Bay Checkerspot Butterfly Dispersal Corridor Modeling for Coyote Ridge
Michael Rochelle and Tanner Harris
The Bay checkerspot butterfly (BCB; Euphydryas editha bayensis) is a federally threatened species endemic to the San Francisco Bay Area. To assess potential impacts to BCB resulting from a proposed development, WRA Inc. modeled dispersal corridors for the species in ESRI’s ArcGIS using Linkage Mapper (McRae and Kavanagh 2011). Physical and biological parameters were used to develop a weighted overlay to create a corridor resistance raster. Cost-distance analysis was used to choose the prime dispersal corridors and identify areas which should be avoided by the project.
Bay checkerspot butterfly is a medium sized butterfly in the brush-footed family, Nymphalidae, with an average wingspan of just over 2 inches. See picture 1. Bay checkerspot butterfly is a short-lived butterfly with adults emerging from pupation in late February to early April; they mate and lay eggs during a four to six week flight season. Adults typically live an average of ten days, and once eggs are laid, they hatch within 10 to 15 days. Larvae (caterpillars) feed for approximately two weeks and then enter a period of dormancy that generally lasts throughout the summer and fall. With the onset of the rainy season and associated wildflower growth, the larvae become active again until reaching a weight of between 250 and 500 milligrams, at which point they pupate and the cycle repeats.
Habitat for BCB exists only on shallow serpentine soils that support larval host plants and a range of native wildflowers used by adult butterflies as nectar sources like the flowers shown in picture 2. Serpentine soils are derived from ultramafic rock, and are characterized as shallow, nutrient-poor (often lacking in nitrogen and phosphorous) soils with elevated magnesium levels, a calcium-magnesium imbalance, high concentrations of toxic heavy metals such as nickel and chromium, and low water holding capacity. Due to these characteristics, serpentine grasslands are inhospitable to many non-native grasses that dominate other California grassland ecosystems (USFWS 2009), whereas native grassland species that are adapted to serpentine soil conditions can persist. The extreme result of this is referred to as edaphic endemism, a condition in which plants and animals have evolved to only occur on a specific soil type.
When BCB larvae emerge, they feed primarily on a single species of plant, referred to as a larval host plant. The primary larval host plant for BCB is a small, annual native plantain (Plantago erecta) which is able to grow on serpentine soils, shown in picture 3. Once adult butterflies emerge, they require nectar from other plants to survive. Adult nectar plants such as desert parsley (Lomatium spp.), California goldfields (Lasthenia californica), owl's clover (Castilleja exserta), tidy-tips (Layia platyglossa), sea muilla (Muilla maritima), scytheleaf onion (Allium falcifolium), false babystars (Leptosiphon androsaceus), and intermediate fiddleneck (Amsinckia intermedia) are important native wildflowers used as nectar sources by BCB (USFWS 2008). See pictures 4, 5, and 6.
BCB Distribution and Decline
Historically, BCB occurred in the vicinity of the San Francisco Bay area from San Bruno Mountain west of the Bay to Mount Diablo in the East Bay to Coyote Reservoir in the South Bay (Murphy and Ehrlich 1980). The current range is greatly reduced, including only San Mateo and Santa Clara counties, with virtually all extant populations occurring on or within a 9-mile radius around Coyote Ridge, which contains the only remaining core population, located roughly 20 miles southeast of San Jose (USFWS 2008). Bay checkerspot butterfly has not been observed in San Mateo County since 1998 (Stanford 2006, CNDDB 2008). See figure 1 for a visual of the geographic distribution of the extirpation time sequence and remaining BCB population.
Some evidence suggests that the primary cause for the decline in BCB populations was the rapid invasion of non-native grasses, which out-competed the butterfly’s native larval host plants. It is thought that nitrogen deposition from car exhaust was responsible for creating soil conditions that allowed non-native grasses to invade serpentine soils where they are otherwise generally not able to grow (Weiss 1999). It is thought that the removal of cattle grazing at several locations resulted in a dramatic decline by allowing for more rapid invasion of non-native grasses.
Bay Checkerspot Butterfly Corridors
Over the last several decades, conservation biologists have focused on the importance that movement corridors or landscape linkages play in maintaining a healthy ecosystem as well as in the recovery of threatened and endangered species. Corridors that link two or more areas of core habitat allow for long-term genetic interchange between populations and opportunities for individuals to re-colonize areas after a local extirpation. The term corridor, at its core, denotes areas that allow movement of organisms or their genetic information (Hitly et al. 2006). “Permeability” is the term used to describe the ability of an organism to travel through the matrix between habitats; “resistance” is described as the ecological or physiological cost of traveling through unsuitable habitat.
