2.1 The documentary record

A search was made for all primary historical records surviving from or directly relevant to every documented European colonial expedition that spent time in the present-day US and Canada from 1510 to 1610 CE. This search resulted in several thousand documents contained in several hundred printed volumes, as well as electronic collections and unpublished archival series in the Archivo General de Indias (Spain) and National Archives (UK), mostly available online (see the Supplement). The author read each of these personally and, where possible, in their entirety. Some documents were consulted in English translations, but in those cases almost all descriptions of weather, climate, or climatic impacts were read again in their original languages (French, Spanish, German, Italian, Latin, and Dutch) to correct any mistranslations and to ensure accuracy. All dates were converted from the Julian calendar into modern calendar dates with the new year beginning on 1 January. Descriptions of weather, climate, and climatic impacts were not simply extracted from their context and compiled separately but always considered contextually, meaning that the study took into account (1) the description's context within the individual document and its larger corpus; (2) the conditions of the observation and of its recording and publication; (3) the background of the author or authors, including their subjective sense of “normal” weather or climate; and (4) the genre of writing in which any description appeared. (For a complete book-length account of the historical context of climate observations in the course of early exploration and colonies, see White, 2017.)

The exploration and colonization of North America proceeded in sporadic, uncoordinated ventures by Spanish, French, and English companies until 1610 CE, by which date enduring colonies were established at St. Augustine (Florida), Santa Fe (New Mexico), Jamestown (Virginia), and Quebec (Quebec). Written first-hand observations potentially capable of providing useful information concerning the presence, absence, or impacts of droughts were left by members of colonial expeditions for the following regions and years: the southeastern US and Gulf Coast, principally Florida (1526, 1528–1529, 1539–1543, 1559–1560, 1560, 1562–1610); the southwestern US, principally New Mexico (1530–1535, 1540–1541, 1581, 1583, 1590, 1598–1610); Virginia and North Carolina (1570, 1585–1588, 1607–1610); the California coast (1542, 1587, 1595, 1596, 1602–1603); the New England coast (1524, 1602–1605, 1607–1608); Nova Scotia (1606–1607); and Quebec (1600–1601, 1608–1610).

The relevant surviving records fall broadly into six genres: private correspondence, official correspondence and memoranda, correspondence of religious orders, pamphlets, travelogues, and finally chronicles and other second-hand compilations of information, which were used only to supplement first-hand information. Each of these genres has been used in historical climatology and each presents particular strengths and weaknesses as a source of weather, climate, and impact evidence (Pfister, 2018; Pfister and White, 2018). However, the unique conditions under which early colonial records were compiled endow them with peculiar advantages and disadvantages compared to other similar written records. On the one hand, events during the exploration of new lands and the colonization of new territories were unusually well recorded for the period, and their records have been carefully sought out, preserved, and analyzed by scholars concerned with the historical and environmental significance of colonization. Moreover, observers present in unfamiliar environments and concerned with prospects for resource extraction, trade, colonization, or missionization usually devoted more attention to environmental features, including weather and climate, than those at home (Taylor, 1993; White, 2015b; Zilberstein, 2016). Europeans on early colonial expeditions were often acutely vulnerable to climatic variability and extremes. On the other hand, the earliest colonial observations in each location necessarily predate the establishment of local institutions or practices that could produce regular records containing climate proxies, such as officially prescribed annual harvest dates. Furthermore, the novelty of colonial environments made it difficult for observers to determine what was normal weather or climate or to identify reliable phenological markers of variability or extremes. Adding to their confusion, early European explorers and colonists of North America expected climates to align with latitudes, overlooking the differences between Europe's predominately maritime climates and North America's predominately continental ones (Kupperman, 1982; Rockman, 2010). This study aims to evaluate whether the advantages of these sources outweigh their potential drawbacks when it comes to reconstructing past weather and climate.

