SNPLMA RESEARCH PROPOSAL ROUND 10 I. Title Page Project title: Wildlife Habitat Occupancy Models for Project and Landscape Evaluations in the Lake Tahoe Basin Primary Science Theme # 1: Forest Health Subtheme 1b: Modeling and decision support tools for multi-objective forest management Secondary Science Theme# 2: Watershed, Water Quality, and Habitat Restoration Subtheme 2b: Special status species and communities and priority invasive species Team members: Dr. Patricia Manley (USFS Pacific Southwest Research Station [PSW]), Dr. James Baldwin (USFS PSW), Dr. Ross Gerrard (USFS PSW) Patricia Manley, Ph.D. (primary contact) Research Wildlife Biologist US Forest Service 1731 Research Park Dr. Davis, CA 95618 530-759-1719 (ph) 530-747-0247 (fx) pmanley@fs.fed.us James Baldwin, Ph.D. Statistician US Forest Service 800 Buchanan St. West Annex Bldg Albany, CA 94710-0011 jbaldwin@fs.fed.us Ross Gerrard, Ph.D. Geographer US Forest Service 1731 Research Park Dr. Davis, CA 95618 530-759-1717 (ph) 530-747-0247 (fx) rgerrard@fs.fed.us Grants contact: Bernadette Jaquint bjaquint@fs.fed.us ph: (510) 559-6309 fax: (510) 559-6440 Total funding requested: $192,000 In-kind contributions: $ 61,000 II. Proposal Narrative a. Abstract A myriad of sources of environmental change and associated land management challenges exist in the Lake Tahoe basin, as is the case for wildland forest ecosystems across the country. Tools exist to evaluate the effects of various sources of change on wildlife, particularly forest management, but they fall short of management’s information needs in a number of important ways. At the present time, the primary multi-species assessment tool, California Wildlife Habitat Relationships (CWHR) Database System (CDFG 2008), is inadequate to the task because it is insensitive to most present-day changes in forest composition, structure, and configuration. The Lake Tahoe basin needs a basin-specific evaluation tool that can provide reliable estimations of species occurrence across landscapes and over time in response to a variety of demands for evaluating current and potential future environmental conditions. The goal of this project is to use existing empirical field data that was collected in a systematic manner in the Lake Tahoe basin to develop geographic range maps and habitat occupancy models for high priority forest-associated vertebrate species in the Lake Tahoe basin. As such, we will be extracting additional information and utility from past agency investments. We propose to build habitat occupancy models because they are more robust than abundance models in that they can account for variation in sampling effort and techniques, and the models are not as sensitive to sample year as abundance models. These models will facilitate site and landscapescale evaluations of management treatments, climate change, and other change agents that affect forest structure and composition today and in the future. b. Justification Statement A myriad of sources of environmental change and associated land management challenges exist in the Lake Tahoe basin, as is the case for wildland forest ecosystems across the country. Sources of change are diverse, including land development, recreational uses, wildfire, insects, disease, forest management, and climate change. The complexity of the challenge is four-parted: 1) the broad suite and multifarious character of influences affecting forest ecosystem conditions; 2) the limited information on historical conditions, particularly in regard to temporal and spatial heterogeneity; 3) the prospect of rapid rates of climate change with uncharted ecological responses; and 4) the uncertainty of how best to achieve targeted conditions through land management. Tools exist to evaluate the effects of various sources of change on wildlife, particularly forest management, but they fall short of management’s information needs in a number of important ways. At the present time, the primary multispecies assessment tool, California Wildlife Habitat Relationships (CWHR) Database System (CDFG 2008), is inadequate to the task because it is insensitive to most present-day changes in forest composition, structure, and configuration. For example, Manley et al. (2007) and Murphy et al. (2008) found in two different landscapes that standard fuels reduction treatments did not affect a change in forest habitat classification as per CWHR, despite substantial changes in forest structure and in bird and small mammal composition and abundance following treatment. Thus, the CWHR database was not effective in predicting or reflecting changes in habitat associated with forest fuels reduction treatments. The Urban Biodiversity study found that the amount of development in the landscape and human disturbance had significant effects on the occurrence and/or abundance of many bird and mammal species (Manley et