A 14,000-year record of fire,climate,and vegetation from the Bear River Range,southeast Idaho,USA

Abstract

The vegetation and fire history of the Bear River Range (BRR), Southeast Idaho has been reconstructed from pollen, plant macrofossils, and macroscopic charcoal from lacustrine sediments. Overall, the BRR record shows independent responses of vegetation and fire regime to climate variation. The reconstructions suggest strong seasonal bias from the proxies evaluated, with the pollen record most sensitive to insolation-driven summer temperature trends, and the charcoal-based fire record more sensitive to winter snowpack variability. Together, the proxies suggest that the early Holocene experienced larger than average snowpacks but very warm summers. Warmer than modern summer temperatures were maintained through much of the mid-Holocene, but snowpacks decreased dramatically, creating the most extreme xeric conditions in the Holocene between ~7100 and 6000 BP. After 6000 BP, summers began to show a consistent cooling trend. Winter precipitation remained low until ~4400 BP, after which higher than average snowpacks are indicated until 2000 BP. Pollen and charcoal data relationships at ~8800 BP and from 1800 to 800 BP suggest periods with anomalously wet summers that created a unique fire regime during those intervals.

Keywords

Bear River Range charcoal fire history Great Salt Lake hydroclimate Idaho paleoclimate paleoecology pollen seasonality vegetation Wasatch Mountains

Introduction

Research conducted over the past few decades has identified a complex relationship between climate, vegetation, and fire regime in terrestrial ecosystems. In most cases, climate has been interpreted to concurrently affect both fire regime and vegetation composition through time, with bilateral feedbacks between fire regime and forest composition (e.g. Brunelle and Anderson, 2003; Long et al., 1998, 2011; Prichard et al., 2009). However, in some cases, climate is identified as the primary control on fire regime, independent of changes in vegetation composition (e.g. Carcaillet et al., 2001; Millspaugh et al., 2000). Understanding of climate-fire-vegetation interactions is important given that changes in forest fire activity have been observed in the Western United States (hereafter West) over the past several decades in response to changes in seasonal hydroclimatology (Westerling et al., 2006). Changes to forest composition and health have also been observed or are predicted to occur in response to global climate changes (Williams et al., 2010).

Paleoenvironmental reconstructions provide data that can be used to inform about ranges of natural variability, rates of change, ecological resiliency, and ecosystem sensitivity to climatological forcings and natural disturbance agents (e.g. Froyd and Willis, 2008). Paleoenvironmental records can be used to identify and understand the mechanisms driving changes in the amount and seasonality of precipitation (Woodhouse, 2004), as has been observed across the West in recent decades (Barnett et al., 2004; Carson, 2007; Regonda et al., 2005).

Here, we present a 14,000-year record of fire and vegetation history from Plan B Pond, Bear River Range (BRR), southeast Idaho, based on pollen, macrofossils, and macroscopic charcoal from a lacustrine sediment core. The primary objectives of this research are to provide insights into the response of forest communities and fire regimes to past changes in climate, as well as add a spatial data point from an important climatic transition zone to further refine our understanding of spatiotemporal changes to hydroclimate in the West over the period of record.

Study site

Plan B Pond (42.148°N, 111.579°W) is a small (~2.5 ha) closed-basin pond in the BRR, the northernmost extension of the Wasatch Mountains. The BRR is a primary hydrologic recharge area for the Bear River, the largest river by volume flowing into Great Salt Lake, Utah. The Bear River provides slightly more than half of the total riverine input to Great Salt Lake today.

Plan B Pond is located at 2500 m elevation at the head of Bloomington Canyon, approximately 50 km north-northwest of Logan, Utah (Figure 1). The pond has a drainage basin to lake surface ratio of approximately 9:1, and had a maximum depth of ~1.1 m at the time cores were collected in July 2007. The hydrology of the pond is primarily groundwater-controlled. Plan B Pond is situated within a compound cirque basin, and surrounded by glacial till mapped as Holocene or Latest-Pleistocene in age (Reheis et al., 2009). Bedrock within and around the drainage basin is Paleozoic quartzite and dolomite (Oriel and Platt, 1980).

Figure 1.

Plan B Pond location map.

Modern climate data from the Natural Resource Conservation Service’s automated Snow Telemetry (SNOTEL) station at Franklin Basin (2450 m elevation and approximately 10 km southwest of the study site) show a high degree of seasonality. Over the period 1983–2010, July temperatures on average ranged from 8°C to 24°C, with a mean of 15.5°C. January temperatures, on average, ranged from −11°C to + 1°C, with a mean of −6.5°C. Total precipitation was approximately 1200 mm annually, with just more than 75% of the annual precipitation occurring from October to April, primarily falling as snow. July and August were the driest months, contributing less than 6% of the annual precipitation, combined.

The study site lies along the transition zone of the north–south winter precipitation dipole of the American West, associated with large-scale ocean–atmosphere teleconnection patterns such as El Niño-Southern Oscillation (ENSO; Wise, 2010). Today, the wet/dry winter precipitation anomalies near the study site tend to correlate with the northern side of the dipole.

Modern vegetation at the study site varies significantly with slope and aspect. Much of the open canyon bottom and steep south facing slopes are composed of open meadows and shrub communities with scattered occurrence of Pseudotsuga menziesii (Douglas fir), Pinus flexilis (limber pine), Abies lasiocarpa (subalpine fir), and Picea engelmannii (Engelmann spruce). North-facing slopes are generally composed of more closed Pseudotsuga or mixed Abies/Picea canopies. Lower in the valley, Pinus contorta (lodgepole pine) has a significant presence, as well as both intermixed and dense patches of Populus tremuloides (quaking aspen). Low-growing dwarf Populus tremuloides is also present in the area directly surrounding Plan B Pond. Examples of other plant species identified from the study site in 2007 include Artemisia tridentata (sagebrush), Thalictrum occidentale (western meadowrue), Symphoricarpos oreophilus (mountain snowberry), Penstamon cyananthus (Wasatch beard tongue), Geranium richardsonii (Richardson’s geranium), Oxyria digyna (alpine mountain sorrel), Agastache urticifolia (nettleleaf giant hyssop), and Lonicera involucrata (twinberry honeysuckle).

Materials and methods

Core collection

A 3.98-m sediment core (PBP0701) was collected from the approximate center of Plan B Pond in ~1.1 m water, using a modified Livingstone corer. Sediments were recovered in 1-m increments and extruded in the field. Extruded core sections were visually described, noting color and dominant grain size, then wrapped in plastic wrap and aluminum foil, and stored in wood storage boxes for transport.