This study identified four main types of evidence concerning drought for early colonial North America. (1) Phenological descriptions, including the growth of plants and conditions of rivers, provide the most objective indicators of drought, but they are not common in all sources. (2) Narrative descriptions, describing a lack of precipitation, are found more frequently but provide less certain indicators of the presence or absence of drought. (3) Societal impacts, particularly crop failures or famine attributed to drought, appear most frequently in the sources and may provide confirmation that conditions were unusually severe. However, these descriptions need to be considered in the context of societal vulnerabilities, which varied according to the colony or Native American society in question. Most of the populations discussed in this study relied heavily on crops of maize, which in turn depended on adequate summer rainfall and a lengthy frost-free growing season. (4) Rain-making ceremonies represent a peculiar but potentially valuable type of evidence for drought. Measuring the occurrence and scale of officially sanctioned rain prayers, known as rogation ceremonies, has been demonstrated as a reliable method of drought reconstruction in Spain and Spanish America (Domínguez Castro et al., 2008, 2018). Historians have identified similar ceremonies outside Spain in many different religious and cultural traditions, but these have not yet been tested in the same fashion. In the colonial North American context, many early European expeditions reported performing rain prayers, being asked by Native Americans to perform rain prayers, or observing indigenous communities performing rain-making ceremonies. Some historians have argued that such accounts could have been biased or even falsified by early European observers eager to present Native Americans as simple pagans ready to be won over to Christianity. However, a close review of the context and consistency of these accounts across sources has indicated that most were probably based on actual events (Kupperman, 2007b; White, 2015a).

Information concerning drought was sorted by each type of evidence. It was evaluated according to whether the information is based on a single observer, two or three observers, or many observers in agreement, and according to whether it provides a definite or indefinite impression of drought. For instance, an observation that it had not rained for a certain number of months was taken as a definite impression, whereas a general description of dry weather was taken as indefinite. Results and further information for specific expeditions are provided in Sect. 3.2.

2.2 Reconstructions from proxies in natural archives

Most high-resolution reconstructions of drought in North America for the pre-instrumental period have relied on measures of the variation in the width of growth rings in trees whose growth has been limited by soil moisture during the growing season. NADA is a set of annual June–August PDSI reconstructions based on 1845 tree-ring chronologies estimated on a 0.5^∘ latitude–longitude grid, which covers most of North America for the past several centuries (Cook et al., 1999, 2010). The PDSI reconstructed in NADA is a function of precipitation and evapotranspiration from the previous winter through the summer growing season. NADA has been found to have a dominant winter precipitation signal in the southwest and Florida and a dominant summer signal in most of the rest of the continent; however, the signal at any given location may depend on local conditions and the proximity of tree-ring samples (St. George et al., 2010). For this reason, variations in PDSI over large regions reconstructed by NADA may not always match local meteorological droughts or agricultural droughts reported in historical observations.

For the area between 30 and 50^∘ N, covering this study area, validation tests produced median CVRE (calibration period reduction of error calculated by leave-one-out cross-validation) of 0.43, VRSQ (verification period square of the Pearson correlation) of 0.33, VRE (verification period reduction of error) of 0.31, and VCE (verification period coefficient of efficiency) of 0.27 (Cook et al., 2010). However, NADA reconstruction skill varies significantly by location. Among the areas covered by this study, reconstruction skill is highest in New Mexico and North Carolina, and high in most of the southeastern US and Virginia but reduced in Florida, and poorer in Quebec, the New England coast, and especially Nova Scotia. (See complete validation statistics for each grid point at https://www1.ncdc.noaa.gov/pub/data/paleo/drought/LBDA2010/LBDA_PMDI_JJA_Recons_cal:ver_stats.txt, last access: 1 October 2019) Other tree-ring-based precipitation reconstructions have produced similar results, with similar timing and magnitude of droughts during the early colonial period, for several locations, including New Mexico (Meko et al., 1995; Grissino-Mayer et al., 2002; Margolis et al., 2011), northern Florida (Harley et al., 2017), Georgia and the Carolinas (Stahle and Cleaveland, 1994), and Virginia (Stahle et al., 1998). Studies of band thickness in a New Mexico speleothem (Asmerom et al., 2013) and ∂¹⁸O in a West Virginia speleothem (Hardt et al., 2010) provide some confirmation at lower temporal resolution for major droughts identified in NADA for those regions during the late 16th century and early 17th century, respectively.

