PLANT- INTERACTIONS Spatial Variability of (: ) Pheromone Trap Captures in Sprinkler Irrigated Corn in Eastern Colorado

1,2 3 1 4 SCOTT C. MERRILL, SHAWN M. WALTER, FRANK B. PEAIRS, AND JENNIFER A. HOETING

Environ. Entomol. 40(3): 654Ð660 (2011); DOI: 10.1603/EN10076 ABSTRACT Strategies for controlling pests are an integral part of any agricultural management plan. Most Þeld crops, such as wheat (Triticum spp.) and corn (Zea mays L.) are managed as if they are homogeneous units. However, pests within Þelds are rarely homogenous. Development of plans that use targeted pest control tactics requires knowledge of the ecological drivers of the pest species. That is, by understanding the spatioÐtemporal factors inßuencing pest populations, we can develop man- agement strategy to prevent or reduce pest damage. This study was conducted to quantify variables Downloaded from inßuencing the spatial variability of adult male western bean cutworm, Striacosta albicosta (Smith). Striacosta albicosta moths were collected in pheromone traps in two center pivot, irrigated corn Þelds near Wiggins, CO. We hypothesized that moth abundance would be inßuenced by the distance from the edge of the Þeld, distance to nearest alternative corn crop and affected by anisotropic effects, such as prevailing wind direction. Greater trap catches of S. albicosta in each of the Þelds were found with increased proximity to the edge of the Þeld, if the nearest neighboring crop was corn. Prevailing wind http://ee.oxfordjournals.org/ direction and directional effects were found to inßuence abundance. Results serve as a Þrst step toward building a precision pest management system for controlling S. albicosta.

KEY WORDS Striacosta albicosta, western bean cutworm, spatial, pheromone, precision pest man- agement by guest on March 16, 2016 Agricultural management strategy for pest control is Precision Agriculture management strategy advocates dependent upon three overarching principles: 1) an using within-Þeld, spatially-explicit characteristics to understanding of pest-crop agroecosystems and phe- target management tactics such as using variable seed- nology will help determine vulnerabilities in crop and ing and fertilization rates to maximize yield potential. insect life stages, and may suggest ways to misalign One underused aspect of Precision Agriculture is vulnerable crop stages with pest incidence; 2) Strategy precision pest management. Like many aspects of must include economically feasible pest detection to cropping systems, spatial dynamics of pest densities inform the timing and location for use of control tac- are heterogeneous (Taylor 1984). Spatial patterns may tics or conÞrm the efÞcacy of control efforts; and 3) be caused by numerous factors including aggregation costs of control tactics need to be less than costs pheromones or landscape covariates such as Þeld associated with uncontrolled pest damage. A multi- slope (Merrill et al. 2009). Precision pest management tude of strategies exist ranging from organic farming to is deÞned as the use of spatially-explicit pest ecology integrated pest management. to directed scouting efforts and inform precision ag- Typically, crops are managed as if they are homog- ricultural strategy. For example, Merrill et al. (2009) enous. Crop managers commonly understand the fal- estimated that adequate control of the Russian wheat lacy of managing their Þelds as if they are each ho- mogenous units. Yet routinely, nitrogen is applied to aphid could be achieved by only spraying within-site areas of the Þeld at levels greater than can be taken up locations modeled to have medium to high aphid in- by the crop and pesticides are sprayed on areas with- cidence thereby reducing pesticide inputs by 30%. In out pests. One potentially dramatic way to reduce Pennsylvania, treating only the infested portions of direct and indirect costs would be to use the hetero- potato (Solanum tuberosum L.) Þelds reduced the geneity that exists within our cropping systems. The amount of pesticide needed to control the Colorado potato beetle by 30Ð40% (Weisz et al. 1996). Finally, 1 Department of Bioagricultural Sciences and Pest Management, Davidson et al. (2007) found differing levels of west- Colorado State University, Fort Collins, CO 80523. ern bean cutworm, Striacosta albicosta (Smith), larvae 2 Corresponding author, e-mail: [email protected]. survival in yield based site-speciÞc management 3 312 Diamond Dr., Fort Collins, CO 80525. 4 Department of Statistics, Colorado State University, Fort Collins, zones, which could have implications for directed CO 80523. scouting. Thus, research describing spatio-temporal