al. 2007). Changes in forest conditions resulting from climate change are likely to be even more subtle, at least initially, and therefore it is unlikely that CWHR will be effective in evaluating climate change effects. The Lake Tahoe basin needs an assessment tool that can be applied to a variety of wildlife and biodiversity evaluation needs to help inform land management planning and implementation in light of multiple interacting change agents acting across the landscape and over time. A rich array of empirical data is available and can be tapped to provide this important tool, and thereby increase the value of past and current agency investments in research and monitoring. c. Background and problem statement Lake Tahoe is treasured place in the Sierra Nevada, and it attracts hundreds of thousands of visitors each year. In addition to the growing resident population, many residents of California and elsewhere have summer homes in Lake Tahoe. The competing pressures to keep Tahoe blue, wild, and safe from wildfire make land management extremely complex. The difficulties are reflected in multiple ongoing planning efforts in the basin. For example, the efforts to revise long-standing and overdue land management plans for the basin as a whole (Tahoe 2 Regional Planning Agency’s 20-yr Regional Plan) and for National Forest System Lands in the basin (Lake Tahoe Basin Management Unit’s Land and Resource Management Plan) have struggled over the past four years to identify key measures and indicators of ecosystem health. Controversy over the derivation and reliability of these measures has served to slow their development. Another example is the large-scale implementation of fuels reduction treatments planned for the basin, as outlined in the Lake Tahoe Multi-jurisdictional Fuel Reduction and Wildfire Prevention Strategy (2007). Perceptions of fire danger are driving forest management activities in the Lake Tahoe basin at the present time, as they are on public lands throughout the west. Although forest management to reduce fuels is designed to significantly alter forest structure, it does not typically change the seral stage classification of a stand. Fuels reduction treatments target the removal of smaller diameter trees, and as such average diameter either stays the same or increases. The CWHR database determines habitat suitability based on proxies for seral stage, namely average tree diameter and canopy cover. Although more refined habitat models exist for a few high interest species (e.g., American marten, California spotted owl), the basin lacks the ability to address many other species of interest (e.g., northern flying squirrel, Trowbridge’s shrew, long-tailed weasel, Pileated woodpecker) or biological diversity as a whole in the assessment of the effects of management at stand or landscape scales. Climate change further complicates land management assessments in that its future effects are unknown but are likely to be ecologically significant, yet it is unclear how the effects of present-day management will interface with future changes in environmental conditions imposed by changes in temperature and precipitation. Research directed at understanding historical vegetation conditions in the basin (Taylor et al. 2008, North 2008) will generate models of landscape conditions and dynamics that can interpreted in terms of how plants and fire may respond to current and future conditions, how to design landscapes to be well adapted to potential future climate conditions, and how wildlife may respond to various future scenarios. The Lake Tahoe basin needs a basin-specific evaluation tool that can provide reliable estimations of species occurrence across landscapes and over time that can be responsive to a variety of demands for evaluating current and potential future environmental conditions. The development of predictive tools for multiple species of wildlife can be accomplished in a variety of ways. Most commonly, quantitative models are developed based on qualitative data, and they are posed as working hypotheses for habitat suitability and quality. Habitat Suitability Index is a fairly common example of this type of model (Tirpak et al. 2009). They are satisfying in that they attempt to address specific habitat needs, such as nest site substrates, other breeding habitat needs, and foraging and resting needs in the quantitative evaluation of site specific habitat suitability. In application, however, they are difficult to validate, and therefore their utility is primarily limited to large-scale relative outcomes. The alternative is to build habitat models based on empirical data, which is likely to limit the scope of model, but greatly improve its reliability and credibility for effectiveness and desired condition monitoring and evaluations. Models that predict abundance are highly dependent upon the