WWETAC Projects

Project Title: Evaluation of models for assessment of threats to wildlands in the Western United States from displacement by cheatgrass and pinyon-juniper woodlands

Status: Completed

Principal Investigators: Mary M. Rowland and Michael J. Wisdom, USDA Forest Service, Pacific Northwest Research Station, La Grande, OR; Lowell H. Suring, USDA Forest Service, Washington Office, Stationed at Rocky Mountain Research Station, Boise, ID; Robin J. Tausch, USDA Forest Service, Rocky Mountain Research Station, Reno, NV

Collaborators: Bryan Endress, Zoological Society of San Diego, Escondido, CA; Jennifer Boyd, Susan Geer, Bridgett Naylor, and Catherine G. Parks, USDA Forest Service, Pacific Northwest Research Station, La Grande, OR; Mark Finco and Ken Brewer, Remote Sensing Applications Center (RSAC), Salt Lake City, UT

Web: http://www.ag.unr.edu/gbem/Members/robin_tausch.htm

E-mail Contact: Mary M. Rowland, mrowland[at]fs.fed.us

Key Issues/Problems Addressed: Assessment of threats to wildlands, especially sagebrush (Artemisia) ecosystems, of the Western United States from displacement by cheatgrass (Bromus tectorum) and pinyon (Pinus spp.) and juniper (Juniperus spp.) woodlands

Study Objectives and Goals: Previously, we developed models that predicted risk of displacement of native vegetation by cheatgrass and pinyon-juniper woodlands in the Great Basin Ecoregion of the western U.S. (Suring et al. 2005). Our current objective is to build on these models by (1) refining them with the use of more recently developed land-cover maps and (2) applying the revised models in the John Day and Mono Basin ecological provinces (fig. 1). We are evaluating the models with data from several sources, including vegetation data collected in the John Day province in 2007, previously collected vegetation data, and aerial photography. Following evaluation of model performance with empirical data, we will revise the models and re-apply them in the two provinces. Resulting predictions of risk of displacement of native vegetation, displayed in maps and tabular format, can be built upon in the future to evaluate their relation to wildland fire risk, fuels management, maintenance of sagebrush and old-growth pinyon and juniper communities, and potential effects on species of concern associated with sagebrush and woodland communities.

General Description: Two of the most prevalent threats to sagebrush and other native shrublands of the western U.S. are displacement by cheatgrass and by pinyon-juniper woodlands. Cheatgrass is an exotic, invasive annual grass that has spread across millions of acres of western rangelands, where it provides a highly flammable fuel. Many native shrubs are fire intolerant, especially sagebrush, so that when cheatgrass invades the understory, subsequent wildfire often destroys the shrublands, leaving behind a cheatgrass monoculture (Chambers et al. 2007). Sagebrush ecosystems also are threatened by pinyon-juniper woodland encroachment that ultimately displaces the sagebrush, altering ecosystem processes and reducing biological diversity (Miller et al. 2005). In response to concerns about these threats, we developed two rule-based models in a GIS (Geographic Information System) to predict risk posed by cheatgrass and pinyon-juniper in 14 ecological provinces (fig. 1). The lack of a sufficiently accurate map of pinyon and juniper woodlands, however, limited initial application of the pinyon-juniper model to just three of these provinces. Moreover, application of the cheatgrass risk model was conducted using a mid-1990s land-cover layer of coarse resolution. Our current analysis uses more recent and more accurate maps of existing vegetation; selection of an appropriate land-cover map is critical for accurately predicting risk, based on our models, and evaluating models (Rowland et al. 2008a).

Figure 1—Ecological provinces in the Intermountain West, adapted from Bailey (1980) and West et al. (1998).
Figure 1—Ecological provinces in the Intermountain West, adapted from Bailey (1980) and West et al. (1998).