0046-225X/11/0654Ð0660$04.00/0 ᭧ 2011 Entomological Society of America June 2011 MERRILL ET AL.: SPATIAL VARIABILITY OF WESTERN BEAN CUTWORM MOTHS 655 agroecosystem dynamics has been shown to effec- tances is less likely (Hanski and Gilpin 1991). Mahrt et tively delineate pest vulnerabilities. al. (1987) found 18% greater catch numbers in traps However, even with spatioÐtemporal pest dynamic located adjacent to beans (an important alternative models found to be highly accurate and precise, mod- food source). els should be used as tools for assisting in planning and We hypothesized that spatial variability within the pest detection and should not be used as a substitute S. albicosta ßight, as monitored by pheromone traps, for observations. Moreover, the economic feasibility would be correlated to directional variables in addi- of many precision spatioÐtemporal modeling efforts tion to the distance from nearby corn crops from the have yet to be demonstrated because of the potential previous year (i.e., locations of previous year corn costs of pest sampling. That is, sampling costs need to crops would act as a source for emerging adult moths) be balanced against beneÞts provided by precision and the distance from the edge of the Þeld. We used pest management. Moreover, the ability to produce an information-theoretic approach to link model in- adequate density maps for pests in a manner ference and model selection to spatially-explicit pest timely enough to be relevant for management pur- dynamics, allowing for the examination of multiple poses has been identiÞed as a serious constraint to the candidate models and variables within each of the development of precision pest management (NRC candidate models (Burnham and Anderson 2002, 1997). Thus, development of spatioÐtemporal models 2004). that use inexpensive inputs, such as pheromone traps

combined with inexpensive remotely sensed data, may Downloaded from Materials and Methods have substantial value for assisting in the development of pest management strategy. S. albicosta ßights were monitored during the 1997 One opportunity for increasing our understanding and 1998 growing seasons in two center pivot-irrigated of spatio-temporal structure of a damaging pest of corn corn Þelds near Wiggins, CO in Morgan County (Þeld (Zea Mays L.) arises with S. albicosta, sampled using 1, 40.332104Њ N, 104.02997Њ W; Þeld 2, 40.30156Њ N, Њ economically feasible pheromone trapping. S. albi- 103.946976 W). Pivot irrigated Þeld sites, each mea- http://ee.oxfordjournals.org/ costa has been recognized as a serious pest of corn and suring Ϸ220 acres, and adjacent corn Þelds were beans (Phaseolus vulgaris L.) for over 50 yr (Douglass planted as part of a continuous corn cropping system. et al. 1955, 1957; Hagen 1962, Blickenstaff and Jolley Traps were placed in Þeld 1 and Þeld 2 on 17 July 1997 1982). Damage to corn is primarily caused by larvae and removed on 20 August 1997, after the ßight had feeding on the ear and indirectly from access to the ceased. In 1998, traps were placed in Þeld 1 and Þeld corn plant provided by the entrance holes (e.g., access 2 on 25 June 1998 and removed from the Þeld on 26 for pathogens) (Douglass et al. 1957, Hagen 1962). August 1998. The sex pheromone used to attract male Unlike many other corn pests, S. albicosta are not S. albicosta moths was identiÞed by Klun et al. (1983)

