Dust-drought interactions over the last 15,000 years: A network of lake sediment records from the San Juan Mountains,Colorado

Abstract

Millions of people in the arid Southwest United States rely on snow-fed Colorado River water. Dust deposition on snow accelerates snowmelt, posing a challenge for water managers who also need to grapple with increased likelihood of drought due to climate change. Dust production is thought to increase during drought, but the impact of drought on dust deposition is unclear. To answer this question, total dust mass accumulation rate (DMAR) reconstructions were developed from sediment cores from three lakes in the San Juan Mountains, Colorado, spanning the last ~15,000 years. Monte-Carlo end-member analysis of particle size and elemental composition, which incorporates measurement and model uncertainties, was combined with age uncertainty to estimate DMAR for each lake. We also synthesize the records providing the first Holocene DMAR reconstruction for the region. The records show little relation between periods of frequent and severe drought (e.g. during medieval megadroughts) and periods of higher DMAR, although there is considerable uncertainty at short timescales. We find instead that sediment availability, modulated by natural or human-mediated geomorphic processes that generate sediment, and transport mechanisms are the key drivers. DMAR was highest during the Late Glacial-Interglacial Transition (LGIT, 15–11 kyr BP) and in the last 250 years, periods when sediment availability was enhanced. DMAR increased by 60% (16–85% 75% highest density region range) compared with the late-Holocene baseline starting in the 1770s and peaking in the 1840s, associated with the intensification of human activities. Human-induced dustiness also represents the highest interval of dust deposition in the last 11,000 years. Our results demonstrate that although the Colorado Plateau is naturally prone to dustiness, drought is a secondary driver of dust accumulation in the mountains. This suggests that land-use management decisions aimed at reducing land disturbance can mitigate future dust accumulation, despite projected increases in regional aridity.

Keywords

Colorado Plateau drought dust Holocene land disturbance mixing model paleoclimate southwest

Introduction

Each stage of the dust cycle (production, transport and deposition) has implications for human health, climate and water resources (e.g. Choobari et al., 2014; Goudie, 2014; Marx et al., 2018). Dust deposition affects soil development (e.g. Lawrence et al., 2011; Reynolds et al., 2001; Simonson, 1995) and ecosystem functioning through the input of major and trace elements (e.g. Ballantyne et al., 2011; Brahney et al., 2014). In alpine regions such as the southern Rocky Mountains in the Southwest United States, dust deposition on snow decreases albedo, accelerates spring melt and reduces snow-cover duration by more than a month, affecting the water availability of millions of people dependent on snow-fed river systems (Maurer and Bowling, 2015; Painter et al., 2018; Skiles et al., 2015). Drought has been linked to increased dustiness by reducing vegetation and destabilizing soil surfaces (Munson et al., 2011; Urban et al., 2009). Projected warming and drying in the Southwest United States is a robust result from climate model studies (Cook et al., 2015; Seager and Vecchi, 2010), suggesting that dust on snow events could increase, further challenging water management (Painter et al., 2007; Routson et al., 2016).

However, in the arid Southwest United States where drought and dust are natural geohazards, quantifying the deposition stage of the dust cycle has remained largely incomplete. Instrumental measurements of dust deposition in the southern Rocky Mountains are limited to the last decade at best. Lawrence et al. (2010) provide directly measured dust deposition estimates from snow for the snow seasons of 2004–2008 for two alpine stations. Indirect estimates indicate an increase in dust deposition over the last three decades in the Inter-Mountain West attributed to emissions from the southwest deserts (Brahney et al., 2013; Clow et al., 2016). However, direct estimates of dust production from southeastern Utah, upwind from the Inter-Mountain West, show no significant changes (Flagg et al., 2014) indicating the increase would be coming from other desert areas, such as northeastern Arizona. Direct measurements of dust production in northeastern Arizona are currently lacking.

A long-term perspective is necessary to quantify the risk of increased dust deposition during prolonged drought. Dust preserved in natural archives extends the record well beyond the limited period of direct observations. The first dust deposition records from lakes in the southern Rocky Mountains lacked the temporal resolution to identify the role of drought (Ballantyne et al., 2011; Neff et al., 2008). Another study from two lakes in the Uinta Mountains had the temporal resolution but did not show variations (Reynolds et al., 2010). On the contrary, Routson et al. (2016) who looked at dust-drought interactions specifically showed increased dust concentrations in a lake from the southern Rocky Mountains during the medieval (1000–1400 CE) and Roman (1–400 CE) megadroughts. Beyond the late Holocene (4000 years ago to the present), palaeoclimatological records suggest large-scale hydrological changes on the Colorado Plateau during the middle Holocene (8000–4000 years before present), including intense aridity (e.g. Anderson et al., 2008; Toney and Anderson, 2006; Weng and Jackson, 1999). However, dust deposition records spanning the Holocene and beyond are lacking from the region. Whereas the link between drought and dustiness appears inconclusive, the impact of human activities on the dust cycle is a robust finding. These long-term records consistently show a rapid increase in dust deposition starting in the 1800s, with this increase attributed to the intensification of human activities in the area.

In this study we test the hypothesis that increased dust deposition in the southern Rocky Mountains coincided with periods of prolonged drought in the Southwest United States using a network of lake sediment records of dust accumulation. We quantify dust in lake sediment and its uncertainty using Monte-Carlo end-member modelling of elemental composition and particle size. To compare records meaningfully, we combine dust abundance quantification with radiocarbon and ²¹⁰Pb and ¹³⁷Cs dating and Bayesian age modelling to calculate dust mass accumulation rates (DMARs) as suggested by Albani et al. (2015). Because lake sediment records are time-uncertain, we propagate age uncertainties through the analysis to quantify their impact on DMAR and our conclusions.

Modern settings, regional climate and dust source

The San Juan Mountains (SJM), located in southwestern Colorado in the southern Rocky Mountains, are composed of two clusters (southeast and northwest) shaped by plutonic and volcanic activity (Figure 1). The northwestern cluster contains Cretaceous plutonic activity and Tertiary volcanic activity (Ellingson, 1996), whereas vent and volcanoclastic facies of the San Juan volcanic field underlies the south-eastern cluster (Lipman and Mcintosh, 2011). The geologic units have geochemistry distinct from the sedimentary desert soils of the dust source (Neff et al., 2008). Modern dust deposition in the SJM originates from a wide variety of locations in the Southwest as determined by back-trajectory experiments (Reynolds et al., 2016).

Figure 1.

Study site maps. Left panel: This study selected three Lakes: (1) Clear Lake, (2) Columbine Lake, (3) Crater Lake in the San Juan Mountains. Previous dust research was conducted in (4) Porphyry Lake (Ballantyne et al., 2011; Neff et al., 2008), (5) Senator Beck Lake (Ballantyne et al., 2011; Neff et al., 2008), (6) Blue Lake (Routson et al., 2019) and (7) Fish Lake (Routson et al., 2016). Letters A–R indicate hydroclimate study locations used in Figure 8. Potential natural dust source areas such as the Little Colorado River (LCR), playas and friable shale bedrock are included, and the black arrow represents the prevailing wind direction. Left panel inset: study location indicated in rectangle. Right panel: lake catchment extent, elevation (m), vegetation cover (green) and bathymetry. Red dots in right panels indicate coring locations. Please refer to the Online version for the color figure.

The climate of Silverton, Colorado, at an elevation of 2865 m and near to the study site, is typified by a biseasonal climate. Over 80% of precipitation falls predominantly as snow from October to March (average 560 mm/month total snowfall) from Pacific frontal storms, and summer rainfall is associated with the northern extent of the North American Monsoon (Jul-Sep, average 70 mm/month). Average winter (DJF) and summer (JJA) temperature ranges are ‒18.8 to 2.5°C and ‒0.1 to 22.8°C, respectively (Western Regional Climate Center, 2018). Like much of the Southwest United States, the El Niño Southern Oscillation (ENSO) teleconnection usually results in wet winters during El Niño and dry winters during La Niña (Sheppard et al., 2002).

Materials and methods

Dust and catchment rock sampling

We selected three different alpine watersheds: Crater Lake (southeast cluster), Columbine Lake and Clear Lake (northwest cluster) based on their common high elevations (>3000 m), but varying catchment characteristics (Figure 1 and Table 1) as we wanted to test the hypothesis that catchment characteristics such as elevation, catchment area, lake surface area, lake to catchment area ratio, vegetation cover, and slope would impact mean DMAR at each site. Representative rock samples were collected from each catchment to constrain local bedrock elemental abundance. Rocks samples were powdered and homogenized using a pestle and mortar after being crushed with a hammer. The rock powders were mounted in powder holders for x-ray fluorescence (XRF) analysis (see section ‘Elemental composition and end-member modelling’). Routson et al. (2016) collected dust samples in April 2012 by sampling snow surfaces at Molas Pass, Wolf Creek Pass and Summitville. Dust samples were mounted in powder holders for XRF analysis, but they were not powdered due to their sufficiently small grain size.

Table 1.

Lake characteristics for the San Juan Mountain lakes from this study and for other lakes used in this study and published previously.

	Clear	Columbine	Crater	Porphyry^a	Senator Beck^a	Blue^b
Lake latitude N	37°49′35.20″	37°51′43.64″	37°23′20.76″	37°53′16.83″	37°51′47.54″	37°14′30.27″
Lake longitude W	107°47′15.78″	107°46′17.81″	106°41′41.50″	107°45′6.59″	107°45′13.9″	106°37′46.82″
Lake elevation (m)	3646	3874	3331	3903	3858	3508
Catchment area (km²)	2.58	1.31	3.6	0.47	0.002	0.22
Lake area (km²)	0.22	0.16	0.13	0.012	0.000116	0.1
Lake: catchment ratio	11.7	8.2	27.7	24	13	2.2
Vegetation cover (%)	25.40	4.03	18.60	8.2	0.2	38.7
Slope (m)	0.001815	0.0004214	0.001396	0.03305	0.106635	0.001025

Fish Lake (Routson et al., 2016) was not included because the study did not include estimates of dust mass accumulation rates.