This study systematically compared the NADA reconstruction to the information in the documentary record of each expedition identified in Sect. 2.1. Using the visualization tools at http://drought.memphis.edu/NADA/ (last access: 1 October 2019), this study produces (1) maps of PDSI for each region of travel where definite indications of drought were found in the documentary record and (2) time series of PDSI variation at the location of specific colonies discussed in the text (Sect. 3.2). Where the quality and quantity of evidence permitted, the routes taken by expeditions and/or the locations of settlements were added to the NADA maps based on information in White (2017).

3.1 Regions where droughts were not documented

Early colonial expeditions left no first-hand written records clearly indicating droughts in any of the following regions: the California coast, the New England coast, Quebec, or Nova Scotia, apart from an observation of “fine weather all winter [1606–1607]” at Port Royal (Nova Scotia). The NADA PDSI reconstruction indicates drought in New England (that is, PDSI between −1 and −2 for most grid points in the region) during 1603, when an English voyage passed by the Massachusetts coast in summer, and in 1604, when a French colony was founded at the mouth of the St. Croix River in autumn; it indicates drought throughout Nova Scotia (PDSI $\le -\mathrm{2}$) in 1606–1607, while a French colony was settled on the west coast of the peninsula; and it indicates drought throughout southern California (PDSI between −1 and −2 for most grid points in the region) during 1542 and 1595, when Spanish expeditions were exploring the coast.

Table 1Summary of written evidence. A capital letter “Y” denotes a definite impression of drought in either the winter or summer half year; a lowercase “y” denotes an indefinite impression. Letters in bold indicate many mutually supporting observations, letters in roman font indicate similar observations in two or three personal sources and/or information contained in an official report, and letters in italics indicate a single observer. A letter “N” indicates information in the documentary record indicating abundant precipitation or lack of drought. An explanation of sources and types of evidence is provided in Sect. 2.1; details for each expedition are given in Sect. 3.2.

Download Print Version | Download XLSX

3.2 Regions where droughts were documented

Early colonial expeditions left first-hand written records indicating droughts in the following regions: the southeastern US, the southwestern US, and Virginia and North Carolina. Table 1 provides a summary of the type of information concerning drought from these expeditions.

3.2.1 Southeast

The first Spanish colonial expeditions to the southeast, led by Juan Ponce de León in 1513 and 1521 and Lucas Vázquez de Ayllón in 1526, left no first-hand evidence concerning climatic conditions during their expeditions. In 1528, Pánfilo de Narváez led an overland expedition along the Gulf Coast from roughly Tampa Bay to the mouth of the Mississippi, where its members, traveling on rafts, were swept by a storm to the Gulf Coast of Texas. A few survivors remained in the region as captives of Native Americans until 1534; their testimonies, recorded years later, described cold and famine among the indigenous populations but left no definite indications of drought during these years. The NADA reconstruction indicates droughts (PDSI $\le -\mathrm{1}$) in parts of Florida and the Gulf Coast traversed by these expeditions in 1521, 1528–1529, and 1532.

During 1539–1543, a much better documented Spanish expedition led by Hernando de Soto landed in northern Florida and traveled throughout the southeastern United States before departing down the Mississippi River. No observers described drought conditions in 1539–1540, during which NADA indicates a positive hydroclimate anomaly. Multiple independent eye-witness accounts describe a 1541 episode in which Native Americans in present-day Arkansas asked Soto to pray for rain in order to end a drought and save their maize crop. In early 1542, the expedition ventured into eastern Texas, where one witness reported frequent wet weather; late that year, the expedition returned to Arkansas, which another source described as experiencing continued drought (specifically, it had not rained for the past month). The NADA map for 1541 (Fig. 1a) indicates a drought (PDSI $\le -\mathrm{1}$) centered on northeast Arkansas during 1541, while most of the region traversed by Soto that year enjoyed moist conditions (PDSI > 0). The NADA map for 1542 indicates a positive hydroclimate anomaly in east Texas and negative hydroclimate in Arkansas (Fig. 1b). The NADA time series for the approximate site of the rain-making ceremony (Fig. 1c) indicates that 1541 was the start of a longer drought at this location, as measured by PDSI.

