Evaluating the Suitability of Roadway Corridors for Use by Monarch Butterflies (2020)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

20 Introduction Roadside ROWs represent potential monarch habitat, but not all areas are likely to be equal in terms of potential suitability for developing or maintaining monarch habitat. Determining which roadsides are best for monarch conservation is a priority when managers are posed with options about where to focus limited resources for higher quality pollinator seed mixes or where to adopt alternative management strategies designed to enhance monarch habitat. The research team’s objective was to develop a standardized roadside prioritization tool that helps state DOTs determine where to allocate effort toward monarch conservation. The main purpose of the model is to provide DOT managers with a roadside monarch habi- tat suitability index to identify roadsides with the greatest potential to contribute monarch habitat that they can use to compare and prioritize roadsides within a given area. Additionally, they may be able to locate areas where roadside habitat may complement high-functioning surrounding habitats or, conversely, to identify places where roadside habitat could be devel- oped as a habitat corridor in areas where the landscape is providing little monarch habitat. If managers are interested in surveying existing habitat in their roadside ROWs, they could use the tool to select sample locations across the spectrum of roadside suitability index values. Finally, if the tool were used in conjunction with manual surveys of ROWs with the RA tool, it could help target roads with high suitability index values but low actual habitat quality scores for enhancement through altered management. The Landscape Prioritization Model is a tool that creates a suitability index using Esri ArcGIS software, widely used within the transportation management community. The tool integrates two sources of data: nationally available standard land cover data that are translated into mon- arch habitat and potential pesticide risk and roads data that are used to evaluate the roadside characteristics. Figure 7 illustrates the conceptual structure of the suitability index. Habitat for monarchs is determined by the potential for a roadside to provide milkweed and nectar plants, augment adjacent habitat, and increase connectivity in the landscape. Risks are determined by the potential exposure to chemicals from traffic, adjacent land use, and risk of collision by cars. Here, each of those components is described and the model integrates them to create the overall suitability index processes by which are discussed. Habitat Modelling Monarchs require two types of resources—one for breeding and one for foraging—during their annual life cycle. Specifically, monarchs require milkweed plants for reproduction and nectar sources for foraging. The model uses land cover data and translates different cover types into milkweed density and nectar quality. These two metrics are combined, in addition to C H A P T E R 3 Product A: Landscape Prioritization Model for Roadside Habitat for Monarchs

Product A: Landscape Prioritization Model for Roadside Habitat for Monarchs 21 potential exposure to pesticides, to determine the overall quality of habitat for monarchs. While the goal is to develop a tool that can be used across the country, the research team could find no single land cover layer that could be used across the full range of the monarch. Instead the team developed separate approaches for three different ecoregions of the monarch (Figure 8): the north core, the south core, and the western region. Milkweed Suitability The ecoregions differ in how estimates for milkweed density are generated, but estimates for nectar and pesticide exposure are the same across the ecoregions. In predicting milkweed stem density in the north core, the team used the U.S. Department of Agriculture (USDA) Cropland Data Layer (CDL) and, following work by Thogmartin et al. (2017b), reclassified each land cover Figure 7. Logical structure and components of the roadside suitability index for monarchs. Figure 8. Monarch bioregions in the habitat model, derived from Anderson et al. 2017 and Dilts et al. 2018.