Neff et al. (2008).

Routson et al. (2019).

Lake sediment sampling

In the summers of 2015–2017, surface cores ranging in length from 91 to 125 cm were collected from Clear, Columbine, and Crater Lakes using inflatable rafts and a universal gravity corer (Table S1, available online). A modified percussion corer from a coring platform was used to collect 3 and 4 m long cores from Clear and Crater Lakes, respectively. Cores were taken from the centres of the lakes in the deepest parts as identified by depth sounding using a HumminBird instrument (model 1199). Attention was paid to preserving the sediment surface using floral foam and Zorbitrol^©.

Chronology

Age control was developed using ¹⁴C, ²¹⁰Pb and ¹³⁷Cs dating in all three lakes (Table 2). For Clear and Columbine Lakes, we used a Canberra Broad Energy Germanium Detector (BEGe; model no. BE3830 P-DET) at the Marine Science Centre at Northeastern University to measure ²¹⁰Pb and ¹³⁷Cs activities on 20 and 16 dried and homogenized samples over the top 25.25 and 12.5 cm of the cores, respectively. For Crater Lake, ²¹⁰Pb and ¹³⁷Cs activities were measured on a gamma detector at the University of Southern California on 12 samples from the top 18.25 cm of a short core. We used the constant rate of supply (CRS) model after Appleby (2001) to calculate ages and standard error based on the ²¹⁰Pb profile. The age models for the different lakes were constrained by AMS ¹⁴C on plant macrofossils but also terrestrial insects in some cases (Table 2). Radiocarbon samples were pre-treated using acid–base–acid pre-treatment procedures at the Sedimentary Records of Change Lab, graphitized in the Schuur Lab, both at Northern Arizona University. The samples were then analysed for radiocarbon activity at the University of California, Irvine. The radiocarbon age-depth model for Clear Lake was spliced from two different sediment cores correlated by magnetic susceptibility profiles. The age-depth model for Columbine was spliced from two different sediment cores correlated visually. Bayesian age-depth modelling R software package BACON was used to calibrate and develop the age-depth models (Blaauw and Christen, 2011), using the IntCal13 calibration curve (Reimer et al., 2013). Discrete sediment layers thicker than 2 cm, interpreted to have been deposited rapidly (<1 year) and accounting for a total of 8.9 cm, were removed from the Crater Lake surface core age-depth model.

Table 2.

Radiocarbon dates for the San Juan Mountain lake cores.

Lake	UCI ID	Depth (cm)	Material	¹⁴C age BP (yr)	± 1-sigma
Clear	196899	31.75^a	Dragonfly wing	1150	30
	188319	54	Plant fragment	900	70
	196897	63	Twig	910	20
	196898	109	Grass blade	1600	90
	202765	291.5	Spruce cone bract, twig, insect chitin, bryophyte	4610	20
Columbine	196901	27.5	Insect wing	520	100
	190157	46.5	Bryophyte twig, Daphnia ephippia	1510	310
	188317	124.65	Bryophyte twig, Daphnia ephippia	2875	20
Crater	169759^b	9.68 (44)^b	Spruce needle, bark, leaf fragment	195	15
	169760^b	17.93 (64)^b	Spruce needle	660	15
	169761^b	52 (83)^b	Spruce needle, leaf fragment	1375	15
	180850	75.75	Woody stem	2770	15
	180849	133.1	Woody stem	6140	15
	202767	165.75^a	Picea needle tip	3455	45
	180851	216.95	Woody stem	9330	20
	180852	233.5	Daphnia ephippia, cladocera, bark	9615	20
	180853	293.5	Spruce needle	10,585	30
	202768	322	Leaf fragment, insect chitin, Daphnia ephippia, chironomid capsules	11,000	60
	188315	404.75	Leaf fragment, insect remains	12,310	70

Depths in italic indicate depths in Crater Lake short record.

Indicates age reversal.

Denotes radiocarbon samples used in Crater Lake short record age-depth model.

Particle size analysis and end-member modelling

Particle size analysis has been widely used to estimate dust contribution to lake sediments (e.g. Chen et al., 2013; Chu et al., 2009; He et al., 2015; Huang et al., 2011; Lim and Matsumoto, 2006; Muhs et al., 2003; Qiang et al., 2014; Routson et al., 2016). Samples of 2 cm³ were collected at 0.5 cm intervals from every core. Samples were weighed wet, then weighed again after freeze-drying for 12 h to calculate water content and dry bulk density (DBD). The samples were then pre-treated with 5 ml of 30% H₂O₂ to remove organics. Biogenic silica was removed by adding 40 mL of 10% Na₂CO₃ and letting the samples react for 5 h at 80°C. Sodium hexametaphosphate was added as a dispersant and shaken for 3 h. Grain size distributions in the 0.004–2000 µm range with 116 classes were analysed using a laser diffraction Coulter LS13-320 and each sample was measured 5 times. Samples of modern dust were also prepared similarly.

Lake sediment is composed of a mixture of sediment populations derived from different sources by various processes (e.g. local catchment processes, in-lake processes, and atmospheric deposition) resulting in polymodal grain size distributions (Tanner, 1964). In this study, an R (R Core Team, 2017) function for modelling of end-member composition, R-based End-member Composition Algorithm (RECA; Seidel and Hlawitschka, 2015), that combines theory- and data-driven approaches was used to decompose the grain size distribution of individual samples from each lake and to compare to modern dust grain size distributions. Briefly, RECA is an ordination approach that describes mixed grain size distributions as a combination of loadings (grain size distributions within the dataset) and scores (the relative contributions of each loading through time) by grouping distributions into a predefined number of subpopulations based on pattern variation similarities. We applied RECA using default values (c5 = 0.8, convexity threshold = ‒6, iterations = 10000). Determining the number of end members is an iterative process based on the coefficient of determination (mean r²) between the original data and the dimensionally reduced modelled matrices. The original code was extended so when a plausible number of end-members was identified, the algorithm would iterate 1000 times to estimate uncertainty in both loadings and scores. The analysis was performed twice, first on the entire dataset and a second time with the last 250 years removed from the dataset to test whether the covariance structure was affected by the dust variability of the historic period.

Elemental composition and end-member modelling

XRF analysis has also been widely used to estimate the contribution of dust to sediment (e.g. Conroy et al., 2013; Mulitza et al., 2010; Routson et al., 2016; Stuut et al., 2014). The cores were prepared, covered with SPEXCerti Ultralene^® and scanned with an Avaatech XRF Core Scanner at Texas A&M University, College Station, at 0.5 and 1 cm resolution at 10, 30 and 50 kV. Dust and powdered rock samples were scanned at three different random points to assess the variability in composition. We gathered the following elements: Mg, Al, Si, P, S, Cl, K, Ca, Ti, Cr, Mn, Fe, Ni, Zn, Br, Rb, Sr, Y, Zr, Rh, Ag and Ba. The XRF results are semi-quantitative and depend on sample geometry, physical properties (e.g. water content, density, organic matter) and instrument settings (Löwemark et al., 2011; Richter et al., 2006; Tjallingii et al., 2007). To overcome these effects, log-ratios of two elements can be interpreted as the relative concentration of the two elements (Weltje and Tjallingii, 2008).

The mineral component of lake sediment is a mixture of catchment terrigenous runoff and wind-deposited dust. To calculate the fraction of dust in the sediment, we applied an R-based (R Core Team, 2017) Monte Carlo one-dimensional end-member mixing model. In a first step, the code detects informative ratios. Considering measurement uncertainty, the code randomly samples n times from the dust dataset, then the rock dataset to create two probabilistic end-member distributions. The code cycles through every permutation of the XRF dataset for individual lakes, or 199 log-ratio possibilities when keeping to associations of elements measured at the same energy level, that is, 10, 30 or 50 kV, estimating the probability distribution of the dust concentration for all samples in a core. Ratios that do not fall between the dust and rock end members are assigned a uniform, uninformative distribution. The resulting probability distributions for each ratio is ranked by level of information provided based on a measurement of relative entropy (Kullback and Leibler, 1951). Those with the highest relative entropy value are most distant to a uniform, uninformative distribution. During each iteration of the first step, we calculate concentration using the formula

Dust fraction = \frac{l o g \frac{A}{B} s e d - l o g \frac{A}{B} r o c k}{l o g \frac{A}{B} d u s t - l o g \frac{A}{B} r o c k}

(1)

where A and B are two different elements. We assume that each ratio is independent. We selected the most informative elemental ratios based initially on the entropy value, but secondarily on whether the elemental ratios were plausibly controlled by dust abundance. Multiple ratios were selected when the down core patterns were similar. The result is an estimation of dust concentration based on several log-ratios considering their probabilistic distributions.

DMAR calculation

The process of calculating DMAR has been described by Albani et al. (2015) in detail and is fundamental to cross-correlate variability among dust archives and sites. DMAR is the product of vertical sediment accumulation rate, DBD, and the dust fraction out of the mineral fraction. Mineral fraction was measured by loss-on-ignition (550°C for 5 h) on 30 samples per core. Because we did not measure the mineral fraction on every sample analysed for grain size, values were estimated from a regression between water content and mineral fraction. Vertical sediment accumulation rate is highly dependent on the age-depth model accuracy of the lakes, which are time-uncertain. In this study, we used GeoChronR (McKay et al., 2018), to quantify age-uncertainty in the estimation of DMAR. GeoChronR uses the age ensemble members produced by BACON, the dust concentration ensemble members produced by RECA and the XRF model, and their uncertainties, as well as DBD to calculate a large ensemble of 1000 iterations of DMAR estimates for each core.