Figure 1(a) NADA JJA PDSI reconstruction for 1541, Soto expedition. (b) NADA JJA PDSI reconstruction for 1542, Soto expedition. (c) NADA JJA PDSI reconstruction time series for 1530–1550 at an approximate location of a 1541 observed rain-making ceremony.

In late summer 1559, Spanish conquistador Tristán de Luna y Arellano led over 1500 soldiers and colonists to settle in Pensacola Bay, Florida, where a hurricane soon destroyed most of their ships and supplies. Unable to live off the land and facing famine, the colony retreated inland to central Alabama, while a detachment explored the region of northeast Alabama and northwestern Georgia, thought to be more promising for settlement. The expedition's breakdown left ample official documentation, including many complaints about the region's environment and climate. These mentioned the frequency and unpredictability of rains – in contrast to those of Mediterranean climates on the same latitude across the Atlantic – but not drought. The NADA composite map of the area in 1559–1560 (Fig. 2a) indicates slightly dry conditions (PDSI between 0 and −1) around Pensacola Bay and slightly wet conditions (PDSI between 0 and 1) in northeastern Alabama and northwestern Georgia, with average conditions in central Alabama, as also indicated in the time series for the approximate location of the colony in that region (Fig. 2b).

Figure 2(a) NADA JJA PDSI reconstruction for 1559–1560, Luna expedition. (b) NADA JJA PDSI reconstruction time series for 1550–1570, central Alabama.

Between 1562 and 1565, French Huguenots attempted to establish colonies on the coast between present-day South Carolina and northern Florida. These were overrun in 1565 by Spanish soldiers, who established the permanent colony of St. Augustine, Florida, as well as several outlying posts including St. Elena, which lasted until 1587. The French colonies, as described in several personal accounts by eye witnesses, suffered from frequent shortages and sometimes famine blamed on poor supplies, shipwrecks, and inability to obtain food from local indigenous communities. The Spanish presence left a much more complete official record, principally letters from the governor to the viceroy and the correspondence of Catholic missionaries. The former describe droughts that led to failure of wheat and maize crops during 1565–1566 in the region between South Carolina and northern Florida; the latter describes Native Americans asking the colony's governor at St. Augustine to pray for rain in 1566 to end a long drought. Official correspondence also indicates that drought led to poor harvests at the St. Augustine colony in 1583, 1588–1589, 1591, and 1598–1599; for instance, the governor reported that 1588 was a “very great drought” (muy gran seca) and maize could not be sown. Food shortages at the colony and surrounding Native American missions were reported for the early 1590s, but these were blamed on late shipments of supplies rather than drought. From 1601 onward, it is possible to calculate maize tithes for most years, and these calculations reveal poor harvests in 1604 and 1609 (Hoffman, 2002). The NADA composite map for 1565–1568 (Fig. 3a) indicates a severe negative PDSI anomaly centered on South Carolina, while the NADA time series at St. Augustine for 1560–1580 (Fig. 3b) and 1580–1610 (Fig. 3c) indicate drought (PDSI $\le -$1) 1566–1570, 1575, 1583–1589, 1591–1595, 1598–1599, 1604, and 1610.

Figure 3(a) NADA JJA PDSI reconstruction for 1565–1568, French and Spanish Florida. (b) NADA JJA reconstruction time series for 1560–1580, St. Augustine. (c) NADA JJA reconstruction time series for 1580–1610, St. Augustine.

3.2.2 Southwest

During 1534–1535, the four survivors of the Pánfilo de Narváez expedition (see Sect. 3.2.1) ventured south into today's Tamaulipas (Mexico), then northwest near the Rio Grande valley. Their testimonies describe dry conditions in the Sierra Madre Oriental but only a single definite indication of drought: survivor Cabeza de Vaca reported that Native Americans in west Texas in late summer 1535 asked them to pray for rain because it had not rained enough to plant maize during the past year. A comparison with the NADA map of PDSI for that year (Fig. 4) indicates average to dry conditions in that region in 1535.

Figure 4NADA JJA PDSI reconstruction for 1535, Cabeza de Vaca expedition.

