Fraxinus pennsylvanica trees growing along the central Assiniboine river floodplain, Central Canada: Flood rings, quantitative wood anatomy and blue intensity.

Abstract

In recent decades, visual identification of flood rings (event years) has been successfully used to document historic high-magnitude spring floods. In Fraxinus spp., flood rings usually have more numerous earlywood vessels and/or earlywood vessels with smaller cross-sectional area than observed in "normal" years. Visual identification of flood rings has also shown to be reproductible.  In more recent times, quantitative wood anatomy of earlywood features (continuous time series) has, among other, been successfully used to reconstruct spring flow and associated flood conditions. In Interior North America, most paleoflood studies have focused on visually identifying flood rings in bur oak (Quercus macrocarpa Michx.) trees growing along river terraces; the idea being that in these “high” elevation sites only high-magnitude floods may be recorded. In this study, we reexamined tree-ring samples collected in 2004-2005 from 87 green ash (Fraxinus pennsylvanica Marsh.) trees growing in four floodplain sites (~20 trees per site) located in the central Assiniboine river watershed i.e., in Spruce Wood Provincial Park, Manitoba. Flood rings were visually searched in all samples and compiled during the crossdating procedure. Earlywood vessels (area ≥ 1000 µ2) were measured in five trees selected from each of the four floodplain sites. In addition, we determined from a subset of these trees the blue intensity from high-resolution scans of tree-rings and compared them, among others, to the earlywood vessel characteristics. Developed chronologies were compared to both regional climate and hydrological records. They were also compared to former flood-ring studies using bur oak trees growing on terraces in the upper Assiniboine river and/or the lower and upper Red river. Results indicated that visual identification of flood rings in F. pennsylvanica was reproductible. Both the relative frequency of flood rings and earlywood mean vessel area were significantly associated with winter precipitations, spring snow cover, spring temperatures and spring runoff records. Some of the years recording the most flood rings were 1948, 1955, 1956, 1976 and 1995. These years corresponded to documented major floods for the central Assiniboine river. They, however, poorly coincided to those observed in bur oak trees growing on terraces along the upper Assiniboine river and the Red river (e.g., 1950, 1979, 1997). These results highlight, among others, the differences between tree species, habitat selection as well as between the rivers’ flood dynamics. We argue that floodplain trees compared to terrace trees, and especially when old individuals are available, may provide a clearer overall picture of flood dynamics and of long-term hydroclimatic changes as they capture conditions leading to both small and large floods. At this time, blue intensity analyses of the floodplain trees remain to be completed. Given than flood rings have been associated with low density earlywood, it is hypothesized that blue intensity (a proxy for tree-ring density) may yield potential in paleoflood studies.