead, we need process-based approaches and flexible management responses.
Key ecosystem processes such as fire are driven by climate at large spatio-temporal scales. Patterns of the fire in the 20th century poorly represent the potential range of fire regimes that have occurred in the past and may occur in the future.
The last 2000 years of environmental change
Primary drivers of change
The primary drivers of climate during the last 2000 years include ocean-atmosphere interactions, volcanic eruptions, changes in incoming solar radiation, and increases in atmospheric greenhouse gases (GHGs) and aerosols due to human activities (fig. 2). Climate model simulations indicate that during the pre-industrial portion of the last 2000 years, solar fluctuations and volcanic eruptions were likely the most strongly varying forcings, and in combination with ocean-atmosphere interactions they likely resulted in periods of relative warmth and cold (Amman et al. 2007; Mann and Jones 2003; Mann 2007; Jones et al. 2009; Esper et al. 2002). GHGs and tropospheric aerosols varied little until around AD 1700 when they began to be significantly impacted by human activities (Solomon et al. 2007; Keeling 1976).The rapid rise in 20th century global temperatures is best explained by the combination of natural and anthropogenic GHG forcings, with GHGs playing an increasingly dominant role during recent decades.
Compared to the conditions driving continental deglaciation and the Pleistocene/Holocene transition, orbital and radiative forcings over the last two millennia have remained relatively constant and can be considered more analogous to modern conditions. However, the major centennial-scale climate variations evident during this period can be linked to changes in solar output, volcanic forcing, and ocean-atmosphere interactions (Amman et al. 2007; Mann et al. 2009; Mann 2007; Jones et al. 2009; Esper et al. 2002). Because the rates of climatic change during the last two millennia were much smaller in magnitude than those associated with the late Glacial and early Holocene, the ecological response to climatic variation was generally less dramatic. For much of the study area considered in this synthesis, climatic variation during this period led to shifts in the extent and abundance of species found in modern vegetation assemblages but rarely to widespread changes in dominant vegetation. Vegetation types since the late Holocene are considered similar to present conditions and, overall, the magnitude and duration of the changes are not comparable to those of the Pleistocene–Holocene transition. Rather, the last 2000 years have had smaller magnitude and shorter duration (centennial, decadal, interannual) climatic controls on ecosystems that have nevertheless resulted in societally and ecologically relevant changes in both ecosystems and natural resources. At these shorter time scales, ocean-atmosphere interactions such as the Pacific Decadal Oscillation, the North Atlantic Oscillation, the Atlantic Multidecadal Oscillation, and the El Niño–Southern Oscillation interact to influence temperature, precipitation, and atmospheric circulation and help explain droughts and wet periods at interannual to interdecadal scales (e.g., Cook et al. 2004; Mann et al. 2009; Gray et al. 2004; Hidalgo 2004; Fye et al. 2003; McCabe et al. 2004; Graumlich et al. 2003; Enfield et al. 2001). Past records of temperature illustrate centennial, decadal, and interannual variation, providing a context for understanding ecosystem changes that occurred in different regions (fig. 12).
Figure 12. Comparison of regional and global temperature reconstructions. Derived from Luckman and Wilson 2005, Salzer and Kipfmueller 2005 (both based on tree-ring records), and Mann et al. 2008 (based on a multi-proxy reconstruction).
Tree-ring records from throughout the western United States show natural variation in temperature, precipitation, and available moisture during the last two millennia, some of which is synchronous across large areas of the four climate regions in this study, while other variations are more representative of local phenomena (figs. 12–13, Pederson et al. 2006; Cook et al. 2004). These records show that decadal and multidecadal fluctuations in precipitation are a defining characteristic of the climate during past millennia and exert important controls on ecosystem processes and species distributions (e.g., Pederson et al. 2006; Cook et al. 2004, 2007). Regionally synchronous wet and dry intervals have been linked to low-frequency variations and state changes in sea surface temperature and pressure anomalies in both the Atlantic and Pacific oceans which are discussed in more detail later (McCabe et al. 2004; Gray et al. 2003; Cayan et al. 1998).
