Influence of habitat quality and resource density on breeding-season female monarch butterfly (Danaus plexippus) movement and space use in north-central USA agroecosystem landscapes

Field Sites Three field sites were selected that contained areas with varying densities of milkweed ramets and blooming inflorescences in different habitat classes (see Table 1 for a glossary of key terms). The selected sites did not have structural boundaries (e.g., forest edges, see Schultz et al. 2012). A detailed description of each site is provided in S1-S4. The mosaic site was a 64-ha area in Floyd County, Iowa that contained a restored prairie (HD), grass-dominated fields (LD), and crop fields (ZD) (Fig 1A). The roadsides were two north-south, 1,000 m by 350 m linear sections (~ 35 ha) of grass-dominated roadside (LD) along secondary gravel roads in Story County, Iowa, with different proximity to prairie habitat. One roadside was bordered by maize to the west (ZD) and a remnant prairie (HD) to the east (RHD; Fig 1B, green box). The other roadside was bordered by maize and soybean fields to the east and west (ZD) and was ≥ 500 m from prairie habitat (RZD; Fig 1B, blue box). Observations occurred from June to August at the mosaic field site in 2019 and the roadside field sites in 2020. Tracking Field-Captured Female Monarch Butterflies Field-captured, female monarchs with minimally worn wings were held for 1-4 days and maintained as previously reported (Fisher et al. 2020, Fisher and Bradbury 2021). Monarch movement was observed on days with a temperature of 18.6-35.4°C and wind speed below 15.0 kph; wind direction was recorded prior to the release of each monarch. Monarchs were anesthetized with CO2 and radio-tagged with a 220 mg LB-2X transmitter (~ half the average weight of a monarch; Holohil Systems Ltd, Ontario, Canada; see Fisher et al. 2020). Untagged individuals served as controls. Monarchs were released on Bromus inermis or Zea mays within a habitat class (Table 1). Those that did not move from their release location (S5) within 15 minutes were considered unresponsive, and the observation period was terminated (see Zalucki and Kitching, 1984). There was no significant difference in responsiveness between tagged and untagged individuals (S6-S8), consistent with previous studies that documented no impact on flight ability (Fisher et al. 2020, Fisher and Bradbury 2021). Consequently, observations of responsive tagged and untagged monarchs were combined for data analyses, resulting in 18-20 responsive individuals for each habitat class (Table 1). Monarchs were observed visually by technicians stationed within a field site (yellow, green, and blue boxes in Fig 1). Locations between natural flight steps were georeferenced (Trimble Geo7x; Sunnyvale, CA). The amounts of time and behaviors performed (i.e., foraging, ovipositing, resting, mating) at each location and the amount of time spent flying between locations were recorded. In cases where monarchs flew beyond the limit of visual detection (ca. 50 m), monarchs were relocated with the aid of handheld radio telemetry equipment (directional 3-element Yagi antennae [164.000–166.000 MHz; Johnson's Telemetry, El Dorado Springs, MO], connected to VHF receivers [Alinco DJ-X11; Toyama, Japan] with 0.9-m coax cables). We also employed an automated radio telemetry system (Fisher et al. 2021) at the mosaic site; however, location estimates from the automated system were unsuccessful (S9). Monarchs that flew within the initial 15 min, but then remained stationary for 30 min, were recaptured. For the remaining responsive monarchs, capture was attempted after an observation period of 3,600-sec (one hour) within a field site. When a monarch was relocated and captured with the aid of handheld radio telemetry outside of a field site, the elapsed time from release to recapture was recorded. Unsuccessful relocation efforts were terminated three hours post-release. Consequently, observation periods ranged from 75 – 7,500 s (1.25 – 125 min). Measurements Space use estimates with continuous-time movement models (ctmm) were created for each monarch using the ctmm package (Calabrese et al. 2016, Fleming et al. 2016, Fleming and Calabrese 2020) in R (R Core Team 2020) with the Ornstein-Uhlenbeck Foraging (OUF) process (see S10 for methodological details). These models produced 0.05 m2 raster cells with occurrence probabilities ranging from 0-1. Straight-line measurements were used to analyze movement paths, assuming direct movement between georeferenced locations where monarchs were observed between flights. Step lengths between two consecutive georeferenced locations, bearing orientation of steps, and turn angles created from three consecutive georeferenced locations within a habitat class were calculated using the 'movement. pathmetrics' package in Geospatial Modeling Environment (GME; Spatial Ecology LLC, www.spatialecology.com/gme/) in combination with ArcMap 10.3. Total distances traveled were calculated by summing step lengths for each monarch. The Euclidian distances and bearings of Euclidian movement were calculated from the release locations to the location at the end of the observation period using ArcMap 10.3. Statistical Analyses All statistical analyses associated with movement characteristics were conducted in R (R Core Team 2020). The total and Euclidian distance measures for each individual were compared with a paired t-test. Distances traveled, step lengths within and between habitat classes, logit transformed proportions (Warton and Hui 2011) of raster cells with non-zero occurrence probability, as well as the time and georeferenced locations in each habitat class, were analyzed by field site and release habitat class with generalized linear models in the package 'emmeans' (Lenth et al. 2020). The Rayleigh test of uniformity, the Watson-Williams test for homogeneity of means, and the Watson-Wheeler test for homogeneity of angles with the 'circular' package (Lund 2017) were used to analyze orientations of movement relative to wind direction and distributions of turn angles within habitat classes. The percent of steps ≥ 50 m that originated in each habitat class were analyzed with multiple chi-square tests of independence and a Bonferroni corrected significance threshold (see Sedgwick 2012 for multiple comparisons).