An 80-cm short core (PBP0702) was collected to recover the uppermost, unconsolidated sediments using a Klein corer. The short core was extruded in 1-cm intervals into Whirlpak bags in the field prior to transporting back to the lab. All samples were immediately stored under refrigeration upon returning from the field site.

Sediments from Plan B Pond consisted of brown to greenish-brown organic rich gyttja in the upper 3.5 m of sediment. Occasional sand layers of 1–2 mm thickness were also observed. The lower 43 cm of the Livingstone core consisted of dense inorganic grayish-brown sandy clay, consistent with glacially derived sediment inputs of rock flour.

Charcoal and macrofossil methods

Sample preparation for charcoal analysis followed procedures and recommendations described by Brunelle and Anderson (2003). A 5 cc sediment sample was collected from each 1-cm depth interval. The sample aliquots were soaked in a solution of sodium hexametaphosphate (50 g/L) to aid in disaggregation. Because of the presence of abundant clay and unidentified mucilaginous material, the sodium hexametaphosphate solution was only marginally effective. The partially disaggregated samples were then rinsed through nested 250 and 125 µm sieves. Remaining samples from each size fraction were rinsed into gridded Petri dishes and scanned under a 10×–70× magnification microscope for macrofossils. Macrofossil identification was verified using known samples from a reference collection. The two size fractions were then recombined into a glass beaker and soaked in a 10% solution of chlorine bleach, heated to 50–60°C for approximately 10 min to eliminate mucilaginous material, and then rinsed through nested sieves a second time. Once again, the two size fractions were placed in gridded Petri dishes and charcoal particles of each size fraction were counted. For plotting, counts from the two size fractions of charcoal were combined.

Charcoal data analysis procedures generally follow those presented by Long et al. (1998), wherein charcoal counts are deconvolved into a low-frequency background component and a high-frequency peaks component with fire events identified by peaks that exceed a statistically significant threshold above the background. The deconvolution and identification of fire events from charcoal data were done using CharAnalysis software (Higuera et al., 2009). Specific input parameters to the CharAnalysis software are shown in Table 1. Based on the CharAnalysis results, a running fire frequency (fire events/1000 years) and mean fire recurrence intervals (average years between fire events) were calculated.

Table 1.

Input parameters to CharAnalysis software for analysis of Plan B pond charcoal record.

	Parameter	Value used	Unit
Pretreatment	Years interpolated	10	Year
Pretreatment	Log transform	None
Smoothing	Method	Lowess smoother
Smoothing	Background smoothing	500	Year
Peak analysis	Peak identification (residuals/ratios)	Ratios
	Threshold type (local/global)	Local
	Threshold method	Base threshold values on a percentile cut-off of a noise distribution determined by a Gaussian mixture model.
	Threshold values	0.990	Percentile of noise distribution
Peak analysis results	Smoothing for fire frequency and return interval	1000	Year

Pollen methods

Pollen sample preparation followed a method modified from Faegri et al. (1989), using a Lycopodium spore tracer (Stockmarr, 1971). Samples were concentrated from 1 cc of sediment, and collected from the core at intervals providing approximately equal age distribution, generally at multiples of 4 cm within the core. Actual sampling interval ranged from 4 to 32 cm (mean ≈ 10 cm), providing a mean temporal resolution of 362 years between samples. On average, pollen samples integrate approximately 34 years with a standard deviation of 11 years. A total of 39 samples were counted. Concentrated pollen samples were preserved in silicone oil. Each sample was counted at 500× magnification. An average of 307 terrestrial grains was counted from each sample, with a minimum sample size of 278 pollen grains for all but two samples, which had very low pollen recovery. Pollen data are presented in relative percentages of the terrestrial taxa sum. Botryococcus algal palynomorphs are calculated as the number of colonies/300 terrestrial grains. CONISS (Grimm, 1987) was used to perform stratigraphically constrained cluster analysis to help inform pollen zonation. Pollen zones are numbered in stratigraphic order, oldest first.

Chronology

Chronological constraints are provided by four radiocarbon dates and a ²¹⁰Pb series measured from the upper 20 cm of core sediments (Table 2). An age model (Figure 2) was created from the ¹⁴C and ²¹⁰Pb data by fitting a best-fit smoothing spline to the data using CLAM software and integrated IntCal 13.14 radiocarbon calibration. The ²¹⁰Pb calendar ages are based on a terminated Constant Rate of Supply (CRS) model that uses the scaling option in the CRS calculation to assign an age of 1962 CE to the depth with highest ¹³⁷Cs content (James Budahn, 2009, personal communication). All ages reported as before present (BP) are in calibrated Calendar Years BP unless specifically noted otherwise. Following standard radiocarbon protocol, present = 1950 CE.

Table 2.

Data used to construct age–depth model for Plan B Pond sediment cores. Calendar age (Cal Age) range for ¹⁴C data points given at 95% confidence interval. The ²¹⁰PB calendar ages are based on a terminated Constant Rate of Supply (CRS) model that uses the scaling option in the CRS calculation to assign an age of 1962 CE to the depth with highest ¹³⁷Cs content.

Depth (cm)	Type	Lab/ID #	Material dated	²¹⁰Pb Age	¹⁴C Age/1−σ error (¹⁴C yr BP)	Cal Age range (cal. yr BP)
0–2	²¹⁰Pb	USGS/–	Sediment	−54.7
2–4	²¹⁰Pb	USGS/–	Sediment	−49.2
4–6	²¹⁰Pb	USGS/–	Sediment	−42.8
6–8	²¹⁰Pb	USGS/–	Sediment	−35.1
8–10	²¹⁰Pb	USGS/–	Sediment	−26
10–12	²¹⁰Pb	USGS/–	Sediment	−15.1
12–14	²¹⁰Pb	USGS/–	Sediment	−3.5
14–16	²¹⁰Pb	USGS/–	Sediment	6.9
16–18	²¹⁰Pb	USGS/–	Sediment	15.8
18–20	²¹⁰Pb	USGS/–	Sediment	51.1
49	¹⁴C	UGAMS/03981	Conifer Needles		910/24	767–913
125	¹⁴C	UGAMS/02397	Wood		3540/30	3719–3900
248	¹⁴C	UGAMS/02398	Wood		8400/40	9305–9500
358	¹⁴C	UGAMS/02399	Conifer Needles		10990/70	12723–13019

Figure 2.

Plan B Pond age–depth model. Black line indicates best-fit smoothing spline model used. Gray shading indicates range of error at 95% confidence interval.