techniques, effort, and year of sampling, thus is it is difficult to develop a generic predictive model for measures of abundance that others can reliably apply in a variety of contexts. Models that predict probability of occupancy, however, are more temporally robust, and they can account for variation in sampling effort and techniques. Fortunately, empirical wildlife occurrence data exist across the basin, which we can employ to build reliable and robust habitat occupancy models. d. Goals, objectives, and predictions The goal of this project is to use existing empirical field data that were collected in a systematic manner in the Lake Tahoe basin to develop geographic range maps and habitat occupancy models for high priority forestassociated vertebrate species in the Lake Tahoe basin. These models will facilitate site and landscape-scale evaluations of management treatments, climate change, and other change agents that affect forest structure and composition today and in the future. This project proposes to develop habitat occupancy models (estimates of probability of occurrence have a range between 0 and 1) based first on geographic range, and second on site-specific environmental conditions. These will be first generation models, and as such, they target simple objectives that can be readily met with existing data. The specific objectives of the project are as follows: 1) gather and groom data from existing projects to enable their combination for this analysis; 2) generate a basin-wide distribution map for each species; 3) derive values for a set of environmental variables for each sample site; 4) develop logistic regression models that predict probability of occurrence for each of approximately 40 wildlife species, with emphasis on those species that are of greatest interest to management and that appear to be most sensitive to recent and expected future changes in forest ecosystems in the basin; 5) develop an electronic atlas for the basin that indicates for each cell the species ranges in which it occurs and the probability of occurrence based on GIS-based environmental variables; and 6) present 3 results in an electronic format that allows users to view distribution maps, observe basin-wide occupancy estimates, and input values for site-specific environmental variables to derive occupancy probabilities for one or more species. e. Approach, methodology and location Species and Datasets The project will draw from two primary datasets: 1) the Multiple Species Inventory and Monitoring data, collected at 100 sites on NFS lands throughout the basin from 2002-2005; 2) the Lake Tahoe Urban Biodiversity Study data collected at 100 sites across multiple land ownerships at lower elevations (<7500 ft) in the basin from 2003-2005. These datasets are similar enough to be combined into one dataset to evaluate habitat relationships. Together they represent the full spectrum of the primary environmental conditions occurring in the basin, meaning locations around the lake and sites from lake level to near the crest. At each site, bird, small mammal, large mammal, and vegetation data were collected in a similar manner. Specific methods are described below. Secondary datasets will be used in the development of distribution maps, since data requirements for this application only pertain to species occurrence, and can be based on positive sightings alone, if necessary. Additional datasets will be identified and obtained through contacts with agency personnel and other researchers in the basin. The proposed species of interest are listed in Table 1. They represent species identified in the wildlife and fisheries chapter of the Pathway 2007 Evaluation report (multi-agency plan revision preparation effort) as species of interest based on past or potential future population declines in response to environmental change. Species listed as sensitive to environmental change are species typically associated with mature or older undisturbed forested conditions, or they are upper trophic level species for which declines or sensitivities to management have been well documented. We will consult with agency personnel to refine the list of species so it best meets their needs to extent possible given the available data. Table 1. Priority species for modeling based on their sensitivity to habitat changes. Sensitive species Adapter species Birds Bald eagle* MacGillivray’s warbler Band-tailed pigeon Black-backed woodpecker Mountain quail Brewer’s blackbird Sooty grouse Nashville warbler Brown-headed cowbird Brown creeper Northern goshawk* Common raven California spotted owl* Olive-sided flycatcher Mountain chickadee Calliope hummingbird Osprey* Mourning dove Cassin’s finch Pileated woodpecker Northern flicker Cassin’s vireo Pine grosbeak* Pygmy nuthatch Chipping sparrow Pine siskin Steller’s jay Dark-eyed junco Townsend’s solitaire