A, WL 0r Y Y Volume 62 Number 1 6 May 2008 ISSN 0024-0966 Journal of the Lepidopterists' Society Published quarterly by The Lepidopterists' Society THE LEPIDOPTERISTS’ SOCIETY Executive Council John H. Acorn, President John Lill, Vice President William E. Conner, Immediate Past President David D. Lavvrie, Secretary Andre V.L. Freitas, Vice President Kelly M. Richers, Ti ■easurer Akito Kawahara, Vice President Members at large: Richard A. Anderson Michelle DaCosta John V. Calhoun John H. Masters Amanda Roe Michael G. Pogue Editorial, Board John W. Brown (Chair) Michael E. Toliver ( Member , at large) Brian Scholtens ( Journal ) Lawrence F. Gall ( Memoirs ) Dale Clark ( Neivs ) John A. Snyder ( Website ) H onorary Life Members of the Society Charles L. Remington (1966), E. G. Munroe (1973), Ian F, B. Common (1987), Lincoln P. Brower (1990), Frederick H. Rindge (1997), Ronald W. Hodges (2004) The object of The Lepidopterists’ Society, which was formed in May 1947 and formally constituted in December 1950, is "to pro- mote the science of lepidopterology in all its branches, ... to issue a periodical and other publications on Lepidoptera, to facilitate the exchange of specimens and ideas by both the professional worker and the amateur in the field; to secure cooperation in all mea- sures" directed towards these aims. Membership in the Society is open to all persons interested in the study of Lepidoptera. All members receive the Journal and the News of The Lepidopterists’ Society. Prospective members should send to the Assistant Treasurer full dues for the current year, to- gether with their full name, address, and special lepidopterological interests. In alternate years a list of members of the Society is is- sued, with addresses and special interests. Active members — annual dues $45.00 within the U.S., $50.00 outside the U.S. Affiliated members — annual dues $10.00 within the U.S., $15.00 outside the U.S. Student members — annual dues $20.00 within the U.S., $25.00 outside the U.S. Sustaining members — annual dues $60.00 within the U.S., $65.00 outside the U.S. Life members — single sum $1,800.00 Institutional subscriptions — annual $60.00, $65.00 outside the U.S. Airmail postage for the News $15.00, $30.00 outside North America (U.S., Canada, Mexico) Send remittances, payable to The Lepidopterists’ Society, to: Kelly M. Richers, Treasurer, 9417 Carvalho Court, Bakersfield, CA 93311; and address changes to: Julian P. Donahue, Natural History Museum, 900 Exposition Blvd., Los Angeles, CA 90007-4057. For information about the Society, contact: Ernest H. Williams, Department of Biology, Hamilton College, Clinton, NY 13323. To order back issues of the Memoirs, write for availability and prices to Kenneth R. Bliss, 28 DuPont Avenue, Piscataway, NJ 08854. Kim Garwood Kenn Kaufman Harry Zirlin The additional cost for members outside the U.S. is to cover mailing costs. Journal of The Lepidopterists' Society (ISSN 0024-0966) is published quarterly by The Lepidopterists’ Society, c/o Los Angeles County Museum of Natural History, 900 Exposition Blvd., Los Angeles, CA 90007-4057. Periodicals postage paid at Los Angeles, CA and at additional mailing offices. POSTMASTER: Send address changes to The Lepidopterists’ Society, % Natural History Museum, 900 Exposition Blvd., Los Angeles, CA 90007-4057. Cover Illustration: Miracavira brillians (Barnes) (Noctuidae) larvae showing their striking coloration and resting posture. Photo Credit: David Wagner, University of Connecticut, email: david.wagner@uconn.edu. Journal of The Lepidopterists’ S OCIETY Volume 62 2008 Number 1 Journal of the Lepidopterists’ Society 62(1), 2008, 1-17 A CHARACTERIZATION OF NON-RIOTIC ENVIRONMENTAL FEATURES OF PRAIRIES HOSTING THE DAKOTA SKIPPER ( HESPERIA DACOTAE, HESPERIIDAE) ACROSS ITS REMAINING U.S. RANGE Ronald A. Royer Division of Science, Minot State University, 500 University Avenue West, Minot, ND 58707, USA; email: ron.royer@minotstateu.edu Rose A. McKenney Department of Geosciences and Environmental Studies Program, Pacific Lutheran University, Tacoma WA 98447, USA AND Wesley E. Newton U.S. Geological Survey, Biological Resources Division, Northern Prairie Wildlife Research Center, 8711 37th Street SE, Jamestown, ND 58401, USA ABSTRACT. Within the United States, the Dakota Skipper now occurs only in Minnesota, North Dakota, and South Dakota. In these states it has been associated with margins of glacial lakes and calcareous mesic prairies that host warm-season native grasses. Preliminary geographic information system (GIS) analysis in North Dakota has indicated a close congruency between historic distribution of the Dakota Skipper and that of specific near-shore glacial lake features and related soil associations. This study analyzed humidity-related non-biotic microhabitat characteristics within three remaining occupied Dakota Skipper sites in each state during the larval growth period in 2000. Measured parameters included topographic relief, soil compaction, soil pH, moisture, and temperature at various depths, soil bulk density, soil texture, and temperature and humidity within the larval nest zone. Results of these efforts reveal two distinctive habitat substrates, one of relatively low surface relief with dense but relatively less compact soils, and another of relatively high relief with less dense but more compact soils. In the low-relief habitat, grazing appears to com- pact soils unfavorably in otherwise similar prairies in the more xeric western portion of the range, potentially by affecting ground- water buffering of larval nest zone humidity. Additional key words: Dakota Skipper, habitat, climate, soils, management. Numerous survey efforts have clearly defined present limits of distribution for the Dakota Skipper ( Hesperia clacotae Skinner, 1911) (McCabe 1979, 1981; Dana 1991; Royer 1988a, 1988b, 2003; Royer & Marrone 1992; Orwig 1995, 1996; Schlieht 1997; Royer & Royer 1997, 1998; Skadsen 1997, 1999, 2000). Some recent work also has characterized this species' habitat floristically at selected sites (Dana 1997, ND Parks and Recreation Department 1999). However, there has been no systematic attempt to define physiographic or other non-biotic features of habitat across the species' entire U. S. range. A primary intention of this project was to identify and characterize non-biotic features that might help habitat managers better understand and more easily recognize favorable sites in areas where the species remains, has recently suffered decline, or is believed historically to have occurred but is now absent. The original range of the Dakota Skipper is believed to have extended from Illinois northwestward as far as southeastern Saskatchewan (Royer 2003) and Manitoba (Klassen et al. 1989). It is known to occur within the U.S. now only in the states of Minnesota, North Dakota, and South Dakota, and a few populations still exist in Canada. The U. S. range originally included Iowa and Illinois, in both of which the species is now believed to have been extirpated (Scott 1986, T. Orwig, Morningside College, pers. comm.). In parts of this range the Dakota Skipper has been specifically associated with the margins of glacial lakes (McCabe & Post 1977, McCabe 1981). Many workers have also associated it with calcareous mesic prairies (McCabe 1981), such indicator plants as smooth camas ( Zygadenus elegans Pursh., Liliaceae) (Royer & Marrone 1992), and warm-season native grasses (ND Parks and Recreation Department 1999). Recently a very close relation has been noted in McHenry County, North Dakota, between recorded distribution of the Dakota Skipper, glacially related o Journal of the Lepidopterists’ Society surface geology, and soil associations defined by the United States Department of Agriculture (USDA) (Royer & Royer 1998, Lord 1988, see Fig. 1). Subsequent preliminary GIS analysis has suggested a statewide congruency of known distribution of the Dakota Skipper with these soil associations (Tom Sklebar, retired USGS NPWRC, pers. comm.). McCabe (1981) proposed that precipitation/evaporation ratios may be an important defining feature of this species' habitat requirements. Presence of "hydrofuge glands" on larval segments 7 and 8 (McCabe 1981) suggests a historic or present need of the species for protection from inundation. This led to our hypothesis that factors limiting Dakota Skipper populations may have more to do with such non-biotic habitat elements as temperature and local humidity during sensitive larval and pupal stages than with such biotic factors as host plant or nectar source availability or predation during the adult flight period, when this species has been most extensively studied. Specifically, we hypothesized that such edaphic features as soil moisture, soil compaction, and soil bulk density, as well as related non-biotic factors such as temperature and relative humidity at and near (within 2.0 cm of) the soil surface, where several authors have noted that early stages abide in a silken nest during most of the summer (cf. McCabe 1981, Dana 1991), may be significant factors in larval survival potential. Microtopography substantially affects soil evaporation rates in the north-central United States (Cooper 1960). Soil compaction and vegetation removal (whether by herbivory, hay mowing, or fire) substantially alter soil water movement and evaporation, thereby altering near-surface humidity (Frede 1985, Miller & Gardiner 1998, Hausenbuiller 1985). Livestock grazing has been shown to increase bulk density (Zhao et al. 2007) and soil compaction (Greenwood et al. 1997), which are correlated with decreased soil water content and hydraulic conductivity (Zhao et al. 2007). In summer months these changes are likely to restrict the movement of shallow groundwater to the soil surface, thus preventing groundwater buffering of surface