cannibalistic and, under infestation conditions result- and the use of pheromone traps for S. albicosta mon- by guest on March 16, 2016 ing in economical losses, more than Þve larvae can be itoring was described by Mahrt et al. (1987). expected per plant (Appel et al. 1993). Natural infes- S. albicosta samples were collected weekly at each tations have been found to reduce kernel yield by as trapping location. Moths were identiÞed, then dis- much as 40% (Hagen 1962). Moreover, additional carded within the Þeld. Roughly half of the traps were losses caused by ear rots have been found in associ- arranged in a 76.2- by 76.2-m (250- by 250-ft) grid ation with S. albicosta feeding (Hagen 1962). Damage overlaying the Þeld, with one trap located at the cen- has been estimated at 3.7 bu/acre per larvae when ter of each grid square. Traps were placed within the corn is at the dent stage (Appel et al. 1993). corn row. Traps were located in the same positions in Many variables are likely correlate with the spatial both Þelds in both years. In 1997, samples were col- distribution of S. albicosta adults within corn Þelds. lected at 91 grid points on Þeld 1 and at 100 grid points Prevailing wind direction, as a vector for pheromone on Þeld 2. In 1998, samples were collected at 86 grid dispersal, likely plays an important role in determining points on Þeld 1 and at 100 grid points on Þeld 2. within-Þeld distribution (Mahrt et al. 1987). For ex- Because the scale of spatial autocorrelation of S. ample, if prevailing winds were from the east, pher- albicosta ßight activity is unknown, additional traps omone traps along the western edge of the Þeld would were placed at random distances between grid points. have their pheromone blown away from the Þeld. That is, to determine if spatial autocorrelation of S. Thus, prevailing wind along with other directional albicosta moths occurs, and with what spatial distri- effects may inßuence moth behavior. Population dy- bution, a range of distances between traps was needed namics may be affected by edge effects. For example, (Turner et al. 2001). Grid-based sampling was aug- insect population density may be affected by temper- mented with additional traps as follows: Two addi- ature gradients from crop edge to crop interior (Chen tional traps were added to each grid square within one et al. 1995), differences in predator abundance and quadrant in each Þeld (the high trap density quad- predator diversity may exist within Þeld margins rant), and one trap was added to each grid square in (Bolger et al. 2000), or simply because once a suitable the remaining three quadrants in both Þelds. A re- location is found, the need to disperse is reduced. striction on the random location of these traps was that In addition, distance to the nearest alternative pop- they were at least 30 m apart to reduce intertrap ulation source may inßuence population dynamics by interference. SufÞciency of intertrap interference re- assuming immigration from locations at greater dis- duction was tested by determining if moth abundance 656 ENVIRONMENTAL ENTOMOLOGY Vol. 40, no. 3 per trap in the high trap density quadrant was lower ically, the Corn Adjacent variable indicated whether than in the low trap density quadrants. Field 1 had 98 the nearest adjacent crop was corn (i.e., a likely pop- and 95 additional traps in 1997 and 1998, respectively. ulation source) or other. If the nearest neighboring Field 2 had 86 and 78 additional traps in 1997 and 1998, crop was corn, the Corn Adjacent variable was delin- respectively. The use of additional traps allowed for eated as 1. If the nearest neighboring crop to the trap more accurate analysis of spatial autocorrelation at was other than corn (e.g., fallow, alfalfa, etcetera), it distances Ͻ76.2 m (250 feet). There were 375 trapping was delineated as 0. The directional variables: NorthÐ locations, in total,on the two Þelds in 1997. In 1998, South component, EastÐWest component, and the in- because of planting differences, the number of trap- teraction term between NorthÐSouth and EastÐWest ping locations was reduced to 359. Components were calculated in meters by using UTM Pheromone traps were similar to those described by coordinates. The directional variable prevailing wind Thompson et al. (1987) and consisted of a 3,785-ml direction was calculated using hourly CoAgMet (1-gal) plastic milk jug with a pheromone lure at- (http://www.coagmet.com/) wind direction data (in tached to the cap. The plastic milk jug was modiÞed degrees) collected at each site during the sampling by cutting a 7.5-cm circle out of the side to the right period for each year. Hourly wind direction data of the jugÕs handle. Two small holes were drilled ap- were binned by increments of 5Њ. For each site and proximately 1 cm apart through the cap. Pheromone each year, the bin with the largest count was con- lures were attached to paperclips which had been bent sidered the prevailing wind direction. Prevailing

into a ÔJÕ shape. The paperclips then were inserted wind direction in 1997 was found to be 100Њ for Þeld Downloaded from through the holes in the cap. Lures consisted of rubber 1 in 1997 (i.e., wind was blowing primarily easterly septa to which had been added 10 ␮l of a 5:5:1 mixture with some northerly wind) and 90Њ (easterly wind) of (Z)5-, 11- and (Z)-7-docenyl acetates (Klun et al. for Þeld 2. In 1998, prevailing wind was found to be 1983). 110Њ on Þeld 1 and 105Њ on Þeld 2. Therefore, for- Traps were attached to electric fence posts placed mulas for calculating prevailing wind effect at each

