National Park Service U. S. Department of the Interior Natural Resource Program Center

Records of past hydroclimatic changes, glacier dynamics, and fire activity in Glacier and Waterton national parks show how interdecadal and longer-term changes in climate interact to alter ecosystem processes in the northern Rockies (fig. 14). Long-term changes in regional temperature (e.g., the relatively cool conditions of the LIA in contrast to warmer temperatures in the 1350s to 1450s and especially over the last half of the 20th century, fig. 12) combined with persistent shifts in summer and winter moisture regimes over decadal to multidecadal time scales have a particularly strong impact on fire regimes in the Glacier National Park region (fig. 14a–c). Regional records spanning the last three centuries show that periods from the 1780s to the 1840s and the 1940s to the 1980s had generally cool wet summers coupled with high winter snowpack, resulting in extended (>20 years) burn regimes characterized by small infrequent fires with relatively little area burned. Conversely, decadal and longer combinations of low snowpack and warm dry summers resulted in burn regimes characterized by frequent fires and large total area burned (e.g., 1910–1940, 1980–present).

Similarly, long-duration summer and winter temperature and moisture anomalies drive glacial dynamics in the northern Rockies (Watson and Luckman 2004: Pederson et al. 2004). Prior to the height of the LIA (ca. 1850), four centuries of generally cool summers prevailed (fig. 12). What drove the glaciers to reach their greatest extent since the Last Glacial Maximum was cool wet summers coupled with generally high snowpack conditions from 1770 to 1840 (Watson and Luckman 2004; Pederson et al. 2004). During subsequent periods when summer drought and snowpack were generally in opposing phases (e.g., 1850–1910) and summer temperatures remained relatively cool (fig. 14), glaciers experienced moderate retreat rates (1–7 m/yr [3–23']). From 1917 to 1941, however, sustained low snowpack, extreme summer drought conditions, and high summer temperatures drove rapid glacial retreat. The Sperry Glacier retreated at 15–22 meters (49–72') per year and lost approximately 68% of its area (fig. 14d). Other glaciers such as the Jackson and Agassiz glaciers at times retreated at rates 100 meters (328') per year (Carrara and McGimsey 1981; Key et al. 2002; Pederson et al. 2004). Climatic conditions from 1945 to 1977 became generally favorable (i.e., level to cooling summer temperatures, high snowpack with variable summer drought conditions) for stabilization and even accumulation of mass at some glaciers (fig. 14a,b). Since the late 1970s, however, exceptionally high summer temperatures (Pederson et al. 2010) combined with low winter snowpack and multiple periods of severe and sustained summer drought have resulted in a continuation of rapid glacier retreat.

Figure 14. Relationship between Glacier National Park summer drought, inferred winter snowpack, fire area burned, and glacial recession 1700 to present. (a) Instrumental and reconstructed summer drought (MSD) normalized by converting to units of standard deviation and smoothed using a 5-year running mean. (b) Measured spring snowpack (May 1 SWE) anomalies (1922–present) and average annual instrumental and reconstructed PDO anomalies. Each time series was normalized and smoothed using a 5-year running mean. For ease of comparison, the instrumental and reconstructed PDO index was inverted due to the strong negative relationship between PDO anomalies and May 1 snowpack. (c) Fire area burned timeline for the GNP region. Timeline is presented with maps of fire activity during periods of interesting winter and summer precipitation regimes. (d) Maps showing the decrease in area of the Sperry Glacier at critical points from 1850 to 2003. The retreat patterns of the Sperry Glacier are representative of regional patterns of recession for glaciers sensitive to regional climate variability. Source: Pederson et al. 2006, American Meteorological Society; reprinted with permission.