Dusky flycatcher Warbling vireo Golden-crowned kinglet Western tanager Green-tailed towhee* White-breasted nuthatch Hairy woodpecker Wilson’s warbler Hermit thrush Yellow-rumped warbler Hermit warbler Small mammals Deer mouse Ermine* Golden-mantled ground squirrel Lodgepole pine chipmunk Long-tailed weasel* Shadow chipmunk Trowbridge’s shrew Western jumping mouse* California ground squirrel Douglas squirrel Long-eared chipmunk Yellow-pine chipmunk Larger mammals American marten* Black bear Bobcat* Spotted skunk* Coyote Raccoon Striped skunk* 4 Species listed in Table 1 as adapters are those that are expected to fair well, and in some cases better, under more disturbed conditions. Adapter species serve as an important point of reference for sensitive species in two ways: 1) sensitive species generally decline in response to unfavorable change, and as their numbers decline, it becomes increasingly difficult to discern their status or validate predicted changes in ecological responses to changing environmental conditions; and 2) adapter species are more abundant overall and increases in their occurrence are readily detected and as such provide a stronger ecological signal as conditions decline. The combination of declines in one group and increases in the other provide a more reliable signal than species in only one group could accomplish. Asterisks indicate species for which the data from the two studies are not likely to have sufficient detections to derive reliable predictions of occurrence. It is likely that existing habitat occupancy models developed for high profile species, such American marten, could be incorporated into the final package of models if they are in a compatible format. Field Methods for Primary Datasets Bird Data - We will compile bird point-count survey data collected at 170 sample sites between MSIM and LTUB projects. Each sample site had 5-7 point count stations set approximately 200 m apart, and each station was visited 2-3 times during the spring of a single year. Although the number of point count stations per site were different between projects, they were both clustered in a circular or oblong formation at a 200-m spacing. We selected four point count stations for each sample site such that sample effort was consistent across studies. Point counts were conducted during the breeding season between mid-May and mid-August between 0530 and 1000 hrs. All stations within a given sample site were sampled on the same day and surveys were not conducted in inclement weather. Observers recorded all individual birds detected within a 10-min count period (Blondel et al. 1981); we used only those detections within 100 m in the analysis. Small Mammal Data - Small mammals were surveyed with Sherman live-traps as part of the MSIM and LTUB projects. A total of 170 sample sites were surveyed once for small mammals. Sampling protocols were similar across the two projects, with trap size (extra long [4” x 4.5” x 15”] for MSIM, extra long and large [3” x 3.5” x 9”] for LTUB), number (64-104), spacing (15 m), bait (oats and seeds), trap duration (3 days and nights), and trap check frequency (twice per day) all being consistent. The grid configuration of traps did vary among projects (Fig. 2.3). In the LTUB project, traps were arrayed in a square 8 x 8 grid (~ 1.1 ha; 2.8 ac). For MSIM, 104 traps were arrayed in a more open transect pattern occupying an area approximately 10 ha. Large Mammal Data - Large mammal data are available for 182 sites: 77 LTUB sites sampled in 2003 or 2004; and 105 MSIM sites (sampled in 2002 or 2005). In both studies, we surveyed for carnivores at each sample site once between 2002 and 2005 from May through September using a combination of track plates (sooted aluminum plates enclosed in a plastic shelter) and Trailmaster 1500 cameras (Goodson and Associates, Lenexa, KS) (Zielinski and Kucera 1995). Use of multiple techniques typically improves the probability of detecting species (Foresman and Pearson 1998, O’Connell et al. 2006). A baited trackplate was placed near the sample unit center and paired with a baited camera located 100 m away on a randomly selected azimuth. Track plates and cameras were baited with chicken and a commercial scent was used as a lure (Gusto, M&M Furs). Detection devices were run for 10 days and were visited every other day to check bait and function for a total of five visits. Species Distribution Maps Species distribution maps are intended to represent primary demarcations in species occurrence in the basin as a function of their geographic range outside the basin and/or their ecological niche in regard to elevational ranges. The basin is located at the intersection of major and minor zoographic regions from west to east and north to south, respectively. As a result, many species are at the edge of their range in the basin or the edge of their ecological niche relative to elevation (high and low). Species distribution maps will be based on positive sightings from reliable sources over the past 10 years. Once sightings are compiled for a given species, they will be analyzed using spatial range estimation techniques such as adaptive kernel estimation (Kernohan et al. 1998) to derive probability isoclines for range boundaries. Clustered sightings will be “thinned” to eliminate any excessive influence on the estimation of range boundaries. Quantitative estimates will be conducted using ArcGIS (8.0) and ABODE software at a 1 ha cell size. For the purposes of the atlas, these estimates will be converted to a 0/1 grid with a cell size of 1 km2. 