Catchment dust focusing effect

Dust deposits unevenly across the lake surface and catchment because of topographic and land-cover effects, and also deposit unevenly on the lake bed due to limnologic processes (Boygle, 1999; Davis and Ford, 1982). Two or more cores were used from each lake and the difference in vertical accumulation rates was noted. Borrowing from concepts in the tephra deposition literature (Lowe, 2011), primary atmospheric dust deposits on the lake surface and settles to the lakebed, but dust also falls on the landscape and may enter the lake as secondary dust. This secondary dust from the catchment may take weeks to years to be added to the lake sediment during spring snowmelt, by soil erosion during high-precipitation events, or as diffuse runoff from the lake margins. Primary dust deposition is also complicated by ice-cover during winter. For these reasons, we assumed that the dust record is affected by secondary dust through the catchment focusing effect, that is an increase in apparent dust flux per square metre due to the addition of secondary dust into the lake. However, the magnitude of this concentration effect, or whether this ratio is stable in time, is unknown.

To estimate the catchment dust focusing effect for each lake, we hypothesized that lake catchment characteristics (percent vegetation cover, slope, elevation, surface area and catchment to lake surface area ratio) would influence the relative amount of dust focusing, and thus DMAR differently at each lake. We estimated DMAR for each lake in this study and three additional lakes from the literature, between the years 250–3000 BP to characterize the late Holocene while avoiding the influence of human impact. We then used correlation matrices and stepwise regression to identify significant predictors. The relationships found were used to estimate an average dust deposition expected. This average DMAR was then used to calculate the focusing factor (FF) for each lake and each core used. A FF is defined as an integer defining the influence of catchment parameters on DMAR that allows us to account for the differences between catchments and estimate regional atmospheric dust deposition rates. To calibrate the records, we applied this FF to each estimation of DMAR before combining records into a composite.

Results

Sedimentology

Sediment type varies by lake. Columbine Lake sediment transitions abruptly from massive grey clay to very fine grey to black laminations then to orange fine laminations near the sediment-water interface. Crater Lake sediments consist of dark brown massive sediment interspersed with thick light brown bands. Clear Lake sediments consist of dark brown massive sediment with light grey bands increasing in frequency and gradually transition to black and light grey fine laminations at the top. Both Crater and Clear Lake sediment contain bands of fining upwards sequences and eroded lower contacts, often containing large rocks (>5 cm diameter). Mineral fraction measured from LOI varies from a minimum of 4% in organic and water rich layers to >95% in many sections of sediment but averages 81%. DBD varies with the mineral fraction for most of the lakes with high mineral fraction increasing DBD. Clear Lake, however, shows an inverse relationship between the two variables.

Chronology

Figure 2 shows the age-depth chronology for each lake. The Clear Lake record spans ~6000 years based on four radiocarbon ages, with one age reversal (UCI ID 196899, Table 3) that was excluded. The excluded date was based on a dragonfly wing. Due to the coring technique, the long core does not contain the sediment-water interface, and the surface was ascribed to 100 ± 20 yr BP based on correlation of magnetic susceptibility data with the short core. The ²¹⁰Pb profile at Clear Lake exhibits a gradual decline in activity, with two negative excursions at 5.75 and 9.75 cm, before reaching equilibrium around 60 Bq kg⁻¹ below 17.5 cm (supplemental material, available online). There is a well-defined peak in ¹³⁷Cs activity at 9.25 cm. The CRS ages calculated using only ²¹⁰Pb activities agrees well with the peak of ¹³⁷Cs activity. The sediment record of Columbine Lake spans ~3000 years based on three radiocarbon ages. Due to the paucity of organic material in the sediment, a mixture of small insects and plant fragments are dated with uncertainty ranging from 20 to 310 years. The ²¹⁰Pb activity in Columbine (supplemental material, available online) exhibits a gradual decline that reaches equilibrium around 50 Bq kg⁻¹ below 8 cm, whereas the ¹³⁷Cs activity peaks at 3.25 cm. As the ²¹⁰Pb profile showed good agreement with ¹³⁷Cs atomic bomb peak, ¹³⁷Cs was not included in the CRS model. The sediment record from the Crater Lake long core continuously spans the last ~15,000 years based on 10 radiocarbon ages, with one age reversal (UCI ID 202767, Table 2) that was excluded. The excluded date was based on a fragment of a Picea needle that may have been washed in later. Due to the coring technique, the long core does not contain the sediment-water interface, and the surface was ascribed to ‒25 (–41, +75) yr BP based on stratigraphic correlation with the short core that does include the sediment interface. The ²¹⁰Pb and ¹³⁷Cs activities in Crater were influenced by a depositional event, so we use only three ²¹⁰Pb age points and the ¹³⁷Cs data were discarded (supplemental material, available online; Yackulic, 2017). The Crater Lake surface core has an independent age-depth model spanning ~1500 years. The youngest radiocarbon sample (UCI ID 169759) had a multimodal profile, and based on the sedimentation profile of the other lakes, we forced BACON to use the mode with the highest probability, giving an age of 167.5 ± 22 uncalibrated years BP (Figures S1 and S2, available online).

Figure 2.

Age-depth models for (a) Clear Lake, (b) Columbine Lake, (c) Crater Lake (long core) and (d) Crater Lake (short core). Probability distribution curves are ²¹⁰Pb and ¹⁴C ages. The thin lines represent an ensemble of 10 randomly selected members. The thick black line and shaded grey area represent 50% and 90% probability, respectively. Note the excluded radiocarbon age for Crater Lake and Clear Lake. Insets show the ²¹⁰Pb models.

Table 3.

Spearman’s Rank correlation between DMAR and lake characteristics.

	Slope	Lake surface area	Catchment area	Lake: catchment	Elevation	Vegetation cover
DMAR	0.6	–0.31	0.26	0.94	–0.14	–0.43
p-value	0.24	0.56	0.66	0.02*	0.80	0.42

DMAR: dust mass accumulation rates.

denotes significant at p < 0.05.

Particle size analysis and end-member modelling

Modern dust-on-snow samples have a unimodal distribution (mean: 15.1 µm, mode maxima: 22.9 µm) with coarse grains <158 µm. Meanwhile, particle size distributions of lake sediment are multimodal suggesting the sediment is composed of material produced by different processes. The mean curve of all samples shows local maxima at 5.6 and 27.4 µm but this varies by lake and sediment depth. The sediment contained coarser material than pure dust, indicating input from local catchment material. Three end members represent sediment particle size in each lake (Figure 3). Applying RECA on a grain size dataset containing samples from every lake reduced the variability of the relative contributions, and likely indicates that the non-dust components are inconsistent from lake to lake. Consequently, we focused on individual RECA analyses for each lake. Our second analysis of RECA with the last 250 years removed shows no significant differences from the full model, suggesting that 20th century dust variability does not change the covariance structure enough that it is affecting RECA prehistorically. Most end members’ particle size distributions have a well-defined dominant mode, although secondary modes did occur (Figure 3). Secondary modes mimic primary modes of other end members that can affect the interpretation and quantification of the mode contribution. End member EM1 has a modal particle size of ~5 µm, end member EM2 of ~20 µm, and end member EM3 has a modal particle size of ~100 µm and was often associated with light coloured bands. We interpret these end members as representing suspended sediment from fluvial and alluvial processes (EM1), windblown dust (EM2) and coarse material from the local catchment (EM3).

Figure 3.

RECA end members with uncertainty for (a) Columbine, (b) Clear and (c) Crater Lake compared with modern dust particle size distribution. An example of a secondary mode is shown in (a). End-members are EM1, EM2 and EM3 from left to right in each panel.

Elemental composition and end-member modelling

Crater, Clear and Columbine Lakes have differing local bedrock compositions, thus different log ratios were necessary to create mixing models with end members sufficiently separated. Log ratios with the highest entropy and high signal to noise ratios were selected. Log ratios of Y/Ba at 50 kV and Zr/Ba at 50 kV are plotted for Clear Lake (Figure 4a) but only log Y/Ba at 50 kV was used to calculate dust fraction. Log Zr/Ba at 50 kV showed a low signal to noise ratio and the catchment material end member showed large variability in composition. Log ratios of K/Ca at 10 and 30 kV were used for Columbine Lake. Dust is enriched in Ca relative to catchment material at Columbine Lake (Figure 4b). Different log ratios were used for Crater Lake short (Rb/Sr 30 and 50 kV) and long (Zr/Ba at 50 kV and Rb/Ba at 50 kV) records (Figure 4c and d).

Figure 4.

Histograms and mixing models for (a) Clear lake, (b) Columbine Lake, (c) Crater Lake (long core) and (d) Crater Lake (short core). Kernel density estimation plots show densities of the samples and their overlap. In the case of Clear Lake, only log ratio Y/Ba at 50kV was used. For every other case, two log ratios were used. Note that Crater Lake long and short cores use different ratios.

FFs

Correlations using Spearman’s method between catchment properties and DMAR for the period 250–3000 BP are given in Table 3. Lake to catchment area ratio has a strong, positive monotonic relationship to average DMAR (r_s = 0.94, n = 6, p = 0.01667). Linear regression indicates lake to catchment area ratio is a significant predictor of DMAR (R²_adj = 0.71, p = 0.0226 at p < 0.05) (Figure S9, available online). Taking into consideration this relationship and within the limits of our small dataset, we use the intercept (DMAR = 21.2 ± 25.9 g m⁻² yr⁻¹) to indicate DMAR at the lowest possible ratio (i.e. 1:1). Based on this estimate of average late-Holocene primary dust deposition, FFs for each of the other lakes varied between 1.3 and 14.6 (Table 4).

Table 4.

Average late-Holocene DMAR calculated from the individual lake records combining XRF and RECA methods for Clear, Columbine and Crater Lakes.

	Clear	Columbine	Crater	Porphyry	Senator Beck	Blue
Mean late-HoloceneDMAR (g m^–2 yr^–1)	115.9 (78, 153.8)	33.6	187 (68.2, 305.7)	119	124	28.4
Focusing factor	5.5 (3.7, 7.3)	1.6	8.9 (3.2, 14.6)	5.7	5.9	1.3

DMAR: dust mass accumulation rates; XRF: x-ray fluorescence; RECA: R-based end-member composition algorithm.