22 Evaluating the Suitability of Roadway Corridors for Use by Monarch Butterflies into an estimated milkweed stem density (Table 2 found in Thogmartin et al. 2017b). Thogmartin et al. (2017b) used expert judgment to estimate how many milkweed stems would be expected to be found on different land cover types within the north core. The CDL classification for land cover works well in areas with mainly agricultural lands but not as well in areas dominated by pasture lands as the classification scheme does not differentiate pasture from grasslands or adequately capture variation in grasslands, e.g., the habitat within the south core. The research team is working to identify more appropriate land cover and then a reclassification scheme so that the same approach can be applied to the south core. For the north core, modelled milkweed stem density was converted to a milkweed quality map using a formula derived from Kasten et al. 2016 and Anderson et al. 2017 (Eq. 1, where Mi = milkweed quality on site i and mij = milkweed stem density per acre for site i from cover type j): 1 2 1 exp 6 2,000 (1)M mi ij = − + ×   The equation was based on the relationship between the per acre density of native milkweed stem and observed immature monarchs (eggs and larvae) on over 1,000 randomly selected, 50-meter swaths of habitat along roadsides in Illinois, Iowa, Minnesota, South Dakota, and Wisconsin (Kasten et al. 2016, Figure 9). For the western region, researchers use a separate milkweed habitat suitability map devel- oped by Dilts et al. (2018) that uses observed milkweed data input into a habitat niche model. Their niche model replaces the expert judgment reclassification approach used in the north core model. Nectar Availability The research team also used the USDA CDL to predict nectar resources by reclassifying each land cover into an estimated nectar suitability, following work by Koh et al. (2016; Table 2). Koh et al. (2016) used expert judgment to estimate the suitability of floral resources for bees on different land cover types across the conterminous United States using the CDL. When monarchs are stopping to forage, rather than travel for migration, they have been found to Figure 9. A representation of milkweed quality for breeding in relation to milkweed stem density (based on Kasten et al. 2016).

Product A: Landscape Prioritization Model for Roadside Habitat for Monarchs 23 travel up to about 240 meters (Zalucki et al. 2016). The research team estimated nectar avail- ability by calculating the average nectar suitability index within a 240-meter foraging radius. The suitability of nectar and milkweed was integrated to develop an overall habitat quality map, such that potential habitat quality on site i, qi is: (2)q M Ni i i= × where M is milkweed quality and N is nectar availability. Pesticide Exposure (Ei) Pesticides, defined here as any chemicals added to control weeds, pests, or fungi on agri- cultural lands, can potentially reduce nearby monarch habitat quality through unintended movement of applied chemicals. Based on the USDA Natural Resources Conservation Service’s Monarch Butterfly Wildlife Habitat Evaluation Guide (USDA 2018), any habitat within 100 feet (∼30 meters) of a pesticide source has reduced habitat potential. While the team knows that pesticide application and unintended drift is highly variable across crops and conditions during application, the team wanted to minimize the risk of exposure in these areas. Thus, the team predicted the potential source of pesticides as any cover type that is agricultural (Table 2). It was assumed that the potential exposure and negative impact on habitat declined with distance. As the CDL raster is at 30 meters resolution, the researchers used a conservative estimate of 60 meters (two CDL cells) to represent potential range of drift exposure. It was assumed that, if the habitat was part of an agricultural land such that the raster cell is expected to have some milkweed and nectar sources, then exposure on site i, Ei, is 1; if monarch habitat was 30 meters away, i.e., a neighboring raster cell, then the exposure is 0.6, and if habitat was 60 meters away, then exposure is 0.2 and 0 otherwise. A Euclidean distance function was used to calculate drift distance to account for distance along diagonal cells. Overall Habitat Quality (Hi) Potential landscape habitat and exposure were combined to create an overall habitat quality index, and it was assumed that as potential exposure to pesticides on site i, Ei, increases, the overall quality of a site should decrease. However, the team also assumed that pesticide drift does not fully remove habitat such that monarch habitat quality on site i, Hi, is: 1 2 (3)H q E i i i= × −   This function indicates that perfect monarch habitat would receive a score of 1 and terrible habitat would receive a score of 0. Core Habitat The overall habitat quality index provides a potential suitability index for monarchs but does not indicate the type of habitat that would support a thriving monarch population. Thus the research team wanted to provide a prediction of where potential “core” patches might be, patches that would represent areas of breeding monarchs. The team took a conservative approach by assuming that core monarch areas are places that have suitability index score, Hi, of 0.95 or higher. The decision-support tool generates core monarch habitat polygons and determines the distance to the nearest core patch. If important monarch areas are known, it would allow a DOT manager to load their own patches.

