Wood in neotropical headwater streams, Costa Rica
Cadol, Daniel D.
MetadataShow full item record
Wood has been shown to be an integral component of forest streams throughout the temperate climate zone, both in terms of the physical structure of the channel and in terms of aquatic ecosystem function, but the function of wood in undisturbed tropical streams has not been studied. This dissertation represents the first systematic analysis of instream wood in a tropical setting to be published. This study was limited to the headwater streams (drainage area <8.5 km2) of La Selva Biological Station, on the Atlantic margin of Costa Rica, a wet tropical site with limited landslide activity. Although the results are instructive and enable comparisons with the vast temperate instream wood literature, they should not be construed as representative of debris flow-dominated wet tropical forest streams or of dry or seasonal tropical forest streams. Wood loads in the thirty 50-m-long study reaches examined ranged from 3.0 to 34.7 m3 of wood per 100 m of channel length and 41 to 612 m3 of wood per ha of channel area. Average values are 12.3 m3/100 m and 189 m3/ha. These values fall generally in the lower range of wood load reported for temperate streams, with values typically lower than those reported from the Pacific Northwest region and the Great Lakes region and within the range of those reported from the Rocky Mountain region and from Southern Hemisphere study sites. Comparisons to study sites in eastern North America, Europe, and Japan are problematic because La Selva is a generally undisturbed forest, whereas studies from those regions are conducted in streams with significant human impact and tend to have very small wood loads. Flow hydraulics appear to be the dominant control on the lateral distribution of wood in the channels of La Selva, but they are only a partial control on the longitudinal distribution of wood, explaining about half of the variation in wood load among the study sites. The remainder of the variation is likely caused by the stochastic nature of large tree fall. In spite of the high temporal variability of lateral input of wood to the channels, spatial variability is small, partially because of the paucity of landslides at La Selva. Therefore, I propose that instream transport has a greater influence on the longitudinal distribution of wood than lateral input variability. Wood in a representative subset of 10 of the 50-m-long study reaches was monitored for 2.3 years. The wood in the streams of La Selva is more transient than wood in most sites studied in the temperate zone, with piecewise mean residence times ranging from 2 to 12 years and volume-wise mean residence times ranging from 2 to 83 years among the 10 sites monitored. Average values were 5 and 7 years, respectively. These are roughly an order of magnitude shorter than mean residence times reported from the Pacific Northwest, but similar to times reported from the Colorado Rocky Mountains. The short residence times may be a result of more frequent large floods caused by the wet tropical climate, higher decay rates caused by the warm tropical climate, or both. Perhaps because of this transience, wood was found to have minimal influence on flow resistance in a subset of 6 of the 50-m-long study reaches. In contrast, wood has been shown to be a major control on flow resistance in temperate mountain streams. It is possible that the channel geometry and bed material size are adjusted to the frequent high discharges, which also mobilize and rework the wood, causing grain and form resistance to overwhelm any resistance contribution from wood. Instream wood at La Selva also appears to have a minimal influence on sediment transport. Jams in sand-bed channels and jams in boulder-bed channels had no associated residual elevation drop. Jams in gravel-bed channels did alter bed elevation by trapping sediment wedges behind them, but analysis of tracer clast movement at one gravel-bed jam resulted in no observable difference in transport distances or mobility between clasts placed upstream of the jam and those placed downstream. An additional forest-stream interaction that was documented is diel cycles in stream discharge associated with groundwater withdrawal by the forest for evapotranspiration. Analysis of the cycles indicates a strong correlation with vapor pressure differential, which previous researchers have found to correlate with sap flow. Further analysis of the cycles suggests that at low-stage conditions transmissivity dominates groundwater flow into the channel, while at high-stage conditions hydraulic gradient is dominant.