Results

Charcoal

Smoothed background charcoal accumulation rates (Figure 3b) are consistently low from the beginning of the record to approximately 7000 years ago. Periods of extremely low charcoal accumulation occur at ~12,500 BP and again between 10,000 and 8,000 BP. After 7000 BP, background charcoal accumulation shows a steady increase until around 1800 BP. From 1800 to 700 BP, charcoal background shows a steady decline to values similar to those seen in the early Holocene. After ~700 BP, charcoal background abruptly returns to higher values and remains high up to modern.

Figure 3.

Plan B Pond fire regime data: (a) Ratio of arboreal to non-arboreal pollen (AP/NAP); (b) charcoal accumulation rate (thin line) and 500-year smoothed background (thick line); (c) charcoal peak occurrence and magnitude; (d) 1000-year smoothed fire frequency; (e) Botryococcus abundance.

Based on the analyses of the peaks component of the charcoal data (Figure 3c and d), long-term fire frequency averages 3.33 fires/1000 years, giving a mean fire return interval of 300 years. Fire frequency is lower than the long-term average over the intervals 13,300–11,800 BP, 9200–7100 BP, 4400–2000 BP, and 1600–300 BP. Fire frequency is higher than average over the intervals 13,900–13,300 BP, 11,800–9200 BP, 7100–4400 BP, 2000–1600 BP, and from 300 BP to modern. The magnitude of charcoal peaks is generally lower in the early half of the record than the latter half, with 7 of the 10 largest peaks occurring after 7000 BP.

Pollen and macrofossils

Pollen results from Plan B Pond (Figure 4), for the most part, show gradual trends in taxa abundance over time, rather than discrete periods bracketed by abrupt composition changes. However, as is common practice with pollen data, the pollen spectra are delineated into discrete pollen zones to simplify discussion of the data. Six pollen zones were defined using CONISS (Grimm, 1987). The uppermost pollen zone (150 BP – modern) likely reflects historical human impacts from livestock grazing, timber harvests, and other activities that are anomalous within the context of the rest of the Holocene record. Therefore, significant discussion of pollen zone 6 has been omitted.

Figure 4.

Plan B Pond pollen and macrofossil data. Pollen data presented as relative percent of total terrestrial taxa pollen sum. White fill on some taxa represents 5x exaggeration for readability. Horizontal lines indicate pollen zones based on stratigraphically constrained cluster analysis. Shaded horizontal bars indicate periods higher than average fire frequency based on charcoal data in Figure 3. Macrofossil data points indicate presence in given sample.

Pollen Zone 1 (13,900–11,100 BP) shows high relative abundances of Artemisia, Pinus, Picea, and Other Asteraceae pollen (exclusive of taxa in Asteraceae that have been identified to genus level). The last few hundred years included in Zone 1 begin to show a modest but notable increase in the abundance of Ambrosia, Pseudotsuga, and Amaranthaceae. The overall ratio of arboreal to non-arboreal pollen (AP/NAP) is quite low over this interval (Figure 3a), and abundance of the algal palynomorph Botryococcus is the lowest at any time in the record (Figures 3e and 4). Macrofossil data confirm the local presence of Abies, Picea, and 5-needle pine by 13,000 BP. It has been assumed that all 5-needle pine macrofossils in our record are from Pinus flexilis as no other 5-needle pines are present in the area today.

Pollen Zone 2 (11,100–9300 BP) continues to be dominated by high abundances of Pinus and Artemisia, although Artemisia declines steadily throughout the interval. Picea declines to its lowest level in the record at the beginning of the zone, and remains uniformly low for the duration of the interval. Abies also reaches its lowest levels in this zone. In contrast, Ambrosia reaches its highest abundance in Zone 2. Amaranthaceae shows a small but steady increase over the interval, and the overall AP/NAP ratio remains low and steady. Botryococcus increases through the zone, reaching a localized peak just after 9500 BP. Macrofossils of Picea, Abies, and Pinus flexilis all continue to be present in Zone 2, although the occurrence frequency of Picea needles is lower than in Zone 1, while the occurrence frequency of Pinus flexilis needles is higher. Pseudotsuga macrofossils are also present regularly in Zone 2, in contrast to their absence in Zone 1.

Pollen Zone 3 (9300–6000 BP) shows a decline in Pinus abundance compared with the previous two zones. Artemisia remains high with a slight increasing trend. Ambrosia remains near the high levels seen in Zone 2, with a very slight declining trend. Abies and Picea both show a gradual increase over the duration of Zone 3. Amaranthaceae is consistently high in Zone 3, reaching its highest abundance during this time period, along with Sarcobatus. Despite the first appearance or more consistent presence of a few arboreal taxa such as Quercus, Populus, Acer, and Juniperus, the AP/NAP declines to its lowest value in Zone 3 (Figure 3a). Botryococcus initially declines in abundance in Zone 3, reaching a local minimum just before 8000 BP. After 8000 BP, Botryococcus abundance increases sharply until 6800 BP, and then levels off. Macrofossils show continued steady occurrence of Abies needles, and regular but less frequent occurrences of Picea and Pseudotsuga. Pinus flexilis needles also occur on a more sporadic basis. Macrofossil evidence for Pinus and Pseudotsuga is absent after 7000 BP.

Pollen Zone 4 (6000–2600 BP) is a period of marked transition. Abies shows a steady increase along with a more stepwise increase in Picea. Artemisia shows a steady decline, as does Ambrosia, although the Ambrosia decline is not as pronounced as that of Artemisia. Amaranthaceae remains near its highest levels in the record. The AP/NAP ratio shows an increasing trend, especially after 4000 BP. Botryococcus abundance hits its highest level in the Holocene at the beginning of Zone 4, just after 6000 BP, then decreases sharply until 4700 BP. After 4700 BP, Botryococcus continues its decline, but at a much slower rate until 2500 BP. Picea and Abies macrofossils are present throughout Zone 4, but no macrofossil evidence for Pseudotsuga or Pinus flexilis is present in the zone. One needle from a 2-needle pine (likely Pinus contorta) occurs at ~2800 BP, and represents the only macrofossil from a 2-needle pine in the entire record.

