Dominique Arseneault

A denser global network of millennial tree ring chronologies is needed to verify whether the warming of the last century is anomalously rapid or of large amplitude in the context of the last millennium. The North American boreal forest in particular has been poorly represented in long climatic reconstructions based on tree rings, because of short tree longevity, as well as frequent and severe wildfires. In this study, we exhaustively sampled tree remains in the littoral zone of two lakes and partially sampled seven additional lakes in the northern boreal forest of eastern Canada to verify if highly replicated millennial tree ring chronologies could be developed. In total, 989 black spruce [Picea mariana (Mill.) BSP.] tree remains were sampled, of which 529 could be crossdated, providing a master chronology spanning the AD 676–1991 time period. Radiocarbon dating of a floating chronology indicates that maximum residence time of tree remains outside of sediments is at least 1500 years. Past fires halted or reduced wood inputs at least once at all but one site and thus influenced the replication of local chronologies. However, fire effects can be minimized using appropriate site selection criteria (old‐growth forest on the leeward shore, sharp deepening of the littoral zone, burial of tree remains by fine sediments), such that it would be possible to develop a network of several highly and evenly replicated multi‐century local chronologies. These tree ring chronologies would have desirable properties for extracting low‐frequency tree growth trends and for reconstructing forest disturbance histories. Copyright © 2013 John Wiley & Sons, Ltd.

Summary: We used dendrochronology to reconstruct the transfer of coarse woody debris across a forest–stream interface in a fire‐prone boreal landscape. A sequence of regulating factors was considered from source to sink of in‐stream woody debris (SWD), including fire history at the landscape scale, patterns of post‐fire recovery of riparian forest and inputs of SWD at the scale of a stream reach and its associated floodplain, and burial of SWD at an excavated site. Fires occurred repeatedly in the studied landscape (at least in 1708, 1733, 1791, 1811, c. 1838, c. 1850, 1882, 1941 and 1998), and were generally patchy on the floodplain because of the firebreak effect of the riparian corridor. Unburned forest remnants were regularly generated at the stream margin, thus permitting temporally continuous but spatially localized transfer of woody material across the forest–stream interface. These remaining forest patches also increased forest resilience by dispersing seeds and promoting conifer re‐establishment in burned areas. Because of higher severity compared with previous fires, the 1941 fire burned almost everywhere on the floodplain, creating only a few widely isolated unburned forest remnants. Consequently, following an abrupt post‐fire increase, SWD inputs almost completely ceased. In addition, post‐fire recovery of the riparian wood source is slow because of the spatially restricted seed source. In this alluvial stream, wood burial is faster than decay and largely determines the residence time of SWD. Because the residence time is about 150 years, the current density of SWD is high and contrasts sharply with the very low tree density at the stream margin. Although this long residence time helps maintain stream integrity while the forest is recovering from the 1941 fire, it is unlikely that SWD inputs would resume extensively before burial of the current SWD pool. Our research exemplifies the potentially complex impacts of disturbances on material transfer between a source and a sink ecosystem. We conclude that when certain components of ecosystems are coupled by unidirectional flow, those components will behave asynchronously if a disturbance impact at the source ecosystem does not propagate rapidly to the sink and the source and the sink recover at different rates.

Summary: We reconstructed the dynamics of a black spruce (Picea mariana) and jack pine (Pinus banksiana) forest stand in northern Québec using a continuous, 5200‐year‐long sequence of stem remains buried in adjacent peatland. Simulations of recruitment of such remains provided guidelines for inferring past ecosystem structure and composition at the stand scale. Compared with the late Holocene (4650–0 cal. year BP (CYBP)), the mid Holocene (5200–4650 CYBP) period was characterized by faster tree growth, larger stems and higher stem density, indicating higher forest productivity in association with a milder climate. The presence of stem remains of both species from 17 out of 20 contiguous 250‐year time intervals suggests that the spruce‐pine stand exhibited high compositional stability, with both species regenerating after fire from canopy‐stored seed banks. Relative species abundance closely followed the duration of past fire intervals deduced from the number of tree rings in buried conifers. Time periods of long (4650–3950, 3400–1850 and 250–0 CYBP) and short fire intervals (4950–4650, 3950–3400 and 1850–250 CYBP) were associated, respectively, with decreasing and increasing pine abundance, probably reflecting faster juvenile growth, lower shade tolerance, earlier sexual maturity and shorter longevity in jack pine compared with black spruce. We conclude that both climate change and climate‐induced fire disturbance have been driving long‐term ecosystem dynamics. Our field evidence supports the idea that interactions between disturbances and the life‐history traits of species modulate the impact of climate change at the scale of forest stands. At the same time, disturbances may result in long‐term stability of disturbance‐adapted ecosystems.