For Clear and Crater Lakes, these values represent the average of the different cores used. DMAR values for Porphyry, Senator Beck and Blue Lake were identified from respective literature. Focusing factor = average late-Holocene DMAR/21.2. Fish Lake was not included because DMAR values could not be calculated.

Numbers in italics denote the DMAR and focusing factors for individual cores.

Temporal patterns

Dust concentration and DMAR for each lake and each approach (XRF and RECA) before and after FFs were applied are shown in Figure 5 and Figures S3–S7, available online. Concentration estimates from RECA and XRF differ (Figures S3–S7, available online). On average (interquartile range (IQR)) with the RECA method, Clear Lake contains 35% (14) dust, Crater Lake short record 38% (22), Crater Lake long record 57% (11) and Columbine Lake 30% (16). On average with the XRF method, Clear Lake contains 49% (27, IQR) dust, Crater Lake short record 61% (21), Crater Lake long record 50% (25) and Columbine Lake 57% (14). Uncorrected DMAR from RECA average 120 (65, IQR), 114 (84), 107 (86) and 48 (36) g m⁻² yr⁻¹ for Clear, Crater short record, Crater Lake long record and Columbine Lake, respectively. Uncorrected DMAR from the XRF method averages 169 (109, IQR), 207 (171), 87 (77), and 76 (52) g m^–2 yr⁻¹ for Clear, Crater short record, long record, and Columbine Lake, respectively. The RECA method tends to estimate lower dust concentrations and have lower uncertainty, and thus lower DMAR, than the XRF method. Neither the Roman nor the medieval megadrought intervals show higher dust concentration or DMAR using either method. The profile from short record at Crater Lake is the exception, showing elevated dust concentrations during medieval times (Figure S5a, available online). However, this is not evident in DMAR (Figure 5c), suggesting that the increase in dust concentration is associated with a reduction in total sediment accumulation. The Crater Lake long record shows a period of elevated dust concentration 12–7 kyr BP but the DMAR records do not show increasing dust deposition during the same period, although increases are visible 9–7 and 4–3 kyr (Figure S4, available online). Sedimentation rates have a strong influence on the estimation of dust deposition as seen from the stark differences between dust concentration and DMAR plots.

Figure 5.

Dust mass accumulation rates (DMARs), and 75% highest-density regions, are determined using the XRF (blue) and RECA (red) methods for (a) Columbine Lake, (b) Clear Lake, (c) Crater Lake short record compared with the long record for the last 1000 years, and (d) Crater Lake long record. DMAR has been corrected for catchment-specific focusing factors. Note the different scales on the x-axis.

Composite DMAR records

A composite record for the last 15,000 years (Figure 6a) was built by combining all three dust records from this study and both RECA and XRF approaches after correction from the FFs. The composite record likely represents a better estimate of regional past DMAR than any one record alone. The late Holocene is characterized by stable DMAR (Figure 6b), punctuated by two increases in DMAR (2000–500 BCE and 1700–2000 CE). Periods with frequent megadrought (1–400 CE and 1000–1400 CE) show no significant increase in DMAR than those with less frequent drought (410–990 CE and 1410–1590 CE) (Figure 7). In contrast, the settlement period (1800–2000 CE) has a significantly higher mean than pre-settlement conditions regardless of aridity.

Figure 6.

Corrected composite DMAR records for the last 15,000 years (a) and the late Holocene (b). Grey bands indicate probabilities of 97.5% and 75 %. Black line represents 50% probability. Other records from the literature are included. (c) Blue Lake dust mass accumulation record (Routson et al. (2019)), (d) Porphyry and (e) Senator Beck dust flux records (Neff et al. (2008)), (f) Fish Lake dust concentration record (Routson et al. (2016)), and (g) 30-year averaged Four Corners (35N-37N, 106W-112W) Palmer Modified Drought Index (PMDI; Cook et al., 2010). Stippled regions highlight the Roman and Medieval intervals.

Figure 7.

Boxplot of late-Holocene DMAR during settlement (1750–2015 CE), Medieval Climate Anomaly megadroughts (1000–1400 CE) and Roman megadroughts (1–400 CE) as compared with non-drought, pre-settlement years (360–540 CE and 960–1540 CE). Significance in the difference in mean is indicated with ** for p = 0.01 and NS. for non-significant using the Wilcox test. Sample size was kept the same for each period. With age uncertainty, only 160 samples were dated to the Roman period, so this was used as the sample number. The dots represent samples from individual age ensemble members falling within each specific time period.

The composite places the late Holocene in the context of the last 15,000 years. Although the composite consists of only one record from 15 to 6 kyr BP, we observe no long-term Holocene trend. Dust deposition was highest prior to 11 kyr when the climate at the study site and at the dust origin were markedly different than during the Holocene, when dust deposition is reduced in comparison. There is no evidence for increased deposition during mid-Holocene aridity (6.2–4.2 kyr BP). The period from 1950 to 2000 CE includes the highest dust deposition rates of the last 11,000 years, having increased by 60% (16–85% in the 75% high density region) compared with the late-Holocene (last 3000 years) baseline.

Discussion

Sources of uncertainty

The mineral component of lake sediments integrates atmospheric dust, secondary dust from the catchment, and local material. Our approaches to estimate atmospheric dust concentration attempt to discriminate between the different sources, but both RECA and XRF end-member mixing models have strengths and weaknesses.

Identification of dust using RECA

Particle size has been used to identify dust in sediment and end-member mixing models have shown success (e.g. Beuscher et al., 2017; Jiang et al., 2017; Prins et al., 2007; Stuut et al., 2014; Varga et al., 2012). EM3 (mode ~100 µm) is interpreted as local catchment material because this mode was most often associated with lighter colour bands of sediment, taken to represent depositional events, and because the coarser particle sizes are less likely transported from other catchments. EM2 (mode ~20 µm) is interpreted as dust because its size distribution corresponds closely to the observed modern dust size distribution. It is likely that EM2 also incorporates some local non-dust material that could lead to an overestimation of dust concentration. EM1 (mode ~5 µm) is interpreted as the deposition of suspended material into the lakes by sheet wash or shoreline erosion (Cohen, 2003). Dietze et al. (2012) examined surface sediments in lakes in China and interpreted the finer clay-sized mode as suspended material and the coarser modes as windblown dust. We interpret EM1 in a similar way because only a small proportion of modern dust is clay-sized. However, the proportion of clay-size dust is dependent on the source material, wind speed, and transport distance (Pye, 1987). Fairlie et al. (2007) found that Asian dust (<2.5 µm) accounted for up to 41% of dust days in Western United States in 2001, and although Reheis et al. (2009) find these estimates overstated, Nd, Sm and Sr isotope data from Neff et al. (2008) do not rule it out. Nevertheless, dust originating from a closer dust source such as the Mojave Desert is more likely than long-distance travel from Asia (Muhs and Benedict, 2006), suggesting a likely dominance of coarser dust sizes due to the relatively short travel distance. In addition, the dust end member may have changed through time based on changes in wind speed, and source distance and size. Such a shift may be reflected in our data, as the relative magnitude of the ‘dust’ mode at ~20 µm has shifted over the last 15,000 years at Crater Lake, becoming coarser through time (Figure S10, available online). Neff et al. (2013) also showed evidence of seasonal changes with winter dust (10 µm) finer than spring (40 µm) as measured in Mesa Verde and Canyonlands. Furthermore, Lawrence et al. (2010) found upon visual inspection that aggregated particles 250–850 µm composed of clay-sized material made up modern dust in the SJM. This would not be captured in our end-member analysis, as deflocculating and sonicating during grain size measurement would break up the aggregates and increase the finer end-member mode as also highlighted by Pye (1987). It is thus possible that EM1 represents disaggregated dust, an additional source of dust, or input from winter dust, indicating that using only EM2 as representing ‘dust’ may be underestimating dust concentration and accumulation rates. However, we do not observe significant covariability between EM1 and EM2 at any of the lakes studied here, suggesting that variability in dust deposition does not dominate down-core variability in EM1.

Identification of dust using XRF end-member modelling

XRF mixing models have been used successfully to estimate dust concentration (e.g. Mulitza et al., 2010; Routson et al., 2016), and this independent estimate generally agrees with the RECA-based estimates of concentration, increasing confidence in our capacity to quantify the dust component of the sediments (Figure 5). Nevertheless, there are discrepancies between the independent estimates, which may have several explanations. First, the modern dust geochemistry used in this project is the integration of eight modern dust-on-snow events collected in April 2012 by Routson et al. (2016), but geochemistry can change through time if the dust source shifts. In fact, Reynolds et al. (2006b) concluded that sources of dust on the Colorado Plateau have shifted during the late Pleistocene and early Holocene, and again over the past 100–150 years (Reynolds et al. 2006a). This likely means our modern dust samples do not capture the full range of dust geochemistry. Second, some of our mixing models use Ca and although sedimentary source regions can be high in carbonates, wind-blown carbonates have a tendency to dissolve, especially in the <0.45 µm fraction (Carling et al., 2012). Dissolution of Ca may impact our grain size observations as well. The loss of soluble material occurs again during wet measurement of grain size (Mahowald et al., 2014). Discrepancies between the two approaches could be resolved using geochemical (major, minor, REE) and isotopic analyses of the sediment, dust source, and catchment geology.

