Development and Validation of Models for Predicting the Seasonal Emergence of Corn Rootworm (Coleoptera: Chrysomelidae) Beetles in Iowa

Abstract

Effective management of adult northern and western corn rootworms, Diabrotica barberi Smith & Lawrence and D. virgifera virgifera LeConte, respectively, requires knowledge of their emergence pattern so that scouting and adult insecticide applications can be accurately timed. The objective of this study was to develop and validate species- and sex-specific models that reliably predicted adult corn rootworm emergence in Iowa. Prediction began from a biofix defined as the date of first beetle emergence in a field. The models were fit with a 3-parameter Weibull function using emergence data collected in 57 Iowa cornfields over 5 yr. Models were validated with emergence data collected in 21 additional fields from a separate year. A single Pherocon CRW Trap per field was as effective as 13 emergence cages per field at detecting the biofix. Air temperature degree-days accumulated from the emergence cage biofix explained 85% of the variability in total corn rootworm emergence over 5 yr. This model explained 89% and 83% of the variability in total beetle emergence observed in the validation year from the emergence cage and Pherocon CRW Trap biofixes, respectively. These models do not eliminate scouting for adult corn rootworms but should improve the scouting efficiency by allowing growers to focus scouting to key periods, such as peak beetle emergence, when populations should be at their maximum abundance in the field.

corn rootworm, Diabrotica, adult emergence, biofix, Weibull function

Northern corn rootworms, Diabrotica barberi Smith & Lawrence, and western corn rootworms, Diabrotica virgifera virgifera LeConte, are perennial insect pests of corn, Zea mays L., across the Corn Belt. In Iowa, economic losses from these pests are most severe when corn is grown in the same field for successive years. Both species are univoltine and overwinter as eggs, typically in the soil of the cornfield. Egg-hatch begins in late May or early June, and the larvae complete three stages feeding on the roots of corn. Adult emergence begins in late June or early July and can extend through August. Western corn rootworms begin emergence before northern corn rootworms, and both species exhibit protandry (Ruppel et al. 1978, Branson 1987).

The primary control tactics used against corn rootworms in Iowa are rotation with a nonhost crop or prophylactic treatment with a soil insecticide at planting. According to Turpin et al. (1972), only 36% of the continuous (corn planted after corn) cornfields in Iowa contained economically damaging populations of corn rootworms. A more recent study conducted by Gray et al. (1993) showed that 45% of the continuous cornfields in Illinois contained economically damaging populations of corn rootworms. To reduce unnecessary applications of soil insecticide, adult sampling techniques that predict potential larval damage the following season have been developed (Steffey et al. 1982; Hein and Tollefson 1984,1985). These techniques also can be used to trigger insecticide applications targeted to prevent oviposition in single fields or multiple fields on an areawide basis (Meinke 1995, Tollefson 1998). Adoption of adult management strategies may be limited in Iowa because of the additional cost for scouting (Foster et al. 1986). Because adult emergence often extends into early September, accurate assessment of a beetle population usually requires multiple visits to the field. Development of models that reliably predict beetle emergence would allow scouting to be focused on key periods, such as peak emergence, rather than over the entire season.

Development of models that accurately predict adult emergence in the field has been difficult, because many factors interact with immature development. For example, delayed planting of corn increases mortality of larvae from early-hatching eggs by starvation and delays the onset of adult emergence (Musick et al. 1980, Bergman and Turpin 1984). Models based on air temperature degree-day (DD) accumulations, with lower thresholds for development ranging from 11 to 16°C, have been poor predictors of adult emergence (Ruppel et al. 1978, Bergman and Turpin 1986, Hein et al. 1988). This is because many edaphic characteristics, such as thermal conductivity of the soil, soil type, and vertical distribution of eggs in the field, all influence the accumulation of heat units by the immature stages (Bergman and Turpin 1986).

Models based on either calendar date or soil degree-day accumulations, with a lower threshold for development of 11°C, have provided the best fit to emergence data collected in the field (Bergman and Turpin 1986, Hein et al. 1988, Fisher et al. 1990). However, models based on calendar date often fail when environmental conditions differ from normal (Ruppel et al. 1978). Soil temperatures vary widely, even over small areas within a field, and can be cumbersome to use because few publicly accessible weather stations record them (Davis et al. 1996).

Davis et al. (1996) improved models based on air temperature by including an upper developmental threshold of 18°C. Although the fit of the models was improved, other studies conducted in the laboratory determined that corn rootworm development was not hindered until temperatures were above 30°C (Jackson and Elliott 1988, Woodson and Jackson 1996).