5 Logistic Regression Modeling We will use logistic regression to determine the environmental variables with the greatest influence on species occurrence and incorporate estimates of probability of detection in estimates of probability of occurrence. The sample unit for presence and non-detection will be determined per point count station for birds, per transect for small mammals, and per site for large mammals. Detections of a given species are a function of presence and detectability (MacKenzie et al. 2002). In cases where the sampling method varied between studies (number of point count stations per site for birds, array and type of traps for small mammals), we will structure the detection probability model to quantify and account for any differences in probability of detection afforded by various methods and use a bootstrap approach to account for dependencies among clusters of sample units (multiple point counts and transects per site). We will use logistic regression adjusted for the detection probability structure (PROC NLMIXED; SAS Institute, 2002) to analyze relationships between species presence and environmental conditions. A fixed set of 34 environmental variables will be used in each logistic regression analysis (Table 2). The environmental variables represent a common set of variables that could be derived for all sample sites across studies, including a combination of field-based measurements and remotely-sensed digital data. Two models will be generated: 1) all 34 environmental variables; and 2) GIS data-only models. It is assumed that the inclusion of field-based variables in predictive models will improve the accuracy of occupancy estimates; however GIS data-only models are valuable for evaluating landscape-scale changes. We will use Akaike’s Information Criteria (AIC) adjusted for small sample sizes (AICc) to rank the relative influence of models (lower AICc indicates greater influence) in environmental analyses, and will use modeling averaging to derive the best predictive model (Burnham and Anderson, 2002). Once the final field-based model is derived, we will rerun the analysis with various subsets of the final set of explanatory variables such that users can still run the field-based models in the event they are lacking one or more input variables. Table 2. Environmental variables used to generate predictive models for species occurrence. Variable Description GIS based: Urban Development Proportion of area occupied by impervious surfaces@100 and 300-m radii Mean Elevation (m) DEM data Precipitation (cm) PRISM data; average annual over past 30 yrs Slope position DEM data Vegetation Zone Lower montane, upper montane, or subalpine Canopy Cover Proportion of area occupied by 5 canopy cover classes within 100 & 300 m Vegetation Type Proportion of area occupied by 6 vegetation series within 100 & 300 m NDVI Productivity index Average Tree Diameter Proportion of area occupied by 4 diameter classes within 100 & 300 m Field based: Canopy Cover Shrub and herb cover Small Trees/ha Medium Trees/ha Large Trees/ha Small Hard snags/ha Large Hard snags/ha Small Soft Snags/ha Large Soft Snags/ha Small Hard Logs/ha Large Hard Logs/ha Small Soft Logs/ha Large Soft Logs/ha Densiometer Line transect 12-28cm dbh >28-61cm dbh >61cm dbh Decay classes 1-2, >28-61cm dbh Decay classes 1-2, >61cm dbh Decay classes 3-5, >28-61cm dbh Decay classes 3-5, >61cm dbh Decay classes 1-2, >28-61cm dbh Decay classes 1-2, >61cm dbh Decay classes 3-5, >28-61cm dbh Decay classes 3-5, >61cm dbh 6 Model Delivery Distribution maps will be available in paper and digital format. The digital format will be in the form of an ArcGIS grid, so that they can be used for analysis if desired. The set of predictive models (GIS-based, field-based global model, and field-based missing-one-variable models) for each species will be packaged in a simple electronic format that will allow users to provide input data for environmental conditions and generate an output of probability of occurrence of each of the species modeled. Documentation for the models will be provided, including an explanation of the assumptions and limitations of the