As demonstrated by these records from Glacier National Park and surrounding regions, biophysical and ecosystem processes of the northern Rockies are strongly influenced by moisture and temperature variability at decadal and longer time scales. This relationship between biophysical and ecosystem processes with hydroclimatic changes that can appear to be static on shorter (e.g., decadal) scales poses challenges to management and sustainability in three ways. First, long-duration proxy reconstructions call into question the conventional strategy of defining reference conditions or management targets based solely on < 100-year records. For example, the use of 30-year climatology for the allocation of natural resources and development of resource management goals is flawed because a 30-year climatic mean may only capture a single mode of climate variability (e.g., an extended regime of wet or dry conditions). Second, abrupt, high-magnitude changes from one climate regime to the next can onset rapidly and have prolonged impacts on ecosystem processes. These persistent and frequent climate-related shifts may amplify or dampen the effects of management activities. Lastly, decadal and longer persistence of either deficits or abundances in climate-related biophysical processes can lead to management policies and economic strategies that, while appropriate during the current regime, may not be robust under subsequent climates. Overall, greater awareness of how ecosystems respond to climate change at longer temporal scales provides important context for future management.

A long-term reconstruction of changing distributions of Utah juniper (Juniperus osteosperma) in the CR-GYA provides an example of how climate variation, complex topography, and spatial distribution of suitable habitat and biotic factors interact to govern plant invasions. It also provides evidence contrary to the popular idea that plant invasions are characterized by a steady and continuous march across landscapes. The distribution of Utah juniper in the mountains of Wyoming, southern Montana, and Utah during the late Holocene has been tracked by radiocarbon-dating fossilized woodrat middens (Lyford et al. 2003). During a dry period in the mid-Holocene (ca. 7500–5400 BP), Utah juniper migrated north into the central Rockies via a series of long-distance dispersal events. Further range expansion and backfilling of suitable habitat was stalled during a wet period from 5400 to 2800 BP (Lyford et al. 2003). In response to warmer, drier conditions that developed after 2800 years BP, Utah juniper populations rapidly expanded within the Bighorn Basin, especially from 2800 to 1000 years BP. The notable absence of significant Utah juniper establishment and expansion during the MCA suggests that long-term climate variations determine the distributions of species with centennial-scale life expectancies. In the case of the Utah juniper, Lyford et al. (2003) noted that establishment rates are significantly more affected by adverse climatic conditions than by individual or population survival where the species is already established. This could explain, in part, the tendency for Utah juniper populations to remain static instead of contracting during the Holocene.

In general, Utah juniper range expanded during periods characterized by warmer, drier conditions, and expansion and establishment ceased during cool, wet periods (Lyford et al. 2003). Thus, the migration of Utah juniper into the central Rockies was at least partly controlled by millennial-scale climatic variations during the Holocene. Although Utah juniper distribution is severely limited by cool temperatures and high precipitation in higher elevations of the central Rockies, the species inhabits only a fraction of the suitable climate space in the region (fig. 15) because it is limited by suitable substrate; present distributions cover more than 90% of the substrate in the region deemed highly suitable for Utah juniper survival.

Figure 15. Habitats for Utah juniper in Wyoming and adjacent Montana. Black areas indicate extremely suitable habitats and gray areas moderately to highly suitable habitats. (a) Climate (ratio of growing season precipitation to growing season temperature). (b) Substrate, including soil, bedrock, and surficial-material type. (c) Climate and substrate combined. (d) Modern distribution (Knight et al. 1987, Driese et al. 1997; available online [see footnote 3]). Note that climate (a) over-predicts the modern distribution, which is strongly constrained by substrate variables (b), and that favorable habitat is patchily distributed (c). Source: Lyford et al. 2003; reprinted with permission.

The case of the Utah juniper shows how climatic controls can influence species distribution, migration, and establishment in the central Rockies within the context of millennial-scale climate change. It also highlights the importance of recognizing other environmental factors that affect species distribution. While suitable climate can allow a species to become established, spatially variable factors such as substrate, dispersal, and competition influence how successfully it can disperse to and colonize areas with suitable climate. This provides an important example for considering how ecosystems and species will respond to changes in the spatial distribution of suitable habitat with changing climatic conditions. In this case, landscape structure and climate variability play key roles in governing the pattern and pace of natural invasions and will be important variables to consider when anticipating future changes in the distribution of plant species. The high temporal and spatial precision provided by this study illustrates that vegetation response to future conditions will be more nuanced than a steady march to newly suitable habitats, and better characterized by episodic long-distance colonization events, expansion, and backfilling (Lyford et al. 2003). This study suggests that models predicting plant invasions based on climate model projections may be oversimplified and encourages a more focused examination of how species dispersal will interact with the spatial distribution of suitable habitat and climate variability to govern future invasions.