in the aforementioned augmented grid-based pattern. trap location were as follows: http://ee.oxfordjournals.org/ An ÔXÕ Ϸ2.5 cm long was cut in the base of the milk Field 1 in 1997: prevailing wind at 100Њ;EW- jugÕs handle though which the post was inserted. Traps 0.173 * NS then were Þlled with Ϸ1500 ml of soapy water (mixed Field 1 in 1998: prevailing wind at 110Њ;EW- at a rate of two drops liquid dish soap to 1 liter water 0.336 * NS [Doyle 1994]), which was replenished weekly. Field 2 in 1997: prevailing wind at 90Њ;EW Trapped S. albicosta moths were counted weekly, re- Field 2 in 1998: Prevailing wind at 105Њ;EW- corded, and discarded in the Þeld. 0.256 * NS Traps initially were placed on the ground. After where EW and NS are the variables eastÐwest and

cultivation operations were complete, traps were northÐsouth components. For example, within Þeld 2 by guest on March 16, 2016 placed on the electric fence posts. Pheromone lures in 1997, the prevailing wind effect increased with were not replaced during the trapping period in either increased easterly trap location. year because previous Þeld experience with this pher- In addition, substantial year-to-year variability as omone indicated that a single lure was sufÞcient to well as Þeld-to-Þeld variability in moth density likely monitor the ßight period (Peairs et al. 2006). exists. Although understanding between-year and be- All sampling points were referenced with Global tween-Þeld variability in S. albicosta density is very Positioning System (GPS) technology by using Om- important, this understanding is beyond the scope of niStar 6300 or OmniStar 7000 differential (OmniStar this study. That is, we sought for relative spatial dif- Inc., Houston, TX). Coordinates were collected in ferences in moth abundance within Þeld, within year. latitude and longitude with MapInfo (MapInfo Corp., Thus, we standardized moth density across years and Troy, NY) and FarmGPS (Agri-Tech Solutions, Onawa, between Þelds, thus allowing for examination of the IA) and then converted to Universal Transverse Merca- within-Þeld spatial structure of the moth ßight. tor (UTM) coordinates. A set of a priori candidate models was developed Data analysis was completed in R(R Development using all subsets of the variables of interest (distance Core Team 2008). Multiple regression models were from edge of Þeld, corn adjacent, prevailing wind used for all analyses. Weekly trap data were averaged direction, north-south component, east-west compo- by sampling point across each Þeld season. Average nent, and the interaction between the north-south and weekly S. albicosta abundance data were skewed. east-west components). An information-theoretic ap- Transformed abundance data were used as the de- proach to model selection and multimodel inference pendent variable. The Box-Cox method (Box and Cox was used (Burnham and Anderson 2001, 2002, 2004). 1964) was used to Þnd the most appropriate transfor- That is, values for AkaikeÕs Information Criterion mation of the dependent variable. SpeciÞcally, a Ϫ0.34 (AIC), ⌬AIC (⌬AIC measures the distance between power transformation was used to normalize the data the best AIC model and the model of interest), and ϩ Ϫ (i.e., (average weekly abundance data 1) 0.34). AIC weight (AIC wr, provides a measure of the prob- The variable distance from edge of Þeld was calcu- ability that the selected model is the best candidate lated in meters. To examine the effect of the distance model given the data) were calculated for each can- of the trap to an alternative population source, we didate model (Burnham and Anderson 2002, 2004). used a binary variable labeled Corn Adjacent. Specif- Models with strong likelihoods of being the best ap- June 2011 MERRILL ET AL.: SPATIAL VARIABILITY OF WESTERN BEAN CUTWORM MOTHS 657

Table 1. Models included in the Model Averaging Procedure

a ⌬ 2 Model Variables AIC AIC L(gr/data) AIC wr normalized Adjusted R No. 1 SY, DistanceEdge, Wind, AdjCorn, East-West, Ϫ667.48 0 1 0.280 0.214 North-South, East-West*North-South No. 2 SY, DistanceEdge, Wind, East-West, North-South, Ϫ667.36 0.12 0.94 0.264 0.213 East-West*North-South No. 3 SY, DistanceEdge, East-West, North-South, Ϫ667.13 0.35 0.84 0.234 0.211 East-West*North-South No. 4 SY, DistanceEdge, AdjCorn, East-West, North-South, Ϫ667.01 0.47 0.79 0.222 0.213 East-West*North-South