models, probability of detection estimates, and the field-based (full model) and GIS-based predictive equations. The format for electronic models will be simple; we will configure a user interface that will allow users to input values for environmental variables, and provide output as to the estimated probability of occurrence for each species and its associated variance. The product will be Excel-based, a readily available platform that can accomplish the simple functions that are required. We may use the graphical interface function provided by Mathematica 7.0.1 (Fig. 2) or a similar product to allow users to get a better understanding of the relative influence of the top three variables on probability of occurrence. The graphical interface allows the user to see changing values for the top three variables affects the probability of occurrence. If a more sophisticated front-end software interface is desired, it can be accomplished at a later date by one or more agencies and designed to meet their specific needs. f. Relationship of research to previous research, monitoring, and management This project utilizes significant investments previously made by local agencies and SNPLMA. The MSIM project was a basin-wide monitoring effort funded by Round 4 SNPLMA and the Lake Tahoe Basin Management Unit. The investment in that project exceeded $1 million. The LTUB study was funded primarily by SNPLMA, but with numerous smaller contributions from various state and federal agencies in the basin, for a total of approximately $750,000. Secondary surveys that may serve as sources of positive sighting data for the distribution maps include (but are not limited to) stream restoration surveys funded by the US Forest Service (re: Morrison), aspen restoration surveys funded by the US Forest Service (re: Richardson), wildlife surveys funded by California Department of Off-Highway Vehicles (re: Manley and Zielinski), and Carson Range surveys funded by Nevada Division of Wildlife (re: Richardson and Catalano). Together these surveys canvas the Lake Tahoe basin, and compiling these data into a usable format for agencies and researchers will extract further value from these initial investments. Some of the data compilation required for this project has already been accomplished through the P7 planning effort. In that effort, we analyzed data to identify potential indicators and an index of forest ecosystem integrity based on richness and abundance data for birds, small mammals, and richness data for large mammals. That analysis was based on a subset of the MSIM data, since not all the MSIM data collection had been completed at the time of that analysis, whereas now the full dataset is available. This project will require a substantial redress of the entire dataset to enable estimates of probability of occupancy for individual species. The product will be useful in supporting a wide range of EIP objectives, as well as the agency planning efforts described earlier that are currently underway. In the EIP, the 10-year target for forest management is to treat nearly 90,000 acres for forest fuel reduction and ecosystem restoration, and to improve wildlife habitat and conserve the natural environment through a variety of activities. The availability of reliable tools to evaluate the effects and effectiveness of management options and activities on wildlife and biological diversity will greatly enhance the ability of agencies to meet the core EIP objectives of ecosystem restoration and conservation. g. Strategy for engaging with managers and obtaining permits We will work closely with agency personnel and wildlife researchers in the basin to use the strongest datasets possible in the analysis of geographic distribution and habitat occupancy. Managers in the basin will be consulted as to the species of greatest interest and need for modeling, and we will engage agency staff in the review of draft maps and models, particularly in the usability of the final package of maps and models. We will hold a workshop at the conclusion of the project to introduce the maps and models to interested users and explain how they can be used to support site and landscape analyses. No permits are necessary for completing this project. 7 h. Description of deliverables The product generated through this project will be the following: 1) electronic distribution maps for each species; 2) a simple software interface that will allow users to input values for environmental variables and generate probability of occurrence (and variance) estimates for one or more of the 40 species; and 3) a users manual that describes how the maps and models were generated and their appropriate use. The maps and models will be made available to agency personnel and any interested individual through the TSC and TIIMS web sites. III. Schedule of major milestones/deliverables Milestone/Deliverables Submit quarterly progress reports Start Date End Date Description Submit brief progress report to Tahoe Science Program coordinator by the 1st of July, October, January, and April. 