RESEARCH ARTICLE ABSTRACT: Ponderosa pine plant community and forest structure were compared among three staI)ds in Grand Canyon National Park, Arizona: one stand had 120 years of artificial fire exclusion (NOBURN) and the other two nearby stands had been frequently burned (BURN-E and BURN-W). These forests are valuable places to gauge-anthropogenic changes associated· with European settlement, due to· their land history of limited livestock grazing and no logging. Precipitation varied greatly between sampling years (260 mm in 2000, 505 mm in 2001). Tree density was significantly higher at NOBURN (1424 • trees ha-1) with significantly higher rotten coarse woody debris (23.2 Mg ha-1) and duff depth (4.3 cm) than at the burned sites, as expected in the absence of fire. Although species richness was not signifi­ Plant Community cantly different among sites (48-89 species), richness differed significantly by year. Shannon's index of diversity increased by approximately 10% from the dry year to the wet year on all sites. Community composition and plant cover at NOBURN differed significantly from the two burned sites in both years Variability in in non-metric multidimensional scaling ordinations. Increasing duff depth was related to decreased plant cover. Two of the three dominant species were different at the fire-excluded site compared to the burned sites. We conclude that although plant community structure was related to fire history, environmental Ponderosa stress and within-stand variability were also important drivers. We suggest selecting reference sites in close proxinlity to the site to be restored and using a multi-scale, multi-year, multi-site approach to Pine Forest Has measure reference conditions in ponderosa pine. Implications Index terms: Grand Canyon, Arizona, fire ecology, Kaibab Plateau, ecological restoration for Reference NOMENCLATlJRE: USDA, NRCS. 2002. The PLANTS Database, Version 3.5 Conditions INTRODUCTION may initiate unprecedented ecological trajectories (Moore et al. 1999). CN. Gildar A diverse and productive understory USDA, Dakota Prairie Grasslands plant community is a vital component Many researchers and land managers have McKenzie Ranger District of the ponderosa pine forest ecosystem. focused on ecological restoration - "assist­ 1901 S. Main St. The understory community contributes ing the recovery of an ecosystem that has Watford City, ND 58854 USA virtually all plant biodiversity (since there been degraded, damaged, or destroyed" 701-842-2393 are usually only 1-4 species of overstory (Society for Ecological Restoration 2002) [email protected] trees), regulates pine regeneration through - as an adaptive mariagement strategy that competition, retains soil, facilitates rapid not only reduces hazardous fuel loads but P.Z. Fule1 nutrient turnover rate due to relatively treats the underlying causes of declining fast decomposition of plant material, and ecosystem health (Covington 2000, Allen W.w. Covington provides wildlife habitat (Rasmussen 1941, et al. 2002). For example, through the Ecological Restoration Institute Cooper 1960, Naumberg and DeWald National Fire Plan, the U.S. Department and School of Forestry 1999, Moore et al. 1999). Understory plants of the Interior allocated over $102 million Northern Arizona University also serve as fuel for frequent surface fires. toward restoration and rehabilitation of Box 15018 Fire, being a key disturbance in ponderosa burned lands in fiscal year 2001, includ­ Flagstaff AZ 86011 USA pine forests, has strong influences at the ing treatment of 911,000 ha to reduce species, community, and ecosystem levels hazardous fuels. • (Cooper 1960, Pearson et al: 1972, Cov­ ington and Moore 1994a, 1994b, Moore et Judgments about ecological degradation al. 1999). In the southwest, the removal of and restorative treatments are often based this key disturbance process through live­ on the range of natural (or "historical") stock grazing, timber harvesting, and fire variability, also known as reference con­ suppression over the last ±120 y resulted ditions (Landres et al. 1999). Reference in dense forests, reduced understory spe­ information is used to establish restoration cies composition and plant cover, higher goals, determine restoration potential of 1 Corresponding author: forest floor accumulation, and low nutri­ sites, and evaluate the success of restora­ [email protected]; 928-523-1463 ent turnover (Arnold 1950, Weaver 1951, tion efforts (Jackson et al. 1995, White Arnold 1953, Cooper 1960, Mitchell and and Walker 1997, Moore et al. 1999). Aldo Freeman 1993, Kolb et al. 1994). The in­ Leopold referred to these conditions as a Natural Areas loumaI24:101-111 creased occurrence of high-intensity crown "base datum" for management decisions fires in these