L(gr/data) is the likelihood of the model (gr) given the data. a where SY is the categorical variable sampling date per site per year, DistanceEdge is Distance from edge of Þeld, Wind is Prevailing Wind Direction, AdjCorn is the variable Adjacent Corn, East-West is the East-West component of the Þeld, and North-South is the North-South component of the Þeld.

Ͼ proximating model (AIC wr 0.05) were used for ing 2,489 moths collected at all sites within the Þeld. multimodel inference. That is, all candidate models In 1997, Þeld 2 total weekly trap catch peaked on 23 Ͼ selected as good models (AIC wr 0.05) were aver- July with 5,808 moths and then diminished through 20 aged using a weighted average based on the value of August, totaling 9,602 moths. In 1998, Þeld 2 total Downloaded from the standardized AIC wr. weekly trap catch peaked on 23 July with 2,550 moths Variable relative importance weight is the sum of and then diminished through 26 August, totaling 5,170 the AIC wrs over all models in which a predictor moths. Total trap catch per plot per season was aver- variable occurs (e.g., if variable X occurs in three of aged to obtain the dependent variable: average weekly the top Þve models, with AIC wrs of 0.3, 0.2 and 0.16, trap catch per sample site. then variable XÕs relative importance weight would Within the high trap density quadrant within each http://ee.oxfordjournals.org/ equal 0.66 and would be compared with the relative Þeld, one extra trap was placed within each grid cell importance weight of the other tested variables) within one quadrant of each Þeld. We tested that (Burnham and Anderson 2002, 2004). The resulting reduced distances between traps and the overall weight can range from 0 to 1. The more important increase in trap density within one quadrant would predictor variables have weights with higher values. not create bias from intertrap interference. On av-

Because variable weights are derived from AIC wrs, erage, traps in the high density quadrant showed they can be interpreted as the likelihood of the lower trap counts than those in the low density predictor variable being an important predictor quadrants. However, the trap catches on average variable for understanding variation in S. albicosta were Ϸ0.01 S. albicosta per trap fewer (i.e., Ϸ0.4% by guest on March 16, 2016 density. fewer moths). Because of this low value, variation attributed to intertrap density was not included in model selection. Results and Discussion Like studying many insect species, substantial vari- Five weekly S. albicosta samples were collected in ation in trap counts was observed throughout our test both Þelds from 23 July 1997 to 20 August 1997, and sites. However, we were able to discern strong struc- nine weekly samples were collected in both Þelds tural components within the moth ßight. Four models from 2 July to 26 August 1998. had high likelihoods for being the best approximating In 1997, Þeld 1 total weekly trap catch peaked on 23 model (Table 1), and thus, were averaged by their July with 2,189 moths and then diminished through 20 model-speciÞc AIC weight. Table 2 dictates parameter August, totaling 2,854 S. albicosta moths for the season estimates, conÞdence intervals for the model with the summed across all plots within the Þeld. In 1998, Þeld lowest AIC value. Small parameter estimates are an 1 total weekly trap catch peaked on 15 July with 934 artifact of the normalization transformation of the moths and then diminished through 26 August, total- dependent variable combined with relatively large

Table 2. Parameter estimates and confidence intervals for Model #1, plus effect direction and AIC variable wt for variables included in modeling efforts