humidity conditions. Water loss from moist soils in contact with diy air occurs rapidly, usually exceeding the rate of upward movement of water through the soil (Hausenbuiller 1985). As a result a dry soil layer forms, inhibiting further evaporation. Formation of a diy soil layer would decrease surface humidity at precisely those times later in the summer when young larvae of the Dakota Skipper are most vulnerable to desiccation. The principal objectives in this study therefore were (1) to characterize non-biotic features related to hydrology and microclimate (microtopography, soil compaction, soil pH, soil moisture, soil temperature, soil bulk density, soil texture, near-surface humidity) and the variability of those features within and across occupied sites in the context of average summer climate conditions generally, and also (2) to compare those features between grazed and hay-mowed sites within the more xerie portion of the range in North Dakota. Study Area and Methods Western Minnesota, eastern North Dakota and eastern South Dakota were shaped by Laurentide ice sheets. This shaping profoundly affected the landforms and materials found at the surface in these areas. The Des Moines lobe cut across Minnesota and the eastern margin of South Dakota (South Dakota Geological Survey 1965). Slightly to the west, the James lobe cut through North Dakota and eastern South Dakota. These lobes deposited extensive moraines that contained unsorted clay to boulder sized material (Agnew et al. 1962, Hobbs and Goebel 1982). During the last glacial retreat, many areas were submerged under melt water lakes (South Dakota Geological Survey 1965, Hobbs and Goebel 1982, Lord 1988). Thus our study area contains relatively level areas with sorted sediment typical of lake bottom and near shore deposits, as well as rolling hills composed of poorly sorted sediment typical of glacial moraine deposits. Original Dakota Skipper habitat across the region ranged from tail-grass to mixed-grass native prairie. Much of the remaining habitat is now privately owned and managed either as hay meadow or pasture. Within this context, we specifically sought sites that were under public ownership or at which conservation is a management goal. Climatically, the study area crosses a transition zone from humid, middle latitude with severe winter type in western Minnesota to mid-latitude steppe in central North and South Dakota (Ackerman 1941). This transition can be seen in summer average monthly temperatures and precipitation for the period of record (1895-2003) and the data collection year (2000, Table 1). South Dakota has average monthly temperatures that exceed Minnesota and North Dakota average monthly temperatures by 1-2°C. Minnesota’s average monthly precipitation exceeds North and South Dakota average monthly precipitation by 20-50mm. Despite these differences in statewide values, temperature patterns are similar at climate stations near the study sites. Precipitation, however, is far more variable throughout the summer season and the region. State averages show that monthly precipitation declines from June through September, and that Minnesota has the largest average precipitation for each month of the Volume 62, Number ] 3 Fig. 1. Superimposition on a surface geology map (Lord 1988) of recently confirmed occurrence sites for Hesperia dacotae in McHenry County, North Dakota (dots) indicating close congruency with distribution of windblown soil units (#3 and #4) in the near-shore environment of glacial Lake Souris. Unit #3 was described by Lord as "silt and sand, fine to medium grained, moderately to well sorted. ... gradational to unit 4.” Unit #4 was described as “Sand, fine to medium grained, well sorted. . . (with dunes) as high as 5 metres.” Both of these were characterized as having been reworked from unit 17, “nearshore lake sediment ... up to 30 metres thick.” The green line represents putative glacial lake margin, and the background map grid indicates square miles. (After Royer and Royer 1998.) 4 Journal of the Lepidopterists’ Society Table 1. Monthly mean temperatures (°C) and precipitation (mm) during summer months for Minnesota, North Dakota, and South Dakota (data from National Climatic Data Center, 2004). Average Temperature (°C) Average Rainfall ( mm) 1895-2003 2000 1895-2003 2000 Minnesota June 17.8 16.6 163.4 192,5 July 20.6 20.4 142.5 150 August 19.4 19.8 136.2 135.8 September 14.1 13.7 113 65.7 North Dakota June 17.1 15.7 136.2 129.9 July 20.4 20.3 103.1 74.4 August 19.2 20.1 82.7 102.8 September 13.4 13.7 64.2 55.9 South Dakota June 18.8 18.9 130.3 107.9 July 22.6 22.8 94.5 99.2 August 21.5 22.8 82.7 58.3 September 16.7 15.8 63.8 25.2 summer (Table 1). Climate stations near the study sites show that in addition to having greater average summer rainfall, the Minnesota site experiences its peak precipitation later in the summer than the North Dakota site and the South Dakota site. In 2000, however, average precipitation patterns were not experienced. North Dakota experienced higher precipitation during August than July in 2000 and South Dakota had much less than average precipitation during both August and September. Because of this variability in precipitation, onsite recording of humidity was deemed necessary. Field sites. Field sites selected for this study all had an extensive history of involvement in earlier work on the Dakota Skipper (McCabe 1979, 1981; Royer 1988a, 1988b; Royer & Marrone 1992; Royer & Royer 1997, 1998; Dana 1991, 1997; Skadsen 1997, 1999). Involving three states, these sites spanned the known remaining U. S. range of the Dakota Skipper (Table 2, Fig. 2). Sampling methods. We first developed a three-state map depicting all known U. S. populations of the Dakota skipper as points (Fig. 2). We then both sampled and monitored habitats at three specific sites in each state that were known to be hosting viable Dakota Skipper populations. (We here use the term “sample” to denote data from a point in time and the term “monitor” to denote continuous data collection with HOBO® loggers.) Sampling was conducted to determine spatial variability within Dakota Skipper habitat; monitoring was conducted to determine temporal variability throughout the most vulnerable period of the larval growth season (eclosure to onset of winter diapause). At all study sites, sampling was conducted in four randomly oriented 50m by 40m gridded plots (Fig. 3), each centered on a monitoring point determined in the field by either (i) directly observing oviposition or (ii) using locations of documented skipper activity within the past three years (Royer & Royer 1997; Schlieht 1997; Skadsen 1997, 1999). Treating each plot as a rectilinear set of five parallel 50m transects, we took Fig. 2. Distribution of all known Dakota Skipper ( Hesperia chcotae ) records from the three states in which the species is known to persist. Site locations for this project are designated as crosses. Volume 62, Number 1 5 Table 2. Dakota Skipper ( Hesperia dacotae) study sites by state, county, ownership, approximate extent (ha), and general soil texture classification (TNC=The Nature Conservancy, DNR=Department of Natural Resources, WMA=Wildlife Management Area, USFWS=LT.S. Fish and Wildlife Service). State/Site County Ownership ha Texture" Minnesota Felton Prairie (FP) Clay County/TNC 200 L/SL Hole-in-the Mountain (HM) Lincoln DNR/TNC/Private 65 SL Prairie Coteau (PC) Pipestone TNC 25 SL North Dakota Mount Carmel Camp (MCC) McHenry ND State School 65 SL/LS Smokey Lake School Sect. (SLS) McHenry ND State School 65 SL Swearson School Sect. (SSS) McHenry ND State School 65 SL South Dakota Scarlet Fawn Prairie (SFP) Roberts Sioux Tribal 30 SL Knapp Pasture (KNP) Roberts Private 65 SL Cox Lake WMA (CXL) Hamlin USFWS 30 SL/LS a L=loam, SL=sandy loam, LS=loamy sand. Site Name . Plot (circle) A 6 C D Al B. 1 C1 D 1 <[> E 1 — A 2<> B2<> C2<> D.2<> E 2<> A 3 B.3<[> C.3 D 3 E.3 Center A.4<> B.4 <> C.4 <> D.4<> E.4<> A. 5 <> B.5 <> C.5 <> D.5 <> E 5<> A. 6 B.6 C.6 D.6 E 6 y* Base line set on random bearing Point designation | < 40 meters s> | 50 meters Fig. 3. Grid design for sampling within each plot. Center was determined by (a) observed oviposition or (b) reference to most recent confirmed adult skipper activity. Samples were taken for compaction and pH at all grid points and for other parameters generally at points Al, A3, A5, B2, B4, B6, Cl, C2, C5, D2, D4, D6, El, E3, and E5. Compass bearing for the grid axis (center transect line) was randomized for each sampling period. probe readings at alternate 10 meter intervals within all four grids in each site. At four points in each grid, soil samples were also taken during one sample period for determining soil texture and hulk density within that grid (a total of 16 samples per site). For possible future GIS reference, precise center-point UTM coordinates (NAD 27) were confirmed during each sampling period. At each of these gridded plots we recorded local surface relief (in meters), soil texture, soil bulk density, and pH; with moisture, temperature, and compaction each measured at three depths (20, 40, 60cm). We also quantified both temperature and humidity within the primary larval nest zone (estimated to be 0-2em above the soil surface). Data loggers were used to monitor surface humidity and larval nest zone temperature continuously, in half- hour intervals, at all study sites from time of oviposition (approximately 5 July) through estimated initiation of larval diapause (23 September). A HOBO® Temp/RH data recorder was placed at the center point of at least two plots at each study site. In North Dakota, data recorders were placed at all plot center points except in grazed habitat. At Minnesota and South Dakota sites, data loggers were placed at two of four plots for each site except Scarlet Fawn Prairie (South