During 1540 and 1541, a large expedition from Mexico led by governor Francisco Vázquez de Coronado occupied first the Zuni Pueblos of today's western New Mexico, then various Puebloan towns of the central Rio Grande valley. Of the many surviving official and personal accounts of the expedition, several mentioned heavy winter snow and none mentioned drought. The NADA composite map for New Mexico 1540–1541 (Fig. 5a) shows exceptionally moist conditions (PDSI between 2 and 3), as does the time series for the approximate location of the Spanish base camp, near present-day Bernalillo (Fig. 5b).

Figure 5(a) NADA JJA PDSI reconstruction for 1540–1541, New Mexico. (b) NADA JJA reconstruction time series for 1530–1550 at the approximate location of Coronado's base, near present Bernalillo, NM.

During 1581, 1583, and 1590–1591, three small unauthorized expeditions entered New Mexico from New Spain. Each traveled through the territory of the present-day state for only a few weeks or months, mainly along the Rio Grande, leaving no enduring colony; and each produced only two or three written testimonies by eye witnesses. A witness of the 1581 expedition described a “climate like that of Castile”: a possible indication of deficient summer rainstorms, since unlike Castile, most of New Mexico usually receives more precipitation in summer than winter. This witness also described a Puebloan rain-making ceremony, although it is not clear whether this was a normal seasonal ritual or a specific response to drought. A witness from the 1583 expedition described heavy snow early in the year but reported Puebloans refusing to share maize due to a “lack of rain” and poor harvest that summer. The 1590–1591 expedition traveled mainly during the winter months and left testimonies of exceptional cold and heavy snow but not drought. The NADA reconstructions indicate near-average conditions in 1581 (Fig. 6a) and negative PDSI anomalies across most of the present-day state in 1583 (Fig. 6b) and in 1591 (Fig. 6c).

Figure 6(a) NADA JJA PDSI reconstruction for 1581, New Mexico. (b) NADA JJA PDSI reconstruction for 1583, New Mexico. (c) NADA JJA PDSI reconstruction for 1591, New Mexico.

Download

A 1598 invasion led by Juan de Oñate occupied present-day New Mexico and subjugated the Pueblos of the region. In 1601, the expedition nearly collapsed when a large number of officers and settlers defected and Catholic missionaries protested the abuse of Pueblos. Narratives of Oñate's campaign, as well as disputes arising from the colony's near collapse, left a large body of personal and official evidence concerning conditions in New Mexico. The earliest indication of a drought in the documentary record occurs in a description of Pueblos asking colonists to pray for rain in 1598; however, this episode appears only in one contemporary source, which contains some fictionalized material. Colonists during 1599 and 1600 compared the climate in New Mexico to that of Castile, with predominately winter precipitation, possibly indicating deficient summer rainstorms. Colonists and missionaries left abundant official testimonies and personal letters describing extreme summer drought during 1600 and 1601, along with descriptions of starvation when the Pueblo maize crop failed and livestock died for want of pasture. However, descriptions also indicate that the winter of 1600–1601 was exceptionally snowy and icy, and that the colony's crop of irrigated winter wheat yielded well. For the years 1602–1610, there is no comparable indication of drought in the documentary record. The NADA composite map for New Mexico during 1598–1601 (Fig. 7a) shows a significant negative PDSI anomaly, particularly in the central and northern parts of the present-day state, where the Oñate expedition settled. The NADA time series for the location of Oñate's first colony at the present-day Okhay Owingeh Pueblo (Fig. 7b) indicates negative PDSI anomalies in 1598 and 1600–1601, although not as significant as those of the “megadrought” during the 1580s.

Figure 7(a) NADA JJA PDSI reconstruction for 1598–1601, New Mexico. (b) NADA JJA PDSI reconstruction time series for 1580–1610, at Oñate's first colony, present-day Okhay Owingeh, New Mexico.