In Pollen Zone 5 (2600 BP–150 BP), Pinus abundance returns to high levels similar to those seen in Zone 1 and Zone 2. Artemisia continues its decline, while Abies and Picea continue to increase in abundance. Pseudotsuga is slightly higher than in the previous two pollen zones, while Ambrosia and Amaranthaceae are lower in this period. A very noticeable increase in Cyperaceae also occurs in Pollen Zone 5. The AP/NAP ratio increases to its highest value in the record in Pollen Zone 5. Botryococcus abundance in Zone 5 initially increases to a local maximum at 1700 BP, before dropping back nearly to levels seen at the 2500 BP minimum. Macrofossil evidence is dominated by Abies needles in Zone 5. Occurrences of Picea, Pinus, and Pseudotsuga macrofossils are limited to one or two samples each, with the Pinus flexilis and Pseudotsuga occurring early in the zone and the Picea occurring late.

Discussion

Vegetation and insolation

Overall, vegetation changes throughout the Plan B Pond record show a close relationship with mid- to late summer (July–August) insolation (Berger and Loutre, 1991). This relationship is clearly evident when the data are plotted as the ratio of cool/moist (CM) to warm/dry (WD) indicator taxa (Figure 5). Here, we use Picea + Pseudotsuga + Abies + Other Asteraceae as the CM taxa and Juniperus + Quercus + Ambrosia + Amaranthaceae + Sarcobatus as the WD taxa with the ratio calculated as [(CM – WD):(CM + WD)]. These taxa are an amalgamation of those used by Davis (1998) to define ‘glacial’ versus ‘interglacial’ taxa in Great Salt Lake pollen records and those used by Jiménez-Moreno et al. (2007) to define ‘cold’ versus ‘warm+dry’ in Bear Lake pollen records. Great Salt Lake and Bear Lake are both in close proximity to Plan B Pond, Bear Lake lying to the east-southeast and Great Salt Lake to the southwest (Figure 1).

Figure 5.

Plan B Pond Cool + Moist (CM) Taxa: Warm + Dry (WD) Taxa Pollen Ratio. Calculated as [(CM − WD)/(CM + WD)], where CM = (Picea + Pseudotsuga + Abies + other Asteraceae), and WD = (Juniperus + Quercus + Ambrosia + Amaranthaceae + Sarcobatus). Pollen zones from Figure 4 are shown by vertical lines. Shaded bar indicates Younger Dryas chronozone. Shaded bars indicate previously published Younger Dryas and MCA chronozones discussed in text.

With the exception of a notable and distinct CM excursion coincident with the Younger Dryas chronozone (12,900–11,600 BP, Rasmussen et al., 2006), the pollen ratio in Figure 5 generally follows mid- to late summer (July + August) insolation changes, and suggests that summer temperatures are the dominant variable controlling vegetation composition for most of the Holocene, either directly or indirectly by affecting evapotranspiration rates and effective moisture. This is similar to the conclusions of Power et al. (2011), which also found that vegetation in high elevation sites in the Northern Rocky Mountains was most sensitive to summer temperature, especially in the early to middle Holocene when summer insolation was highest.

Summer temperatures

The direct importance of summer temperature is particularly well demonstrated by the taxa Picea and Ambrosia. Throughout the record, relative abundances of Picea and Ambrosia generally behave in an offsetting manner, with one increasing at the apparent expense of the other. An examination of published relationships between climate variables and relative pollen abundance in modern sediments (Williams et al., 2006) shows that mean July temperature is the one climate variable where Picea and Ambrosia show little overlap in their distribution, each with abrupt thresholds at approximately 17.5°C (Figure 6a). Picea shows a sharp increase in abundance when mean July temperatures are below this threshold and Ambrosia increases sharply when temperatures are above the threshold. Currently, mean annual temperature at Plan B Pond is ~15.5°C, supporting a high ratio of Picea to Ambrosia (Figure 6b).

Figure 6.

Picea and Ambrosia temperature distributions and abundance ratio. (a) Scatter plot of pollen abundance versus mean July temperature from the North American Pollen Atlas (Williams et al., 2006). Black dashed line indicates modern mean July temperature at Plan B Pond. Gray dotted line indicates threshold temperature at approximately 17.5°C. (b) Ratio [(Picea − Ambrosia):(Picea + Ambrosia)] from Plan B Pond. Pollen zones from Figure 4 are shown by vertical lines. Shaded bars indicate previously published Younger Dryas and MCA chronozones discussed in text.

The decline in Picea to minimal levels at ~10,500 BP and lasting until 9500 BP, as Ambrosia was reaching its peak Holocene abundance (Figure 4), may suggest that summer temperatures at Plan B Pond peaked near or above the 17.5°C threshold, equal to or greater than 2°C warmer than modern. This is consistent with midge-based summer temperature reconstructions from Windy Lake, SE British Columbia, that suggest peak temperatures 3–4°C warmer than modern from 10,500 to 9000 BP (Chase et al., 2008).

Based on the CM:WD ratio, near-peak or at least warmer than modern summer temperatures were sustained throughout Pollen Zones 2 and 3 (11,100–6000 BP). This is also consistent with the midge inferred temperature record of Chase et al. (2008) which shows continued warmer than modern temperatures by ~2°C until ~6500 BP. However, the bulk of this period of maximum summer warmth at Plan B Pond (from ~11,100–7100 BP) is characterized by lower than average fire frequency. This suggests that while summer temperatures have been the dominant driver of vegetation change, other variables may have been more important for determining fire activity, despite the large effects that such warm summer temperatures would have had on evapotranspiration and fuel/soil-moisture.

Background charcoal and fuels

Marlon et al. (2006) showed background charcoal accumulation rate is largely controlled by the availability of woody fuel (proportion of woody taxa on the landscape), and is generally unaffected by changes in fire frequency or sedimentation rate. A comparison of charcoal accumulation rate and AP/NAP from Plan B Pond confirms this relationship (Figure 3a and b) at our site, given that arboreal species represent the bulk of woody fuel. Background charcoal trends correspond well with the fuel available on the landscape, with the exception of a few notable departures such as ~8800 BP and just after 1000 BP when fuel availability increases as charcoal background decreases. These unique time intervals may represent periods of time when climatic controls limited fire activity and allowed woody fuels to accumulate. Fire frequency is in decline during each of the two anomalous intervals and suggests a more or less synchronous response between fire activity and vegetation composition, or a perhaps a causal response with fire causing a vegetative change. However, the relationship at these two discrete time periods is not representative of the record as a whole.

Charcoal peaks/fire frequency and winter snowpack

Today, fire frequency in the West is strongly linked to the length of fire season, a function of the timing of spring snowmelt (Westerling et al., 2006). The timing of spring snowmelt is affected by both spring temperatures and winter precipitation, but Westerling et al. (2006) found that when comparing early and late spring snowmelt years in the West, there was a strong association between warm spring temperatures and reduced winter precipitation. This association was also noticed by Cayan (1996) when exploring the interannual variability in snowpack across the West. Thus, the two variables, winter precipitation and spring temperature, do not tend to behave independently. Rather, one can infer that periods of higher than average fire activity likely exhibit both low snowpack and warmer than average spring temperatures, while low fire activity is associated with higher than average snowpacks and cool spring temperatures.