Dakota), where there was only one plot and hence only one data logger was needed. One data logger failed at the Prairie Coteau site in Minnesota, and loss of another necessitated reducing the total number of useful Minnesota data sets to four. The resulting array of monitoring devices provided a continuous record of 6 Journal of the Lepidopterists’ Society both spatial and temporal variability in larval nest zone temperature and humidity across the range of the Dakota Skipper in all three states. Sampling was conducted at approximately two-week intervals, from the beginning of the mating flight (ca. 1 July 2000) until the estimated beginning of larval diapause in the fall (the first significant frost in North Dakota sites was on 23 September 2000). Each site was subjected to at least four rounds of sample data collection. For the first sampling period at each site, all 30 grid points were sampled for all parameters. For subsequent temperature and moisture readings, half the points were sampled by alternating sample points as follows: Al,3,5; B2,4,6; Cl,3,5; D2,4,6; El,3,5. Soil samples for determining composition, texture, and bulk density were taken similarly at compass-randomized points B2, B4, D2 and D4 within each plot. For relief, we determined the minimum elevation for each plot within each site and then subtracted this minimum from each elevation within the plot to define the response variable “relief ,” scaled to the minimum elevation value within each plot. Instrumentation. Equipment included (a) for relief a total station with data logger, (b) for soil compaction a DICKEY-jolm® Soil Compaction Tester (indicating compaction pressure in lbs/in2), (c) for soil pH a Kelway® soil pH and moisture meter, (d) for soil moisture content both a Kelway® soil pH and moisture meter (surface moisture) and an Aquaterr® soil moisture, temperature, and salinity probe (moisture at various depths), (e) for soil temperature at various depths an Aquaterr® soil moisture, temperature, and salinity probe, and (f) for temperature, relative humidity, and absolute humidity within the larval nest zone a HOBO® RH data logger programmed to read continuously in 30-minute intervals. To determine soil bulk density samples of known volume were dried to a constant weight. To determine soil texture these same samples were subjected to settling and mechanical analysis in order to define percent sand, silt, and clay. Data were compiled by study site and stored in tabular form in Microsoft Excel®. All were archived electronically at the USGS Northern Prairie Wildlife Research Center in Jamestown, North Dakota. Data Analysis. To gain an understanding of how variation in the various non-biotic response variables might be partitioned and to take advantage of the completely nested design structure of the study (i.e., 40x50m plots nested within study sites, grid sampling points nested within plots, with repeat sampling considered nested within grid sampling points), we first conducted a variance components analysis using the variance components procedure (PROC VARCOMP) of SAS (1999). This allowed us to compute site-to-site, plot-to-plot, point-to-point, and sampling time-to- sampling time variance components (where applicable) and assess their relative contribution to the total variation for each non-biotic response variable. Variance components are useful descriptive summaries and have their greatest value in planning future studies (e.g., if there is more plot-to-plot variation relative to variation among points within plots for a particular response variable then sampling effort should focus on establishing more plots within sites with less effort focused on the number of grid sampling points within plots to fully characterize Dakota Skipper sites). We were also interested in isolating specific differences in the various non-biotic response variables among the nine study sites, and if applicable, how those differences might vary with soil depth (20, 40, and 60cm for soil compaction, temperature, and moisture only). To do so, we used analysis of variance (ANOVA) techniques using the mixed linear models procedure (PROC MIXED) of SAS (1999). For the ANOVAs, and as with the variance components analysis described above, we considered the 40x50m plots to be a random factor nested within study sites, with grid sampling points also as a random factor nested within plots. Repeat sampling effort, where applicable, was also considered as a random factor and nested within grid points. We compared not only mean responses among the nine study sites but also mean variances, where variances were calculated across the sampling grid points within each plot, and mean variances then computed by averaging across plots within sites. We examined these mean variances because variation in abiotic response variables may be as important as or more important than mean responses for characterizing Dakota Skipper habitat. For the responses soil compaction, soil temperature, and soil moisture measured at three depths the ANOVA design structure was considered to be a split-plot with depth being the sub-unit (Littell et al. 1996). All other ANOVAs were considered to be one-ways, and where applicable, with sub-sampling (Steel and Torrie 1980). For those response variables measured in the “larval nest zone” as described earlier, we did not conduct an ANOVA because of the small sample sizes for most of the sites, but we do report the mean responses for each plot and site, where the means are seasonal means. In North Dakota we also compared these characteristics at three known Dakota Skipper sites (two hay meadows and one grazed site with a similar plant community and topography) in order to assess possible differing effects of hay mowing and grazing on these features. All means reported, unless stated otherwise, are least squares Volume 62, Number ] 7 Table 3, Summary statistics for selected physical response variables (RV) measured at occupied Dakota Skipper ( Hesperia dacotae) study sites in Minnesota, North Dakota, and South Dakota (see Table 2 for study site descriptions and abbreviations). RVa Minnesota North Dakota South Dakota Metric1' FP HM PC MCC SLS SSS SFP KNP CXL Relief Mean 0.38 3.99 4.36 0.38 0.37 0.45 2.02 3.16 3 SD 0.21 2.2 2.37 0.34 0.25 0.31 1.48 2.27 1.86 Min. 0 0 0 0 0 0 0 0 0 Max. 0.76 8.68 9.01 1.26 0.98 1.29 4.28 8.67 8.19 n 3 3 4 4 4 4 1 4 4 BD Mean 0.86 0.86 0.91 1.04 1.14 1.28 0.78 0.96 0.92 SD 0.13 0.09 0.1 0.16 0.18 0.23 0.05 0.21 0.13 Min. 0.65 0.68 0.76 0.73 0.7 0.77 0.73 0.53 0.74 Max. 1.12 1 1.14 1.21 1.35 1.55 0.84 1.41 1.23 n 4 4 4 4 4 4 1 4 4 pH Mean 6.26 6.28 6.61 6.4 6.73 6.39 6.45 6.66 6.4 SD 0.25 0.22 0.3 0.55 0.58 0.46 0.22 0.27 0.28 Min. 5.4 5.8 6 4.9 5.6 5.5 6 5.9 5,8 Max. 7 7 7.4 7.8 8 7.6 6.80 7 7 n 4 4 4 4 4 4 1 4 4 Clay Mean 8.3 9.2 7.7 6.9 9 11.7 5.8 4.8 3.7 SD 4.6 6.3 4.3 5.2 5.9 4 3.2 4.2 3.2 Min. 3.3 0 3.3 0 0 3.3 3.3 0 0 Max. 16.7 23.3 16.7 20 23.3 20 10 16.7 10 n 4 4 4 4 4 4 1 4 4 Sand Mean 53.3 61.7 60.8 65.6 61 74.4 56.7 56.2 61.5 SD 8 8.3 11.1 12.7 8.6 5.9 8.6 9.8 8.8 Min. 40 46.7 40 33.3 46.7 60 46.7 40 50 Max. 66.7 80 76.7 86.7 73.3 80 66.7 80 86.7 n 4 4 4 4 4 4 1 4 4 Silt Mean 38.3 29.2 31.5 27.5 30 14 37.5 38.9 34.8 SD 5.2 6.1 9.2 11.8 6.7 5.1 6.9 8.3 7.8 Min. 30 16.7 20 6.7 16.7 6.7 30 20 10 Max. 46.7 40 46.7 60 40 26.7 46.7 53.3 43.3 n 4 4 4 4 4 4 1 4 4 “RV=response variable; relief in meters above lowest elevation, BD=bulk density (g/cm3 bMean=arithmetic mean of all data (as distinguished from least squares means reported 40x50 meter plots within each site ), texture (clay, sand, silt) as percent composition, in later tables), SD=standard deviation, n=number of means (LS MEANS) with separations among LSMEANS done using Fisher’s protected least significant value (LSD) as recommended by Milliken and Johnson (1984) and only for significant site effects at a =0.05. All statistical tests were considered significant at the 0.05 level. Because of the correlated nature of many of the response variables, we also conducted a principal components analysis (PCA) (McCune and Grace 2002) to help visualize separations among the study sites along the principal component gradient variables. For the PCA, we did not include any of the responses measured in the “larval nest zone” because of small sample sizes and because no data were collected on the Swearson School Section study site. Although no soil compaction data were collected at the Prairie Coteau (PC) site, anti 8 Journal of the Lepidopterists’ Society Table 4. Variance components for site-to-site, plot-to-plot within sites, point-to-point within plots, and sampling time-to-sampling time across the season at occupied Dakota Skipper ( Hesperia dacotae) study sites in Minnesota, North Dakota, and South Dakota (see Table 2 for study site descriptions and abbreviations); values in parentheses are within row percents of total variation attributed to that variance component. Response Variable Site (%) Plot (%) Point (%)a Sampling (%)“ Total relief (m) 10.39(91) 1.06 (9) nm nm Mean relief (m) 2.91 (55) 0.03(1) 2.32 (44) nm Bulk density (g/cm3) 0.021 (43) 0.000 (<1) 0.028 (57) nm pH 0.023(13) 0.027(15) 0.126 (72) nm Clay (%) 5.24(18) 0.00 (<1) 24.11 (81) nm Sand {%) 29.71 (25) 20.01 (17) 70.13(58) nm Silt (%) 55.81 (47) 12.07(10) 50.42 (43) nm Compaction 20 cm (kg/cm2) 18.39 (47) 5.74(15) 7.23 (19) 7.43 (19) Compaction 40 cm (kg/cm2) 26.02 (51) 8.96 (17) 8.79 (17) 7.64 (15) Compaction 60 cm (kg/cm2) 30.03 (52) 9.62 (17) 10.28(18) 8.02(14) Temperature 20 cm (°C) 2.51 (14) 0.21 (1) 0.00 (<1) 15.40 (84) Temperature 40 cm (°C) 2.00 (17) 0.14(1) 0.00 (<1) 9.83 (82) Temperature 60 cm (°C) 1.12 (13) 0.19(2) 0.00 (<1) 7.03 (84) Moisture surface (% sat.) 23.07 (9) 47.21 (18) 196.44 (73) nm Moisture 20 cm (% sat.) 28.59 (7) 13.27 (3) 0.00 (<1) 357.57 (90) Moisture 40 cm (% sat.) 47.70 (14) 11.74(3) 