| Abstract: | Dated records of ice-cap growth from Arctic Canada recently suggested that a succession of strong volcanic eruptions forced an abrupt onset of the Little Ice Age between A.D. 1275 and 1300 Miller GH, et al. (2012) Geophys Res Lett 39(2):L02708, 10.1029/2011GL050168]. Although this idea is supported by simulation experiments with general circulation models, additional support from field data are limited. In particular, the Northern Hemisphere network of temperature-sensitive millennial tree-ring chronologies, which principally comprises Eurasian sites, suggests that the strongest eruptions only caused cooling episodes lasting less than about 10 y. Here we present a new network of millennial tree-ring chronologies from the taiga of northeastern North America, which fills a wide gap in the network of the Northern Hemisphere''s chronologies suitable for temperature reconstructions and supports the hypothesis that volcanoes triggered both the onset and the coldest episode of the Little Ice Age. Following the well-expressed Medieval Climate Anomaly (approximately A.D. 910–1257), which comprised the warmest decades of the last millennium, our tree-ring-based temperature reconstruction displays an abrupt regime shift toward lower average summer temperatures precisely coinciding with a series of 13th century eruptions centered around the 1257 Samalas event and closely preceding ice-cap expansion in Arctic Canada. Furthermore, the successive 1809 (unknown volcano) and 1815 (Tambora) eruptions triggered a subsequent shift to the coldest 40-y period of the last 1100 y. These results confirm that series of large eruptions may cause region-specific regime shifts in the climate system and that the climate of northeastern North America is especially sensitive to volcanic forcing.Tree-ring chronologies are the type of proxy record most used to develop climate reconstructions covering the last millennium (1). These chronologies have been integrated into large-scale networks, often with additional proxies, to document the amplitude, duration, and forcing mechanisms of the Medieval Climate Anomaly, the Little Ice Age, and the recent warming trend. However, the spatial coverage of long tree-ring records must be improved to allow a better understanding of regional variations in past climate (1, 2). For example, in eastern North America, millennial climate reconstructions have been constructed from tree species and sites sensitive to drought and precipitation (3), whereas temperatures were inferred solely from low-resolution proxies, such as pollen data (4, 5). Furthermore, only tree-ring-based climate reconstructions shorter than a millennium or using chronologies poorly replicated before A.D. 1500 have been published for the entire North American boreal forest (6, 7), whereas several millennial, highly replicated, temperature-sensitive tree-ring records have been developed across the Eurasian boreal zone. This lack of data is an important issue that causes the poor representation of North America in long-term, large-scale temperature reconstructions (1, 4).The feasibility of reconstructing volcanic forcing from tree-ring data has been debated, especially in regards to large and successive eruptions. Two of the largest eruptions of the last millennium, the A.D. 1257 Samalas and A.D. 1815 Tambora events, were both closely followed and preceded by additional large eruptions in 1227, 1275, 1284, 1809, and 1835 (8−11). Whereas general circulation model experiments suggest that the impacts of large and successive eruptions might have influenced climate systems for periods ranging from 20 y to several decades, or even centuries (12–16), Northern Hemisphere tree-ring-based temperature reconstructions only display negative temperature anomalies lasting between 2 and 10 y (17–20). Region-specific responses of the climate system to volcanic forcing may in part explain this discrepancy (17). For example, large and successive eruptions may have had stronger impacts on summer temperatures in northeastern North America (hereafter NENA) than elsewhere. An extensive Northern Hemisphere network of tree-ring density chronologies supports this idea, showing that the coldest 1816 temperature anomalies occurred over the Quebec-Labrador Peninsula (21), where they may have persisted for several decades (7). The idea is also supported by the abrupt acceleration of ice-cap growth in the Eastern Canadian Arctic during A.D. 1275–1300, at the onset of the Little Ice Age, as a consequence of a series of eruptions (22). However, the lack of millennial, well-replicated, and temperature-sensitive tree-ring chronologies in the NENA sector precludes the examination of the volcano−temperature relationship in a long-term context with an annual resolution.In this study, we have built a network of six highly replicated millennial tree-ring chronologies from large stocks of black spruce Picea mariana (Mill.) B.S.P.] subfossil trees preserved in lakes of the NENA taiga from which we developed a millennial reconstruction (A.D. 910–2011) of regional July−August temperatures. For this purpose, we selected homogeneous sites with infrequent and well-documented ecological disturbances (23), and sampled homogeneous subfossil and living samples to maximize the robustness of our reconstruction. We then used a Bayesian mixture of probability distributions with dependence (also referred to as hidden Markov models or Markov switching models; see refs. 24 and 25) to detect possible regime shifts in summer temperatures triggered by series of large eruptions and to provide new insights concerning the climate history of NENA during the last 1,100 y. |