Tree-ring records from Wyoming’s Green River basin provide a reconstruction of drought conditions for the last millennium and reveals how natural variations in dry and wet periods are a defining characteristic of the CR-GYA. Tree rings were used to develop a 1100-year record of the Palmer Drought Severity Index, a measure of drought that includes precipitation and temperature trends, for southwestern Wyoming (Cook et al. 2004). This record is typical of many areas of the CR-GYA, showing above-average precipitation effective moisture in the early 20^th century (Woodhouse et al. 2006; Gray et al. 2004, 2007; Meko et al. 2007) and the potential for severe, sustained droughts far outside the range of 20^th century records, including several multidecadal droughts prior to 1300 (fig. 16).

Figure 16.

Although using the last century as a reference for climate conditions would suggest that the Green River Basin is wet and relatively free of drought, this longer-term reconstruction indicates that some of the most severe droughts in the 1930s and 1950s were relatively minor compared to many dry periods in past millennia, and that the second half of the 20^th century was relatively wet with no prolonged droughts (fig. 16). This study provides strong evidence that drought periods are a natural feature of the regional climate and that long-term records are critical for understanding natural variability of climate conditions in the western United States.

Gray et al. (2006), who used woodrat midden and tree-ring data to track the spatial and temporal patterns of pinyon pine (Pinus edulis Engelm.) distribution in the Dutch John Mountains (DJM) of northeastern Utah, showed that the distribution during the Holocene has been strongly controlled by multidecadal precipitation patterns (Gray et al. 2006). The DJM population is an isolated northern outpost of pinyon pine that established ca. 1246. Similar to Utah juniper in the central Rockies, the pinyon pine probably reached the DJM via long-distance dispersal from the Colorado Plateau (Jackson et al. 2005) during the transition from the warmer, drier MCA to the cooler, wetter LIA (fig. 17). DJM pinyon pine expansion stalled in the late 1200s and significant recruitment did not resume until the 14^th century pluvial, when regionally mesic conditions promoted establishment (fig. 17).

Figure 17.

The case of DJM pinyon pine demonstrates the importance of episodic, multidecadal climatic variation in controlling rates of ecological change in the southern Rockies during past millennia. Records suggest that the development of the DJM population was not a steady movement associated with improving climate conditions but rather a markedly episodic invasion regulated by fluctuations in precipitation (Gray et al. 2006). In particular, this example highlights the consequences of having short-term, episodic climatic variation superimposed on centennial to millennial scale climate change, a pattern that can significantly affect species migration and establishment in ways that are more complex than a simple wave-like expansion.

As with previous studies of plant invasions at longer time scales (Lyford et al. 2003), this study suggests that climatic variation can amplify or dampen the probability of survival and reproduction after a species colonizes new areas. Climate can also modify the density and distribution of favorable habitats across the landscape and influence competitive interactions and disturbance processes (Gray et al. 2006). For example, different locations within a region may experience similar changes in average precipitation or temperature during a particular period, yet differences in the variability of precipitation could easily produce different disturbance dynamics and different end states. It is often assumed that vegetation responds to climate change with a steady wave-like movement to better growing conditions, yet the DJM example reveals that species are influenced by other factors, including the dynamics of long-distance dispersal and climate variability at different scales (Gray et al. 2006). Anticipating ecological response to climate change will require a better understanding of how natural climate variability regulates species migrations and invasions at smaller spatial and temporal scales.