0.00 (<1) 276.13(82) Moisture 60 cm {% sat.) 79.94 (26) 14,59 (5) 0.00 (<1) 213.00 (68) Larval zone temperature (°C) 0.75(1) 0.00 (<1) nm 52.02 (98) Larval zone dew point (°C) 0.29(1) 0.24 (1) nm 29.20 (98) Larval zone abs. hum. (g/m3) 0.21 (1) 0.19(1) nm 18.49 (98) Larval zone rel. hum. (%) 3.08 (1) 5,50(1) nm 314.14(98) anm= not measured at that level. Table 5 elevations study sites Analysis of variance results for total relief (i.e., maximum elevation - minimum elevation within plots), relief (all - minimum elevation within plots), and variance in relief within plots at occupied Dakota Skipper {Hesperia dacotae ) in Minnesota, North Dakota, and South Dakota; all units are in meters. Total relief Relief Variance in relief SVa df Fb df Fb df Fb s 8 27.41” 8 35.48” 8 10,89” P(S) 22 - 22 - 22 T(P S) - - 505 - - Total 30 535 30 aSV=sources of variation; S=site, P(S)=plot nested within site, T(P S)=sampling point within plot. ||P(S) served as the appropriate error term for testing significance of S based on expected mean squares; °=significant at a=0.05, °°= significant at ct=0.01, ns=not significant. Volume 62, Number ] 9 Table 6. Least squares means (LSMEAN ± SE) for total relief (i.e., maximum elevation-minimum elevation within each plot), relief (all elevations-minimum elevation within plots), and variance in relief within plots at occupied Dakota Skipper ( Hesperia dacotae) study sites in Minnesota, North Dakota, and South Dakota; all units are in meters. LSMEANs within columns followed by die same letter are not significantly different using Fishers protected LSD value at a=0.05 (see table 2 for study site descriptions and abbreviations). Site na Total relief Relief Variance in relief LSMEAN SE LSMEAN SE LSMEAN SE FP 3 0.66 a 0.66 0.38 a 0.31 0.03 a 0.78 HM 3 7.64 c 0.66 3.99 de 0.31 4.68 c 0.78 PC 4 7.89 c 0.57 4.38 e 0.28 5.37 c 0.68 MCC 4 0.96 a 0.57 0.38 a 0.27 0.10 a 0.68 SLS 4 0.79 a 0.57 0.37 a 0.27 0.06 a 0.68 SSS 4 1.02 a 0.57 0.45 a 0.27 0.10 a 0.68 SFP 1 4.28 b 1.14 2.02 b 0.54 2.20 b 1.35 KNP 4 6.35 be 0.57 3.17 cd 0.27 4.78 c 0.68 CXL 4 6.06 be 0.57 3.00 c 0.27 3.51 be 0.68 an=number of 40x50 m plots within each site. Table 7. Analysis of variance results for bulk density (g/m3), Results variance of bulk density, pH, variance of pH, surface moisture (% saturation), and variance of surface moisture at occupied Dakota Skipper ( Hesperia dacotae) study sites in Minnesota, North Dakota, and South Dakota (no surface moisture data were available for study site=HM). Bulk density pH Surface moisture Response SV“ df Fb df Fb df Fb Mean S 8 13.30°“ 8 3.26° 7 4.76°° P(S) 24 24 18 T(P S) 99 956 749 Total 131 988 774 Variance S 8 1 .76 ns 8 4.58°° 7 0.86 ns P(S) 24 24 18 TPS) - - - Total 32 32 25 aSV=sources of variation; S=site, P(S)=plot nested within site, T(P S)=sampling point within plot. '’P(S) served as the appropriate error term for testing significance of S based on expected mean squares; °=significant at a=0.05, °°= significant at a=0.01, ns=not significant. because all other responses were collected there, we chose to include the PC study site in the PCA. We therefore substituted the mean soil compaction values from all of the other study sites for soil compaction at PC. We realize this is not ideal, but for descriptive visualization we believe it suffices. We also conducted a separate PCA for the eight sites using only the “larval nest zone” variables. We used the principal components procedure (PROC PRINCOMP) of SAS (1999) to conduct the PC As. General. Table 3 presents the arithmetic means, standard deviations (SI)), and ranges (minimum and maximum) for selected physical non-biotic attributes (non-climatic) measured at each of the nine study sites. Table 4 presents the results of the variance component analyses with each of the non-biotic response variable results described below. In general and as would be expected, most of the variation in climatic variables (temperature and moisture) relates to sampling time across the season, with mixed results for the more physical attributes. Relief. Nearly all of the variation in total relief (maximum elevation minus minimum elevation within each plot) is attributable to site-to-site (91%) differences (Table 4), implying consistency of plot-to- plot total relief within sites (i.e., plots, once established, all had nearly identical total relief within sites but substantial differences among sites). However, relief (all elevations within a plot minus minimum elevation within each plot) from site-to-site accounted for 55%, with less than 1% of the variation in relief being plot-to- plot, and 44% from point-to-point within plots. These results imply that the relief, or “roughness” in microtopography within plots, was consistent from plot- to-plot within sites, while still maintaining substantial variation in relief from site-to-site. Table 5 presents the ANOVA table results for comparing specific differences among the nine sites with respect to total relief, relief, and variance of relief (all F-tests for the main effect [site] are highly significant, implying that differences exist among sites). Table 6 presents the LSMEANS and mean separations using Fisher’s protected LSD test. In 10 Journal of the Lepidopterists’ Society Table 8. Least squares means (LSMEAN ± SE) bulk density (g/m3), variance of bulk density (g/m3), pH, and variance of pH at occupied Dakota Skipper ( Hesperia dacotae ) study sites in Minnesota, North Dakota, and South Dakota. LSMEANs within columns followed by the same letter are not significantly different using Fishers protected LSD value at a=0.05 (see Table 2 for study site descriptions and abbreviations). Site” Bulk density Variance in bulk density PH Variance in pH LS MEAN SE LS MEAN SE LS MEAN SE LS MEAN SE FP 0.86 ab 0.04 0.02 a 0.01 6.26 a 0.09 0.04 a 0.05 HM 0.86 ab 0.04 0.01 a 0.01 6.28 a 0.09 0.05 a 0.05 PC 0.91 ab 0.04 0.01 a 0.01 6.61 be 0.09 0.06 a 0.05 MCC 1.04 e 0.04 0.03 a 0.01 6.39 ab 0.09 0.27 b 0.05 SLS 1.14c 0.04 0.04 a 0.01 6.73 c 0.09 0.27 b 0.05 SSS 1.28 cl 0.04 0.06 a 0.01 6.39 ab 0.09 0.21 b 0.05 SFP 0.78 a 0.08 0.00 a 0.03 6.45 ab 0.18 0.05 a 0.09 KNP 0.96 be 0.04 0.04 a 0.01 6.66 be 0.09 0.06 a 0.05 CXL 0.92 ab 0.04 0.02 a 0.01 6.40 ab 0.09 0.07 a 0.05 an=4 40x50 m plots within each site, n=l for SFP general, ND sites had less relief and variation in relief than those in either MN or SD. Soil Hulk density, pH, and surface moisture. Nearly 60% of the total variation in bulk density, and an even greater percentage of the variation in pH (72%) and surface moisture (73%) is attributable to point-to- point samples within plots, with consistency in this variation from plot-to-plot among all sites (i.e., all plot- to-plot variation < 18%), with some even less so site-to- site (Table 4). This implies high micro-scale variation in these attributes within the plots (e.g., bulk density varies substantially from point-to-point within a plot, and this variation is fairly constant from plot-to-plot, and to a lesser extent, site-to-site). Table 7 presents the ANOVA table results, showing that significant differences occur among site mean responses and for variance in pH (all F-tests for the main effect “Site” are significant; no surface moisture data were collected at the Hole-in-the Mountain study site). Specific differences in LSMEANS among the sites using Fisher’s protected LSD test are presented in Table 8 for bulk density, variance in bulk density, pH, and variance in pH (mean surface moisture comparisons are presented with other moisture responses below). In general, MN and SD sites had the lowest mean bulk density with ND sites having the highest (no differences were observed among sites with respect to variance in bulk density). While there was no consistent difference in LSMEANS for pH among sites with respect to states, ND sites showed consistently higher variance in pH than the other study sites. Soil texture (% clay, sand, and silt). Samples across all plots and study sites generally were classified as sandy loams, occasionally as loamy sand, with occasional plot points yielding soils that would be classified strictly as loams. Variance component results (Table 4) show great variation in clay from point-to- point within plots (81%), little to no plot-to-plot variation (<1%), with the remainder variation in clay site-to-site (18%). Sand and silt also show approximately half of the variation attributable to point-to-point comparison (58% and 43% respectively), but with more plot-to-plot variation (17% and 10% respectively) than clay, while much more variation is attributable to site-to- site comparison (25% and 47% respectively). These results imply that while there is substantial variation within each plot with respect to soil texture, there is also substantial variation within sites from plot to plot (sand and silt), and even more from site to site. ANOVA results indicated that mean % clay, % sand, and % silt varied significantly among sites with no significant differences in mean variances (Table 9). Further comparisons among LSMEANS indicated a tendency for SD sites to have lower % clay, whereas ND sites tended to have more sand and less silt (Table 10). Soil compaction, temperature, and moisture. Variance components analyses were conducted separately for each of these response variables and separately for each depth (20, 40, and 60cm), with results presented in Table 4. As mentioned above, almost all of the variation in the two climatic variables, temperature and moisture, is attributable to sampling time across the season, with the remaining variation mostly attributable to site-to-site differences. However, with increasing soil depth, more and more variation is attributable to site-to-site differences, particularly for moisture, than to sampling time, the latter nevertheless still accounting for 