2010 Data compilation 1-Apr 30-Sep Data analysis 1-Oct 31-Dec 15-Jan 30-Jun 15-Jan 31-Aug Develop logistic regression models for each species and package them for use 1-May 30-Sep 1-Oct 15-Dec Distribute draft species predictive models for agency and peer review and then finalize Offer to present results to interested parties and at local symposium opportunities 15-Jan 30-Jun 2011 Review and finalize species distribution maps Generate logistic regression models Review and finalize predictive models Present findings and review of draft documentation 2012 Deliver final model package and conduct a workshop Compile primary and secondary survey site locations and detections and environmental data for primary datasets Generate species distribution maps, derive values for all environmental variables for regression analyses, and develop statistical programming language for logistic regression analysis Distribute draft species distribution maps for agency and peer review and then finalize Model documentation and electronic models posted on the TSC and TIIMS websites for use and a workshop conducted to demonstrate how to use the models 8 IV. Literature Cited Burnham, K. P., and D. R. Anderson. 2002. Model selection and inference: a practical information-theoretic approach. Springer-Verlag, New York, New York. CDFG California Department of Fish and Game. 2008. California Wildlife Habitat Relationships Database 8.1. California Department of Fish and Game, Rancho Cordova, CA. Foresman, K.R., Pearson, D.E.,1998. Comparison of proposed survey procedures for detection of forest carnivores. Journal of Wildlife Management 62:1217-126. Kernohan, B.J., J.J. Millspaugh, J.A. Jenks, D.E. Naugle. 1998. Use of an adaptive kernel home-range estimator in a GIS environment to calculate habitat use. Journal of Environmental Management 53(1):83-89. MacKenzie, D.I, Nichols, J.D., Lachman, G.B., Droege, S., Royle, J.A., Langtimm, C.A., 2002. Estimating site occupancy rates when detection probabilities are less than one. Ecology 83, 2248-2255. Manley, P.N., D.D. Murphy, M.D. Schlesinger, L.A. Campbell, S. Merideth, M. Sanford, K. Heckmann, and S. Parks. 2007. The role of urban forests in conserving and restoring biological diversity in the Lake Tahoe basin. Final report, SNPLMA grant CCA01-CS-11051900-022, submitted to US Forest Service Lake Tahoe Basin Management Unit, South Lake Tahoe, CA. Manley, P.N., R. Gerrard, S. Parks, T. Kirk, C. Damiani, and K. McIntyre. 2007. Biomass to energy project: wildlife habitat evaluation. Report to the California Energy Commission, Sacramento, CA Murphy, D.D., P.N. Manley, A.E. Stanton, and B.M. Pavlik. 2008. Lake Tahoe Upland Fuels Research Project: 2006-2007 Report. Progress report submitted to Nevada Division of State Lands, Carson City, NV. North, M. 2008. Restoration and fuel treatment of riparian forests. Round 9 SNPLMA grant, Pacific Southwest Research Station, Davis, CA. O’Connell, A.F., Talancy, N.W., Bailey, L.L., Sauer, J.R., Cook, R., Gilbert, A.T., 2006. Estimating site occupancy and detection probability parameters for meso- and large mammals in a coastal ecosystem. Journal of Wildlife Management 70, 1625-1633. Taylor, A., C. Skinner, and H. Safford. 2008. Identifying spatially explicit reference conditions for forest landscapes in the Lake Tahoe basin. Round 9 SNPLMA grant, Pacific Southwest Research Station, Davis, CA. Tirpak, J.M., D.T. Jones-Farrand, F.R. Thompson III, D.J. Twedt, and W.B> Uihlein, III. 2009. Multiscale habitat suitbility index models for priority landbirds in the central hardwoods and west Gulf Coastal plain/Ouachitas Bird Conservation Regions. USDA Forest Service General Technical Report GTR-NRS-49, Newtown Square, PA. Zielinski, William J.; Kucera, T.E. 1995. American Marten, Fisher, Lynx, and Wolverine: Survey Methods for Their Detection. Albany, CA: U.S. Department of Agriculture, Forest Service, Pacific Southwest Research Station; Gen. Tech. Rep. PSW-GTR-157: 163p. 9 V. Figures Figure 1. Locations of sample sites for the two primary datasets (MSIM and LTUB) and one of the secondary datasets (RIP). 10 . Figure 2. Example of a graphical output for a single species model prediction that shows the relationship between the standard error of the estimator of the probability of occurrence (Psi) and one or more predictor or design variables. This software allows one to explore the effects of changing variable values on the standard error of the estimator of Psi through the use of “sliders”. As the sliders are moved, the line representing the standard error shifts in response, reflecting the effects of changing values for one or more explanatory or design variables. 11