Download

3.2.3 Virginia and North Carolina

In 1570, a small group of Spanish Jesuit missionaries attempted to establish an outpost along the lower James River in present-day Virginia. In the scarce documentation that survives from the expedition, they reported that the land had been “punished with six years of sterility and death” and was “very parched (muy agostada)” such that both maize and wild plants had died. Native Americans massacred the priests in early 1571. The NADA composite map for 1565–1570 indicates a significant negative PDSI anomaly for the region of the outpost (Fig. 8a), and the time series of PDSI variation for the approximate location (Fig. 8b) indicates persistent if not always severe drought since 1562, with the worst year (PDSI of −3) occurring in 1566.

Figure 8(a) NADA JJA PDSI reconstruction composite map for 1565–1570, Virginia. (b) NADA JJA PDSI reconstruction time series for 1560–1580, approximate location of Ajacán.

Between 1585 and 1587, English investors led by Sir Walter Raleigh made several attempts to colonize Roanoke Island in the Carolina Outer Banks, before losing all contact with the final “lost colony.” A settler described how in 1586 the Native Americans' corn “began to whither by reason of a drought which happened extraordinarily” and reported being asked to pray for rain to end that drought. During the following year, the colony was unable to obtain maize from indigenous communities, who reported scarcity, and the colonists therefore suffered famine. The NADA composite map for 1585–1588 (Fig. 9) shows a major drought throughout eastern North Carolina, although not specifically for Roanoke Island, which is not covered by the atlas.

Figure 9NADA JJA PDSI reconstruction for 1585–1588, eastern North Carolina.

The English colony at Jamestown, Virginia, suffered through multiple environmental disasters and conflict with Native Americans from its founding in 1607 through 1610, when it was rescued by a large infusion of new settlers and supplies. Several contemporary personal letters, as well as pamphlets and histories later written by eye witnesses, left indications of exceptional summer drought throughout these first years at the colony. In summer and autumn of 1607, saltwater apparently intruded into the James River as far as Jamestown, where the water was described as “at flood very salt, at low tide full of slime and filth”; this unusual occurrence may have caused salt poisoning at the colony (Earle, 1979). Settlers' summer grain crops failed repeatedly for unspecified reasons. Multiple colonists also recorded complaints from Native Americans in villages throughout the region during 1608–1610 that they had no maize to share with colonists or even for their own consumption. Two sources reported that in 1608 or 1609 a leader of a nearby indigenous community asked a colonist to “pray to his god for rain, for his [own] god would not send him any”. During the summer of 1610, sturgeon did not swim up the James River as usual, which may indicate high salinity in that waterway. The NADA composite map for 1606–1610 (Fig. 10a) indicates a moderate drought over southeastern Virginia. The NADA time series of PDSI variation for 1600–1620 at the location of Jamestown (Fig. 10b) indicates a negative PDSI anomaly each summer during 1606–1612, with a severe drought (PDSI of −3) in 1610, making this span of years Virginia's longest continuous drought of the past seven centuries, as reconstructed from local tree-ring width (Stahle et al., 1998).

Figure 10(a) NADA JJA PDSI reconstruction for 1606–1610, eastern Virginia and North Carolina. (b) NADA JJA PDSI reconstruction time series for 1600–1620, Jamestown, VA.

Table 2Synthesis of comparison between drought information in the documentary record of early North American colonial expeditions and the NADA PDSI reconstruction. Green indicates droughts identified in both the documentary record and NADA reconstruction; blue indicates droughts identified in neither; red indicates droughts identified in the NADA reconstruction but not the documentary record; yellow indicates instances of partial or possible agreement between the documentary record and NADA, including instances where the documentary record identifies the impacts of a long drought but does not specify each year of its occurrence; dark grey indicates that there is no first-hand written evidence concerning climatic conditions.

3.3 Synthesis

Table 2 presents a synthesis of the results described in Sect. 3.1 and 3.2. It indicates the years and regions where documented expeditions were present, and one of the following results: (1) droughts identified in both the documentary record and NADA reconstruction; (2) droughts identified in neither; (3) droughts identified in the NADA reconstruction but not the documentary record; (4) instances of partial or possible agreement between the documentary record and NADA, including instances where the documentary record identifies the impacts of a long drought but does not specify each year of its occurrence; (5) no first-hand written evidence concerning climatic conditions.