24 Evaluating the Suitability of Roadway Corridors for Use by Monarch Butterflies Roadside Potential The second component of the Roadside Monarch Habitat model is the roadside character- istics. The goal of this component is to apply conceptual logic about roads that maximize the potential benefits of creating habitat along roadsides while minimizing risks to monarchs. Benefits The team identified three types of benefits: 1) direct habitat along a roadside, 2) the potential to augment habitat next to the road, and 3) the potential to connect to core habitat patches in the landscape. Potential Roadside Habitat (Ai) Here, the research team made the simple assumption that as the width of roadside ROWs (edge of pavement to the edge of DOT-managed area) increases, the potential habitat area increases. The research team is not aware of studies that precisely examine how ROW width affects monarch habitat use, so instead applied the goal to minimize the risk associated with roadside habitat and the assumption that wider ROWs have the potential to provide habitat farther from traffic. The team sought to capture the expected range of potential widths by using Google Earth to estimate the width of ROWs. With this approach, a relationship was developed between roadside width of site i, wi, and the potential habitat area along road i, Ai (Equation 4, Figure 10): = − +    1 2 1 exp 40 (4)A wi i Augmenting Adjacent Habitat (Qi) Roadside habitat has the potential to increase existing adjacent habitat, and the team assumed that, as the quality of the habitat increased, the greater the benefit of the roadside habitat. The team represented the quality of adjacent habitat around roadside i, Qi, by calculating the Figure 10. A representation of potential ROW habitat quality in relation to ROW width.

Product A: Landscape Prioritization Model for Roadside Habitat for Monarchs 25 average beneficial habitat within a radius of 120 meters, a distance derived from the single maxi- mum step distance from Zalucki et al.’s (2016) paper on monarch movement. Roadside Connectivity (Di) The research team suggested that roadsides could provide additional supporting monarch habitat if they were closer to core habitat patches, such that the value of creating or maintain- ing monarch habitat would decrease as the distance to core habitat patches increases. Monarch movement studies by Zalucki et al. (2016) and Grant et al. (2018) suggest benefits would become negligible beyond about nine miles (15 km). Using this logic, researchers created a distance decay function that is scaled from 0 to 1 and represents the benefit of connecting core patches from roadside habitat i to a core patch, Di (Equation 5, Figure 11): = + ( )− 1.03 1 0.9994 , (5) 3.4 Di dij where dij is the distance in miles between roadside habitat i and the nearest core patch j (Figure 11). Risks The researchers identified two types of risks of roadsides—collision with vehicles and expo- sure to chemicals associated with roads. Vehicle Collision risk (Vi ) High-quality habitat along roads can be beneficial, but they could also bring monarchs close to vehicles and lead to mortality from collisions with cars. The researchers assumed that, as traffic volume increases, the risk of being hit also increases and that this risk is mediated by the speed limit of the road. McKenna et al. (2001) found that roadside mortality was higher with inter- mediate speed limits (between 15 and 55 mph). At slow speeds, monarchs can avoid cars, and, at higher speeds, the aerodynamics appear to push the monarch out of the way. The researchers represent this logic by creating two relationships with traffic volume along roadside i, ti, one Figure 11. A model of a Roadside Connectivity Index relative to miles to nearest core habitat area.

26 Evaluating the Suitability of Roadway Corridors for Use by Monarch Butterflies for intermediate speeds and one with slow or fast speeds. The team also assumed that collision risk along roadside i, Vi, increases with traffic volume (McKenna et al. 2001, Skorka et al. 2013, Soluk et al. 2011, Martin et al. 2018) up to an annual average daily traffic (AADT) of 26,000 where risk is maximized (Equation 6, Figure 12): 26,000 , 26,000 , 26,000 (6) where: 1, 15 55 0.8, V t t t l else i i i i i = θ ∗ < θ ≥     θ = < <   where li represents the speed limit (mph) along roadside i. Chemical Exposure from Cars (Ci) Emissions from cars have chemicals that could be toxic to insects (Bukowiecki et al. 2010, Carrero et al. 2013, Wang et al. 2013), so the research team included an additional risk to monarchs as a function of traffic volume. In addition, runoff from car wear-and-tear and road salt application creates additional chemical risks along high traffic roads (Snell-Rood et al. 2014, Lagerwerff and Specht 1970, Jaradat and Momani 1999). The team assumed that, as traffic volume increases, the potential exposure to chemicals that could be detrimental to monarchs foraging along roadside i, Ci, also increases: max , (7)C t t i i i )( = where ti is the average annual daily traffic along roadside i, and the maximum value references the maximum traffic volume for the area of analysis, likely a state. This means the index will be scaled from 0 to 1, where 0 would be no traffic and 1 would be along the road with the highest traffic volume in the area of analysis. Figure 12. A model of vehicle collision risk relative to traffic volume.