Estimation of DMARs and FFs

Dust concentrations were converted into DMAR to make meaningful comparisons between records and with modern observations. However, DMAR is largely dependent on the accuracy of the age-depth model. Unless varved sediment is used (e.g. Chu et al., 2009), lacustrine archives have large time uncertainties at short timescales. Consequently, we quantified the impact of age uncertainty on the fluxes, and when appropriate, averaged total deposition fluxes over longer periods of time (hundreds to thousands of years) to reduce the impact of age certainty. We hypothesized that catchment parameters (percent vegetation cover, slope, elevation, surface area and catchment to lake surface area ratio) would have some influence on DMAR. For example, Munroe (2007) determined that certain watershed characteristics had a strong influence on organic sedimentation. Ballantyne et al. (2011) also attempted correcting DMAR for the catchment focusing effect by normalizing to the total watershed area. Linear regression analysis indicates that mean late-Holocene DMAR increases with increasing lake to catchment area ratio. Although imperfect, this positive relationship suggests that lakes with smaller lake to catchment area ratios would be better locations for dust studies, as they would have limited secondary dust input. With large uncertainty, the analysis suggests that we should expect 21.2 g m^–2 yr⁻¹ primary dust deposition for the late Holocene on average. This is consistent with Lawrence et al. (2011) who estimated DMAR in soils over the last 12,000 years in the SJM to be 10–22 g m^–2 yr⁻¹, but higher than Ballantyne et al. (2011) who estimated dust flux values of ~10 g m^–2 yr⁻¹ at Porphyry and Senator Beck lakes. The uncertainty in this estimate propagates uncertainty into the FFs, and into the final DMAR. However, patterns and relative changes remain intact. More problematic is the assumption that the relationship remained stable through time which is unlikely to be true. Although the lakes are hydrologically open and lake level fluctuations are minimal, a local glacier was likely present at Crater Lake possibly until 12 kyr BP (Guido et al., 2007).

Records of dust deposition in the SJM

Despite the uncertainty discussed previously, our new records support previous work: steady dust accumulation over the late Holocene with an abrupt increase in the most recent time (Figure 6). For Columbine Lake, the steady dust accumulation appears to be more of a slight decline over the last 2000 years, comparable to Blue Lake and Fish Lake. However, it is unclear whether this is a significant trend considering the large uncertainty. Nevertheless, our records highlight that the Colorado Plateau is naturally dusty.

Impact of human activities

The most significant changes in DMAR during the late Holocene occur following European settlement in the 1500s. Three periods of increased accumulation are visible from our composite: 1770–1910 CE, 1955–1990 CE and 2005–2015 CE (Figure S11, available online). The increase begins earlier than shown in some other studies (~1770s compared with ~1885–1905 CE for Porphyry, Fish and Blue Lakes), but is later than at Senator Beck, suggesting both age uncertainty and local influences play a role in the timing of the increase. Although human activity did not intensify until later in the 1800s, the earlier increase seen in the composite is not incompatible with historical records of land use as the Spaniards reported innumerable sheep flocks by the end of the 18th century (White, 1983). The more gradual increase in our record also suggests that the overall impact of land-use activities might have gradually increased until culmination in ~1850 CE. This increase in DMAR has been previously associated with the introduction of sheep, mining, as well as the completion of the railroad (Abruzzi, 1995; Neff et al., 2005, 2008; Reynolds et al., 2010; Routson et al., 2016; Sayre, 1999). It may also reflect local changes in population of nearby mining towns of Silverton and Telluride that grew by 58–220%, respectively, during the 1900–1910 mining rush (US Census Bureau, 2002). The second and third increases are more abrupt. The second increase is likely related to the doubling in irrigated farmland, the construction of three coal fire plants and of a major paper mill, the tripling in population and the 340% increase in groundwater exploitation that occurred around the Little Colorado River (Abruzzi, 1995). Indirect observations over the past 17 years, coinciding with part of the third peak, indicate increasing dust deposition is likely partly attributed to agricultural, residential and industrial activities expanding into sensitive desert areas of the southwest (Brahney et al., 2013).

The dust-drought nexus

The dust-drought nexus is the concept that greater aridity is the main driver of increased dust flux through the reduction of vegetation cover (Pye, 1989; Rea, 1994). Our triad of dust accumulation records provide additional insights on the Southwestern dust-drought nexus. The hypothesis of increased dust accumulation during periods of megadrought (medieval or Roman, as identified by Routson et al., 2011 for the SJM) in the late Holocene is generally not supported by our observations (Figures 5–6 and Figures S12–S14, available online). This question was addressed by Routson et al. (2016) for the Colorado Plateau who concluded higher dust concentration in lake sediment coincided with these megadroughts (Figure 6f). However, as noted by Routson et al. (2016), the records were not converted to DMAR. Dust concentration versus accumulation makes comparisons with our records difficult and introduces uncertainty as whether the increases in dust concentration was due to changes in dust or changes in local sediment input. Routson et al. (2019) used the high-resolution record at Blue Lake to compare drought years and DMAR on short timescales (Figure 6c). They concluded that long-term DMAR changes were difficult to attribute to drought and that age uncertainty might be blurring assessments on annual timescales, since 22% of their age ensembles could be yielding higher DMAR during drought years. Neff et al. (2008) did not look at the dust-drought nexus directly, but produced two DMAR records from the same area. Although their records have low temporal resolution, they do not show increased dust deposition during the megadroughts (Figure 6d and e). Further north, Reynolds et al. (2010) inferred dust deposition from magnetic properties of two lake sediment records. They concluded that human disturbance had a much larger impact on dust deposition than droughts as their records do not appear to show increased dustiness before the industrial revolution. Finally, Aarons et al. (2016) measured dust deposition over the last 270 years in ice cores from Wyoming. Their records do not encompass the megadroughts, but they conclude that human activities combined with periodic drought may have switched the dust source in recent times. Overall, conclusive evidence supporting higher dust deposition during periods of prolonged drought remains elusive for the late Holocene.

The steadiness in dust accumulation during the late Holocene was a feature of the last 11,000 years (Figure 6a). Certain periods (4–3, 8–7 kyr) experienced high dust deposition, but not as high as in the last 250 years. It is challenging to make inferences on the Holocene dust-drought nexus because our understanding of climate during the last 15,000 years is largely incomplete for the Colorado Plateau (Figure 8). Hydroclimate records are asynchronous, but arid conditions might have existed on the Colorado Plateau during the middle Holocene. Dust accumulation did not increase during this period and we cautiously take this as supporting our late-Holocene dust-drought interpretation. Higher DMAR was a feature of the Late Glacial-Interglacial Transition (LGIT, 15–11 kyr BP) based on the record from Crater Lake that covers that period (Figure 6a and Figure S4, available online). The LGIT was a period of intense climatic and environmental change when local glaciers were retreating as temperature increased (Guido et al., 2007) and shorelines of Pleistocene Great Basin pluvial lakes were regressing as winter precipitation decreased (e.g. Barron et al., 2012; Reheis et al., 2014). The contrast between the steadiness in DMAR during the Holocene and the high DMAR during the LGIT suggest the magnitude of the forcing plays a role in determining the response in dust accumulation.

Figure 8.

A summary of postglacial hydroclimate and geomorphic records from the Colorado Plateau. The records show low coherence possibly due to low resolution, seasonality bias in proxies and inhomogeneity in precipitation patterns.

We present reconstructions of total dust accumulation rate, but we do not have the resolution to quantify changes in event frequency. Painter et al. (2012) found that dust events had increased from 2003 to 2010, whereas total annual accumulation had not. If intervals of increased aridity cause an increase in frequency without increasing long-term deposition rates, we would not expect to observe a drought-dust connection in our reconstructions. Whether a dry year results in higher accumulation or in higher frequency of events on the Colorado Plateau in modern times is unclear, and we would expect long-term source-region aridity to be an important driver of long-term dust deposition (Rea, 1994). However, we find little evidence for this and infer that other drivers must be controlling long-term dust deposition variability in the region.

Sediment availability and transport mechanisms as the drivers of dust accumulation in the SJM

The dust cycle is influenced by the interplay between climate, wind, vegetation, and sediment availability (Lancaster, 1997). Kocurek’s (1998) conceptual model of the temporal relationship between climate and dust suggests that sediment production mostly occurs during humid phases, and sediment availability and dust flux increase during increasing aridity, although the rapidity at which this occurs is unclear. The Southwest was wetter than today during the Last Glacial Maximum (LGM, Bartlein et al., 2011) as the Laurentide Ice Sheet steepened moisture gradients (Morrill et al., 2018) and winter storms travelled further south (Wong et al., 2016). Sediment, produced by local glaciers in the SJM and in Pleistocene Great Basin pluvial lakes during the LGM, would have become available during the retreat of local glaciers (Guido et al., 2007), as well as from the regressing shorelines of pluvial lakes in the Western United States as they dried during the LGIT (Goudie and Middleton, 2006; Reheis et al., 2014). The fine grain size of the sediment, and the lack of vegetation and soil development, would make these features preferential dust sources (Prospero et al., 2002). A steeper pressure gradient between the poles and the equator likely increased wind speeds during the LGM (Fischer et al., 2007) and possibly remained stronger than today during the LGIT. The combination of sediment availability and higher wind speeds increased dust accumulation in the SJM.

The consistency in Holocene dust accumulation likely reflects the depletion of easily erodible sediment as well as the establishment of soil crusts and grasses increasing landscape stability. Biogenic and non-biogenic soil crusts and desert pavement have been shown to make desert surfaces resistant to wind erosion (Belnap et al., 2014) and have very slow establishment rates (Belnap, 2003). Perennial grasses also protect arid land surfaces from wind erosion (Okin et al., 2011). Mid-Holocene hydroclimate on the Colorado Plateau is likely to have been drier (Figure 8) in winter, but a stronger monsoon (Barron et al., 2012, Figure 8) likely kept perennial grasses alive for most of the year. Nevertheless, our records do show a baseline of dust accumulation over the Holocene, suggesting the Colorado Plateau landscape is naturally prone to some dustiness. This baseline in accumulation likely represents the regional rate of particle formation (Pye, 1989).