Status:

Model development, application, and evaluation

  • Acquired all pertinent spatial data layers and associated metadata for model application in the John Day and Mono Basin provinces, e.g., existing vegetation, digital elevation models (DEMs), and precipitation, and created associated grids.
  • Downloaded additional spatial data layers for analysis, e.g., land ownership, administrative boundaries, wildland-urban interface (WUI).
  • Investigated three currently available land-cover maps in the John Day Province to evaluate their utility for our project: 1) Shrubmap, a land-cover map developed expressly for assessment of shrublands and associated types in Oregon and Washington; 2) LANDFIRE existing vegetation type (evt;); and 3) PNW ReGAP, a land-cover map for Oregon and Washington under development, which is virtually identical to Shrubmap in its depiction of non-forested communities in Oregon.
  • Mapped and quantified the areal extent of sagebrush and pinyon-juniper communities in both study areas (fig. 2, fig. 3).
  • Compiled additional summary statistics for the John Day and Mono Basin provinces, e.g., dominant ecological systems (Comer et al. 2003), sagebrush and pinyon-juniper distribution and extent by landowner.
  • Re-created the pinyon-juniper risk model in GIS software (ModelBuilder, ArcGIS 9.2) and applied the model in the John Day province, creating maps and summaries of predicted risk to sagebrush communities in this province (fig. 4).
  • Exploration of model performance is underway, comparing predicted risk in field-sampled plots with observed risk as characterized by combinations of western juniper (J. occidentalis) density, age class, and relative canopy cover (pinyon pines and other juniper tree species are absent from the John Day province).
  • The cheatgrass model is currently being revised in ModelBuilder and will be available for application in our study areas in June 2008.
  • Will conduct sensitivity analysis for both models when the cheatgrass model is completed.
  • Will continue to acquire auxiliary data for model evaluation from various sampling efforts in the western U.S. (e.g., LANDFIRE reference database, BLM ecosite data).

Figure 2—Sagebrush and western juniper woodlands in the John Day Ecological Province, as mapped by LANDFIRE existing vegetation type.

Figure 2—Sagebrush and western juniper woodlands in the John Day Ecological Province, as mapped by LANDFIRE existing vegetation type.

Figure 3—Sagebrush and pinyon-juniper woodlands in the Mono Basin Ecological Province, as mapped by LANDFIRE existing vegetation type.
Figure 3—Sagebrush and pinyon-juniper woodlands in the Mono Basin Ecological Province, as mapped by LANDFIRE existing vegetation type.

Figure 4—Predicted risk of displacement of sagebrush by western juniper in the John Day province, central Oregon, by risk category.
Figure 4—Predicted risk of displacement of sagebrush by western juniper in the John Day province, central Oregon, by risk category.

Field sampling

Collaboration with RSAC

  • Lowell Suring and Mary Rowland met several times with Ken Brewer, Mark Finco, Kevin Megown, and others at RSAC to discuss application of remote sensing products and techniques in our project.
  • RSAC completed an extensive cross-walk matrix between key land-cover types from the original land-cover map (“SageStitch”) used to develop the two models to Shrubmap, in order to revise the model rule sets with a more current land-cover map.
  • RSAC also compiled all resource photography available from the Aerial Photography Field Office (APFO) in the John Day province, for potential use in model evaluation and retrospective analysis of juniper conditions in the province.
  • Large disparities were seen in juniper and sagebrush distribution and extent as mapped by LANDFIRE vs. Shrubmap; to better understand these differences, RSAC completed a pilot sampling of juniper and other vegetation using a dot grid method with Digital Mylar and NAIP aerial photography (Rowland et al. 2008a).

Key Findings:

  • The choice of a land-cover map for model application and evaluation was not as straightforward as anticipated:
    • Crosswalks completed by RSAC between “SageStitch” and Shrubmap in the John Day province revealed little consistency in mapping of sagebrush and woodland types between these coverages. For example, areas mapped as mountain big sagebrush by “SageStitch” were mapped as 26 different cover types in Shrubmap, with the dominant type being Columbia Plateau Western Juniper (38.9%), followed by Inter-mountain Basins Montane Sagebrush (21.3%) and Inter-mountain Basins Big Sagebrush Steppe (16.3%).
    • We found marked differences in the distribution and extent of sagebrush and juniper vegetation among the three recently available maps we considered for use in the John Day province (LANDFIRE existing vegetation type, Shrubmap, and NW ReGAP) (Rowland et al. 2008a).
    • These disparities profoundly affect model predictions and evaluation; we decided to use LANDFIRE for our initial analyses, but recognize that it likely underestimates western juniper distribution and overestimates sagebrush in the province.
    • The occurrence of multiple sagebrush taxa within single ecological systems (e.g., basin [A. tridentata tridentata] and Wyoming big sagebrush [A. t. wyomingensis] in Inter-mountain Basins Big Sagebrush Shrubland) used by both LANDFIRE and Shrubmap (Comer et al. 2003) is problematic for application of the current woodland model, which relies on more clear spatial and taxonomic discrimination among sagebrush types.
    • Mountain big sagebrush (A. t. vaseyana), an important sagebrush taxon with different susceptibility to juniper encroachment than other sagebrush types, appears to be widely under-represented in current vegetation maps in the John Day province. Mean canopy cover of this subspecies across all plots was greater than that of any other single sagebrush taxon measured in our study area during 2007.
  • Preliminary summaries of field data from John Day revealed a more complex sagebrush ecosystem in the study area than anticipated, with three or even four sagebrush taxa co-occurring at the plot level (about 1 ha).
  • Canopy cover of total sagebrush was negatively correlated with canopy cover of western juniper (r = -0.41; fig. 5) in our plots, a phenomenon widely reported in juniper/sagebrush ecotones as juniper dominance increases.
  • Densities of western juniper varied widely among plots in the John Day province but were consistently high enough to pose substantial risk to sagebrush communities:
    • Only a fourth of our plots contained pre-settlement junipers (>140 years), and <5% of the total junipers (n = 2,733) tallied were pre-settlement trees. The low percentages of mixed-age stands and of old trees resemble those reported for pinyon-juniper woodlands in Idaho, Nevada, Oregon, and Utah (Miller et al. 2008). Thus, most junipers now present in the John Day province have been established since the late 1800s.
    • Mean densities of pre-settlement junipers were low, ranging from zero in low sagebrush (A. arbuscula)/mountain big sagebrush plots far from juniper stands to 16.7 trees/ha in other sagebrush sites far from juniper stands, suggesting that historically junipers were widely scattered across the landscape.
    • In contrast to pre-settlement trees, densities of post-settlement junipers were uniformly higher across the province, ranging from 76 to 231 trees/ha in plots from all strata except juniper to 427 trees/ha in plots from the juniper stratum. More than 20% of plots sampled had densities exceeding 300 trees/ha.
    • Across all plots (n = 206), mean density of junipers in the smallest size class (0-30 cm tall) was 21.6 trees/ha, suggesting continued expansion and in-fill of juniper trees in the future (Azuma et al. 2005).
    • Canopy cover of western juniper trees was highest in the juniper stratum, as expected; outside this stratum (i.e., in sagebrush and other shrublands) juniper canopy cover was lowest in the low/mountain big sagebrush sites, and highest in the “other sage” stratum near existing juniper stands.
  • Application of the pinyon-juniper model in the John Day province yielded the following predictions of risk to sagebrush communities: 1) low risk = 876,600 ha (71.2%); 2) moderate risk = 68,800 ha (5.6%); and 3) high risk = 286,200 ha (23.2%)
    • These proportions are similar to the values reported for the three provinces in Nevada and Utah to which the woodland model was first applied (Suring et al. 2005).
    • The largest area of high-risk sagebrush is managed by private landowners (167,200 ha), followed by the BLM (98,000 ha).
    • Among sagebrush types, low sagebrush was at highest risk based on proportion of total sagebrush at risk, followed by “other sage” (primarily basin and Wyoming big sagebrush) and mountain big sagebrush.
  • Initial analyses of 2007 field data suggest relatively poor performance of the pinyon-juniper model:
    • Of the 178 sagebrush plots that have a predicted risk value from the model, at least 67% (119) were observed to be at high risk, based on our field data and the criterion of total juniper density exceeding 50 trees/ha (draft model evaluation rules on file). Of these 119 observed high-risk plots, the majority (69%) were predicted by the model to be at low risk (n = 82); only 25% (n = 30) were predicted to be high risk.
    • Overall, risk to sagebrush from woodland encroachment appears to be underestimated by our model predictions in the John Day province.
    • Reasons for poor model performance are not clear yet, but likely relate to poor map performance, i.e., the failure LANDFIRE to accurately portray the current distribution of sagebrush and juniper woodland communities in the John Day Province, especially those types required for model application (e.g., mountain big sagebrush, western juniper).
    • In particular, the areal extent of juniper woodlands appears to be underestimated by LANDFIRE, which can drastically affect the area mapped as juniper stands as well as area affected by juniper (e.g., the area in the first buffer zone adjacent to juniper stands) and thus affect risk predictions of the model
  • Cheatgrass was present in >75% of the plots sampled. Mean absolute canopy cover of cheatgrass was 12.6%, and in 23 plots (11.1%), cheatgrass canopy cover exceeded 30%.
  • These values pose a high risk of future cheatgrass dominance in susceptible sites (e.g., lower elevation, south aspects), especially following fire.