Ruppel et al. (1978) compared calendar date, air temperature degree-day accumulations, and days after the first beetle was captured for their reliability as indices of cumulative emergence for adult corn rootworms. They concluded that days after first beetle emergence was the most accurate index for describing emergence, but no predictive models were developed. The objective of our study was to develop and validate species- and sex-specific models that reliably predicted adult corn rootworm emergence in Iowa. The models were based on air temperature degree-day accumulations beginning from a biofix, defined as the date the first beetle was captured in a field. First beetle emergence was incorporated as a biofix, because the accuracy of degree-day models can be significantly enhanced by use of relevant biofixes that correspond to predictable biological events (Welch et al. 1981).

Materials and Methods

Sampling.

Northern and western corn rootworm adult emergence was monitored in 78 continuous cornfields across Iowa in 1992, 1993, and 1997-2000. The number of fields monitored each year and locations, by county, are listed in Table 1. Conventional tillage methods were used in all fields. Row spacing was 76 cm in 67 fields, and 96.5 cm in 11 fields. Only three fields were treated with a soil insecticide at planting. The remaining 75 fields were either not treated or had untreated strips within the field. All other agronomic practices were applied at the discretion of the grower and were typical of Iowa field corn production.

Table 1.

Summary of data collected in 78 Iowa continuous cornfields used to develop and validate models predicting adult corn rootworm emergence

a

Data from 1992, 1993, 1997, 1998, and 2000 combined to develop models. Models validated with data from 1999.

b

Number of fields monitored in each year. Each field contained 13 emergence cages.

c

Mean biofix determined with emergence cages. Means followed by the same letter are not significantly different (GLM, PDIFF test, P > 0.05).

d

Proportion of total beetles that were northern (NCR) and western (WCR) corn rootworm males and females.

Thirteen emergence cages, each with dimensions of 76 by 43 cm, were placed in the untreated portion of each field in late June. The cages were similar to those designed by Hein et al. (1985), but had a zipper sewn into one side of the screen to aid in collection of the beetles. Six cages were placed in one row of corn and seven cages in another. The first cage in each row was placed at least 23 m from the field edge. Rows containing cages were separated by 4-50 rows, and cages within each row were spaced 19-57 m apart, depending on size of the untreated area.

Placement of each cage required the cutting of three consecutive corn plants near the soil surface. The two outer plants were removed with a shovel, and the cage was placed over the center plant so it extended from row-center to row-center. The base of each cage was sealed with loose soil. Beetles were captured in a 0.47-liter paper cup that was coated on the inside with Tangle-Trap Insect Trap Coating (The Tanglefoot Company, Grand Rapids, MI), and positioned on top of a wooden stake in the center of each cage. Thirteen emergence cages per field provided an estimate of the corn rootworm population with a standard error ± 25% of the mean (Hein et al. 1985).

Biofix Determination.

In 1992 and 1993, emergence cages were monitored at 7-d intervals after placement to determine the biofix, or date of first beetle emergence, for each field. In 1997-2000, cages were monitored at 2- to 3-d intervals after placement to obtain more accurate biofixes. A biofix for each field was recorded as the date midway between the last date when no beetles were captured and the date when the first beetle of either sex or species was observed in an emergence cup. After the biofix was determined, cups were changed at least weekly, and the total number of beetles captured in the 13 cages was recorded by species and sex. Emergence was completed in all fields by late August or early September.

A more efficient method for determining the biofix was tested in a total of 55 cornfields in 1999, 2000, and 2001 using the Pherocon CRW Trap (Trecé, Salinas, CA). These are kairomone traps that contain 4-methoxycinnamaldehyde and eugenol as primary components in the lure. Trecé lure #8391 (recommended for areas with predominately western corn rootworms) was used in 1999, and lure #8275 (recommended for areas with predominately either western or Mexican corn rootworms, Diabrotica virgifera zeae Kryson & Smith) was used in 2000 and 2001. In each field, a single Pherocon CRW Trap was hung 1 m above the ground from a wooden stake. The trap was placed at least 23 m into the field between two rows of corn and centered between rows of emergence cages. The Pherocon CRW Traps were checked for beetles on the same dates as the emergence cages in each field. Biofixes between the Pherocon CRW Traps and emergence cages were compared with analyses of variance.

Temperature.

Air temperature data were obtained from weather station summaries published by the National Climatic Data Center (National Climatic Data Center, Asheville, NC 28801). Daily minimum and maximum temperature for each field were interpolated from weather stations nearest each field, and adjusted for latitude, longitude, and elevation differences using a combined optimal interpolation and multiple regression technique developed by ZedX, Inc. (Bellefonte, PA). Air temperature degree-days (Celsius) accumulated from the biofix of each field were calculated with the sine-wave method (Allen 1976). Lower and upper developmental thresholds were set at 11.7 and 35°C, respectively (Kuhlman et al. 1970, Jackson and Elliott 1988).

Development of Models.