The last 250 years of our record further demonstrate that sediment availability is the dominant driver of dust accumulation in the SJM. Both biogenic and non-biogenic soil crusts that protected soils from wind erosion over the past 11,000 years have been disturbed by a range of human activities (Belnap, 1995; Belnap and Eldridge, 2003). In addition, dust limiting vegetation systems (perennial grasses and taller vegetation) have been transforming into shrublands (Ravi et al., 2010) by overgrazing (Grover and Musick, 1990; Okin et al., 2011) and by invasive species (Reheis and Urban, 2011) such as Russian thistle (Draut et al., 2012).

There appears to be inconsistency between the theory of higher dust production during drought and our result of insignificant change in dust accumulation during arid intervals that might be related (1) to the drought-dust production nexus not holding true for some parts of the Colorado Plateau, and (2) to transport mechanisms of dust from source to sink. From modern studies, it appears that the drought-dust nexus might hold true for some source regions for the SJM and not for others. Northeastern Arizona shows greater aeolian sand transport during drought due to reduced vegetation cover (Draut et al., 2012). Alternatively, in southeastern Utah, sustained drought conditions will accelerate the likelihood of dust production in the future on disturbed soil surfaces (Munson et al., 2011). Meanwhile for the same region, Flagg et al. (2014) found no relation between dust production and drought for sites with low disturbance. The Colorado Plateau dust source is so large that even with a substantial reduction in sediment mobility, wind still supplies ample material to the SJM (Draut et al., 2012). Finally, different types of dust sources, for example, playas, dunes, ephemeral washes and bedrock, may be triggered under different conditions and not necessarily simultaneously. This is apparent in our record as a constant background of dust accumulation, although short-term (e.g. <10 year) variability in dust deposition likely occurs but is obscured by age uncertainty and sampling resolution. Observations of dust emission, transport and deposition in the Southwest United States over the past have indicated that all phases of the dust cycle are complex and modulated by a myriad of factors and boundary conditions (Draut et al., 2012; Flagg et al., 2014; Munson et al., 2011).

Transport mechanisms from source to sink have been used to explain discordance in records (e.g. Marx et al., 2018; Pye, 1989). For example, Pye (1989) explains the disagreement between Chinese loess deposits and dust accumulation in marine cores in the Northwest Pacific as reflecting the frequency and effectiveness of frontal depressions. Pacific frontal storms are most likely the transport mechanism for dust deposition in the Rocky Mountains as they travel through the Southwest and reach the SJM during Nov-Mar (Keen, 1996). The lowest rate of dust accumulation in our record occurs during the middle Holocene. As mentioned previously, winters during the middle Holocene might have been drier than today (Barron et al., 2012) indicating Pacific storms would have been less frequent thus reducing the effectiveness of dust transport to the mountains. In summary, interactions between sediment supply and transport mechanisms appear to be the dominant drivers of dust accumulation in the SJM, whereas aridity likely plays a secondary and complex role.

Conclusions and implications for the future

New dust deposition records from three lakes in the SJM suggests that changes in dust deposition on the Colorado Plateau is primarily driven by changes in sediment availability and transport mechanisms rather than hydroclimate. Our results further demonstrate that while being naturally prone to dustiness, dust deposition on the Colorado Plateau has remained relatively stable over the course of the Holocene. DMAR was highest during the LGIT when the climate of the Colorado Plateau was transitioning from humid to arid and sediment produced during the LGM was made available to wind erosion. In contrast, DMAR was lower during the Holocene as biogenic soil crusts and desert pavements stabilized the surfaces and limited dust production. The middle Holocene saw the lowest DMAR as winter Pacific storms were likely reduced potentially decreasing the effectiveness of dust transport. Finally, dust deposition increased by 60% compared with the late-Holocene baseline following European settlement in the southwest. The greater impact of human disturbance compared with climate over the last 11,000 years suggests the dust deposition phase of the dust cycle on the Colorado Plateau is highly sensitive to human activities. This suggests that land-use management decisions aimed at reducing land disturbance can mitigate future dust accumulation, despite projected increases in regional aridity.

Supplemental Material

ARCUSAetal_Supplement_revised – Supplemental material for Dust-drought interactions over the last 15,000 years: A network of lake sediment records from the San Juan Mountains, Colorado

Supplemental material, ARCUSAetal_Supplement_revised for Dust-drought interactions over the last 15,000 years: A network of lake sediment records from the San Juan Mountains, Colorado by Stephanie H Arcusa, Nicholas P McKay, Cody C Routson and Samuel E Munoz in The Holocene

Supplemental Material

Measurements_supplement_revised – Supplemental material for Dust-drought interactions over the last 15,000 years: A network of lake sediment records from the San Juan Mountains, Colorado

Supplemental material, Measurements_supplement_revised for Dust-drought interactions over the last 15,000 years: A network of lake sediment records from the San Juan Mountains, Colorado by Stephanie H Arcusa, Nicholas P McKay, Cody C Routson and Samuel E Munoz in The Holocene

Footnotes

Acknowledgements

We thank E Yackulic, C Mogen, C Wiman, W Wiman, C Webster, A Wong, A Platt, J Chaffeur, M Caron and C Stielstra for the help in the field. We thank C Wiman, A Wong and E Broadman for collecting bathymetry data at Columbine Lake. We are grateful to D Buscombe for letting us use the bathymetric equipment and to D Hammond for analysing the Crater Lake lead and cesium samples. We would like to acknowledge D Kaufman and M Redsteer for valuable comments and insights. We thank RS Anderson for the identification of macrofossils for ¹⁴C dating and K Whitacre for lab assistance.

Author contribution

SHA, NPM and CCR designed the research. NPM developed the code for the Monte Carlo XRF modelling. SEM measured the ²¹⁰Pb and ¹³⁷Cs of Columbine and Clear Lake and contributed to the interpretation of those data and their integration into the age models. CCR provided the dust samples and helped with fieldwork and logistics. SHA led the study, performed the sedimentological and computer analyses and wrote the paper. All authors commented on the drafts.

Data availability

1. DMAR time series and their uncertainties are archived at the World Data Service for Paleoclimatology at: .

2. Code: code for the modified R-based end-member composition algorithm and for the end-member modelling of XRF data will be available upon reasonable request.

3. Third party datasets: upon publication, we will include data citations to the third-party datasets. Some of the third-party datasets are not currently archived.

Fish lake: .

Blue lake: .

PMDI: .

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship and/or publication of this article: Braudy Foundation. Grant number: Dust-Drough Nexus.

ORCID iD

Stephanie H Arcusa

Supplemental material

Supplementary word document contains Figures S1–S14 and Table S1. Supplementary measurement document contains lead measurements for each lake.

References

Aarons

Aciego

Gabrielli

et al. (2016) Ice core record of dust sources in the western United States over the last 300 years. Chemical Geology 442: 160–173.

Abruzzi

(1995) The social and ecological consequences of early cattle ranching in the Little Colorado River Basin. Human Ecology 23(1): 75–98.

Albani

Mahowald

Winckler

et al. (2015) Twelve thousand years of dust: The Holocene global dust cycle constrained by natural archives. Climate of the Past 11(6): 869–903.

Anderson

Jass

Toney

et al. (2008) Development of the mixed conifer forest in northern New Mexico and its relationship to Holocene environmental change. Quaternary Research 69(2): 263–275.

Appleby

(2001) Chronostratigraphic techniques in recent sediments. In: Last

Smol

(eds) Tracking Environmental Change Using Lake Sediments, vol. 1. Dordrecht: Kluwer Academic Publishers, pp.171–203.

Ballantyne

Brahney

Fernandez

et al. (2011) Biogeochemical response of alpine lakes to a recent increase in dust deposition in the Southwestern, US. Biogeosciences 8(9): 2689–2706.

Barron

Metcalfe

Addison

(2012) Response of the North American monsoon to regional changes in ocean surface temperature. Paleoceanography 27(3): PA3206.

Bartlein

Harrison

Brewer

et al. (2011) Pollen-based continental climate reconstructions at 6 and 21 ka: A global synthesis. Climate Dynamics 37(3): 775–802.

Belnap

(1995) Surface disturbances: Their role in accelerating desertification. Environmental Monitoring and Assessment 37: 39–57.

10.

Belnap

(2003) The world at your feet: Desert biological soil crusts. Frontiers in Ecology and the Environment 1(4): 181–189.

11.

Belnap

Eldridge

(2003) Disturbance and recovery of biological soil crusts. In: Belnap

Lange

(eds) Biological Soil Crusts: Structure, Function, and Management. Ecological Studies (Analysis and Synthesis). Berlin: Springer, pp.363–383.

12.

Belnap

Walker

Munson

et al. (2014) Controls on sediment production in two U.S. deserts. Aeolian Research 14: 15–24.

13.

Beuscher

Krüger

Ehrmann

et al. (2017) End-member modelling as a tool for climate reconstruction – An Eastern Mediterranean case study. PLoS ONE 12(9): e0185136.

14.

Blaauw

Christen

(2011) Flexible paleoclimate age-depth models using an autoregressive gamma process. Bayesian Analysis 6(3): 457–474.

15.

Boygle

(1999) Variability of tephra in lake and catchment sediments, Svı́navatn, Iceland. Global and Planetary Change 21(1–3): 129–149.

16.

Brahney

Ballantyne

Kociolek

et al. (2014) Dust mediated transfer of phosphorus to alpine lake ecosystems of the Wind River Range, Wyoming, USA. Biogeochemistry 120(1–3): 259–278.

17.

Brahney

Ballantyne

Sievers

et al. (2013) Increasing Ca^2⁺ deposition in the western US: The role of mineral aerosols. Aeolian Research 10: 77–87.

18.

Carling

Fernandez

Johnson

(2012) Dust-mediated loading of trace and major elements to Wasatch Mountain snowpack. Science of the Total Environment 432: 65–77.

19.

Chen

Qiang

Zhou

et al. (2013) A 2000-year dust storm record from Lake Sugan in the dust source area of arid China. Journal of Geophysical Research Atmospheres 118(5): 2149–2160.

20.

Choobari

Zawar-Reza

Sturman

(2014) The global distribution of mineral dust and its impacts on the climate system: A review. Atmospheric Research 138: 152–165.