Upper Columbia Basin

Climate variation and fire-related sedimentation

Fire-related sediment deposits in central Idaho reveal millennial to centennial scale climate variation and its control on fire regimes. Pierce et al. (2004) dated charcoal in alluvial fan deposits to reconstruct fire-related erosion events in dry forests dominated by ponderosa pine (Pinus ponderosa) and frequent, low-severity fires. They found that small sedimentation events occurred more frequently during the late Holocene, especially during the LIA, and suggested that these were associated with frequent fires of low to moderate severity. Large sedimentation events were associated with prolonged periods of drought and severe fires that were most pronounced during the MCA (Pierce et al. 2004, fig. 18).

These results were compared with a similar record from northern Yellowstone National Park, where mixed conifer forests are associated with infrequent, stand-replacing fires (Meyer et al. 1995, fig. 18). Changes in inferred fire occurrence were synchronous between central Idaho and northern Yellowstone during the warmer, drier MCA. In central Idaho, longer intervals of warm dry conditions allowed for sufficient drying of large fuels to increase the frequency of large stand-replacing canopy fires and fire-related erosion events. Hence, strong climatic controls changed a fuel-limited, infrequent, low-severity fire regime to a fuel-rich, high-severity, stand-replacing fire regime. Large stand-replacing fires also increased in mixed conifer forests of northern Yellowstone during the MCA (fig. 18), coinciding with large pulses of fire-related sedimentation ca. 1150 AD. Meyer et al. (1995) inferred that this activity resulted from increased intensity and interannual variability in summer precipitation. During the LIA, ca. 650–501400-1700 ADcal yr BP, cooler wetter conditions in both northern Yellowstone and central Idaho are inferred to have maintained high-canopy moisture that inhibited canopy fires and facilitated the growth of understory grass and fuels required to sustain frequent low-severity fires in central Idaho (Pierce et al. 2004).
Figure 18. Fire-related sedimentation in the South Fork Payette River (SFP) area in central Idaho (SFP) and Yellowstone National Park (YNP), Wyoming. Probability distributions were smoothed using a 100-year running mean to reduce the influence of short-term variations in atmospheric ¹⁴C but retain major peaks representing the most probable age ranges. The general trend of decreasing probability and overall lower probability before ca. 4000 cal yr BP reflects fewer sites and decreased preservation of older deposits. (a) SFP small events (blue line) are thin deposits probably related to low or moderate severity burns. Note correspondence of peaks with minima in YNP fire-related sedimentation and major peak during the Little Ice Age. Fewer near-surface deposits since 400 cal yr BP were selected for dating because of bioturbation and large uncertainties in ¹⁴C calibration. (b) SFP large events are major debris-flow deposits probably related to severe fires. Note correlation with the YNP record and prominent peak in large-event probability during the Medieval Climate Anomaly. Source: Pierce et al. 2004; reprinted with permission.

Paleoenvironmental investigations like these provide further evidence that millennial and centennial scale variations in temperature and precipitation have influenced biophysical conditions and important disturbance processes across the Upper Columbia Basin and CR-GYA. The results of these and similar studies (e.g., Whitlock et al. 2003) suggest that natural climatic variability acts as a primary control of ecosystem processes, which has important implications for management. Efforts to manage fuels in different stand types to restore specific fire regimes may be trumped by future climate variations, and actively managing for stand conditions that supported what are considered 20^th century structural and disturbance characteristics may have limited impacts under future climatic conditions.

What can we learn from the last 2000 years about decadal and centennial scale climate change?

Paleoenvironmental records of the last 2000 years provide information on decadal and centennial scale climate variability, the associated ecological responses, and the context for anticipating future change. These records suggest that ecosystems have responded to climatic variation primarily through shifts in ranges, ecotonal position, and community composition and structure. High-resolution records indicate that the rate, magnitude, and duration of climate change strongly govern the ecological response. Species respond to changing environmental conditions by moving up or downslope and by increasing or decreasing in density (e.g., treeline “infill”) and abundance. In contrast to the widely held assumption that changes in plant distribution are characterized by steady advances across landscapes, paleoenvironmental evidence suggests that such range adjustments are episodic in response to climatic conditions, occurring rapidly when conditions are suitable and slowly or not at all otherwise (Gray et al. 2006: Lyford et al. 2003). Predicting plant response to future climate change will require consideration of the rate and magnitude of climate change, spatial heterogeneity in biophysical conditions, the catalytic and synergistic role of ecosystem responses (e.g., fire, nutrient cycling, insect outbreaks), and intrinsic biotic limitations (Whitlock and Brunelle 2007: Jackson et al. 2009a).

