1 Appendix S1 Library development We developed a novel microsatellite library for the cactus wren using a modification of the standard techniques of Hamilton et al. (1999). Libraries were constructed by excising genomic DNA using the restriction enzyme HincII, and these fragments were ligated to an SNX linker. Using biotinylated oligonucleotide probes with trinucleotide and tetranucleotide repeats, microsatellite repeat regions were isolated in the genome. These fragments were PCR amplified and sequenced on a 454 GL FLX (Roche) in the Evolutionary Genetics Core Facility (EGCF) at Cornell University. In 3,350 resulting sequences, 414 had microsatellite regions. We mapped these to the Zebra Finch (Taeniopygia guttata) genome to identify the physical locations of each marker, and eliminated any that mapped to sex chromosomes to avoid complications in statistical analysis. We screened 52 loci for variation using a three-primer technique (Schuelke 2000). Genotyping runs were performed on an ABI 3730 DNA Analyzer using the GS600 size standard (Life Technologies) either in the CSUPERB Microchemical Core Facility at San Diego State University or at Bio Applied Technologies Joint, Inc. in San Diego, CA. References Hamilton MB, Pincus EL, DiFiore A, Fleischer RC (1999) Universal linker and ligation proceedures for construction of genomic DNA libraries enriched for microsatellites. BioTechniques, 27, 500-507. Schuelke, M (2000) An economic method for the fluorescent labeling of PCR fragments. Nature Biotechnology, 18, 233-234. 2 Appendix S2 Landscape Variables and Habitat Suitability Model We used cactus wren occurrence data and an environmental dataset characterizing southern California to construct a partitioned Mahalanobis D2 model predicting suitable habitat for cactus wren dispersal (Rotenberry et al. 2002, 2006). The environmental data set consists of climatic, topographic, and vegetation variables calculated for a grid of points spaced 150 m apart using Geographic Information Systems (GIS) ESRI ArcMap 10.2 software (ESRI 2013). The environmental dataset encompasses southern California from the US-Mexico border north to Maricopa (Kern Co., CA) and from the Pacific Ocean to the east of El Centro (Imperial Co., CA). We calculated climate variables at each grid point using normalized 30-year modeled estimates (1981-2010; Daly et al. 2008, PRISM 2013) for minimum January temperature (°C), maximum July temperature (°C), and annual precipitation (mm). Topographic variables included elevation, slope (degrees), northness (sine of aspect), and eastness (cosine of aspect), extracted for each point from a 10 m digital elevation model (Gesch et al. 2002, Gesch 2007). We derived topographic heterogeneity, a measure of ruggedness, for a 30 m x 30 m neighborhood of 10 m cells at each grid point (Sappington et al. 2007). Cactus wrens disperse through shrublands and we calculated the percent of the landscape within 1 km of each grid point that was classified as coastal sage scrub or as chaparral using the Fire Resource Assessment Program vegetation map published in 2002 (FRAP 2002) and updated in 2006. Mahalanobis D2 measures the similarity between the multivariate mean for a set of environmental variables calculated at locations where a species occurs and at points in the landscape being modeled (Clark et al. 1993). For model calibration, we randomly selected 65% of cactus wren records from a dataset of 353 spatially distinct locations (only one location per 150 m x 150 m grid cell) and used the remaining 35% to evaluate performance. To avoid biasing 3 the model towards environmental conditions at locations with relatively large concentrations of cactus wrens, we bootstrapped the calibration data by limiting each genetic cluster to 10 or fewer occurrences per iteration and performed a Principal Components Analysis (PCA) on each of 1000 samples (Knick et al. 2013). The final model consisted of the average of the PCA output for all iterations after correcting for sign ambiguity (Bro et al. 2008). A continuous habitat similarity index (HSI) was calculated by rescaling D2 from 0 – 1 with an HSI value calculated for each 150 m x 150 m grid cell in the environmental dataset for southern California. An HSI of “0” represents environmental conditions least similar to the multivariate mean for the species occurrence data set used to calibrate the model (i.e., unsuitable), while a “1” represents those conditions that are most similar (i.e., suitable). A second GIS layer representing urban areas in southern California was used to create a fragmentation index among aggregations and to calculate the percent of urban development surrounding populations. Urban areas were derived from an impervious surfaces layer with 30 x 30 m grid cells with more than 20% of their surface area covered by at least 20% imperviousness categorized as urban (Fry et al. 2011). We delineated the amount of suitable dispersal habitat available to cactus wren populations as well as the amount of urban area. Population polygons were calculated by merging contiguous 10-km buffers for each cactus wren location that fell into a particular population. The 10 km distance represents a maximum dispersal distance for individual cactus wrens. We calculated the percentages and amounts (ha) of suitable dispersal habitat (HSI ≥ 0.3), unsuitable habitat (HSI < 0.3), and urban area for each population polygon. To calculate distances among aggregations, first a habitat resistance surface was calculated from the habitat model as the –ln (HSI scores + 1) for 150m grid cells (following Spear et al. 2010). Least cost paths and weighted cost distances among all pairs of aggregations 4 were calculated using the Landscape Genetics ArcToolbox (Etherington 2011) in ArcGIS 10.2. Resistances were calculated in Circuitscape 4.0 (McRae et al. 2008) under the “connect eight neighbors” and average conductance settings. While least cost path measures a single connection route between all pairs of populations, Circuitscape incorporates information on all pathways between pairs. All of these “habitat” distance measures, as well as Euclidean distance were compared to genetic distances among sites using Mantel Tests in IBDWS (Jensen et al. 2005). We compared resulting correlation coefficients to assess the relative fit of each distance measure. To further assess the effects of urban fragmentation, we calculated a binary index based on whether least cost paths crossed through urban areas (1) or not (0) and used partial Mantel tests to compare this matrix to pairwise genetic distances while accounting for geographic distance. Model Results Overall, the cactus wren dispersal habitat model performed well with a median HSI of 0.71 for the calibration dataset (n = 0.71) and 0.69 for the validation dataset (n = 123). The exception was near the Aguanga population where the calculated second-lowest amount of suitable habitat was likely due to poor model fit for the area. REFERENCES Bro R, Acar E, Kolda TG (2008) Resolving the sign ambiguity in the singular value decomposition. Journal of Chemometrics 22, 135-140. Clark JD, Dunn JE, Smith KG (1993) A multivariate model of female black bear habitat use for a geographic information system. Journal of Wildlife Management 57, 519-526. Daly C, Halbleib M, Smith JI, Gibson WP, Doggett MK, Taylor GH, Curtis J and Pasteris PA (2008) Physiographically-sensitive mapping of temperature and precipitation across the coterminous United States. Journal of Climatology 28, 2031-2064. Etherington TR. (2011) Python based GIS tools for landscape genetics: visualising genetic 5 relatedness and measuring landscape connectivity, Methods in Ecology and Evolution 2, 52-55. Fire Resource Assessment Program, Department of Forestry and Fire Protection (FRAP) (2002) Methods for Development of Habitat Data: Forest and Range 2002 Assessment. Technical Working Paper 8-19-02. http://frap.fire.ca.gov/projects/frap_veg/methods/Methods_Development_Habitat_Data_0 2_2.pdf Fry J, Xian G, Jin S, Dewitz J, Homer C, Yang L, Barnes C, Herold N, Wickham J (2011) Completion of the 2006 National Land Cover Database for the coterminous United States, Photogrammetric Engineering and Remote Sensing 77, 858-864. Gesch DB (2007) The National elevation dataset, In: Digital Elevation Model Technologies and Applications: The DEM Users Manual, 2nd Edition (ed, Maune, D). American Society for Photogrammetry and Remote Sensing, Bethesda, Maryland, 99-118. Gesch D, Oimoen M, Greenlee S, Nelson C, Steuck M, Tyler D (2002) The national elevation dataset. Photogrammetric Engineering and Remote Sensing 68 5-11. Jensen JL, Bohonak AJ, Kelley ST (2005) Isolation by distance, web service. BMC Genetics 6, 13. Knick ST, Hanser SE, Preston KL (2013) Modeling ecological minimum requirements for distribution of greater sage-grouse leks: implications for population connectivity across their western range, USA. Ecology and Evolution 3, 1539-1551. McRae, BH, Dickson BG, Keitt TH, Shah VB (2008) Using circuit theory to model connectivity in ecology and conservation. Ecology 10, 2712-2724. PRISM. 2013. PRISM climate GIS dataset. PRISM Climate Group, Northwest Alliance for Computational Science and Engineering. http://www.prism.oregonstate.edu/ Rotenberry JT, Knick ST, Dunn JE (2002) A minimalist approach to mapping species’ habitat: Pearson’s plane of closest fit. In: Predicting Species Occurrences: Issues of Accuracy and Scale. (eds, Scott JM, Heglund PJ, Morrison ML, Haufler JB, Raphael MG, Wall WA, Samson FB). Island Presss, Washington, D.C., U.S.A., 281-289. Rotenberry JT, Preston KP, Knick ST (2006) GIS-based niche modeling for mapping species’ habitat. Ecology 87, 1458-1464. Sappington JM, Longshore KM, Thompson DB (2007) Quantifying landscape ruggedness for animal habitat analysis: a case study using bighorn sheep in the Mohave Desert. Journal of Wildlife Management 71, 1419-1426. Spear SF, Balkenhol N, Fortin MJ, McRae BH, Scribner K (2010) Use of resistance surfaces for landscape genetic studies: considerations for parameterization and analysis. Molecular Ecology 19, 3576-3591. 6 Appendix S3 We would like to acknowledge each of these partners and cooperators individually for their assistance in this study: AECOM Audubon California Starr Ranch Sanctuary Bureau of Land Management Cactus Wren Working Group California Department of Fish & Wildlife California State Parks & Recreation California State Polytechnic University, Pomona California State University, Channel Islands California Department of Transportation Center for Natural Lands Management City of Chula Vista City of Diamond Bar City of El Cajon City of Fullerton City of Glendora City of Irvine City of Los Angeles, Dept. of Recreation and Parks City of Moorpark City of San Diego City of San Dimas City of Thousand Oaks City of Whittier Conejo Open Space Conservation Authority Conejo Recreation and Parks District Conservation Biology Institute Cooper Ecological Monitoring, Inc. County of Los Angeles, Dept. of Parks and Recreation County of San Diego Crystal Cove State Park Fallbrook Naval Weapons Station Helix Water District Irvine Ranch Conservancy Many Private Landowners Marine Corps Base Camp Pendleton Orange County Water District Orange County Parks Outdoor Resorts Rancho California, Inc. Pala Band of Mission Indians Palos Verdes Peninsula Land Conservancy Puente Hills Habitat Preservation Authority San Bernardino County Dept. of Public Works 7 San Bernardino County Flood Control District San Bernardino County Water Conservation District San Bernardino Valley Municipal Water District San Diego Audubon Society San Diego Gas & Electric San Diego Monitoring and Management Program San Diego National Wildlife Refuge San Diego Zoo Institute for Conservation Research San Dieguito River Park San Dieguito River Valley Conservancy Santa Ana Watershed Association Southern California Edison Viejo Conservation Easement Sweetwater Authority The Nature Conservancy UC-Irvine Ecological Preserve US Fish & Wildlife Service Vulcan Materials Company Western Riverside County MSHCP Western Riverside County Regional Conservation Authority Western Foundation for Vertebrate Zoology