Multiproxy reconstructions of climate for three sites in the Canadian High Arctic using Cassiope tetragona

We developed calibration models and reconstructed climate for sites in the central and eastern Canadian High Arctic using dendroclimatological and stable isotope analysis techniques on the dwarf-shrub, Cassiope tetragona. Our results may suggest complex temporal and spatial patterns of climate change in the region over the past century. For sites on Bathurst and Devon Islands, we reconstructed fall mean and June-July mean temperature using multiple linear regression analysis that explained 54 % and 40 % of the variance, respectively. The predictor variables included annual growth, annual production of leaves, flower buds and annual delta A(1)A(3)C values for the Bathurst Island model, and annual growth and delta A(1)A(3)C values for the Devon Island model. Both models revealed warmer than average temperatures throughout the mid-20th century, followed by a cooling trend from the early 1960s and mid-1970s at the Devon and Bathurst Island sites, respectively. Temperatures remained cool until the early 1980s and then increased until 1998/1999 at both sites. Our models are supported by other paleoclimate proxies and the instrumental record from the Canadian Arctic. For sites on Axel Heiberg and Bathurst Islands, we developed models using multivariate regresssion for February and March total precipitation that explained 44 % and 42 % of the variance, respectively. The Axel Heiberg Island model included annual production of flowers and flower buds, as well as annual delta A(1)A(3)C values as predictor variables, while the Bathurst Island model only included the annual production of flower buds as a predictor. Both models showed lower than average precipitation from the early to mid-1900s, followed by increasing precipitation from the late 1980s to 1998/1999. Our precipitation models, supported by instrumental and proxy data, suggest a trend of increasing late-winter/early spring precipitation in the late 20th century. The lack of a single detectable climate signal across the study sites suggests local climate, topography, genetic variation and/or ecological conditions may dictate, in part, site responses and result in a heterogeneous climatescape over space and time. Yet, like other arctic paleoclimate proxies, chronology error and temporal discrepancies may complicate our interpretations. However, comparisons with other arctic proxies and the meteorological record suggest our models have also registered a regional climate signal.