densely forested landscapes (1941). Reference conditions are often Volume 24 (2),2004 Natural Areas Journal 1 01 poorly understood because of the decay of and site-specific variation (Anderson and median probability intervals, a measure of historical evidence, inherent limitations in Dugger 1998). We measured the following fire frequency, ranged from 3.4-3.9 y (all the quality and quantity of paleoecological attributes that might show strong responses fires) to 6.1-7.5 y (fires scarring ~ 25% evidence, and changes in climate (Swetnam to fire exclusion and to forest restora­ of sampled trees). No fires had burned et al. 1999). Poor information about plant tion treatments: forest structure, plant at Galahad Point since 1879, the time at community reference conditions impedes community structure, and above-ground which the Kaibab Plateau was affected forest restoration planning. For example, herbaceous biomass (Aronson et al. 1993, . by land management practices associated Crawford et al. (2001) showed that non­ Anderson and Dugger 1998). We chose to with European settlement, such as heavy native plants dominated a site severely sample many attributes in a relatively small livestock grazing (Rasmussen 1941, Ful6 burned by wildfire on the Kaibab Plateau, number of plots, instead of measuring only et al. 2003b). The other sites had several Arizona, but they did not have information a few attributes in many plots, to look for large spreading surface fires and many on the characteristics of the native plant relationships among forest structure and smaller fires since 1879 (Fu16 et al. 2003a). community, making it difficult to assess plant community. Assessing fire (or fire Although the fire regimes are altered at the ecological degradation and recovery except exclusion) effects often presents statisti­ burned sites, they still represent relatively by the crude yardstick of native vs. non­ cal difficulties due to pseudoreplication, undisrupted areas (Fu16 et al. 2002). native composition. because study plots are usually nested within a single burned or fire-excluded area The average elevation and slope of sample Compared to trees, which can live for (Hurlbert 1984, van Mantgem et al. 2001, plots were 2,314 m elevation and 11 % slope centuries and decompose slowly, under­ Heffner et al. 1996). Carefullymatchedim­ on Powell Plateau (BDRN-W), 2,308 m, story plants have short life spans, and pacted/reference area studies create useful 20.5% slope on Rainbow Plateau (BURN­ most evidence of past plant communities comparisons between specific burned and E), and 2,388 m, 16.3% slope on Galahad disappears quickly (see Kerns et al. 2001 unburned sites that limit the role of envi­ Point (NOBURN). Soils are derived from for an example of persistent grass evi­ ronmental variability as a causal agent for Kaibab Limestone (Bennett 1974). Soils on dence). A valuable alternative approach is differences (van Mantgem et al. 2001). Our all sites have been tentatively classified as to measure relict areas that are influenced sites were close in proximity, topography, Elledge Family; well drained, 20-40 inches by present-day climatic and other global and elevation and had the same soil type to bedrock, with about 10 percent gravel changes but relatively untouched by the and contemporary vegetation. (A.Dewall, pers. comm., U.S. Department agents of disruption (Thatcher and Hart of Agriculture, NRCS Soil Scientist 2002). 1974, Johnson 1986). Unfortunately, relict We hypothesized that the fire-excluded site Annual precipitation averages 57.9 cm, sites that have not been grazed by domestic would exhibit traits commonly associated with an average annual snowfall of 326.9 livestock and/or logged, and where fires with fire exclusion, such as different species cm (White and Vankat 1993, GCNP Fire have not been suppressed, are very rare. composition as well as lower understory Management Plan 1992). Temperatures The few relict areas in southwestern forests species richness, cover, and biomass. We measured at the North Rim ranger station tend to be isolated mesas and plateaus often expected that among-site variability be­ ranged from an average July maximum of located in National Parks or other protected tween the burned sites and the fire-excluded 26°C to an average January minimum of -2 natural areas (Leopold 1924, Madany and site would be higher than within-site vari­ °C (Bennett 1974). Precipitation differed in West 1983, Belsky and Blumenthal 1997, ability on any site in all responses. Finally, the two years of this study. Grand Canyon Rowlands and Brian 2001, Fu16 et al. 2002) we expected that the two frequently burned weather station records were incomplete, or within grazing exclosures (Stohlgren et sites would be more similar to each other