Figure 5—Mean percent canopy cover (absolute) of all sagebrush taxa combined compared to percent canopy cover of western juniper, estimated using line-intercept methods in the John Day province, central Oregon (n = 206).
Figure 5—Mean percent canopy cover (absolute) of all sagebrush taxa combined compared to percent canopy cover of western juniper, estimated using line-intercept methods in the John Day province, central Oregon (n = 206).


Management Implications:

  • Increasing distribution and abundance of cheatgrass are likely, given the predicted response of this species to global climate change, especially increased CO2 concentrations (Smith et al. 2000, Nielson et al. 2005, Ziska et al. 2005). Moreover, changing precipitation regimes (especially milder, wetter conditions) could increase the rate of woodland expansion and infill into sagebrush communities (Nielson et al. 2005, Miller et al. 2008). Such increases pose continued threat to sagebrush ecosystems.
  • Approaches to predict and quantify the continued loss and degradation of sagebrush habitats are especially needed owing to the current re-evaluation of greater sage-grouse (Centrocercus urophasianus) for listing under the Endangered Species Act. Sage-grouse in Oregon are a key component of the sagebrush ecosystem, and sagebrush habitats in the state are considered critical strongholds for the species.

Next Steps:

  • We will apply the cheatgrass model, when available, in the John Day and Mono Basin provinces and quantify and map results. Then we will evaluate model performance by comparing predicted risk with observed risk (derived from relative canopy cover of cheatgrass to all perennial herbaceous vegetation).
  • To further investigate implications of increased fire risk from threats associated with woodlands and cheatgrass, we will overlay predicted risk from each model with the interface and intermix classes of the WUI in each of our test provinces.
  • To obtain more values of observed risk for the woodland model, we will collaborate with RSAC to quantify juniper densities, using Digital Mylar and the 2005 0.5-m NAIP photography available state-wide, at the GPS-based centers of our field sampling plots in 2007. If we can demonstrate a strong relationship between observed risk in field plots and vegetation characteristics measured using aerial photographs, we can calculate observed risk for a much larger area and thus greatly increase our sample size for evaluation of model performance (observed vs. predicted risk).
  • A second goal is to collaborate with RSAC to conduct a retrospective analysis of juniper. Such an analysis would confirm that the field-level conditions we describe as posing risk of displacement of sagebrush 30 years in future are accurate predictors of risk. If we can demonstrate a strong relationship between current, plot-derived estimates of risk in sagebrush with a suite of vegetation characteristics (e.g., juniper density and cover, shrub cover) interpreted from current aerial photography, we can then apply those characteristics to estimate risk to sagebrush with aerial photographs taken ~30 years ago.
    • Areas of sagebrush predicted to be at high risk of woodland encroachment ~30 years ago, as determined by interpretation of the 1970s resource photography, are more likely to be mapped as juniper woodlands in current land-cover maps and/or have a higher percent canopy cover of juniper than areas predicted to be at low risk 30 years ago.

Deliverables:

  • Updated versions of the cheatgrass and pinyon-juniper models for application across all 14 ecological provinces (pinyon-juniper model complete);
  • Maps and associated quantitative data depicting risk of displacement of sagebrush communities by pinyon-juniper woodlands in the Mono Basin and John Day Ecological Provinces (completed for John Day province);
  • Maps and associated quantitative data depicting risk of displacement of native vegetation by cheatgrass in the Mono Basin and John Day Ecological Provinces;
  • Catalogued photographs of field sites in the John Day province (complete);
  • A final report;
  • Manuscripts to be submitted to scientific journals or published in proceedings.