Parameter Lower conÞdence Upper conÞdence Back-transformed Model averaged Variablesa estimates (Model 1) interval (Model 1) interval(Model 1) effect direction variable AIC wt DistanceEdge Ϫ0.0002967 Ϫ0.0004041 Ϫ0.0001894 ϩ 1 East-West Ϫ0.0004765 Ϫ0.0009159 Ϫ3.704E-05 ϩ 1 North-South Ϫ0.0001095 Ϫ0.0002549 3.597E-05 ϩ 1 East-West*North-South 4.894E-07 2.179E-07 7.609E-07 Ϫ 1 Wind Ϫ0.0003167 Ϫ0.0007237 9.044E-05 ϩ 0.544 AdjCorn Ϫ0.01969 Ϫ0.04639 0.007001 Ϫ 0.501

a where DistanceEdge is Distance from edge of Þeld, Wind is Prevailing Wind Direction, AdjCorn is the variable Corn Adjacent, East-West is the East-West component of the Þeld, and North-South is the North-South component of the Þeld. 658 ENVIRONMENTAL ENTOMOLOGY Vol. 40, no. 3 values of many of the variables (EW, NS, EW*NS, distance from edge of the Þeld and prevailing wind direction). It is important to note that back-trans- formed parameter values will switch effect direction. That is, as transformed moth counts decrease, the corresponding back-transformed modeled values will increase. For example, the adjacent corn variable was parameterized as having a negative effect on the trans- formed S. albicosta moth abundance, which means that adjacent corn will have a positive effect on moth incidence when back-transformed. As expected, broad-scale spatial and temporal vari- ability were noted for pheromone trap catches in both Þelds in both years (i.e., there was a year*Þeld effect in each of the four selected models). This variation was addressed and removed using a categorical Þeld by year variable. Striacosta albicosta average trap counts were inßu-

enced by their distance from the edge of the Þeld (AIC Downloaded from variable weight ϭ 1), with more moths caught closer to Þeld margins (i.e., moth density decreasing toward the center of the pivot), indicative of an edge effect. Distance from Þeld edge was included in all of the Fig. 1. Modeled results of a pivot-irrigated corn Þeld models selected for model averaging indicating that it (Field 2) in 1998. Average S. albicosta trap catch density is depicted with darker areas indicating higher moth in- is a good variable for spatial delineation of S. albicosta. cidence with modeled results showing a high of 2.35 moths http://ee.oxfordjournals.org/ Because of the layout of our grid-based pheromone per trap and lighter gray areas have low to zero moth traps, this edge effect may result from males being incidence. attracted to the closest trap, or closest available mate and thus, not traveling into the interior past perceived mates. costa moth activity. Thus, a posteriori recalculations The directional variables EW, NS, and the interac- were not completed based on wind direction during tion term between EW and NS were included in all of nocturnal hours or a set of hours selected for high the selected for model averaging, indicating that these moth activity because reanalysis with multiple addi-

