To better plan for and implement long-term ecological monitoring, we measured riparian vegetation and fluvial geomorphic features at pilot study sites on four wadeable perennial stream reaches, representative of drainages across the Colorado Plateau. Our primary objectives were to (1) collect field data, (2) evaluate the efficiency and effectiveness of various ecological measures and measurement techniques for riparian ecosystems, and (3) use field-based sampling to inform and refine the development of standard operating procedures for use in implementing integrated, long-term monitoring of riparian ecosystems. Ultimately, this work was aimed at providing NPS staff with some of the information and methods needed to design and implement long-term monitoring of NPS riparian resources, which is both relevant to management, and fully operational within institutional resource constraints.

Our results suggest that selecting sampling reaches and establishing a sampling frame of 11 transects, across a range of stream types, is feasible given a limited set of decision rules. A distinctive feature of richness across all sites was the high percentage of rare species, defined here as species having a single occurrence at a site. Rare species represented from 33 percent to 47 percent of the species total across the four pilot sites. Our data show that the two smallest quadrat sizes, 0.01 m² and 0.1 m² , rarely had any species that occurred in the desired frequency range and can be omitted from the monitoring protocol. Few species fell within the 30–70 percent range in the 1-m² quadrats, but this quadrat size appears to be useful at the Tsaile Creek (CACH) site. We recommend continuing to collect information at the 1-m² scale and reevaluating its usefulness after more data are available from different types of sites. The 10-m² quadrat is adequate for monitoring changes in frequencies of very common species at all sites. Based on pilot study results, we conclude that at sites with low total species numbers (< 60 species), 40–60, 10-m² quadrats, would be sufficient to characterize overall species diversity for relatively common species. At sites with higher total numbers of species (> 100), 60–80, 10-m² quadrats would be required to characterize overall species diversity. Rare species of interest should be monitored using alternative approaches, such as a site inventory and/or mapping (see Elzinga and others, 1998). A large number of the systematically placed 10-m² quadrats span two or more geomorphic surfaces, especially adjacent to the channel. This makes resolution of species affinities with distinct geomorphic landforms difficult. Thus, we provide an amendment to improve characterization of herbaceous and shrub species on narrow, near-channel surfaces by sampling additional 0.5-m by 1-m quadrats on those surfaces. It appears that for sites in narrow valley settings where riparian zones average less than approximately 40 m, the number of 10-m² quadrats systematically placed on 11 transects will not provide shrub cover estimates at 20 percent precision. In such cases, additional sample reaches should be added in order to attain a minimum of 130 to 140 10-m² shrub quadrats. 

The line-intercept technique can provide a relatively rapid, reach-scale quantification of proportional cover for woody vegetation and geomorphic surface types and that variance in these measures stabilizes by the eighth or ninth transect sampled. An overlay of the distribution of geomorphic surface data derived from line-intercept sampling on topographic survey information indicates that delineation of geomorphic surfaces could be done in conjunction with the topographic survey of each transect, obviating the need to record surface breaks using the line intercept. To include geomorphic surface identifications with the topographic survey, surface breaks and transitional surfaces should be included and identified in the survey, in addition to systematically placed survey points.

Compared to 5-m by 20-m tree quadrats, belt transects were shown to provide similar estimates of stand structure (stem density and stand basal area) in less than 30 percent of the time. Further, for the streams sampled, there were no statistically significant differences in stem density and basal area estimates between 10-m and 20-m belt transects and the smaller belts took approximately half the time to sample. There was, however, high variance associated with estimates of stand structure for infrequently occurring stems, such as large, relict or legacy riparian trees. Legacy riparian trees occurred in limited numbers at all sites sampled. A reachscale population census of these trees indicated that the 10-m belt transects tended to underestimate both stem density and basal area for these riparian forest elements and that a complete reach-scale census of legacy trees averaged less than one hour per site.