Presentations completed or planned:

Select Background Citations:

Azuma, D. L., B. A. Hiserote, and P. A. Dunham. 2005. The western juniper resource of eastern Oregon, 1999. USDA Forest Service Resource Bulletin PNW-RB-249. 18 p.

Chambers, J. C., B. A. Roundy, R. R. Blank, S. E. Meyer, and A. Whittaker. 2007. What makes Great Basin ecosystems invasible by Bromus tectorum? Ecological Monographs 77: 117-145.

Comer, P., D. Faber-Langendoen, R. Evans, S. Gawler, C. Josse, G. Kittel, S. Menard, M. Pyne, M. Reid, K. Schulz, K. Snow, and J. Teague. 2003. Ecological systems of the United States: A working classification of U.S. terrestrial systems. Arlington, Va.: NatureServe.

Gedney, D. R., D. L. Azuma, C. L. Bolsinger, and N. McKay. 1999. Western juniper in eastern Oregon. USDA Forest Service General Technical Report PNW-GTR-464.

Miller, R. F., J. D. Bates, T. J. Svejcar, F. B. Pierson, and L. E. Eddleman. 2005. Biology, ecology, and management of western juniper. Oregon State University, Agricultural Experiment Station, Technical Bulletin 152, Corvallis, OR, USA.

Miller, R. F., R. J. Tausch, E. D. McArthur, D. D. Johnson, and S. C. Sanderson. 2008. Age structure and expansion of piñon-juniper woodlands: a regional perspective in the Intermountain West. USDA Forest Service Research Paper Report RMRS-RP-69.

Neilson, R. P., J. M. Lenihan, D. Bachelet, and R. J. Drapek. 2005. Climate change implications for sagebrush ecosystems. Transactions of the North American Wildlife and Natural Resources Conference 70: 145-159.

Rowland, M. M., L. H. Suring, and M. J. Wisdom. In press. Assessment of habitat threats to shrublands in the Great Basin: a case study. In: Pye, J. M., H. M. Rauscher, Y. Sands, D. C. Lee, and J. S. Beatty, eds. Advances in Threat Assessment and their Application to Forest and Rangeland Management. Gen. Tech. Rep. PNW-xxx. Portland, OR: U.S. Department of Agriculture, Forest Service, Pacific Northwest Station & Southern Research Station: xxx-xxx.

Smith, S. D., T. E. Huxman, S. F. Zitzer, T. N. Charlet, D. C. Housman, J. S. Coleman, L. K. Fenstermaker, J. R. Seeman, and R. S. Nowak. 2000. Elevated CO2 increases productivity and invasive species success in an arid ecosystem. Nature 408: 79-82.

Suring, L. H., M. J. Wisdom, R. J. Tausch, R. F. Miller, M. M. Rowland, L. Schueck, and C. W. Meinke. 2005. Modeling threats to sagebrush and other shrubland communities. Pages 114-149 in Wisdom, M. J., M. M. Rowland, and L. H. Suring, editors. 2005. Habitat threats in the sagebrush ecosystem: methods of regional assessment and applications in the Great Basin. Alliance Communications Group, Lawrence, KS, USA.

Tausch, R. J., N. E. West, and A. A. Nabi. 1981. Tree age and dominance patterns in Great Basin pinyon-juniper woodlands. Journal of Range Management 34: 259-264.

Thompson, J. 2007. Sagebrush in western North America: habitats and species in jeopardy. Science Findings, USDA Forest Service, Pacific Northwest Research Station, Issue 91. http://www.fs.fed.us/pnw/sciencef/scifi91.pdf

Ziska, L. H., J. B. Reeves, and B. Blank. 2005. The impact of recent increases in CO2 on biomass production and vegetative retention of cheatgrass (Bromus tectorum): implications for fire disturbance. Global Change Biology 11: 1325-1332.

Additional Information:

Ecological Systems
http://www.natureserve.org/explorer/

LANDFIRE
http://www.landfire.gov/index.php, http://www.landfire.gov/NationalProductDescriptions21.php

Managing Disturbance Regimes Program, USDA Forest Service, PNW: http://www.fs.fed.us/pnw/about/programs/mdr/index.shtml

Shrubmap
http://sagemap.wr.usgs.gov/Shrubmap.aspx

 

 

Cheatgrass Encroach
   
prov R. Miller
   
Juniper  

Project ID: FY06JB13