Paleoenvironmental records from throughout the western United States demonstrate that ecosystem processes are strongly influenced by climate variability occurring at decadal and longer timescales whereas management planning is often based on climate “regimes” that are determined using a few decades of climate data. For example, fuel management goals are often developed from fire patterns over the last century even though paleoenvironmental data show that this only captures a single mode of climate variability (i.e., an extended regime of warm and dry or cool and wet conditions, e.g., Pierce et al. 2004, Whitlock et al. 2003). Records of change in the U.S. Rocky Mountains and Upper Columbia Basin provide context for better understanding current changes and the character of ecosystem response we might expect in the future. In particular, these records highlight:

Decadal to centennial scale climate variability are related to ocean-atmosphere interactions such as ENSO, volcanic eruptions, and solar variability. These events include the Medieval Climate Anomaly (ca. 950–1250), the Little Ice Age (ca. 1400–1700), and numerous decadal and multidecadal droughts and pluvials.

Prolonged droughts in the last millennia, which rival 20^th century droughts in duration and magnitude, influenced vegetation in many ways, including:

• 
  Changes in treeline position.
  
• 
  Increases and decreases in treeline density (e.g., the “infill” phenomenon experienced in some areas during the MCA).
  
• 
  Multidecadal to centennial scale climate variability that results in episodic species dispersal, colonization, and establishment.
  

Human activity was superimposed on climatic variability and altered prehistoric fire regimes in some areas of the western United States, but the impacts were mostly local (Whitlock and Knox 2002). As evidenced by changes in fire regimes of the late 19^th and early 20^th centuries, human activities left a strong imprint on the landscape and can rapidly influence ecosystems and ecosystem processes such as fire.

Key findings:

1. 
   Using the last century as a baseline for climate conditions does not capture important scales of natural climate variability and is often an inadequate reference for considering future climate change.
   

Rapid climate changes and associated ecosystem responses have occurred in the past and will likely occur in the future.

Throughout much of the western United States, the expression of natural variations in the climate system can differ greatly across elevational, latitudinal, and longitudinal gradients and at different spatial and temporal scales. Likewise, the footprint of broad-scale climate changes will vary across finer spatial and temporal scales, and from one area to another. As a direct measurement of climate conditions for the last century, instrumental observations provide the most accurate and reliable data available, but they are influenced by local biases (e.g., differing slopes or aspects) along with the signature of finer-scale processes. Thus, individual instrumental records should not be interpreted as representing the region as a whole, but rather as an indication of local conditions or, at most, conditions at similar locations (e.g., with comparable elevation and vegetation cover).

For all climate regions in this study, 20^th century climate change is characterized by high spatial and temporal variability. At broad spatial and temporal scales, however, it is possible to summarize trends in climatic conditions impacting large portions of the four climate regions. Since 1900, temperatures have increased in most areas of the western United States from 0.5‍ºC to 2ºC (Pederson et al. 2010; Mote 2003; Ray et al. 2008), although cooling has occurred in a few areas, e.g., southeastern Colorado (Ray et al. 2008) and individual sites in the Northwest (CIG 2010). Where temperatures are trending higher, the rate of change varies by location and elevation, but is typically 1ºC since the early 20^th century (Hamlet et al. 2007). In most of the northern portions of the study area, temperatures generally increased from 1900 to 1940, declined from 1940 to 1975, and have increased since then (Parson 2001). Similarly, in the southern Rockies, temperatures generally increased in the 1930s and 1950s, cooled in the 1960s and 1970s, and have consistently increased since then (Ray et al. 2008). The rate of increase for much of the study area doubled since the mid-20^th century, with most of this warming occurring since 1975 (fig. 19).