68% of the variation. Nearly half of the variation in soil compaction can be attributable to site-to-site differences, with the other 50% distributed nearly equally among the other variance components. Volume 62, Number ] 11 Table 9. Analysis of variance results for texture composition (clay, sand, silt) and variance in texture composition at occupied Dakota Skipper ( Hesperia dacotae) study sites in Minnesota, North Dakota, and South Dakota. Clay (%) Sand (%) Silt (%) Response SVa df Fb df Fb df Fb Mean S 8 3.97°° 8 3.64°° 8 8.68°° PCS) 24 24 24 - T(P S) 99 99 99 - Total 131 131 131 Variance S 8 1.08 ns 8 0.93 ns 8 0.88 ns P(S) 24 24 24 - T(P S) - ~ - - Total 32 32 32 “SV=sources of variation; S=site, P(S)=pIot nested within site, T(P S)=sampling point within plot. bP(S) served as the appropriate error term for testing significance of S based on expected mean squares; "= significant at a=0.05, ° at a=0.01, ns=not significant. °=significant regardless of the depth. ANOVA results for comparison among sites and how those differences might vary with soil depth are presented in Table 11, with the interaction of depth and site being significant in all cases (no soil compaction data were available for the Prairie Coteau site where equipment failure precluded collection of data). Because of these significant interactions and the numerous possible pair-wise comparisons, we plotted the LSMEANS (± I SE) for soil compaction (Fig. 4), soil temperature (Fig. 5), and soil moisture (Fig. 6), noting in the legend the approximate Fisher’s FSD values that can be used for specific pair-wise comparisons. Pair-wise comparisons of FSMEANS indicated that soil compaction increases with depth at all sites, and that this rate of increase varies depending on site. With o. a) o 55 FP ?0 40 m HM to 40 60 PC ?0 40 60 MCC 70 40 60 SLS 70 40 m SSS 70 40 60 SFP 70 40 60 KNP 70 40 60 CXL 70 40 60 10 14 18 22 26 Compaction (kg/cm2) 30 34 Fig. 4. Least squares means (LSMEAN ± SE) for soil compaction (kg/cm2) at depths of 20, 40, and 60cm at Dakota Skipper ( Hesperia dacotae) study sites in Minnesota, North Dakota, and South Dakota, USA (Fisher’s LSD=4.5 for n=n=4 and LSD=7.1 for n=4, n=l; see Table 2 for study site descriptions and abbreviations, no compaction data were collected at PC, n=l for SFP, n=4 for all other sites). the exception of the Swearson School Section study site, ND sites tended to have the lowest soil compaction values, with SD having among the highest at all depths. Although not significant, soil temperatures tended to increase with depth at the MN sites whereas temperatures declined significantly with depth at the ND and SD sites. Minnesota sites tended to have substantially higher soil temperatures on average. In general, soil moisture tended to stay the same at various depths or in some cases to decline with depth, depending on the site. Soil temperature tended to be consistent within depth for all sites, with MN sites tending to have higher soil temperatures for all depths than either ND or SD. We did not compute and compare mean variances using ANOVA among sites for soil compaction, soil temperature, or soil moisture because of the added complexity of incorporating soil depths, and because we did not think it would add substantially to understanding these response variables. Principal component analysis (PCA). Table 12 presents the results of the PCA using the listed 15 response variables. The first two principal components accounted for 66% of the variation, with the first three principal components accounting for 80%. Examination ol the principal component variable coefficients, or “loadings,” reveals that the first component variable (PC-1) can be considered a physical-moisture gradient, the second component variable (PC-2) a temperature- relief gradient, and the third component variable (PC-3) a textural gradient. Figure 7 is a plot of the mean principal component values illustrating separation among sites along PC-1 and PC-2. All sites separate out well along the PC-1 axis, with ND sites to the far right and SD sites to the far left, and the MN sites centered (note: the Prairie Coteau study site lies at zero, most 12 Journal of the Lepidopterists’ Society Table 10. Least squares means (LSMEAN ± SE) for texture composition (clay, sand, silt) and variance in texture at occupied Dakota Skipper ( Hesperia dacotae ) study sites in Minnesota, North Dakota, and South Dakota. LSMEANs within columns followed by the same letter are not significantly different using Fishers protected LSD value at a=0.05 (see Table 2 for study site descriptions and abbreviations). RV“ Siteb Clay (%) Sand (%) Silt (%) LS MEAN SE LS MEAN SE LS MEAN SE Mean FP 8.3 cd 1.2 53.3 a 3.1 38.3 c 2,5 HM 9.2 cd 1.2 61.7 ab 3.1 29.2 b 2.5 PC 7.7 be 1.2 60.8 ab 3.1 31.5 be 2,5 MCC 6.9 abc 1.2 65.6 be 3.1 27.5 b 2,5 SLS 9.0 cd 1.2 61.0 ab 3.1 30.0 b 2.5 SSS 11.7 d 1.2 74.4 c 3.1 14.0 a 2,5 SFP 5.8 abc 2.4 56.7 ab 6.2 37.5 c 5.1 KNP 4.8 ab 1.2 56.2 a 3.1 38.9 c 2,5 CXL 3.7 a 1.2 61,5 ab 3.1 34.8 be 2,5 Var. 1LM 47.5 a 12.3 66.2 a 39.1 34.7 a 42.1 FP 16.2 a 12.3 53.4 a 39.1 28.3 a 42.1 PC 18.8 a 12.3 48.7 a 39.1 22.9 a 42.1 MCC 26.7 a 12.3 163.3 a 39.1 149.0 a 42.1 SLS 42.2 a 12.3 63.4 a 39.1 40.1 a 42.1 SSS 18.5 a 12.3 34.1 a 39.1 23.9 a 42.1 SFP 10.3 a 24.5 74.1 a 78.3 47.4 a 84.1 KNP 14.7 a 12.3 48.4 a 39.1 38.3 a 42.1 CXL 11.6 a 12.3 82,3 a 39.1 67.0 a 42.1 aRV=response variable, Var.=variance in texture (clay, bn=4 40x50 m plots within each site, n=l for SFP. sand, and silt). E o Q. 50). A portion of the chorion was consumed at eelosion. Larvae passed through five instars. The first three instars each lasted about 3-5 days, the fourth instar 6-8 days, and the final instar 7-14 days (Table 1). Most larvae matured in 4 weeks. Not surprisingly, given this rate of rapid development, larvae often fed both day and night (and remained at rest adjacent to feeding site). Prepupae tunneled into leaf litter or below ground, where they fashioned a loose cocoon of off-white silk. Pupation occurs 3-4 days after the cocoon is completed. Duration of the pupal stage is expected to be close to 10.5 months for those individuals hatching after a single year, although other psaphidines are known to overwinter multiple times and up to 7 years (Wagner et al. 2009, Dale Schweitzer unpublished data). Life History Notes. Miracavira brillians is a specialist on Ptelea trifoliata (and perhaps other Ptelea in Mexico) (Family Rutaceae). While new foliage is preferred, especially by early instars, mature leaves, including those that are somewhat blighted, are ingested and satisfactoiy for development. Such is not the case for many eastern psaphidines which will struggle and starve if not offered young, not-yet- hardened foliage (Wagner 2005, Wagner et al. 2009). The first through at least the third instars spin a thin sheeting of silk along a leaf edge and then feed on adjacent tissues, keeping the prolegs engaged in silk. Disturbed first instars may balloon downward on a line of silk. The first two instars skeletonize the upper side of the blade over and adjacent to a leaf edge, although towards the end of the second instar some larvae chew through the blade. Third instars largely confine their feeding to a leaf edge, either eating small holes through the blade or carving out cavities from a leaf edge. Some fourth instars also spin a silken sheet over the lamina into which the crochets are engaged, especially prior to Table 1. Head capsule widths and development times for Miracavira brillians. Stage/ Instar Head capsule widths in mm: range, mean, # obs. Approx, length in days; stragglers excluded1 Egg Aug 16 - Aug. 23 1st 0.45-0.48, 0.47, 11 Aug. 23 -Aug.26 2nd 0.73-0.79, 0.77, 15 Aug. 26 - Aug. 31 3rd 1.14-1.18, 1.2,9 Aug. 29 - Sept. 5 4th 1.82-1.98, 1.8, 18 Sept. 1 - Sept. 10 5th 2.88-2.90, 2.89, 2 Sept. 7 - Sept. 30 Pupa 8 Sept 15- 1 Data combined from single clutch reared as two cohorts: one indoors in Hereford Arizona at ambient temperature and a second cohort reared at 23° C in a lab at the University of Connecticut. A third cohort of larvae from the same female sleeved (outdoors) in Hereford had accelerated development with larvae maturing after only 3-3.5 weeks. Figs. 15-16. Miracavira brillians e gg. ( 15) Chorion sculpturing, note adhesive. (16) Micropylar area. Volume 62, Number 1 45 Figs. 17—23. Miracavira brillians (17-22) and Amphipyra pyramidoides (23) larvae. (17) First instar. (18) Second instar. (19) Third instar. (20) Fourth instar. (21) Fifth instar. (22) Fifth instar, head. (23) Fifth instar Amphipyra pyramidoides. a molt. Last instars typically rest off the blade, firmly grasping petioles or shoot tips. Larvae of all instars are difficult to remove from their perch, either because they securely engage the prolegs into their silken sheet (first four instars) or because they hold onto the petiole or raehis tenaciously (last instar). Early instar caterpillars spin silk in advance of any change in position. Most remarkably, two of three preserved (boiled) last instars retained their grip on leaf tissue throughout a five-minute boiling period and to this writing remain firmly attached (in 70% alcohol) to the petiole to which they had initially secured themselves. It is remarkable that the larvae would hold on with such leviathan force, and one must wonder it this behavior has evolved, at least in part, to help the larvae maintain their purchase in the violent squalls ot the American Southwest’s monsoon season. Silk also aids molting as larvae secure the anal prolegs into the sheeting prior to molting. Almost without exception, cast skins are consumed following the molts. First through third instars, when disturbed, sometimes vibrate rapidly from side to side. This behavior was most often noted in first instars and could sometimes be induced with a wisp of air. Vibrating