Friday, June 10, 2022

There Was A Megadrought In The Colorado River Basin In The 100s CE

Colorado is currently in a 1200 year drought (the worst since 800 CE). But, in the 100s CE, around the time that the Roman Empire was at its peak and Christianity was in a formative period, there was a far more severe drought than the current one in the Colorado River basin or the one experienced around 800 CE.

The drought of ca. 800 CE was probably a decisive factor in the collapse of the Anasazi a.k.a. Ancient Puebloan culture. It predated by about two centuries the arrival of the Navajo and Apache people in the American Southwest from what is now Southern central Canada. It also coincided with the arrival of the Thule ancestors of the modern Eskimo people of Alaska and Canada in Western Alaska from Siberia which was a major push factor behind the southern mass migration of the ancestors of the Navajo and Apache people.

The Colorado River is in an extremely severe drought and has been for the last 22 years. To better understand this drought, researchers looked at the drought history within the Colorado River Basin. Previous studies have gone back 1,200 years, but this paper goes back 2,000 years. The findings, using paleo hydrology, show that there was an even more severe drought in the Colorado River Basin in the second century. . . . . The research finds that compared to the current 22-year drought in the Colorado River, with only 84% of the average water flow, the water flow during a 22-year period in the second century was much lower, just 68% of the average water flow.
From Science Daily citing Subhrendu Gangopadhyay, et al., 'Tree Rings Reveal Unmatched 2nd Century Drought in the Colorado River Basin." 49(11) Geophysical Research Letters (June 9, 2022). DOI: 10.1029/2022GL098781

The abstract and other prefatory materials from the paper are as follows:

The ongoing 22-year drought in the Upper Colorado River Basin (UCRB) has been extremely severe, even in the context of the longest available tree-ring reconstruction of annual flow at Lees Ferry, Arizona, dating back to 762 CE. 
While many southwestern drought assessments have been limited to the past 1,200 years, longer paleorecords of moisture variability do exist for the UCRB. Here, gridded drought-atlas data in the UCRB domain along with naturalized streamflow data from the instrumental period (1906–2021) are used in a K-nearest neighbor nonparametric algorithm to develop a streamflow reconstruction for the Lees Ferry gage starting in 1 CE. 
The reconstruction reveals a second-century drought unmatched in severity by the current drought or by well-documented medieval period droughts in the UCRB. Although data are sparse, analysis of individual long tree-ring records and other paleoclimatic data also support the occurrence of an exceptional second-century drought. 
Key Points

A new tree-ring based ensemble streamflow reconstruction spanning the last two millennia was developed for the Colorado River

The new reconstruction reveals a second-century drought unmatched in severity by any past droughts in the Upper Colorado River Basin

The ongoing 22-year period of low Colorado River streamflow is a rare event, but is not the most severe drought in the past 2,000 years

Plain Language Summary

The Colorado River drought we currently are experiencing is severe in the context of the 116-year gage record (1906–2021), but how severe is it in a long-term context? 
Existing tree-ring based reconstructions of Colorado River streamflow have suggested that the 22-year period 2000–2021 could be the worst drought in the southwestern United States in 1,200 years. The purpose of this study is to extend the Colorado River reconstruction back 2,000 years and to evaluate the current drought in a long-term context. 
We find that an even more extreme drought occurred and persisted over much of the second century. Data are sparse this far back in time, but evidence from both tree-ring data and paleoclimate data from lakes, bogs, and caves supports the existence and severity of this drought in the context of the last two millennia. 
Additional work is needed to learn more about this drought and its causes, but we now know that drought more persistent than even the well-documented medieval period droughts occurred in the past, expanding our understanding of the range of natural climate variability.

 The body text goes on to explain that:

Other paleoclimatic data provide additional support for second-century drought (e.g., Routson et al., 2021; Shuman et al., 2018). A collection of 37 non-tree-ring hydroclimate records (e.g., pollen, diatoms, etc.) was assessed for evidence of the second-century drought. Resolution and age control limit these records' ability to faithfully record events that occurred within multidecadal timescales. However, non-tree-ring hydroclimate records are useful because they are not limited by detrending and short segment lengths inherent in tree-ring chronologies (Cook et al., 1995). This collection of lower resolution records shows dry conditions occurred during the second century over a spatial domain broadly consistent with the tree-ring based evidence. Furthermore, these records show a gradual pattern toward wetter conditions over the last two millennia, with the driest conditions occurring in the early portion of the first millennium, coincident with the second century. This period also coincides with the higher frequency of longer and more severe droughts over the first millennium in the tree-ring records.
Local Cultural and Historical Context

In part due to my thinner understanding of North American prehistory, some of which is my own fault, and some of which is due to a lack of a good comprehensive chronology in academia itself, I struggle to recall anything terribly significant or notable going on in the American Southwest in that time period, that would overshadow the well known climate impacts of the less severe drought that was experienced six or seven centuries later. 