Bay checkerspot butterfly is described as having a metapopulation dynamic (Ehrlich et al. 1975), which is a group of spatially distinct populations that occasionally exchange individuals (USFWS 1998), hence the importance of preserving dispersal corridors in maintaining the metapopulation dynamic between distinct populations of the species. Bay checkerspot butterfly is considered a relatively localized insect, with most movements occurring within a single patch of serpentine grassland (Ehrlich 1965; Ehrlich et al. 1980; Ehrlich and Murphy 1981; Harrison 1989). It has been estimated that less than 5% of BCB individuals travel greater than 1,600 feet from where they pupate (McKechnie et al. 1975) and less than 10% of BCB individuals ever leave serpentine habitat (ICFI 2012). Despite the species’ propensity to remain within a single habitat patch, several authors have documented long-distance dispersal. Harrison (1989) documented movements of up to 3.5 miles by one individual. Other authors have reported movements of up to 5 miles (USFWS 1998).
The Santa Clara Valley Habitat Plan (SCVHP) established a conservation strategy for the species aimed at preserving large blocks of high-quality habitat and maintaining linkages (i.e., dispersal corridors) between such blocks of habitat; the Plan also allows for development that is consistent with the conservation strategy. The proposed development, referred to as Young Ranch, will help meet the SCVHP’s conservation goal by proposing a 79-unit, low-density conservation community that would preserve 1,950 acres (90% of the Young Ranch property) and site all homes away from high quality serpentine habitat and BCB dispersal corridors.
The SCVHP defines landscape linkages as “areas that allow for the movement of species from one area of suitable habitat to another” (ICIF 2012), and identifies 20 aquatic and terrestrial linkages within its coverage area. One linkage, Linkage 6, extends northwest to southeast along the length of Coyote Ridge, approximately 9.5 miles along the ridgeline from Silver Creek Hills in the north to Anderson Reservoir in the south. The lands at Young Ranch represent a critical part of Linkage 6, the purpose of which is to “provide connectivity for serpentine species within core habitat along Coyote Ridge” (ICIF 2012, Table 5-9). Linkage 6 is the focus of this assessment, with Coyote Ridge representing the Study Area shown in figure 2.
Corridor Modeling and Software
To determine where to site the development, Linkage Mapper was used to model dispersal corridors in GIS and identify the prime corridors for BCB. Linkage Mapper is a GIS tool developed by The Nature Conservancy to support wildlife connectivity analysis. It consists of several Python scripts packaged into an ArcGIS Toolbox to automate corridor analysis.
Linkage Mapper only requires two inputs for analysis: a species’ core habitat areas and a resistance to movement layer used to map least-cost linkages between the core habitat areas. Each cell in the resistance layer is attributed with values reflecting the energetic cost, difficulty, or mortality risk of moving across that cell (McRae and Kavanagh 2011). The resistance layer is usually created from a weighted overlay of several biological and physical factors identified as important to corridor movement for a specific species.
With core habitat areas and resistance to movement for inputs, Linkage Mapper performs a cost-weighted distance analysis for each core area, then normalizes and mosaics the individual corridor maps to create a single composite corridor map (McRae and Kavanagh 2011). The cost-weighted distance raster for each core habitat only shows the accumulated cost from moving away from that core habitat in every direction; it does not model a specific corridor from one area of core habitat to another. However, when you mosaic together two or more cost-weighted distances from multiple core habitat areas, then you get the cost-weighted distance between habitats, resulting in a corridor layer and the ability to distinguish different percentages of suitable corridor “slices” between habitats.
Corridor Model Parameters and Results
Based on the ecological requirements of BCB (determined from an extensive literature search) and the distribution of the species at Young Ranch (determined from six years of adult butterfly surveys at the site), WRA identified the following parameters as the most significant influences on the likelihood that individuals will move through potential corridors between core habitat areas: (1) topographic position, (2) elevation, (3) thermal stratification, (4) wind interaction, (5) land cover, and (6) distance from occupied BCB habitat. Resistance values are given to each factor based on statistically analyzing occurrences of BCB observations during WRA’s annual, 21 person-day surveys at Young Ranch. See figure 3 for survey results. For each resistance factor, the density of BCB was calculated for each category or class break within the variable, then normalized based on the proportion of that variable present on Young Ranch. Each category or class was assigned resistance numbers on a scale of 0 to 100, with a score of zero representing prime “permeability” and a score of 100 for complete “resistance.” Following the initial development of the model, WRA consulted with Dr. Alan Launer, a BCB expert from Stanford University’s Center for Conservation Biology, to help refine the weighting factors for each of the parameters as they relate to their importance for BCB.