was not observed in fourth and fifth instars. As in other trifid noetuids, the early instars scarcely use the first two pairs ot abdominal prolegs when crawling. The anterior pair (on A3) is only about half the size of those on A5 and A6. Prolegs on A4 are also reduced in size. Even while perched, first, and to some 46 Journal of the Lepidopterists’ Society extent second instars, elevated the anterior end of the body such that the first two pairs of prolegs were either not in contact with the leaf/silk or only weakly secured. Miracavira is exceedingly sedentary, often occupying the same perch for three instars. The caterpillars site fidelity contrasts markedly with Amphipyra pyramidoides Guenee, a familiar eastern species that Miracavira somewhat resembles. A. pyramidoides was cited by Heinrich (1979) as a species that plays "the shell game” with its (avian) predators by frequently changing its location, especially after feeding, and in so doing, removing itself from leaves that it has damaged and which might reveal its whereabouts to natural enemies. The first two instars perch extended along a leaf margin where their coloration is stunningly cryptic (we found it difficult to accurately count larvae without the aid of reading glasses or a lens). Fourth and fifth instars perch with the head, partially drawn into the prothorax, craned back over and held above or pressed against the abdomen; the forelegs are commonly folded across the mouthparts. In middle instars the head is held over the dorsum of the middle abdominal segments. In the last instar the head is pushed even farther rearward, and in the extreme, the frons is held against the anterior face of the abdominal hump (segments 7 and 8) (Fig. 21) or drawn to one side. Again, the first two pairs of legs are held forward and flat against the body; the metathoracic legs are held outward. The anal prolegs are mostly covered by the anal plate. This resting (not alarm) posture presumably provides a case of protection through non-resemblance — the larva is most uncaterpillar-like in appearance. In the fourth and especially fifth instars, the larva becomes increasingly blue-green and a whitish bloom develops over the dorsum, enhancing the insects countershading (Cott 1940, Edmunds 1974, Ruxton et al. 2004) (the caterpillar s pale dorsum is directed downward when the insect is perched on a petiole or twig). Whether Miracavira , in fact, enjoys the evolutionary benefits of non-resemblance and/or countershading will require testing, but there can be little argument that the insect’s posture protects the head from direct strikes: at rest the head is pulled beneath the horn-like rim of the prothorax and the front is held proximate to the abdominal hump. Taxonomic Placement. In 2005 Fibiger and Lafontaine transferred Miracavira into the Acronictinae on the basis of the heavily selerotized, apically positioned clasper, and pattern similarities with the Old Figs. 24-26. Pupa of Miracavira brillians. (24) dorsal. (25) lateral. (26) ventral. Volume 62, Number 1 47 World acronictine genera Nacna Fletcher and Diphtherocome Warren. Neither author had early stages of the insect for examination. The larva of M. 1) rill i a ns lacks acronictine features as defined by Crumb (1956), Kitching and Rawlins (1998), and Wagner (2007a, 2007b): i.e., Miracavira bears only primary setae, verrucae are absent, there is only one seta on the L3 pinaculum on A1-A8, and the dorsal pinacula are distant on both the meso- and metathorax. The caterpillar of Miracavira shares a number of features common to the Psaphidinae (and Amphipyrinae): A8 is humped, the spiracular stripe continues around the anal plate, the dorsal pinacula are whitened, and the head is partially retracted into the thorax (Wagner et al. 2009). The pupa of Miracavira possesses dorsal pits on A10 (Fig. 30), a feature regarded to be synapomorphic for the subfamily Psaphidinae by Kitching and Rawlins ( 1998). Below we expand on our argument that Miracavira is a Amphipyrinae: Psaphidini, and best fits within the subtribe Feraliina. Poole (1995) tentatively associated the Psaphidini and Feraliini on the basis of four characters: the thick, hairy vestiture of the adults; spring flight of the adults; irregular spilling of the tarsi; and enlarged bulla posterior to the tympanal hood. The first of these are common among spring-flying noctuids; the fourth character also was noted by Poole (1995: 162) to occur in other subfamilies. Kitching and Rawlins (1998) identified the shared dorsal pits Alt) of the pupa as an additional feature strengthening tire association between the two tribes. Many of the genera that we examined over the course of this study were found to possess a dorsally lengthened, almost hood-like tegumen. Beyond these few characters, the Psaphidini and Feraliini are rather structurally divergent. Psaphidini have a “claw” at the apex of the foretibia (actually a spine-like seta), a character common Figs. 27—30. Line drawings of female pupa. (27) ventral. (28) lateral. (29) A8-A10, ventral. (.30) A8-A10, dorsal: arrow points to A10 pits. 48 Journal of the Lepidopterists’ Society throughout the Oncocnemidinae, Psaphidinae, and Stiriinae, but frequently lost secondarily (the tibial “claw” in the Cuculliinae is a spine not a seta). The male abdomen has the seventh tergite greatly enlarged and heavily sclerotized, a peculiar character shared with two other psaphidine tribes, Nocloini Poole and Triocnemidini Poole (but absent in the Feraliini). In the male genitalia (e.g., Fig. 31), the uncus is simple, tapered at the apex into a spine; the coronal setae at the valve apex are weak; the clasper is a slightly more heavily sclerotized area on the ventral margin of the valve with an elongated, lightly sclerotized, setose ampulla; the vesica is a simple expanded tube covered with spike-like cornuti with a single larger cornutus at the apex in most species. We consider most of these features to be plesiomorphie within the Psaphidinae because they are also present in the Oncocnemidinae and Stiriinae. Typical Feraliini (only the genus Feralia , Figs. 32, 33) depart from the Psaphidini in several ways: the apical spine on the tibia is lost; the uncus is divided apically into a pincer-like structure; the apical corona on the valve is weak or absent; the clasper and ampulla are absent; a heavily sclerotized, tapered digitus is fused to the inner surface of the valve and narrows into a subapical pollex-like process; the vesica typically is rounded with two diverticula (e.g., Fig. 33b), each covered with long spike-like cornuti. In some species one or both of these diverticula are reduced (e.g.. Fig. 32b). Both larval and adult characters indicate that Mi racavira has a close phylogenetic affinity to the Feraliini Poole. The emerald green and, more importantly, transparent, second and third instars of Miracavira resemble those of Feralia. Like Miracavira , larvae of Feralia are exceedingly sedentary in habit (McFarland 1963), and at least in later instars, caterpillars of both genera accept older foliage, a trait not shared with spring-active genera of Psaphidini. Adult coloration of Miracavira and Feralia are similar — both M. brillians and M. sylvia (Dyar) were originally described as members of the genus Feralia (or its synonym Momaphana ); evidently, the principal reason that the two species were removed by Franclemont (1937) was because the adults lacked eye lashes. Adults lack the apical digging claw on the foretibia common to Psaphadini. Miracavira has highly divergent male genitalia (Fig. 34, note we figure M. sylvia , the type species of the genus), but within the psaphidine is structurally more similar to genera in the Feraliini than to those in the Psaphidini. Miracavira and other genera have diverticula in the vesica covered with spike-like cornuti (in Miracavira the vesica has three large diverticula, each covered with long spike-like cornuti). In both Feralia and Miracavira the ampulla of the clasper and the corona are lost. Differences in genitalia between the two genera are extreme and seem to overshadow the similarities: Miracavira has no trace of a digitus, the uncus is typical of other psaphidines, not highly modified as in Feralia , the dorsal part of the tegument is highly modified, and the clasper is massive (lost in Feralia). We associate three other genera ( Paratrachea Hampson, Fig. 35; Apsaphicla Franclemont, Fig. 36; and Viridemas Smith, Fig. 37) with the Feraliini on the basis of the loss of the tibial “claw,” the loss of the clasper and ampulla on the valve, the dorsally expanded tegumen, and the presence of two cornuti-eovered diverticula in the vesica. These three genera can be associated with each other by a brush-like structure formed by a tight clustering and reduction in length of the cornuti at the apex of the diverticulum closest to the ductus ejaculatorius. Two of these genera, Paratrachea, based on P. viridescens (B. & McD.), and Apsaphicla , can be associated as sister taxa by the close similarity of the shape of the vesica. Connections to the Amphipyrinae. Intriguing are the similarities between the larvae of Miracavira brillians and Amphipyra pyramidoides (Amphipyrinae) (Figs. 21, 23). Shared features include the raised and rather angulate eighth abdominal segment; a yellow middorsal wart on A8; a similar set of middorsal, subdorsal, and spiracular stripes, with the latter weakening over the anterior abdominal segments; and bulging yellow excrescences (Fig. 11) over the upper half of the body and smaller warting below the level of the spiracles. Miracavira caterpillars and those of some Amphipyra (including the Palearetic species A. pyramidea L. and A. berbera Rungs) often have a decided blue-green aspect to the ground color — an unusual coloration among cateipillars. In both genera the head is partially retracted into the thorax (as is the case with many psaphidines). An especially striking similarity is the spiracular coloration: both Miracavira brillians and Amphipyra pyramidoides have a broad black ring (Pperitreme) about the spiracle that, in turn, is surrounded by a pale halo (Figs. 21, 23). Late instars of the two genera rest with the anterior end of the body lifted and well removed from the perch (Figs. 21, 23). Members of both Amphipyra and Feralia also bridge the phenotypic gap between the Amphipyrinae and Psaphidinae. The larval coloration and patterning of Amphipyra tragopoginis (Clerck), and in particular its striping and humped eighth abdominal segment are reminiscent of North American Feralia species. Feralia Volume 62, Number ] 49 Figs. 31-34. Male genitalia: (a) genital capsule; (b); aedeagus with vesica everted. (31a, b) Psaphida resumens Walker (32a, b) Feralia jocosa (Guenee). (33b) Feralia saaberi (Graeser). (34a, b) Miracavira sylvia (Dyar). 