This drought coincides with the time period in the region is known among archaeologists as the Late Basketmaker II era (from 50 CE to 500 CE), which is "a cultural period of Ancient Pueblo People when people began living in pit-houses, raised maize and squash, and were proficient basket makers and weavers. They also hunted game and gathered wild foods, such as pinyon nuts." 

There isn't a lot that is known about these people, but what is known is summarized as follows:

The primary dwellings of this era were round or circular pit-houses that were built on open land and partially below the ground surface. The entrance to the house faced east or south. Logs and rocks were often used for the dwellings foundation. The building materials for the walls could include stacked logs, jacal or poles and brush. In the center of the dwelling was a fire pit.

Some early people built their dwellings within the natural protection of rock shelters, especially during the beginning of this period. 

The Basketmaker II people raised maize and squash, the first people of the northern American southwest to do so, which required them to be located near sources of water and good soil. Carbon isotope analysis of bones of Archaic people compared to Basketmakers indicates that the Basketmakers' diet was rich in maize.

Manos and metates were used to grind maize and other foods. Food was stored below ground in storage cists, often lined with slabs of stone.
Material goods

Excavated items from this period include: 
* good quality, tightly woven baskets
* woven yucca bags, sandals and blankets
* robes and blankets made of feather and fur
* stone projectile points, scrapers and knives
* atlatl and throwing spears (the main tools for hunting)
* bone stitching awls, whistles, and gaming pieces
* cord made from yucca and cedar bark
* oval-shaped cradles
* stone pipes
About AD 200, the middle of this period, there was some experimentation with a crude form of brown pottery.
Wikipedia's Timeline of North American Prehistory provides the following entries between and around these two droughts in this part of North America also sometimes called Oasisamerica (all dates CE except as noted in the first two entries):

There is a two centuries gap in the prehistory of the American Southwest that includes the time period of this megadrought. 

The archaeological record would be consistent with the hypothesis that human civilization basically ended as a result of this megadrought in the American Southwest starting about 0 CE and restarted again with Hohokam culture around 200 CE when the Colorado River basin had recovered again. Or, it could simply be that the archaeological record as filtered through to me through secondary sources is just incomplete or inaccurate.

This excerpt from the Wikipedia Hohokam article describes the early phase of this culture and suggests that a 200 CE date may be a bit early for it:
Pioneer/Formative period (AD 1–750)

As farmers of corn and beans, early Hohokam founded a series of small villages along the middle Gila River. The communities were located near arable land, and dry farming was common early in this period. Water wells, usually less than 10 feet (3 m) deep, were dug for domestic water supplies. Early Hohokam homes were built of branches that were bent, covered with twigs or reeds and heavily applied mud, and other available materials.
Crop, agricultural skill, and cultural refinements increased between AD 300 and 500 as the Hohokam acquired a new group of cultivated plants, presumably from trade with peoples in the area of modern Mexico. These acquisitions included cotton, tepary bean, sieva and jack beans, cushaw and warty squash, and southwestern pigweed. Agave species had been gathered for food and fiber for thousands of years by southwestern peoples, and around 600, the Hohokam began cultivating agave, particularly Agave murpheyi ("Hohokam agave"), on large areas of rocky, dry ground. Agave became a major food source for the Hohokam to augment the food grown in irrigated areas.  
Engineering improved access to river water and the inhabitants excavated canals for irrigation. 
Evidence of trade networks include turquoise, shells from the Gulf of California, and parrot bones from central Mexico. 
Seeds and grains were prepared on stone manos and metates. 
Ceramics appeared shortly before AD 300, with pots of unembellished brown used for storage and cooking, and as containers for cremated remains. Materials produced for ritual use included fired clay human and animal figures and incense burners.

The Supplemental Information discusses the Second Century drought in the American Southwest further:

The 2nd century drought in this study's Colorado River reconstruction occurs in the earliest part of the time span identified using tree-ring data for the Upper Colorado River basin region. A total of 11 tree-ring chronologies in or near the Upper Colorado River basin extend back into the 2nd century. Data from most of these sites were included in the tree-ring network used to reconstruct PDSI. Species range from high elevation bristlecone pine (Pinus aristata) to relatively low-elevation pinyon (Pinus edulis). 
In order to evaluate the 2nd century drought at these sites, tree-ring chronologies were standardized, converted to z-scores, and then smoothed with a 25-year running average. Time series plots show a period of severe drought occurring at most of these sites over the middle to the end of the 2nd century. In some cases, this drought is not the most severe and sustained in the record, but it is often one of the worst. 
The 2nd century drought is particularly notable at the Summitville (Routson et al., 2011), Harmon Canyon (Knight et al., 2010), and Mammoth creek sites. Based on these sites, the heart of this drought appears to be centered over Utah and Colorado. This pattern is generally reflected in the gridded PDSI values averaged over the years 120–180 CE (Cook et al., 2010). 
From a regional perspective, the severity of the 2nd century drought does stand out in the context of the past 2,000 years. A composite of the 11 long tree-ring records, again 7 based on standardized chronologies (z-scores), smoothed with a 25-year moving average, was generated. Bootstrapping was used to test how subsets of the chronologies might influence the composite, particularly since several chronologies showed the 2nd century to be an extremely severe event. In this process, the pool of 11 smoothed records was randomly sampled with replacement. These randomly sampled smooth records were averaged together to create a composite. This step was repeated 1,000 times, creating 1,000 composites. The median of the composites clearly shows the 2nd century to be the worst drought in this region. 
Chronology development typically includes all samples available for a site, regardless of the start and end date of each sample. In order to determine the number of trees that actually provide evidence for the 2nd century drought, we examined the tree-ring measurement series within each of the chronologies. There are approximately 63 individual tree samples from the 11 sites that include at least 4 years in the 2nd century. Numbers of trees were estimated from the sample codes (typically, a common tree number is used with differentiation of the within-tree cores). All of these samples come from dead trees, except for two trees at the Red Canyon site. While the number of samples is not large, these data provide evidence that site chronologies are reflecting tree-growth patterns from multiple trees (3-13) at these sites. There is clear evidence in the tree-ring data, sparse as it is, for the 2nd century drought. However, it is difficult to precisely assess the severity and duration of this drought relative to other severe, sustained droughts because only a small subset of chronologies in the Upper Colorado River basin extend to the 2nd century. 
Furthermore, in the subset of chronologies that do extend to the 2nd century drought, reductions in sample depth (fewer samples during the 2nd century) make direct comparisons with other events challenging. 
Finally, very few of the samples that reflect the 2nd century drought extend through other droughts. There are only two series (both from Summitville) that extend from the 2nd century through the 1100s, which contains the most severe, sustained drought in the Colorado River in the past 12 centuries (Meko et al., 2007). There also are three series from Red Canyon, but this site is the one site in this set that does not reflect the 2nd century drought. 
Other paleoclimatic data are available that have the potential to confirm the existence of the 2nd century drought. We evaluated hydroclimatic records published in two recent syntheses efforts for western North America (Routson et al., 2021; Shuman et al., 2018). Records from those studies were screened using the following criteria: 
1) Are within the geographic region 30°N–50°N, and 95°W–125°W. 
2) Are interpreted as hydroclimate variables (precipitation or precipitation minus evaporation). 
3) Span at least 1,000 years during the last 2,000 years. 
4) Have a median sample resolution of 100 years. 
5) Include data points during the 2nd century CE. 
6) Have a minimum of 2 age control points in the last 2,000 years. 
Thirty-seven (37) individual records met the screening criteria. Hydroclimate anomalies during the 2nd century CE were identified by truncated records to span the last 2,000 years, then averaging data points into 20-year windows. The 20-year binned records then were normalized by their mean and standard deviation (resulting in zscores). Anomalies were calculated as the average of the binned zscores during the 2nd century. The anomalies plotted show a region that includes the UCRB that was generally dry during the 2nd century. Wetter conditions in the Great Basin also are consistent with tree-ring data (Hughes & Graumlich, 1996). A composite index of the 37 hydroclimate records was developed, generated over 1,000 iterations, and sampling with replacement over each iteration. Sampling with replacement helps to quantify the impact of the record selection. The final composite, the median of the ensemble, shows a period of negative anomalies centered around 100 CE. However, there are many caveats around using these kinds of records for evaluating relatively short events. The records were mostly generated to characterize Holocene timescales, and the age control is not sufficient to confidently put data points within 100-year windows. 
Given this evidence, what is the bottom line? 
The tree-ring data analyzed here are essentially the same data that are the basis for the Colorado River reconstruction developed in this study, so they do not provide independent verification of this drought. However, tree-ring chronologies enter the reconstruction via gridded PDSI, which does not readily allow an assessment of the footprint of 2nd century drought on individual trees and chronologies. With a sparse tree-ring network in the 2nd century, it is especially important to appreciate where these trees were located, and which contributed to what we have inferred as a singularly severe drought. The extended analyses here provide some insights on the numbers of records that have the potential to document this drought. While not a large number, the samples come from 11 different sites, with most documenting the 2nd century as at least one of the most severe droughts in the past 2,000 years. The three sites that most strongly document the 2nd century drought are from three different tree species and are spatially distant (Harmon Canyon, Mammoth Creek, and Summitville), indicating that the severity is not site- or species-specific. Independent paleoclimate proxy records support the occurrence of this drought, although the dating uncertainty make it challenging to determine temporal concurrence. 
These data suggest that this period of drought may have been centered on a region that includes the UCRB. A challenging aspect of this evaluation is determining the relative severity of this drought in the context of the past 2,000 years. Without more tree-ring data, and particularly series from trees that lived through both the 2nd century and medieval period droughts, there remains uncertainty regarding drought severity.


Guy said...

Hi Andrew, You must have considered posting this in WPP, since there is a dangling link to such.

andrew said...

I posted there in error, moved it here, and the constant loss amnesia problem I've had with the blogger interface struck again. I killed the bad link and put in a verbal description until I can track down the right one again.