Figure 13. Long-term aridity changes in the West as measured by the percent area affected by drought (PDSIb−1, thick black line), 95% boostrapped confidence intervals (light-blue dotted lines) and the long-term mean (thin horizontal black line). ). The four most significant (p<0.05) dry and wet epochs since 800 are indicated by arrows. The 1900–2003 interval is highlighted by the yellow box. The average drought area during that period and for the 900–1300 interval is indicated by the thick blue and red lines, respectively. The difference between these two means is highly significant (p<0.001). Source: Cook et al. 2004, 2007; reprinted with permission.
Two major, well-documented examples of centennial-scale climate change during the last two millennia are the Medieval Climate Anomaly 950–1250) and the Little Ice Age (1400–1700). As indicated by figure 13 (Cook et al. 2007), a number of anomalous warm dry periods and cool wet periods occurred (see Biondi et al. 1999; Meko et al. 2007; Salzer and Kipfmuller 2005; Cook et al. 2007), and resulted in extensive hydrological and ecological impacts. During the MCA, warm temperature dry precipitation anomalies persisted across western North America (fig. 13). While the general climatic conditions during this time were defined by warmer, drier conditions, the local effects were highly variable. Elevated aridity and “mega-droughts” (lasting at least ~50 years) were common across the western United States, with more areas experiencing drought simultaneously than during the LIA or most of the 20th century (fig. 13, Cook et al. 2007).
Data from a number of sites suggest that regionally synchronous drought events occurred regularly during the MCA, with durations and extents unmatched in the late Holocene (Cook et al. 2007, 2004). Glacial retreat occurred in mountainous areas of Colorado, Wyoming, Montana, and the Cascades, with substantial reductions in streamflow (e.g., Meko et al. 2007; Gray et al. 2003) and lake levels throughout the study area (Millspaugh et al. 2000; Brunelle and Whitlock 2003). However, spatial and temporal variations in the generally warm dry conditions were widespread, as the central U.S. Rockies, Greater Yellowstone Area, and parts of the southern Rockies experienced increased summer moisture (Whitlock and Bartlein 1993). Upper treelines in some areas increased in elevation and areas now covered by krummholz were occupied by arborescent trees (Graumlich et al. 2005; Whitlock et al. 2002; Fall 1997; Rochefort et al. 1994; Winter 1984; Kearney and Luckman 1983). Additionally, alpine larch expanded 90 kilometers north of its current range (Reasoner and Huber 1999, Reasoner and Hickman 1989) ca. 950–1100 BP. The mechanisms and drivers leading to the MCA are still debated, but there is increasing evidence that low-frequency variation in ocean-atmosphere interactions was an important factor (Mann et. al 2009).
The LIA was a period of anomalous Northern Hemisphere cooling when mountain glaciers throughout the western United States expanded, many reaching their Holocene maximum (e.g., Pederson et al. 2004; Luckman 2000; Watson and Luckman 2006). While temperatures across the study area were persistently cooler than the long-term average (>1˚C [2°F] cooler), some data suggests that the magnitude of the cooling decreased with latitude (Whitlock et al. 2002). In the northern Rockies where the most pronounced cooling occurred, conditions in the late LIA may have approached those of the late Pleistocene (Pederson et al. 2007; Luckman 2000; Clark and Gillespie 1997). Both tree ring and glacier data indicate sustained cool summer conditions and increased winter precipitation across the northern Rockies resulted in the major glacier advance during the LIA (Watson and Luckman 2004; Pederson et al. 2004). Advances of this magnitude did not occur in the southern and central Rockies, demonstrating the spatial variability in LIA climate anomalies (Clark and Gillespie 1997). As with the MCA, the drivers and mechanisms that influenced cooler conditions during the LIA are not well understood. Decreased sunspot activity during a period called the Maunder Minimum which led to decreased incoming solar radiation from ca. 1645 to 1715 is considered one of the main variables explaining cooler conditions for this interval of the LIA (Eddy 1976; Luckman and Wilson 2005).
The following case studies from the four climate regions highlight examples of biophysical and biotic response to climate change during the last two millennia and provide clues to the timing and extent of future biotic changes.