Topographic position was included as a resistance factor due to the documented preference of BCB for ridge tops and shoulders (Baughman et al. 1989; Ehrlich and Wheye 1988). Topographic position was categorized using the tool of the same name in Corridor Designer (Majka et al. 2007) in which categories of canyon bottom, ridge top, and the sloped surface between canyon bottom and ridge top are derived from a neighborhood analysis of elevation data. Topographic position was weighted at 10% of the overall resistance layer for the model.
Elevation was included as a resistance factor due to the high concentrations of BCB at certain elevation ranges. Since there is an energetic cost to moving up and down elevations, elevation was included in the resistance layer and given a weight of 5%.
Thermal stratification, in the form of solar radiation, was included as a resistance factor because of its documented importance in the ecology of BCB (Weiss et al. 1988). ESRI’s Solar Radiation tool was used to calculate insolation due to changes in slope, aspect, and time of day and year for the BCB flight season and was given a weight of 15% due to the importance of this factor for BCB.
Wind interaction was included as a resistance factor due to the influence of wind on BCB flight, whereby areas with greater average wind interaction are less favorable than those with lower wind interaction. Wind data from a local weather station was transformed following a method that integrates wind speed and direction with local topography to estimate interaction with the land surface as outlined by Zack and Minnich (1991). Wind interaction was weighted at 10% for the resistance layer.
Land cover (i.e., vegetation) on Coyote Ridge was included as a resistance factor for its potential to either encourage or impede movement through habitat areas. Land cover was derived from the SCVHP (ICFI 2012) land cover data and further revised by adding more detailed site-specific habitat mapping at Young Ranch and additional modifications outside Young Ranch based on development and other features apparent on aerial imagery. Land cover provides an indicator of both disturbance levels (e.g., four-lane roads with high traffic levels) and physical barriers to movement (e.g., buildings). Due to its importance in defining suitable BCB habitat (i.e., serpentine grassland) land cover was weighted relatively high, at 20% for the resistance layer.
Distance from occupied BCB habitat was included as a resistance factor because of the high habitat fidelity demonstrated by BCB. Although BCB have been, on rare occasions, documented to disperse distances of between 3.5 and 5 miles (Harrison 1989; USFWS 1998), presumably through large expanses of non-serpentine habitat, BCB movement is generally restricted to serpentine habitats (Ehrlich 1965; Ehrlich et al. 1980; Ehrlich and Murphy 1981; Harrison 1989). Given the highly restricted nature of BCB distribution and dispersal, distance from occupied BCB habitat was given relatively high importance in the weighted model. The determination of resistance scores for each class was based on the number of observations of BCB within pre-defined distance classes radiating away from habitat centers. The BCB habitat polygon itself was scored as the “zero” distance class. The vast majority of BCB were found on or in the immediate vicinity of occupied BCB habitat, demonstrating the importance of this factor in determining areas most likely to be accessed by BCB. Distance from occupied BCB habitat was weighted at 40% for the resistance layer.
Figure 4 shows the resistance values and weighting factors for each parameter. A weighted overlay of these six factors creates a corridor resistance raster model which reflects the energetic cost/difficulty/mortality risk faced by an individual moving across a cell. The weighting was performed with simple mathematics in Raster Calculator by multiplying each factor by its weight, then summing together, producing a composite of overall resistance for BCB dispersal on a scale of 0 to 100. See figure 5.
The resistance raster and BCB population centroids were then used as inputs for Linkage Mapper, resulting in a composite cost-distance raster between BCB populations. The cost-distance raster was then reclassified to derive corridor suitability slices, where the best 1%, 10%, 20%, etc., of the modeled dispersal corridor was visualized and used for further analysis, as shown in figure 6. Corridor suitability was modeled for both pre and post-project conditions, and prime corridor suitability slices were determined to be at the top 20% threshold. Figure 7 shows the corridor suitability slices and 20% prime threshold for both pre and post-project conditions, resulting in a comparison in which it is difficult to see any visual difference.
The main area of concern was a bottleneck in the northeast corner of the property, adjacent to a portion of the development footprint. Although there is no visual difference at this bottle neck for pre- and post-project conditions, upon closer inspection within GIS, the 20% prime corridor threshold does not change in diameter under the post-project condition, but moves north by 1 cell size (~10 ft. resolution) in some areas through the bottleneck. This minimal shift in the prime corridor suggests that the development would not have a significant impact on the dispersal of BCB across the site. This is also reflected in the overall corridor resistance scores for Young Ranch, which only rose by 0.5 (from 48.9 to 49.4) and the same for the entire Coyote Ridge Study Area (from 51.00 to 51.05).