variables inßuenced S. albicosta spatial dynamics tional prevailing wind variables verged upon data by guest on March 16, 2016 (each of these variables had an AIC variable weight ϭ mining. 1). Directional variation may be because of differ- The adjacent corn variable was included in two of ences in insolation or other unexamined directional four models (AIC variable weight equals 0.501) as a factors. variable for elucidating location and abundance of The prevailing wind direction variable was only S. albicosta moths. This effect was parameterized in found in two of the four models (AIC variable the direction as expected from our hypothesis (Ta- weight ϭ 0.544). Interestingly, prevailing wind direc- ble 2). Because of the variableÕs AIC weight, adja- tion inßuenced moth abundance in the opposite di- cent corn is the least likely of the selected variables rection as expected. Two mechanisms could explain to be an important variable for elucidating variation this result: 1) prevailing wind direction in all years at in this system. Potential reasons that adjacent corn both sites was primarily or entirely the variable EW was not selected highly by this data set as a top (i.e., primarily an easterly wind). The variable EW has predictor of the spatial dynamics of S. albicosta a large effect size in the opposite direction as the moths include: 1) The signal correlating moth vari- prevailing wind variable and may be assuming some of ability associated with adjacent corn Þelds was de- the variation associated with prevailing wind. 2) In tected more strongly by another variable (e.g., dis- each year at both Þeld sites, there was wind blowing tance from edge of Þeld). 2) The scale at which in the opposite direction (approximately westerly) for adjacent corn inßuences moth distribution was not a substantial proportion of the sampling time. Moth captured by this trap pattern, 3) There is a threshold behavior interacting with the timing of wind may distance after which corn adjacency becomes irrel- affect spatial distribution. That is, moth trap catches evant, and 4) The data were biased, reducing the may be more closely associated with night time pre- effect size of nearby corn Þelds. vailing wind direction (e.g., because of higher noc- Fig. 1 depicts the predicted S. albicosta moth rela- turnal moth activity) as compared with average pre- tive abundance map by using the model-averaged re- vailing wind direction. Recalculating prevailing wind sult for Þeld 2 in 1998. Figure 1 was created using direction solely during nocturnal hours may provide a model results in combination with GIS layers for each different value as that calculated from the observed of the covariates (e.g., prevailing wind direction) by daily average prevailing wind direction. However, lit- using the raster calculator tool (spatial analyst) in tle is known about the hourly timing of peak S. albi- ARCGIS 9.1 (ESRI 1995Ð2007). Maps, such as Fig. 1, June 2011 MERRILL ET AL.: SPATIAL VARIABILITY OF WESTERN BEAN CUTWORM MOTHS 659 could be used to as a Þrst step toward developing in southern California: area, age, and edge effects. Ecol. spatial explicit management strategy. For example, Appl. 10: 1230Ð1248. Fig. 1 depicts areas expected to have high and low Box, G.E.P., and D. R. Cox. 1964. An analysis of transfor- moth incidence. Results depicting structured moth mations (with discussion). J. R. Stat. Soc. Series B. 26: abundance, if found to be associated with oviposition- 211Ð252. ing females, could be used to help pesticide applica- Burnham, K. P., and D. R. Anderson. 2001. Kullback- Leibler information as a basis for strong inference in tion patterns or for resistance management. Alterna- ecological studies. Wildl. Res. 28: 111Ð119. tively, information on spatial structure of male ßight Burnham, K. P., and D. R. Anderson. 2002. Model selection may be used to optimize releases of sterile males if a and multimodel inference: the practical information the- sterile insect tactic is found to be economically viable. oretic approach. Springer, New York. Model results and the derivative maps should inßu- Burnham, K. P., and D. R. Anderson. 2004. Multimodel in- ence future research. For example, research efforts ference - understanding AIC and BIC in model selection. may prove more enlightening by concentrating on Sociol. Methods Res. 33: 261Ð304. areas of the Þeld that consistently showed higher un- Chen, J. Q., J. F. Franklin, and T. A. Spies. 1995. Growing- explained variability in trap counts. Moreover, be- season microclimatic gradients from clear-cut edges into cause directional variables were found to have such old-growth Douglas-Þr forests. Ecol. Appl. 5: 74Ð86. Davidson, S. A., F. B. Peairs, and R. Khosla. 2007. Insect pest power in elucidating spatial population structure, ad- densities across site-speciÞc management zones of irri- dition research could explore mechanism creating this gated corn in northeastern Colorado. J. Econ. Entomol. direction structure. 100: 781Ð789. Downloaded from Pest agroecosystem dynamics are not economically DiFonzo, C. D., and R. Hammond. 