Based on this relationship between fire activity and hydroclimatic conditions, we hypothesize that the time periods 13,300–11,800 BP, 9200–7100 BP, 4400–2000 BP, and 1600–300 BP were characterized by higher than average snowpacks and cool springs, while the periods 13,900–13,300 BP, 11,800–9200 BP, 7100–4400 BP, 2000–1600 BP, and from 300 BP to modern were characterized by lower than average snowpacks and warm springs.

The accuracy of our overall fire frequency reconstruction is bolstered by an obvious correlation between Botryococcus abundance and our reconstructed fire frequency curve (Figure 3d and e). The Botryococcus generally increases during periods of high fire activity (dry winters/warm springs) and declines during periods of low fire activity (wet winters/cool springs). The correspondence of fire activity and Botryococcus abundance is most likely explained by a common dependence of each variable on winter snowpack. Groundwater hydrology in the BRR is snowmelt dominated (Spangler, 2001), and lake level in Plan B Pond is primarily groundwater-controlled, so it is expected that during periods of higher than average snowpacks lake level would remain higher. In addition, water temperatures would likely be cooler, and the ice-free period would be shorter. In lacustrine deposits, especially from small lakes, Botryococcus is often an inverse indicator of water level (Clausing, 1999), and has been shown to be weakly, but positively correlated with temperature (Rull et al., 2008). Thus, higher than average snowpacks at Plan B Pond would likely lead to reduced Botryococcus abundance. Decreased fire activity associated with high snowpacks could also limit nutrient delivery to the pond as compared with periods of high fire activity, so some of the correlation between fire activity and Botryococcus may in fact be causal.

Summer precipitation and fire

One might argue that an alternative explanation for the low fire frequency in the BRR during the early Holocene was increased summer precipitation because of enhanced convective precipitation (‘summer monsoon’) resulting from higher than modern summer insolation. This mechanism has been widely applied to explain climate, fire, and vegetation reconstructions across much of the Western United States (e.g. Brunelle et al., 2005; Whitlock and Bartlein, 1993; Whitlock et al., 1995). However, in the course of said investigations, it has consistently been shown that the effective spatial footprint of summer monsoonal precipitation has remained fixed throughout the Holocene, such that sites which today lack significant summer precipitation (‘summer dry/winter wet’ sites in Whitlock and Bartlein, 1993) were not significantly affected by summer precipitation in the early Holocene either. Rather than a change in spatial footprint, the summer precipitation gradient increased between the ‘summer-wet’ and ‘summer-dry’ sites (Brunelle et al., 2005; Whitlock et al., 1995). As Plan B Pond is a definitive summer-dry site today, we do not consider intensified summer convective/‘monsoon’ activity in the early Holocene a likely cause of reduced fire activity from 9200–7100 BP.

Climate/fire/vegetation relationships

Overall, the data from Plan B Pond suggest that pollen assemblages and macroscopic charcoal data are each primarily explained by one dominant climatic variable, but not the same dominant climatic variable. Vegetation composition as indicated by pollen assemblages is most responsive to mid- to late summer temperature, a function of summer insolation. In some cases, temperature itself may have been limiting, as discussed previously with the taxa Picea and Ambrosia. In other cases, the effect of temperature on evapotranspiration and effective growing season moisture may have been more important. Variability in fire frequency at Plan B Pond is independent of vegetation change, and most likely has been controlled by the length of the fire season throughout the Holocene.

Holocene paleoenvironments

Since fire season is a function of winter precipitation and spring/early summer temperature, low fire frequency during much of the early Holocene (9200–7100 BP) suggests that winter precipitation was higher than the long-term average during that time, offsetting the enhanced evapotranspiration that would be expected during the peak of summer insolation.

With higher than average winter snowpack partially or completely offsetting higher summer temperatures, we speculate that peak hydrologic drought conditions would have occurred during the interval of time when summer temperatures were still near their peak (before ~6000 BP), but after winter precipitation decreased (~7100 BP) from the high levels of the early Holocene. This approximately 1000-year window corresponds well with an intense dry episode recorded in nearby Minnetonka Cave from 7200 to 6200 BP (Lundeen et al., 2013) and coincides or just slightly lags many regional paleoenvironmental indicators of intense drought conditions. For example, the Great Salt Lake is thought to have nearly desiccated between approximately 6800 and 7500 BP (6700–6000 ¹⁴C yr BP; Murchison, 1989); Lundeen (2001) documented an interval of lake desiccation in the Sawtooth Mountains of Central Idaho from ~7500 to 7000 BP (6800–6200 ¹⁴C yr BP), and both palynological and isotopic indicators from Pyramid Lake, Nevada, suggest the dominance of intensely dry conditions from 7500 to 6300 BP (Benson et al., 2002; Mensing et al., 2004). Model results also show minimum Holocene moisture index values, a measure of annual effective precipitation for plant growth, occurred from ~7800 to 5700 BP (7000–5000 ¹⁴C yr BP) in the northeastern Great Basin (Broughton et al., 2008).

Beginning at ~6000 BP, pollen data suggest the initiation of a consistent trend toward cooler than previous summer conditions and increased effective moisture, consistent with declining summer insolation. This can be seen as an inflection point in the AP/NAP (Figure 3a), the CM:WD ratios (Figure 5), and as a dramatic shift in the Picea:Ambrosia ratio (Figure 6). This inferred cooling trend is consistent with several other regional records that indicate a CM trend after approximately 6000 BP, or shortly thereafter (e.g. Beiswenger, 1991; Munroe, 2003; Reinemann et al., 2009). Although the effective moisture was generally increasing over the entirety of Pollen Zone 4 (6000–2600 BP), higher than average fire frequency continued until 4400 BP, suggesting winters were still relatively dry up to that point. A stepwise increase in Picea abundance, a drop in Botryococcus, and a slight increase in herbaceous pollen also occur at ~4400 BP all indicating a shift to wetter conditions.