50 Journal of the Lepidopterists’ Society Figs. 35-38. Male genitalia: (a) genital capsule; (b) aedeagus with vesica everted. (35 a, b) Paratrachea viridescens (Bames & McDunnough). (36 a, b) Apsaphida eremna Franclemont. (37 a, b) Viridemas galena Smith. (38 a, b) Amphipyra tragopoginis (Clerck). Volume 62, Number ] 51 februalis Grote, a western oak-feeding member of the genus, has an exaggerated, sharply angulate, hump on A8, comparable to that of Miracavira brillians and Amphipyra pyramidoides. The male genitalia of the Amphipyrinae and Psaphidinae also share many characters. In the Amphipyrinae the clasper and ampulla may be lightly selerotized with the ampulla finger-like and setose (e.g., Amphipyra tragopoginis , Fig. 38); or similar to those of the Psaphidinae with the ampulla large, spike-like, and heavily selerotized (e.g., Pyrois ejfusa (Boisduval)); or lost (e.g., Amphipyra pyramidoides and many Feraliini). Also, in both the Amphipyrinae and Psaphidinae the vesica is covered with long, spike-like cornuti arising from stout bases. Two derived amphipyrine character states (not shared by Psaphidinae) are the large, broad, flat pleural sclerite and the disproportionately massive uncus. In sum the similarities between the Amphipyrinae and Psaphidinae show that the Psaphidinae would be best subsumed within the Amphipyrinae as the tribe Psaphidini, and the Feraliine as a subtribe of the latter. Evolutionary relationships among the currently recognized amphipyrine-psaphidine tribes, and inparticular the Nocloini and Trioenemidini, need study. Towards this end, we encourage others to secure and preserve early stages of the Nocloini and Trioenemidini (which are all but unknown) and preserve tissue for molecular studies. Acknowledgements George Godfrey alerted us to the fact that he and Jack Franelemont had reared Miracavira on hop tree ( Ptelea trifoliata ) in 1967 from the Chiricahuas. Jim Romanow assisted with the scanning microscopy. Andrea Farr and Rene Twarkins prepared the line art. Rene Twarkins cleaned the images, “inked” the seta] map, and assembled the plates. Jocelyn Gill prepared the male genitalia plate. Pupae of Feralia and Amphipyra were sent to us by Ben D. Williams and Steven Passoa, respectively. Glenn Dryer and the Connecticut College Arboretum supplied Ptelea leaves to BC and Clinton Morse of the University of Connecticut Greenhouse propagated seedlings for DLW. Financial support came from the U.S. Department of Agriculture, Forest Services, Forest Health Technology Enterprise Team, cooperative agreement number 01-CA-l 1244225-215 to DLW. Literature Cited Barnes, W. 1901. Descriptions of some new species of North American Lepidoptera. Canad. Entomol. 33: 53-57. Cott, H.B. 1940. Adaptive coloration in animals. Methuen, London, England. 508 pp. Crumb, S.E. 1956. The caterpillars of the Phalaenidae. Technical Bulletin 1135. USDA, Washington. DC. 356 pp. Edmunds, M. 1974. Defence in animals. Longman, England. 357 pp. Fibiger, M. & J.D. Lafontaine. 2005. A review of the higher classification of die Noctuoidea (Lepidoptera) with special reference to the Holarctic fauna. Esperiana. Buchreihe zur Entomologie. Volume 11: 1-205. Franclemont, J.G. 1937. Descriptions of new genera (Lepidoptera, Noctuidae, Cuculliinae). J. New York Entomol. 69: 127-130. & E.L. Todd. 1983. Noctuidae, pp. 120-159. In Hodges, R.W. et al. (eds.), Check list of the Lepidoptera of America north of Mex- ico, E.W. Classey Ltd. & The Wedge Entomological Research Foundation, Cambridge Univ. Press, Cambridge, United Kingdom. Heinrich, B. 1979. Foraging strategies of caterpillars: leaf damage and possible predator avoidance strategies. Oeeologia 42: 325-337. Kitching, I.J. & J.E. Rawlins. 1998. The Noctuoidea, pp. 355-401. In Kristensen, N. P. (ed.), Lepidoptera, moths and butterflies. Volume 1: Evolution, Systematics, and Biogeography. Handbook for Zoology. Volume IV. Arthropoda: Insecta, Walter de Gruyter, Berlin, Germany. McDunnough, J. 1938. Checklist of the Lepidoptera of Canada and the United States of America. Part 1. Macrolepidoptera. Mem. Southern Calif. Acad. Sci. 272 pp. McFarland, N. 1963. The Macroheterocera (Lepidoptera) of a mixed forest in Western Oregon. Oregon State University, Corvallis. Unpublished Pli.D. Thesis. 154 pp. Poole, R.W. 1995. Noctuoidea. Noctuidae (Part), Cuculliinae, Stiriinae, Psaphidinae (Part). In Dominick, R. B. et al. (eds.), The Moths of America north of Mexico. Fasc. 26.1, Wedge Entomolog- ical Foundation, Washington DC. 249 pp. Ruxton, G.D., M.P. Speed, & D.J. Kelly. 2004. What, if anything, is the adaptive function of countershading? Anim. Behav. 68: 445- 451. Wagner, D.L. 2005. Caterpillars of eastern North America: A guide to identification and natural history. Princeton Lhiiversity Press, Princeton, New Jersey. 512 pp. . 2007a. Larva of Cerma Hiibner and its enigmatic linkages to the Acronictinae (Lepidoptera: Noctuidae). Proe. Entomol. Soe. Wash. 109: 198-207. . 2007b. Barking up a new tree: Ancient pupation behavior suggests Cerma Hiibner is an aeronictine noctuid (Lepidoptera). Syst. Entomol. 32: 407-419. — , D.F. Schweitzer, J.B. Sullivan, & R.C. Reardon. 2009. Caterpillars of eastern North American Noctuidae. Princeton Uni- versity Press. Received for publication 25 April 2007; revised and accepted 28 September 2007. 52 General Notes Journal of the Lepidopterists’ Society 62(1), 2008, 52-53 SUMMER AZURE ( CELASTRINA NEGLECTA W. H. EDWARDS, LYCAENIDAE) NECTARING ON POISON IVY ( TOXICODENDRON RAD1CANS , ANACARDIACEAE) The purpose of this communication is to report on the ecological relationship between poison ivy ( Toxicodendron radicans [L.] Kuntze) and Summer Azure ( Celastrina neglecta , W. H. Edwards; Papilionoidea: Lycaenidae) as discovered during a systematic survey of poison ivy pollination during the summer of 2005. Daily observations of at least one hour in length were conducted at a central Iowa site (East River Valley Park/Carr Woods, Ames, Iowa; Stoiy County) from June 6-June 20, 2005. June 6 was the day of the first recorded open inflorescence and pollination event and June 20 the last recorded pollination event. This site harbors both climbing and nonclimbing individuals of eastern poison ivy ( Toxicodendron radicans subsp. negundo, Anacardiaceae; Gillis 1971). Each pollination event was photographed using an Olympus D-540 (either still shots or video) and was accompanied by field notes indicating length of visit and time of day. Celastrina neglecta visited inflorescences on three of the fifteen days that viable inflorescences were available (Fig. I). Five distinct nectaring observations were recorded on June 8, eleven on June 9, and one on June 10. All events occurred between 13:00 and 18:00 hours, and the observation period on each of the three days was approximately the same (~2 h). These days were towards the beginning of the flowering period when inflorescences were most abundant throughout the population (pers. obs.). Multiple individuals were observed visiting the same plants simultaneously on both | une 8 and 9, indicating visits were not by a single Fig. 1. Celastrina neglecta nectaring at an inflorescence of poison ivy ( Toxicodendron radicans) on June 8, 2005. butterfly that repeatedly visited the same site. Total length of time spent per visit on a single inflorescence was recorded on both June 9 and June 10 (n = 12). Mean time per visit was 39.3 s (standard deviation = 38 s; median = 37.6 s). During this observation period, Celastrina neglecta would only nectar at an inflorescence if it was the sole visitor; when a competing visitor (such as a bee) alighted on the same inflorescence, the butterfly would immediately leave. Celastrina neglecta was persistent in its visits even when strong wind was present. Previously, the only known relationships between Lepidoptera and poison ivy and its relatives ( Toxicodendron section Toxicodendron, Anacardiaceae) were for larval feeding and shelter (Criddle 1927; Dyar 1904; Eastman and Hansen 1991; Gillis 1971; Richers 2007; Robinson et al. 2007; Tietz 1972). Nectar-seeking at poison ivy (T. radicans) by Celastrina neglecta represents a novel relationship between adult Lepidoptera and poison ivy previously unrecognized, and enhances our understanding of Lepidoptera- Toxicoclendron interactions. This observation also adds to our understanding of the diversity of plant lineages for which Lepidoptera may provide pollination service. Insects from two other orders are also known to pollinate poison ivy, including multiple coleopteran families (e.g., Cantharidae, Cerambycidae, and Cleridae; Senchina 2005) and the ubiquitous honeybee (Apis mellifera , I4ymenoptera:Apidae; Gillis 1971; Lieux 1981). The identification of Celastrina neglecta as a poison ivy floral associate suggests that adults from multiple insect orders may be important in poison ivy pollination ecology. Literature Cited Criddle, N. 1927. Lepidoptera reared in Manitoba from poison ivy. Can. Entomol. 