The corridor model was used to refine the development plans to avoid conflicts with the BCB dispersal corridors at the site and along the larger Coyote Ridge. The corridor analysis helped to develop a plan for the proposed conservation community at Young Ranch that would have nominal impacts to BCB dispersal along Coyote Ridge. The analysis, which was peer reviewed by experts in the ecology and life history of BCB, determined that any impacts from the low-density housing would be almost imperceptible in nature and would not be likely to impact long-term trends in BCB movement between core habitat units on Coyote Ridge. In addition to the avoidance of all serpentine grassland habitats and prime dispersal corridor habitat, it was also determined that the spacing of the houses, combined with management of non-native grassland, would allow for effective butterfly passage both within the lands at Young Ranch and along the greater Coyote Ridge corridor.
Baughman, J.F., D.D. Murphy, and P.R. Ehrlich. 1989. A reexamination of hilltopping in Euphydryas editha. Oecologia 83: 259-260.
CNDDB: California Department of Fish and Game, Natural Diversity Data Base. 2008. Element Occurrence Reports for Euphydryas editha bayensis. Unpublished cumulative data current to August 2, 2008.
Ehrlich, P.R. 1965. The Population Biology of the Butterfly, Euphydryas editha. II. The Structure of the Jasper Ridge Colony. Evolution 19: 327-336.
Ehrlich, P.R. and D.D. Murphy. 1981. The Population Biology of Checkerspot Butterflies (Euphydryas). Biologisches Zentralblatt 100: 612-629.
Ehrlich, P.R. and D. Wheye. 1988. Hilltopping checkerspot butterflies revisited. The American Naturalist 132: 460-461.
Ehrlich, P.R., D.D. Murphy, M.C. Singer, and C.B. Sherwood. 1980. Extinction, reduction, stability and increase: the responses of checkerspot butterfly (Euphydryas) populations to the California drought. Oecologia 46: 101-105.
Ehrlich, P.R., R.W. White, M.C. Singer, S.W. McKechnie, and L.E. Gilbert. 1975. Checkerspot Butterflies: A Historical Perspective. Science 118(4185): 221-228.
Harrison, S. 1989. Long distance dispersal and colonization in the Bay checkerspot butterfly, Euphydryas editha bayensis. Ecology 70: 1236-1243
Hilty, J.A., W.Z. Lidicker Jr., and A.M. Merenlender. 2006. Corridor Ecology: The Science and Practice of Linking Landscapes for Biodiversity Conservation. Island Press, Washington.
ICFI. 2012. Final Santa Clara Valley Habitat Plan. Prepared for: City of Gilroy, City of Morgan Hill, City of San Jose, County of Santa Clara, Santa Clara Valley Transportation Authority, Santa Clara Valley Water District. August, 2012.
Majka, D., J. Jenness, and P. Beier. 2007. CorridorDesigner: ArcGIS tools for designing and evaluating corridors. Online at: http://corridordesign.org. Accessed January 2013.
McKechnie, S.W., P.R. Ehrlich, and R.R. White. 1975. Population genetics of Euphydryas butterflies. I. Genetic variation and the neutrality hypothesis. Genetics 81: 571-594.
McRae, B.H. and D.M. Kavanagh. 2011. Linkage Mapper Connectivity Analysis Software. The Nature Conservancy, Seattle, Washington. Online at: http://www.circuitscape.org/linkagemapper. Accessed January 2013.
Murphy, D.D. and P.R. Ehrlich. 1980. Two California checkerspot butterfly subspecies; one on the verge of extinction. Journal of Lepidopterists’ Society 34: 316-320.
Stanford University. 2006. Jasper Ridge Biological Preserve: annual report 2005-2006. Unpublished report. pp. 36.
U.S. Fish and Wildlife Service. 1998. Recovery Plan for serpentine soil species of the San Francisco Bay Area. Portland, OR. 330 pp.
U.S. Fish and Wildlife Service. 2008. Endangered and Threatened Wildlife and Plants: Designation of Critical Habitat for the Bay Checkerspot Butterfly (Euphydryas editha bayensis), Final Rule. Federal Register 73: 50406-50452. August 26, 2008.
U.S. Fish and Wildlife Service. 2009. Bay Checkerspot Butterfly (Euphydryas editha bayensis) 5-Year Review: Summary and Evaluation. Sacramento Fish and Wildlife Office, Sacramento, California. August.
Weiss, S.B. 1999. Cars, cows, and checkerspot butterflies: Nitrogen deposition and management of nutrient-poor grasslands for a threatened species. Conservation Biology 13: 1476-1486.
Weiss, S.B., D.D. Murphy, and R.R. White. 1988. Sun, slope, and butterflies: Topographic determinants of habitat quality for Euphydryas editha. Biological Conservation 46: 183-200.
Zack, J.A. and R.A. Minnich. 1991. Integration of geographic information systems with a diagnostic wind field model for fire management. Forest Science 37: 560-573.
Spring 2016 Volume 9 Issue 1