Based on the fire frequency reconstruction, wetter than average winters persisted from ~4400 BP, through the remainder of the Pollen Zone 4 chronozone, and into Pollen Zone 5, finally returning to drier conditions ~2000 BP. Pollen data suggest a continued cooling trend after 4500 BP, and an overall increase in effective moisture. The transition from Pollen Zone 4 to Pollen Zone 5 at ~2600 BP marks a shift to a climatic regime with a more stable overall trend, but more point-to-point variability than seen in Zone 4. This shift is most evident in the CM:WD pollen ratio (Figure 5). In addition to the increased variability in the pollen ratio, a sharp increase in wetland taxa pollen (mostly Cyperaceae) can be seen after 2600 BP. We interpret the increased Cyperaceae after 2600 BP to be indicative of a more variable lake level that would consistently create new lake-margin habitat for Cyperaceae in both transgressive and regressive phases.

Increased climatic variability in the late Holocene is consistent with the results of several studies, especially as it relates to variability in ENSO (e.g. Barron and Anderson, 2011; Conroy et al., 2008; Moy et al., 2002). The primary effects of ENSO variability in the study region are changes to the amount of winter precipitation. Specifically, winters in negative ENSO phases (La Niña) tend to be wetter than average and winters in positive ENSO phases (El Niño) tend to be drier than average (Cayan, 1996; Jain and Lall, 2000). Reconstruction of ENSO activity by Moy et al. (2002) and Conroy et al. (2008) both suggest that the frequency of El Niño events increased in the late Holocene, with the highest frequency at ~2000 BP (Conroy et al., 2008) or shortly thereafter (Moy et al., 2002). Based on a compilation of records, Barron and Anderson (2011) suggest that ENSO variability in the eastern North Pacific began to intensify after ~4000 BP in Southern California, followed by a northerly migration of the ENSO teleconnection pattern over the next couple of millennia.

Underlying likely intensified climate variability, Pollen Zone 5 is reflective of effectively wetter conditions, as compared with the three previous pollen zones. This is reflected by the highest AP/NAP values in the record, driven in part by high contributions of pine in conjunction with declining Artemisia (Figure 7). The ratio of Pinus:Artemisia is often interpreted as being an indicator of lower treeline in montane systems, primarily controlled by effective moisture (Fall, 1997; Jiménez-Moreno et al., 2011). A declining trend in summer insolation explains part of the effective moisture increase; however, the changes in winter precipitation because of increased ENSO variability are likely to have impacted fire regime, lake level, and vegetation at shorter timescales, making paleoclimatic interpretations more complex in Pollen Zone 5 than earlier portions of the record.

Figure 7.

Pinus/Artemisia ratio. Ratio is calculated as [(Pinus − Artemisia)/(Pinus + Artemisia)]. Higher values are interpreted to indicate lowering of lower treeline and effectively wet conditions.

Medieval Climate Anomaly

One sub-interval in Pollen Zone 5 (~1800–700 BP) has a particularly interesting suite of proxy indicators, suggesting unique environmental conditions at that time. This sub-interval at Plan B Pond includes the time period widely recognized as the Medieval Climate Anomaly (MCA), occurring ~1000–700 BP (Mann et al., 2009). Anomalous conditions at Plan B Pond begin to occur considerably earlier than the traditionally accepted onset of the MCA, but the peak temperature and moisture anomalies do occur during the MCA, as discussed below.

Background charcoal accumulation at Plan B Pond declines from 1800 to 700 BP, despite the AP/NAP ratio (indicating fuel availability) reaching its highest values between 1500 and 700 BP (Figure 3a and b). The disconnect between fuel availability and background charcoal production suggests that fires were either limited to small ground fires rather than larger magnitude crown fires or that fires were completely absent in the study area. In either case, wet conditions would be inferred. The ratio of Pinus:Artemisia also increases to peak values between 1500 and 700 BP, suggesting effectively wet conditions. However, increased Botryococcus suggests reduced winter snowpack and/or warm spring/early summer temperatures, especially during the first portion of the sub-interval. The Picea:Ambrosia ratio also shows an increase at this time, indicating a period of warm summer temperatures disrupting a general cooling trend.

Together, these data suggest that from 1800 to 700 BP, but especially from 1600 to 700 BP, climate was warmer and effectively wet overall. Because at least part of this interval shows evidence of reduced snowpack, we infer that the time period must have experienced anomalously high summer precipitation. This interpretation is supported by a correspondence with the isotope record from carbonates in Bison Lake, Colorado, that also suggests a shift in precipitation regimes, with greater relative importance of rain from ~1600 to 600 BP (Anderson, 2011). The possibility of significantly increased summer precipitation in the late Holocene is at odds with our earlier dismissal of intensified monsoonal precipitation in the early Holocene. However, the combination of proxy indicators in the early Holocene does not indicate the same conditions seen from 1800 to 700 BP with the possible exception of a few centuries ~8800 BP. Thus, if summer precipitation did increase in the early Holocene, it does not appear that it sufficiently offset the higher rates of evapotranspiration associated with higher insolation and summer temperatures, and did not greatly affect moisture availability for vegetation.

Effective moisture likely increased further after ~1500 BP when a drop in fire frequency and Botryococcus suggest snowpacks returned to normal or above normal until ~300 BP. Together with increased summer precipitation, one could infer that the time period from 1500 to 700 BP would have been exceptionally wet, with long moist summers ideal for plant productivity. It is likely not a coincidence that this interval corresponds with the peak occupation period of the Fremont Culture in the northern Great Basin and adjacent areas (Coltrain and Leavitt, 2002). Other local to regional evidence of extreme wetness during this period includes faunal evidence at Homestead Cave in the Bonneville Basin, suggestive of freshening and/or increased water levels in the Great Salt Lake (Broughton et al., 2000), and a growth hiatus in the Minnetonka Cave record from ~1750 to 750 BP, interpreted to indicate anomalously wet conditions (Lundeen et al., 2013).

After ~800 BP, the declining Pinus:Artemisia ratio at Plan B Pond suggests more xeric conditions than in the preceding millennia. This is corroborated by an increasing trend in fire frequency from 700 BP to modern, and is synchronous with the abandonment of many Fremont sites throughout the region.

Conclusion

Prior to ~13,000 BP, evidence from Plan B Pond suggests that small cirque glaciers were still present in Bloomington Canyon. By 13,000 BP, macrofossil evidence shows local presence of Picea, Abies, and Pinus flexilis, and pollen ratios indicate that conditions were ameliorating. A distinct CM excursion can be observed in Plan B Pond pollen data, coincident with the Younger Dryas period. By ~11,500 BP, climate in the BRR had mostly returned to its pre-Younger Dryas trend of warming temperatures.