21.

Chu

Sun

Zhaoyan

et al. (2009) Dust records from varved lacustrine sediments of two neighboring lakes in northeastern China over the last 1400 years. Quaternary International 194(1–2): 108–118.

22.

Clow

Williams

Schuster

(2016) Increasing aeolian dust deposition to snowpacks in the Rocky Mountains inferred from snowpack, wet deposition, and aerosol chemistry. Atmospheric Environment 146: 183–194.

23.

Cohen

(2003) Paleolimnology: The History and Evolution of Lake Systems. New York: Oxford University Press.

24.

Conroy

Overpeck

Cole

et al. (2013) Dust and temperature influences on glaciofluvial sediment deposition in southwestern Tibet during the last millennium. Global and Planetary Change 107: 132–144.

25.

Cook

Ault

Smerdon

(2015) Unprecedented 21st century drought risk in the American Southwest and Central Plains. Science Advances 1(1): e1400082.

26.

Cook

Seager

Heim

et al. (2010) Megadroughts in North America: Placing IPCC projections of hydroclimatic change in a long-term palaeoclimate context. Journal of Quaternary Science 25(1): 48–61.

27.

Davis

Ford

(1982) Sediment focusing in Mirror Lake, New Hampshire. Limnology and Oceanography 27(1): 137–150.

28.

Davis

Shafer

(1992) A Holocene climatic record for the Sonoran Desert from pollen analysis of Montezuma Well, Arizona, USA. Palaeogeography, Palaeoclimatology, Palaeoecology 92(1–2): 107–119.

29.

Dietze

Hartmann

Diekmann

et al. (2012) An end-member algorithm for deciphering modern detrital processes from lake sediments of Lake Donggi Cona, NE Tibetan Plateau, China. Sedimentary Geology 243–244: 169–180.

30.

Draut

Redsteer

Amoroso

(2012) Recent seasonal variations in arid landscape cover and aeolian sand mobility, Navajo Nation, southwestern United States. Geophysical Monograph Series 198: 51–60.

31.

Ellingson

(1996) Volcanic rocks. In: Blair

(ed.) The Western San Juan Mountains: Their Geology, Ecology, and Human History. Boulder, CO: University Press of Colorado, pp. 68–79.

32.

Ellwein

Mahan

McFadden

(2015) Impacts of climate change on the formation and stability of late Quaternary sand sheets and falling dunes, Black Mesa region, southern Colorado Plateau, USA. Quaternary International 362: 87–107.

33.

Ely

(1997) Response of extreme floods in the southwestern United States to climatic variations in the Late-Holocene. Geomorphology 19(3–4): 175–201.

34.

Fairlie

Jacob

Park

(2007) The impact of transpacific transport of mineral dust in the United States. Atmospheric Environment 41: 1251–1266.

35.

Fischer

Siggaard-Andersen

Ruth

et al. (2007) Glacial/interglacial changes in mineral dust and sea-salt records in polar ice cores: Sources, transport, and deposition. Reviews of Geophysics 45: RG1002.

36.

Flagg

Neff

Reynolds

et al. (2014) Spatial and temporal patterns of dust emissions (2004–2012) in semi-arid landscapes, southeastern Utah, USA. Aeolian Research 15: 31–43.

37.

Goudie

(2014) Desert dust and human health disorders. Environment International 63: 101–113.

38.

Goudie

Middleton

(eds) (2006) Desert Dust in the Global System. Berlin: Springer.

39.

Grover

Musick

(1990) Shrubland encroachment in southern New Mexico, U.S.A.: An analysis of desertification processes in the American Southwest. Climatic Change 17: 305–330.

40.

Guido

Ward

Anderson

(2007) Pacing the post-last glacial maximum demise of the Animas Valley glacier and the San Juan Mountain ice cap, Colorado. Geology 35(8): 739–742.

41.

Hasbargen

(1994) A Holocene paleoclimatic and environmental record from Stoneman Lake, Arizona. Quaternary Research 42: 188–196.

42.

Zhao

Song

et al. (2015) Onset of frequent dust storms in northern China at ~ AD 1100. Scientific Reports 5: 17111.

43.

Hevly

(1985) A 50,000 year record of Quaternary environments: Walker Lake, Coconino Co., Arizona. In: Jacobs

Fall

Davis

(eds) Late Quaternary Vegetation and Climates of the American Southwest (AASP Contribution Series 16). Dallas: AASP, pp. 141–154.

44.

Huang

Oberhänsli

Von Suchodoletz

et al. (2011) Dust deposition in the Aral Sea: Implications for changes in atmospheric circulation in central Asia during the past 2000 years. Quaternary Science Reviews 30(25–26): 3661–3674.

45.

Jiang

Wan

et al. (2017) End-member modeling of the grain-size record of Sikouzi fine sediments in Ningxia (China) and implications for temperature control of Neogene evolution of East Asian winter monsoon. PLoS ONE 12(10): e0186153.

46.

Keen

(1996) Weather and climate. In: Blair

(ed.) The Western San Juan Mountains: Their Geology, Ecology, and Human History. Boulder, CO: University Press of Colorado, pp.113–126.

47.

Kocurek

(1998) Aeolian system response to external forcing factors – A sequence stratigraphic view of the Saharan region. In: Alsharan

Glennie

Whittle

et al. (eds) Quaternary Deserts and Climatic Change. Rotterdam; Brookfield: Balkema, pp. 327–338.

48.

Kullback

Leibler

(1951) On information and sufficiency. The Annals of Mathematical Statistics 22(1): 79–86.

49.

Lancaster

(1997) Response of eolian geomorphic systems to minor climate change: Examples from the southern Californian deserts. Geomorphology 19: 333–347.

50.

Lawrence

Neff

Farmer

(2011) The accretion of aeolian dust in soils of the San Juan Mountains, Colorado, USA. Journal of Geophysical Research 116(F2): F02013.

51.

Lawrence

Painter

Landry

et al. (2010) Contemporary geochemical composition and flux of aeolian dust to the San Juan Mountains, Colorado, United States. Journal of Geophysical Research: Biogeosciences 115(G3): G03007.

52.

Lim

Matsumoto

(2006) Bimodal grain-size distribution of aeolian quartz in a maar of Cheju Island, Korea, during the last 6500 years: Its flux variation and controlling factor. Geophysical Research Letters 33(21): L21816.

53.

Lipman

Mcintosh

(2011) Tertiary volcanism in the Eastern San Juan Mountains. In: Blair

Bracksieck

(eds) The Eastern San Juan Mountains: Their Geology, Ecology, and Human History. Boulder, CO: University Press of Colorado, pp. 17–37.

54.

Lowe

(2011) Tephrochronology and its application: A review. Quaternary Geochronology 6: 107–153.

55.

Löwemark

Chen

Yang

et al. (2011) Normalizing XRF-scanner data: A cautionary note on the interpretation of high-resolution records from organic-rich lakes. Journal of Asian Earth Sciences 40(6): 1250–1256.

56.

McKay

Emile-Geay

Heiser

et al. (2018) GeoChronR development repository. Available at: https://dx-doi-org.web.bisu.edu.cn/10.5281/zenodo.60812.

57.

Mahowald

Albani

Kok

et al. (2014) The size distribution of desert dust aerosols and its impact on the Earth system. Aeolian Research 15: 53–71.

58.

Marx

Kamber

McGowan

et al. (2018) Palaeo-dust records: A window to understanding past environments. Global and Planetary Change 165: 13–43.

59.

Maurer

Bowling

(2015) Dust effects on snowpack melt and related ecosystem processes are secondary to those of forest canopy structure and interannual snowpack variability. Ecohydrology 8(6): 1005–1023.

60.

Menking

Anderson

(2003) Contributions of La Ninã and El Ninõ to middle Holocene drought and Late Holocene moisture in the American Southwest. Geology 31(11): 937–940.

61.

Morrill

Lowry

Hoell

(2018) Thermodynamic and dynamic causes of pluvial conditions during the last glacial maximum in western North America. Geophysical Research Letters 45(1): 335–345.

62.

Muhs

Benedict

(2006) Eolian additions to late Quaternary alpine soils, Indian Peaks Wilderness Area, Colorado Front Range. Arctic, Antarctic, and Alpine Research 38(1): 120–130.

63.

Muhs

Ager

Been

et al. (2003) A late Quaternary record of eolian silt deposition in a Maar lake, St. Michael Island, western Alaska. Quaternary Research 60(1): 110–122.

64.

Mulitza

Heslop

Pittauerova

et al. (2010) Increase in African dust flux at the onset of commercial agriculture in the Sahel region. Nature 466(7303): 226–228.

65.

Munroe

(2007) Exploring relationships between watershed properties and Holocene loss-on-ignition records in high-elevation lakes, southern Uinta Mountains, Utah, U.S.A. Arctic, Antarctic, and Alpine Research 39(4): 556–565.

66.

Munson

Belnap

Okin

(2011) Responses of wind erosion to climate-induced vegetation changes on the Colorado Plateau. Proceedings of the National Academy of Sciences 108(10): 3854–3859.

67.

Neff

Ballantyne

Farmer

et al. (2008) Increasing eolian dust deposition in the western United States linked to human activity. Nature Geoscience 1(3): 189–195.

68.

Neff

Reynolds

Belnap

et al. (2005) Multi-decadal impacts of grazing on soil physical and biogeochemical properties in southeast Utah. Ecological Applications 15(1): 87–95.

69.

Neff

Reynolds

Munson

et al. (2013) The role of dust storms in total atmospheric particle concentrations at two sites in the western U.S. Journal of Geophysical Research Atmospheres 118(19): 11201–11212.

70.

Okin

Bullard

Reynolds

et al. (2011) Dust: Small-scale processes with global consequences. Eos 92(29): 241–248.

71.

Painter

Barrett

Landry

et al. (2007) Impact of disturbed desert soils on duration of mountain snow cover. Geophysical Research Letters 34(12): L12502.