59: 99-101. Dyar, H.G. 1904. Poison ivy caterpillars. J. New York Entomol. Soc. 12: 249-250. Eastman, J. and A. Hansen. 1991. The book of forest and thicket: trees, shrubs, and wildflowers of eastern North America. Stack- pole. Mechanicsburg. xi + 212pp. Gillis, W.T. 1971. The systematics and ecology of poison-ivy and the poison-oaks (Toxicodendron, Anacardiaceae). Rhodora 73: 72-159, 161-237, 370-443, 465-540. Lieux, M.H. 1981. An analysis of Mississippi USA honey: pollen, color, and moisture. Apidologie 12: 137-158. Richers, K. 2007. California Moth Specimen Database. Accessed on July 11, 2007. http://bscit.berkeley.edu/eme/cal- moth_species_list.html. Volume 62, Number 1 53 Robinson, G.S., P.R. Ackery, I.]. Pitching, G.W. Baccaloni, and L.M. Hernandez. HOSTS - A Database of the World’s Lepi- dopteran Hostplants. Accessed on |uly 11, 2007. http://www.nlim.ac.uk/researcli-curation/projects/liostplants. Senchina, D.S. 2005. Beetle interactions with poison ivy and poison oak (Toxicodendron P. Mill. sect. Toxicodendron, Anacardiaceae). Coleopt. Bull. 59: 328-334. Tietz, H.M. 1972. An index to the described life histories, early stages and hosts of the Macrolepidoptera of the continental United States and Canada, vol. II. A. C. Allyn; Sarasota. 1041pp. David S. Senchina, 2507 University Ave., Biology Department, Drake University, Des Moines, I A 50311- 4516, email: dssenchina@drake.edu. Received for publication 26 June 2006, revised and accepted 6 December 2007. Journal of the Lepidopterists’ Society 62(1), 2008, 53-56 ROAD CROSSING BEHAVIOR OF AN ENDANGERED GRASSLAND BUTTERFLY, ICAR1CIA ICARIOIDES FENDERI MACY (LYCAENIDAE), BETWEEN A SUBDIVIDED POPULATION Additional key words: conservation, Lupinus, Oregon. As high quality grasslands dwindle from degradation, habitat fragmentation increases, and urbanization expands butterflies must cope with the encroachment of human modified landscapes if they are to survive. Some butterflies have incorporated exotic larval host plants and non-native nectar resources to survive in urbanized habitats (Shapiro 2002, Graves & Shapiro 2003) while others occupy the isolated vestiges of historically dominant habitats (Severns el al. 2006). For butterflies to survive in human modified habitats they must successfully navigate amongst an array ol unnatural physical structures like residential areas, roads, vacant lots, agricultural fields, orchards, to find adult resources, mates, and larval host plants. While some vagile, polyphagous butterflies appear to be successful in urban situations (Blair & Launer 1997) others with narrow host plant breadth and specific- habitat requirements suffer as habitat modification increases. If we are to conserve, create, and maintain Fig. 1. Photograph of narrow, two-lane paved road, and hedgerow (3m - 5m tall x 100m long) separating the southern subpopulation habitat (left) and the northern subpopulation (behind the hedgerow). areas for butterflies with specialized habitat requirements, then understanding how these species respond to human modified habitats is important for conservation planning. Icaricia icarioides fenderi Macy (Lycaenidae), hereafter Fender’s blue, is an endangered, endemic- species to remnant Willamette Valley upland prairies of western Oregon, U.S.A. Fender’s blue is presently known from about 15 remnant upland prairie sites (Wilson et al. 2003) and most of these are fragmented and isolated. About half of the remaining Fender's blue butterflies are located within the city limits and just west of Eugene, Oregon (Schultz et al. 2003), suggesting that conservation of this species will likely involve butterfly movement through human modified habitats (McEntire et al. 2007). Furthermore, Fender’s blue appears to Ire limited to primarily local movements (Schultz 1998) and its primary larval host, Lupinus sulphureus Dough ex Hook. ssp. kincaidii [C.R Smith] Phillips (Fabaceae), Kincaid’s lupine, is also a locally restricted, threatened species that can be difficult to establish (Schultz 2001, Severns 2003). In the near future, Fender’s blue will face the pressures of navigating through a matrix human modified habitats as open areas surrounding remnant native prairies are becoming increasingly urbanized. An understanding of how Fender’s blue responds to roads and physical barriers that isolate butterfly populations and suitable grassland habitat will contribute important information to aid landscape level butterfly conservation planning. I selected a population of Fender’s blue butterfly that occupies remnant upland prairie in western Oregon, USA to study if a road and hedgerow were barriers to butterfly movement. This study site, -10km west of Eugene, contains one of the larger remnant butterfly populations that is bisected by a paved, narrow two- lane road, bordered on the east side by a 3-5m tall 54 Journal of the Lepidopterists’ Society hedgerow that extends for circa 100m (Fig. 1). On either side of the road habitat conditions are similar, excepting that host plant abundance in the southern subpopulation is about 10 times greater than in the northern subpopulation. Both subpopulations are surrounded by residential areas, open water, and Populus balsamifera L. ssp. trichocarpa [Torr & Gray ex Hook] Brayshaw (Salicaceae) forests. In the spring of 2007, I recorded butterfly behavior on four separate occasions on the 7th, 8th, 26th, and 28th of May on clear, sunny days above 22°C, totaling 2 hrs and 35 minutes of observation. I recorded butterfly sex and the height from the ground, lm, that butterflies flew as they left the southern subpopulation and crossed the road. Since all but three of the butterflies that I observed flying onto the road also crossed the width of the road (» 8m), I recorded the flight behavior of the butterflies when they reached the hedgerow (=T00m long x 3m-5m tall). I grouped the behavior into three flight patterns; 1) those individuals that immediately returned across the road to the prairie after encountering the hedgerow (immediate returns), 2) individuals that flew over the top of the hedgerow into the next field (emigrants), and 3) those individuals that when encountering the hedgerow tracked the length of the hedgerow for at least 5 meters before returning across the road to the original field (eventual returns). Additionally, I noted the flight heights of individuals flying from the northern subpopulation (over the hedgerow) as they flew across the road (immigrants). It is likely that individual butterflies were observed more than once and that the lack of independence was likely to be substantial enough that any statistical tests on butterfly road crossing behavior would be inappropriate, so I present the percentage of observations having recorded behaviors. In the combined observation time of 155 minutes there were 185 road-crossing events, 161 occasions were by males and 21 occasions by females (Table 1). Under the observation conditions and duration, a Fender's blue butterfly crossed the road about once eveiy 50 seconds. Most of the butterflies observed crossing the road from the southern subpopulation also returned to the source field when encountering the hedgerow (Table 1). All of the immigrating males that flew over the hedgerow (from the north) did not turn around when they crossed the road to head back towards the hedgerow, but rather continued on into the southern subpopulation. Most males and females from the southern subpopulation flew along the base of the hedgerow for at least 5 m before returning across the road to the original field (Table 1). Since less than 10% of females and 2% of males flew over the hedgerow from the south (Table 1), it appears that hedgerow was a more substantial barrier to movement between the two subpopulations than the road. Several other studies have demonstrated that roads do not appear to substantially restrict butterfly movement (Mungira & Thomas 1992, Ries & Debinski 2001, Bies et al. 2001, Saarinen et al. 2005, Valtonen & Saarinen 2005). However, in these studies butterflies with different dispersal tendencies also differed in their behavioral response to road edges. The more vagile, strong-flying species were less sensitive to road barriers (Mungira & Thomas 1992, Ries & Debinski 2001) than butterflies that were either habitat specialists (Ries & Debinski 2001) or those that were not efficient dispersalists (Mungira & Thomas 1992, Valtonen & Saarinen 2005). Although I did not directly measure the proportion of Fender’s blue butterflies that turned before encountering the road habitat, the high frequency of road crossings suggests that the road at the study site is not likely to impact dispersal, but the hedgerow was a substantial barrier to dispersal. Since grassland butterflies have been demonstrated to be sensitive to linear objects like lines of flagging (Dover & Fiy 2001), forest edges (Haddad 1999), and abrupt changes in vegetation structure (Summerville et al. 2002, Ries & Debinski 2001), it is not surprising that the hedgerow was a substantial barrier to emigration. One of the primary concerns with roads, besides being a potential barrier to movement, is that roads may lead to significant butterfly mortality (Munguira & Thomas 1992, Mckenna et al. 2001, Ries et al. 2001). I only observed three occasions when cars were present on the road simultaneously with Fender’s blue butterflies. On all three occasions the vehicles were traveling around 40km/hr and butterflies detected the Table 1. Summary of male and female Fender behavior while road crossing. s blue flight 6 9 total observation # 161° 21 % emigrants (southern subpopulation to north) 1.2% 9.5 % % immediate returns 1.9% 4.8% % eventual returns 96.9 % 85.7 % % road crossing flights