2008. Range Expansion valuable unless they are recorded in a timely and cost of Western Bean Cutworm, Striacosta albicosta (Noctu- efÞcient manner, thus allowing for the development of idae), into Michigan and Ohio. Plant Management Net- precision pest management strategy. Our results re- work Online. ßect S. albicosta moth dispersal in an irrigated con- Douglass, J. R., K. E. Gibson, and R. W. Portman. 1955. Western bean cutworm and its control. Idaho Agriculture tinuous corn production system in a region commonly http://ee.oxfordjournals.org/ observed to relatively contiguous corn (and bean) Experiment Station Bulletin 233. Douglass, J. R., J. W. Ingram, K. E. Gibson, and W. E. Peay. moth emergence sites. Our work should be considered 1957. The western bean cutworm as a pest of corn in a Þrst step toward directed scouting efforts and to Idaho. J. Econ. Entomol. 50: 543Ð545. focus future research efforts. Future research should Doyle, M. 1994. Pheromone trapping and laboratory rearing emphasis reduced sampling costs and the develop- of western bean cutworm (Lepidoptera: Noctuidae). ment of pest density maps that are relevant for use in M.S. thesis, Department of Entomology. Colorado State precision pest management. Moreover, additional re- University, Fort Collins, CO. search is needed to assess how spatial relationships ESRI, A. 1995–2007. ArcGIS 9.1 computer program, version By ESRI, A., Redlands, CA. developed in this work hold for nonirrigated regions by guest on March 16, 2016 of corn and soybean production especially given the Hagen, A. F. 1962. Biology and control of western bean recent expansion of S. albicostaÕs habitat range into the cutworm in dent corn in Nebraska. J. Econ. Entomol. 55: 628Ð631. eastern midwest of the United States (DiFonzo and Hanski, I., and M. Gilpin. 1991. Metapopulation dynamics - Hammond 2008). Continued research in this Þeld Brief - history and conceptual domain. Biol. J. Linn. Soc. could allow growers to enjoy the potential beneÞts of 42: 3Ð16. precision agriculture approaches for pest manage- Klun, J. A., C. C. Blickenstaff, M. Schwarz, B. A. Leonhardt, ment inputs, as well as those beneÞts already being and J. R. Plimmer. 1983. Western bean cutworm, Lox- realized for nutrients and water. agrotis-albicosta (Lepidoptera: Noctuidae) - female sex- pheromone identiÞcation. Environ. Entomol. 12: 714Ð 717. Acknowledgments Mahrt, G. G., R. L. Stoltz, C. C. Blickenstaff, and T. O. Holtzer. 1987. Comparisons between blacklight and Funds supporting this research were provided by Colo- pheromone traps for monitoring the western bean cut- rado State UniversityÕs Agricultural Experiment Station Proj- worm (Lepidoptera, Noctuidae) in south central Idaho. ect 753. J. Hoeting gratefully acknowledges the support of the J. Econ. Entomol. 80: 242Ð247. USDAÐARSÐAgricultural Research Service, including con- Merrill, S. C., T. O. Holtzer, F. B. Peairs, and P. J. Lester. tract no. 58-5402-6-343. 2009. Modeling spatial variation of Russian wheat aphid overwintering population densities in Colorado winter wheat. J. Econ. Entomol. 102: 533Ð541. References Cited [NRC] National Research Council. 1997. Precision agricul- ture in the 21st century: geospatial and information tech- Appel, L. L., R. J. Wright, and J. B. Campbell. 1993. Eco- nologies in crop management. National Acad. Press, nomic injury levels for western bean cutworm, Lox- Washington, DC. agrotis albicosta (Smith) (Lepidoptera: Noctuidae), Peairs, F. B., J. B. Rudolph, T. L. Randolph, S. M. Walters, and eggs and larvae in Þeld corn. J. Kans. Entomol. Soc. 66: A. Gebre-Amlak. 2006. 2005 Colorado Þeld crop insect 434Ð438. management research and demonstration trials. TR06Ð Blickenstaff, C. C., and P. M. Jolley. 1982. Host plants of 05, pp. 1Ð30. Colorado State University Agriculture Ex- western bean cutworm. Environ. Entomol. 11: 421Ð425. periment Station Technical Report, Fort Collins, CO. R Bolger, D. T., A. V. Suarez, K. R. Crooks, S. A. Morrison, and Development Core Team. 2008. R: A language and en- T. J. Case. 2000. in urban habitat fragments vironment for statistical computing. R Foundation for 660 ENVIRONMENTAL ENTOMOLOGY Vol. 40, no. 3

Statistical Computing computer program, version By, Vi- Turner, M. G., R. H. Gardner, and R. V. O’Neill. 2001. enna, Austria. Landscape ecology in theory and practice: pattern and Taylor, L. R. 1984. Assessing and interpreting the spatial process. Springer, New York. distributions of insect populations. Annu. Rev. Entomol. Weisz, R., S. Fleischer, and Z. Smilowitz. 1996. Site-spe- 29: 321Ð357. ciÞc integrated pest management for high-value crops: Thompson, D. C., J. L. Capinera, and S. D. Pilcher. 1987. impact on potato pest management. J. Econ. Entomol. Comparison of an aerial water-pan pheromone trap with 89: 501Ð509. traditional trapping techniques for the European corn- borer (Lepidoptera, pyralidae). Environ. Entomol. 16: 154Ð158. Received 26 March 2010; accepted 8 April 2011. Downloaded from http://ee.oxfordjournals.org/ by guest on March 16, 2016