72.

Painter

Skiles

Deems

et al. (2012) Dust radiative forcing in snow of the Upper Colorado River Basin: 1. A 6 year record of energy balance, radiation, and dust concentrations. Water Resources Research 48(7): W07521.

73.

Painter

Skiles

Deems

et al. (2018) Variation in rising limb of Colorado River snowmelt runoff hydrograph controlled by dust radiative forcing in snow. Geophysical Research Letters 45: 797–808.

74.

Pederson

(2000) Holocene paleolakes of Lake Canyon, Colorado Plateau: Paleocimate and landscape response from sedimentology and allostratigraphy. GSA Bulletin 112(1): 147–158.

75.

Poore

Pavich

Grissino-Mayer

(2005) Record of the North American southwest monsoon from Gulf of Mexico sediment cores. Geology 33(3): 209–212.

76.

Prins

Vriend

Nugteren

et al. (2007) Late Quaternary aeolian dust input variability on the Chinese Loess Plateau: Inferences from unmixing of loess grain-size records. Quaternary Science Reviews 26(1–2): 230–242.

77.

Prospero

Ginoux

Torres

et al. (2002) Environmental characterization of global sources of atmospheric soil dust identified with the Nimbus 7 Total Ozone Mapping Spectrometer (TOMS) absorbing aerosol product. Reviews of Geophysics 40(1): 1022.

78.

Pye

(1987) Mechanisms of fine-particle formation. In: Pye

(ed.) Aeolian Dust and Dust Deposits. London: Academic Press, pp.10–28.

79.

Pye

(1989) Processes of fine particle formation, dust source regions, and climatic changes In: Leinen

Sarnthein

(eds) Paleoclimatology and Paleometeorology: Modern and Past Patterns of Global Atmospheric Transport. NATO ASI Series (Series C: Mathematical and Physical Sciences), vol. 282. Dordrecht: Springer, pp. 3–30.

80.

Qiang

Liu

Jin

et al. (2014) Holocene record of eolian activity from Genggahai Lake, northeastern Qinghai-Tibetan Plateau, China. Geophysical Research Letters 41: 589–595.

81.

R Core Team (2017) R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing. Available at: https://www.r-project.org/.

82.

Ravi

Breshears

Huxman

et al. (2010) Land degradation in drylands: Interactions among hydrologic – Aeolian erosion and vegetation dynamics. Geomorphology 116(3–4): 236–245.

83.

Rea

(1994) The paleoclimatic record provided by eolian deposition in the deep sea: The geologic history of wind. Reviews of Geophysics 32(2): 159–195.

84.

Reheis

Urban

(2011) Regional and climatic controls on seasonal dust deposition in the southwestern U.S. Aeolian Research 3(1): 3–21.

85.

Reheis

Adams

Oviatt

et al. (2014) Pluvial lakes in the Great Basin of the western United States – A view from the outcrop. Quaternary Science Reviews 97: 33–57.

86.

Reheis

Budahn

Lamothe

et al. (2009) Compositions of modern dust and surface sediments in the Desert Southwest, United States. Journal of Geophysical Research 114(1): F01028.

87.

Reheis

Reynolds

Goldstein

et al. (2005) Late Quaternary eolian and alluvial response to paleoclimate, Canyonlands, southeastern Utah. GSA Bulletin 117(7–8): 1051–1069.

88.

Reimer

Bard

Bayliss

et al. (2013) IntCal13 and Marine13 radiocarbon age calibration curves 0–50,000 years cal BP. Radiocarbon 55(4): 1869–1887.

89.

Reynolds

Belnap

Reheis

et al. (2001) Aeolian dust in Colorado Plateau soils: Nutrient inputs and recent change in source. Proceedings of the National Academy of Sciences 98(13): 7123–7127.

90.

Reynolds

Mordecai

Rosenbaum

et al. (2010) Compositional changes in sediments of subalpine lakes, Uinta Mountains (Utah): Evidence for the effects of human activity on atmospheric dust inputs. Journal of Paleolimnology 44(1): 161–175.

91.

Reynolds

Munson

Fernandez

et al. (2016) Concentrations of mineral aerosol from desert to plains across the central Rocky Mountains, western United States. Aeolian Research 23: 21–35.

92.

Reynolds

Neff

Reheis

et al. (2006a) Atmospheric dust in modern soil on aeolian sandstone, Colorado Plateau (USA): Variation with landscape position and contribution to potential plant nutrients. Geoderma 130(1–2): 108–123.

93.

Reynolds

Reheis

Neff

et al. (2006b) Late Quaternary eolian dust in surficial deposits of a Colorado Plateau grassland: Controls on distribution and ecologic effects. Catena 66(3): 251–266.

94.

Richter

Van der Gaast

Koster

et al. (2006) The Avaatech XRF Core Scanner: Technical Description and Applications to NE Atlantic Sediments. London: Geological Society of London (Special Publications 267), pp. 39–50.

95.

Routson

Arcusa

McKay

et al. (2019) A 4500-year-long record of southern Rocky Mountain dust deposition. Geophysical Research Letters 46: 8281–8288.

96.

Routson

Overpeck

Woodhouse

et al. (2016) Three millennia of southwestern North American dustiness and future implications. PLoS ONE 11(2): e0149573.

97.

Routson

Woodhouse

Overpeck

(2011) Second century megadrought in the Rio Grande headwaters, Colorado: How unusual was medieval drought? Geophysical Research Letters 38(22): L22703.

98.

Sayre

(1999) The cattle boom in southern Arizona: Towards a critical political ecology. Journal of the Southwest 41(2): 239–271.

99.

Seager

Vecchi

(2010) Greenhouse warming and the 21st century hydroclimate of southwestern North America. Proceedings of the National Academy of Sciences 107(50): 21277–21282.

100.

Seidel

Hlawitschka

(2015) An R-based function for modeling of end member compositions. Mathematical Geosciences 47(8): 995–1007.

101.

Sheppard

Comrie

Packin

et al. (2002) The climate of the US Southwest. Climate Research 21: 219–238.

102.

Simonson

(1995) Airborne dust and its significance to soils. Geoderma 65(1–2): 1–43.

103.

Skiles

Painter

Belnap

et al. (2015) Regional variability in dust-on-snow processes and impacts in the Upper Colorado River Basin. Hydrological Processes 29(26): 5397–5413.

104.

Smith

McFaul

(1997) Paleoenvironmental and geoarchaeologic implications of late Quaternary sediments and paleosols: North-central to southwestern San Juan Basin, New Mexico. Geomorphology 21(2): 107–138.

105.

Stokes

Breed

(1993) A chronostratigraphic re-evaluation of the Tusayan Dunes, Moenkopi Plateau, and southern Ward Terrace, Northeastern Arizona. Geological Society, London, Special Publications 72: 75–91.

106.

Stuut

JBW

Temmesfeld

De Deckker

(2014) A 550ka record of aeolian activity near North West Cape, Australia: Inferences from grain-size distributions and bulk chemistry of SE Indian Ocean deep-sea sediments. Quaternary Science Reviews 83: 83–94.

107.

Tanner

(1964) Modification of sediment size distributions. Journal of Sedimentary Research 34(1): 156–164.

108.

Tjallingii

Röhl

Kölling

et al. (2007) Influence of the water content on x-ray fluorescence core-scanning measurements in soft marine sediments. Geochemistry, Geophysics, Geosystems 8(2): Q02004.

109.

Toney

Anderson

(2006) A postglacial palaeoecological record from the San Juan Mountains of Colorado USA : Fire, climate and vegetation history. The Holocene 16(4): 505–517.

110.

Urban

Reynolds

Fulton

(2009) The dynamic interaction of climate, vegetation, and dust emission, Mojave Desert, USA. In: Fernandez-Bernal

De La Rosa

(eds) Arid Environments and Wind Erosion. New York: Nova Science Publishers, Inc, pp.1–25.

111.

US Census Bureau (2002) Measuring America: The Decennial Censuses from 1790 to 2000. Washington, D.C: US Census Bureau.

112.

Varga

Kovács

Újvári

(2012) Late Pleistocene variations of the background aeolian dust concentration in the Carpathian Basin: An estimate using decomposition of grain-size distribution curves of loess deposits. Geologie En Mijnbouw/Netherlands Journal of Geosciences 91(1–2): 159–171.

113.

Wells

McFadden

Schultz

(1990) Eolian landscape evolution and soil formation in the Chaco dune field, southern Colorado Plateau, New Mexico. Geomorphology 3(3–4): 517–546.

114.

Weltje

Tjallingii

(2008) Calibration of XRF core scanners for quantitative geochemical logging of sediment cores: Theory and application. Earth and Planetary Science Letters 274(3–4): 423–438.

115.

Weng

Jackson

(1999) Late Glacial and Holocene vegetation history and paleoclimate of the Kaibab Plateau, Arizona. Palaeogeography, Palaeoclimatology, Palaeoecology 153(1–4): 179–201.

116.

Western Regional Climate Center (2018) Cooperative climatological data summaries. Available at: https://wrcc.dri.edu/cgi-bin/cliMAIN.pl?co7656.

117.

White

(1983) The Roots of Dependency. Lincoln, NE: University of Nebraska Press.

118.

Wong

Potter

Montañez

et al. (2016) Evolution of moisture transport to the western U.S. during the last deglaciation. Geophysical Research Letters 43: 3468–3477.

119.

Yackulic

(2017) Productivity and temperature variability over the past 15000 years at a small Alpine Lake in the southern San Juan Mountains, Colorado. ProQuest Dissertations, Publishing no. 10688325, Northern Arizona University.

120.

Yuan

Koran

Valdez

(2013) Late glacial and Holocene record of climatic change in the southern Rocky Mountains from sediments in San Luis Lake, Colorado, USA. Palaeogeography, Palaeoclimatology, Palaeoecology 392: 146–160.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

5.58 MB

0.05 MB