9598 - Laserfiche WebLink

Home Browse Search

276 M. C. MOLLES, JR., C. S. CRAWFORD AND L. M. ELLIS METHODS Study area The study area was located at Bosque del Apache National Wildlife Refuge, approximately 160 km south of Albuquerque, New Mexico at an elevation of about 1400 m. The refuge includes about 14.5 km of the Rio Grande and its associated riparian vegetation, including cottonwood forest. During the summer of 1991, we established experimental and control study sites in cottonwood forest. We chose the two study sites because they were similar to each other in size and vegetative composition and because they had not been flooded for approximately 50 years. The forest floor of both the experimental and control study sites contained large amounts of leaf and woody plant litter. The area of each study site was approximately 3.1 ha. The canopy of both study sites is dominated by Rio Grande cottonwood (Populus fremontii var. wizlizeni) 8-15 m in height, with subcanopies of Goodding wil- low (Salix gooddingii) and tamarisk (Tamarix chinensis). The dominant understory shrubs are seepwillow (Baccharis glutinosa) and New Mexico olive (Forestiera neomexicana), with scattered Russian olive (Elaeag- nus angustifolia), screwbean mesquite (Prosopis pubescens), wolfberry (Lycium torrevi) and false indigo (Amorpha fruticosa). Experimental design We monitored the experimental and control study sites for two years before flooding the experimental site. The purpose of this pre-flood monitoring was to compare leaf litter production, decomposition and forest floor organic matter at the two study sites under non-flooding conditions before the experimental flood. This pre-flood comparison was performed to assess whether changes followed flooding of the experimental site. Such before-after comparisons are now widely applied to large-scale unreplicated experiments (Carpenter et al., 1989; Carpenter, 1990; Rasmussen et al., 1993). Manipulative flooding The water used for flooding the experimental site was diverted from a riverside irrigation canal which is, itself, diverted from the Rio Grande at San Acacia, New Mexico, about 45 km upstream from the experimen- tal site. Water diverted from the canal flows about 200 m down a roadside borrow ditch to the experimental site. Groin dikes and variable-height spillways controlled the water level at the experimental site during flooding. We timed the experimental flood to peak on about 31 May, the average date of peak flow in the 100 year hydrograph for the USGS gauging station on the Rio Grande at Embudo, New Mexico. In 1993, flooding was begun on 17 May, reached a maximum height on 27 May and was held at that level until 3 June. Flood waters were then gradually drained off the experimental site over the next two weeks. Litter production Litter production at each site was monitored with 12 circular rubber tubs, 50 x 10 cm deep. The tub con- tents were collected monthly from September through March, oven-dried for 48 hours at 60°C and weighed. The cumulative dry weight for each site was plotted against time to determine the timing and amount of annual leaf litter production in 1991, 1992 and 1993. Cumulative litter fall from September-March at the two sites was compared using a Wilcoxon rank sum test (NPARIWAY procedure, SAS Institute, 1989b). Decomposition Litter bags (15 x 15 cm) were constructed of fibre glass window screening (1 mm mesh) and filled with 5 g of air-dried cottonwood leaves. These leaves were collected in October as senescent leaves from 10 P. fremontii growing near the study sites. Each year, 20 decomposition bags were placed at the study sites and then retrieved in sets of five in late October or early November, on the initial day of placement, mid-April, late June and one year after placement. In the laboratory, the contents of the decomposition bags were oven-dried at 60°C for 48 hours, weighed and ashed. Ash-free dry weights of the litterbag contents at control and flood sites were compared on each collection date using a Wilcoxon rank sum