Summer temperatures continued to climb into the early Holocene, driven by increases in summer insolation. Peak summer temperatures were reached just after 10,500 BP, and may have been >2°C warmer than modern. Summer temperatures remained close to 2°C warmer than modern until ~6000 BP. Despite warm summers and the expected effects of increased evapotranspiration, annual moisture balance was partially offset by larger than average snowpacks that kept fire activity low until ~7100 BP. Peak hydrologic drought in the BRR would have occurred from approximately 7100 to 6000 BP when summer temperatures were still much higher than modern, and no longer offset by higher than average winter precipitation.

After ~6000 BP, declining summer temperatures drove effective moisture up, initiating a steady increase in the relative proportion of arboreal vegetation; this, despite lower than average snowpacks until ~4400 BP. From 4400 to 2000 BP, increased winter precipitation kept fire frequency lower than average, and effective moisture continued to climb. After ~2600 BP the overall climate variability may have increased at the study site, consistent with an intensification of ENSO activity. Continued declines in summer insolation, driving a trend toward effectively wetter conditions, were overlain by higher frequency oscillations in winter precipitation.

An anomalous warm, but wet, period began ~1800 BP, and climaxed between 1500 and 700 BP, coincident with the MCA. This warm, wet interval was unique in that the extra moisture was seemingly not from excess snowpack, but rather summer precipitation. After 700 BP, data suggest cooler but drier conditions at Plan B Pond than during the MCA.

The primary objectives of this study were to provide insights into the response of forest communities and fire regimes to past changes in climate, as well as add a spatial data point from an important climate transition zone to further refine our understanding of spatiotemporal changes to hydroclimate in the West over the period of record. The paleoenvironmental record from Plan B Pond, BRR, southeast Idaho, offers a nearly 14,000-year record of vegetation, fire, and climate relationships. Throughout the record, the dominant vegetation changes are driven primarily by summer insolation, either directly through phytological temperature limits, or indirectly by affecting rates of evapotranspiration. In a select few instances at ~8800 and 1000 BP, climatically controlled reductions in fire activity may have influenced vegetation composition. Otherwise, fire frequency and dominant vegetation change have largely been independent of one another in the BRR. Fire frequency is primarily driven by the magnitude of winter snowpack and spring/early summer temperatures, which both affect the timing of spring snowmelt and the length of the fire season.

Footnotes

Acknowledgements

We would like to thank Vachel Carter, Shawn Blissett, Jessica Spencer, Doug Bird, Carrie Spruance, Sharon Leopardi, and Stephanie Cobabe for assistance in the field and lab. Ken Peterson, Jesse Morris, Vachel Carter, and Jennifer Watt provided valuable feedback at various stages of the project.

Funding

Funding was provided in part by NSF Doctoral Dissertation Improvement Grant# 0926715.

References

Anderson

(2011) Holocene record of precipitation seasonality from lake calcite δ¹⁸O in the central Rocky Mountains, United States. Geology 39(3): 211–214.

Barnett

Malone

Pennell

. (2004) The effects of climate change on water resources in the West: Introduction and overview. Climatic Change 62: 1–11.

Barron

Anderson

(2011) Enhanced late Holocene ENSO/PDO expression along the margins of the eastern North Pacific. Quaternary International 235(1–2): 3–12.

Beiswenger

(1991) Late quaternary vegetational history of Grays Lake, Idaho. Ecological Monographs 61: 165–182.

Benson

Kashgarian

Rye

. (2002) Holocene multidecadal and multicentennial droughts affecting Northern California and Nevada. Quaternary Science Reviews 21: 659–682.

Berger

Loutre

(1991) Insolation values for the climate of the last 10 million of years. Quaternary Sciences Review 10(4): 297–317.

Broughton

Madsen

Quade

(2000) Fish remains from Homestead Cave and lake levels of the past 13,000 years in the Bonneville Basin. Quaternary Research 53(3): 392–401.

Broughton

Byers

Bryson

. (2008) Did climatic seasonality control late Quaternary artiodactyl densities in western North America? Quaternary Science Reviews 27(19): 1916–1937.

Brunelle

Anderson

(2003) Sedimentary charcoal as an indicator of late-Holocene drought in the Sierra Nevada, California, and its relevance to the future. The Holocene 13(1): 21–28.

10.

Brunelle

Whitlock

Bartlein

. (2005) Holocene fire and vegetation along environmental gradients in the Northern Rocky Mountains. Quaternary Science Reviews 24: 2281–2300.

11.

Carcaillet

Bergeron

Richard

. (2001) Change of fire frequency in the eastern Canadian boreal forest during the Holocene: Does vegetation composition or climate trigger the fire regime? Journal of Ecology 89: 930–946.

12.

Carson

(2007) Temporal and seasonal trends in streamflow in the Uinta Mountains, Northeastern Utah, U.S.A. Arctic, Antarctic, and Alpine Research 39(4): 521–528.

13.

Cayan

(1996) Interannual climate variability and snowpack in the Western United States. Journal of Climate 9: 928–948.

14.

Chase

Bleskie

Walker

. (2008) Midge-inferred Holocene summer temperatures in Southeastern British Columbia, Canada. Palaeogeography, Palaeoclimatology, Palaeoecolology 257: 244–259.

15.

Clausing

(1999) Palaeoenvironmental significance of the green alga Botryococcus in the lacustrine rotliegend (upper carboniferous – lower permian). Historical Biology: An International Journal of Paleobiology 13(2–3): 221–234.

16.

Coltrain

Leavitt

(2002) Climate and diet in Fremont prehistory: Economic variability and abandonment of maize agriculture in the Great Salt Lake basin. American Antiquity 67: 1–33.

17.

Conroy

Overpeck

Cole

. (2008) Holocene changes in eastern tropical Pacific inferred from a Galápagos lake sediment record. Quaternary Science Reviews 27: 1166–1180.

18.

Davis

(1998) Palynological evidence for vegetation cycles in a 1.5 million year pollen record from the Great Salt Lake, Utah, USA. Palaeogeography, Palaeoclimatology, Palaeoecology 138: 175–185.

19.

Faegri

Iverson

Kaland

. (1989) Textbook of Pollen Analysis. Caldwell, NJ: Blackburn Press.

20.

Fall

(1997) Timberline fluctuations and late Quaternary paleoclimates in the Southern Rocky Mountains, Colorado. Geological Society of America Bulletin 109: 1306–1320.

21.

Froyd

Willis

(2008) Emerging issues in biodiversity & conservation management: The need for a paleoecological perspective. Quaternary Science Reviews 27: 1723–1732.

22.