Product A: Landscape Prioritization Model for Roadside Habitat for Monarchs 27 Roadside Suitability Overall Overall Roadside Suitability (Ri) The research team determined the overall Roadside Suitability Index by taking the average of the components of the habitat and roadside potential metrics separately and then weight- ing the two major factors (habitat and roadside potential) to calculate on overall weighted average. The team’s experts felt that the habitat factors were more important for the overall weighting than the roadside risks such that the overall roadside habitat suitability index for monarchs at road segment i, Ri, is: 3 2 3 2 1 3 (8)R Q D A C V i i i i i i= + +    × − +    × where the first term represents the three components of the habitat quality metrics and the second term represents the two components of roadside potential. Data Sources The U.S. Geological Survey’s National Transportation Dataset (NTD) was used as the basis for all roadside analyses as it provides a standard source across the country. The National Map Feature Road Class (TNMFRC) attribute field in the NTD was used to predict ROW width, traffic volume, and speed limit (Table 2). The team assumed that ROW width and speed limits are likely to be predicted by road type but also recognized that they are likely to vary state to state (AASHTO 2001). The team made the general assumption that traffic volume and roadside width increase with increasing road type, i.e., county roads have lower traffic volume and smaller roadsides than interstate highways (Table 2). The research team recognizes that there is likely substantial variation in the observed data (Setton et al. 2005), however, so these are meant to be exploratory assessments. The tool allows users to override the basic reclassification of NTD road type by providing optional inputs if state-specific data are available. The NTD values remained the default coverage, replaced by the user-inputted roads only for overlapping coverage. Discussion The national scale Landscape Prioritization Model is a tool that was designed to give trans- portation managers insight into which roadsides within a given area have the greatest potential to contribute to monarch habitat, whether through connectivity to other core areas or because TNMFRC Generalized Road Characteristics (Road Classification) Right-of-way Width (ft) Traffic Volume (AADT) Speed Limit (mph) Controlled-access Highway 400 26,000 75 Secondary Highway or Major Connecting Road 100 15,000 55 Local Connecting Road 75 5,000 45 Local Road 40 2,600 25 Ramp 50 15,000 45 Four-Wheel Drive 20 1,000 45 Ferry Route; Tunnel; Unknown NA NA NA Table 2. Reclassification of NTD’s TNMFRC field into ROW width, traffic volume, and speed limit.

28 Evaluating the Suitability of Roadway Corridors for Use by Monarch Butterflies they are more suitable from a risk mitigation standpoint. Built with Esri ArcGIS software to foster usability, the tool combines national-level datasets, published literature, and expert opin- ion into a model that estimates monarch habitat quality across the entire landscape and relates those estimates to individual roadsides. The final output, a map of potential roadside suitability for monarch habitat, can be used by departments of transportation to optimally allocate habitat maintenance and restoration funds with the study area or as the basis for more intensive roadside assessments such as those detailed later in this report. The Landscape Prioritization of Roadside Suitability tool provides information about the context in which roadside habitat exists. Further exploration about field-level habitat quality values (such as derived from the Habitat Calcula- tor) and monarch use relate to the landscape factors depicted in the model. Further research into the risks from roads (e.g., chemicals and collisions) could be structured using a sampling design from the Landscape Prioritization tool. The model provides road managers supplemental information to make decisions at scales ranging from individual counties to entire states.