Grimm

(1987) CONISS: A FORTRAN 77 program for stratigraphically constrained cluster analysis by the method of incremental sum of squares. Computers and Geosciences 13: 13–35.

23.

Higuera

Brubaker

Anderson

. (2009) Vegetation mediated the impacts of postglacial climatic change on fire regimes in the south-central Brooks Range, Alaska. Ecological Monographs 79: 201–219.

24.

Jain

Lall

(2000) Magnitude and timing of annual maximum floods: Trends and large-scale climatic associations for the Blacksmith Fork River, Utah. Water Resources Research 36(12): 3641–3651.

25.

Jiménez-Moreno

Anderson

Fawcett

(2007) Orbital- and millennial-scale vegetation and climate changes of the past 225 ka from Bear Lake, Utah-Idaho (USA). Quaternary Science Reviews 26: 1713–1724.

26.

Jiménez-Moreno

Anderson

Atudorei

. (2011) A high-resolution record of climate, vegetation, and fire in the mixed conifer forest of northern Colorado, USA. Geological Society of America Bulletin 123(1–2): 240–254.

27.

Long

Power

Bartlein

(2011) The effects of fire and tephra deposition on forest vegetation in the Central Cascades, Oregon. Quaternary Research 75: 151–158.

28.

Long

Whitlock

Bartlein

. (1998) A 9000-year fire history from the Oregon Coast Range, based on a high-resolution charcoal study. Canadian Journal of Forest Research 28: 774–787.

29.

Lundeen

(2001) Refined Late Pleistocene Glacial Chronology for the Eastern Sawtooth Mountains, Central Idaho. Unpublished MSc Thesis, Idaho State University.

30.

Lundeen

Brunelle

Burns

. (2013) A speleothem record of Holocene paleoclimate from the northern Wasatch Mountains, southeast Idaho, USA. Quaternary International 310: 83–95.

31.

Mann

Zhang

Rutherford

. (2009) Global signatures and dynamical origins of the Little Ice Age and Medieval Climate Anomaly. Science 326: 1256–1260.

32.

Marlon

Bartlein

Whitlock

(2006) Fire-fuel-climate linkages in the northwestern U.S. during the Holocene. The Holocene 16: 1065–1077.

33.

Mensing

Benson

Kashgarian

. (2004) A Holocene pollen record of persistent droughts from Pyramid Lake, Nevada. Quaternary Research 62: 29–38.

34.

Millspaugh

Whitlock

Bartlein

(2000) Variations in fire frequency and climate over the past 17000 yr in central Yellowstone National Park. Geology 28: 211–214.

35.

Moy

Seltzer

. (2002) Variability of El Niño/Southern Oscillation activity at millennial time scales during the Holocene epoch. Nature 420(6912): 162–165.

36.

Munroe

(2003) Holocene timberline and palaeoclimate of the northern Uinta Mountains, northeastern Utah, USA. The Holocene 13(2): 175–185.

37.

Murchison

(1989) Fluctuation history of Great Salt Lake, Utah, during the last 13,000 years. PhD Thesis, University of Utah.

38.

Oriel

Platt

(1980) Geologic map of the Preston 1 degree x 2 degrees quadrangle, southeastern Idaho and Western Wyoming. U.S. Geological Survey, Miscellaneous Investigations Series Map I-1127, scale 1:250000. Reston, VA: U.S. Geological Survey.

39.

Power

Whitlock

Bartlein

(2011) Postglacial fire, vegetation, and climate history across an elevational gradient in the Northern Rocky Mountains, USA and Canada. Quaternary Science Reviews 30: 2520–2533.

40.

Prichard

Gedalof

Oswald

. (2009) Holocene fire and vegetation dynamics in a montane forest, North Cascade Range, Washington, USA. Quaternary Research 72(1): 57–67.

41.

Rasmussen

Andersen

Svensson

. (2006) A new Greenland ice core chronology for the last glacial termination. Journal of Geophysical Research: Atmospheres 111(D6): D06102.

42.

Regonda

Rajagopalan

Clark

. (2005) Seasonal cycle shifts in hydroclimatology over the Western United States. Journal of Climate 18: 372–384.

43.

Reheis

Laabs

BJC

Kaufman

(2009) Geology and geomorphology of Bear Lake Valley and upper Bear River, Utah and Idaho. In: Rosenbaum

Kaufman

(eds) Paleoenvironments of Bear Lake, Utah and Idaho, and its Catchment. Boulder, CO: Geological Society of America (Special Paper 450), pp. 15–48.

44.

Reinemann

Porinchu

Bloom

. (2009) A multi-proxy paleolimnological reconstruction of Holocene climate conditions in the Great Basin, United States. Quaternary Research 72(3): 347–358.

45.

Rull

López-Sáez

Vegas-Vilarrúbia

(2008) Contribution of non-pollen palynomorphs to the paleolimnological study of a high-altitude Andean lake (Laguna Verde Alta, Venezuela). Journal of Paleolimnology 40: 399–411.

46.

Spangler

(2001) Delineation of recharge areas for karst springs in Logan Canyon, Bear River Range, northern Utah. In: Kuniansky

(ed.) U.S. Geological Survey Karst Interest Group Proceedings: Water Resources Investigations Report 01–4011. Reston, VA: U.S. Geological Survey, pp. 186–193.

47.

Stockmarr

(1971) Tablets with spores used in absolute pollen analysis. Pollen et Spores 13: 615–621.

48.

Westerling

Hidalgo

Cayan

. (2006) Warming and earlier spring increase Western U.S. forest wildfire activity. Science 313: 940–943.

49.

Whitlock

Bartlein

(1993) Spatial variations of Holocene climatic change in the Northern Rocky Mountain Region. Quaternary Research 39: 231–238.

50.

Whitlock

Bartlein

Van Norman

(1995) Stability of Holocene climate regimes in the Yellowstone region. Quaternary Research 43: 433–436.

51.

Williams

Allen

Millar

. (2010) Forest responses to increasing aridity and warmth in the southwestern United States. Proceedings of the National Academy of Sciences 107(50): 21289–21294.

52.

Williams

Shuman

Bartlein

. (2006) North American Pollen Atlas. IGBP PAGES/World Data Center for Paleoclimatology Data Contribution Series # 2007–001. Boulder, CO: NOAA/NCDC Paleoclimatology Program.

53.

Wise

(2010) Spatiotemporal variability of the precipitation dipole transition zone in the western United States. Geophysical Research Letters 37: L07706.

54.

Woodhouse

(2004) A paleo perspective on hydroclimatic variability in the Western United States. Aquatic Sciences 66: 346–356.