Data collected in 1992, 1993, 1997, 1998, and 2000 were combined to develop adult emergence models for northern and western corn rootworm males and females, and for total beetle emergence combined across both species and sexes. For each model, fields with less than five beetles emerging during the growing season were removed from each respective data set. In these fields, where corn rootworm populations were extremely low, the data points were clumped together either very early or very late and did not create a representative curve across the entire emergence period. The northern corn rootworm male and female models used data from 34 and 45 fields, respectively. The western corn rootworm male and female models used data from 47 and 57 fields, respectively. All 57 fields were used to develop the model for total beetle emergence.

Emergence in each field was expressed as a cumulative proportion of total emergence in degree-days accumulated from the emergence cage biofix. Scatter plots of these data showed that corn rootworm emergence followed a sigmoid curve, typical of insect development. The distribution was asymmetrical, however, with rapid emergence early that tailed off slowly near the end of the emergence period. A 3-parameter Weibull function (equation 1) was fit to each data set, where F (x) = the probability of emergence at degree-days postbiofix (x), gamma (γ) = the lag in the onset of emergence, eta (η) = an emergence rate constant, and beta (β) = a shape parameter.

$$ F(x) = 1\exp (- [x - \gamma)/\eta ]^\beta)\,\,\,[1] $$

The Weibull function was chosen because Wagner et al. (1984) demonstrated that it provided an accurate description of insect developmental curves similar to that observed for adult corn rootworm emergence. Values for each parameter were calculated with nonlinear regression using PROC NLIN, DUD method, in SAS (SAS Institute 1999). The NLIN procedure requires the user to provide the cumulative emergence data, degree-days, the function, and an initial range of values for each parameter. The program then tests values within each range at assigned increments using the Gauss iterative method to identify the best-fit value for each parameter and approximate 95% CL. Coefficients of determination (R² values) for each model were determined with the computer program DataFit 7.0 (Oakdale Engineering, Oakdale, PA).

Validation of Models.

Each model was validated with a separate data set collected in 1999 (Table 1) using both biofix determination methods. The data collected in 1999 were chosen to validate the models because emergence was monitored in a large number of fields, and locations were chosen from several different areas of the state to accurately assess the fidelity of the models under different environmental conditions. Fields also were removed from each validation data set if less than five beetles emerged during the growing season. Data from 17 and 19 fields were used to validate the northern corn rootworm male and female models, respectively. The western corn rootworm male and female models were validated with data from 20 and 21 fields, respectively. All 21 fields in 1999 were used to validate the model for total beetle emergence. Emergence in each field was expressed as a cumulative proportion of total emergence in degree-days accumulated from both the emergence cage biofix and the Pherocon CRW Trap biofix.

The degree-day accumulations from each field in 1999 and the parameters for each model were inserted into the Weibull equation to calculate the predicted rate of emergence for 1999. Relationships between predicted and observed emergence from both the emergence cage biofix and the Pherocon CRW Trap biofix were determined with linear regression. Validation was achieved by comparing the slope and intercept values (±95% CL) between observed and predicted emergence in 1999 and by calculating the coefficient of determination (R² values) for the relationships.

Results

A total of 19,921 corn rootworm beetles were captured in 78 cornfields over the 6 yr of this study (Table 1). Western corn rootworms were the predominant species, making up 82% of the total number of beetles that were captured. In most years, populations were also skewed toward females for both species. The mean number of beetles captured per field across all 6 yr was 255 and total beetles captured per field ranged from 11 to 867.

Biofix Determination.

The average biofix determined with 13 emergence cages per field for adult northern and western corn rootworms was between 8 and 9 July over the 6 yr of this study (Table 1). In 4 of 6 yr, the average biofix occurred between 4 and 5 July, with significant delays of 11 and 21 d in 1997 and 1993, respectively. There was a moderate, but significant, correlation between biofixes determined with emergence cages and the Pherocon CRW Trap (r = 0.44, P = 0.0008, n = 55). There were, however, no significant differences in average biofixes between emergence cages or the Pherocon CRW Trap within each year, or when data from all 3 yr were combined (Table 2).

Table 2.

Comparison of the mean ± SD dates of biofix (date first corn rootworm beetle captured) between emergence cages and the Pherocon CRW trap hi 1999, 2000, 2001, and combined across all years

a

Within rows, means followed by the same letter are not significantly different (LSD, P > 0.05).

b

Number of cornfields in which the two biofix determination methods were compared. Each field contained 13 emergence cages and a single Pherocon CRW trap.

The large sphere of influence of the lure used in the Pherocon CRW Traps probably explains why a single trap per field was as effective as 13 emergence cages at determining the biofix. Preliminary studies in Kansas showed the lure in a single Pherocon CRW Trap had an effective plume-radius of 30 m (G. E. Wilde, personal communication). This is equivalent to an area of 0.28 ha and is much larger than the 4.3 m² of soil covered by 13 emergence cages.

Development of Models.

The Weibull equation for each model of corn rootworm emergence and values for each parameter (±95% CL) are shown in Table 3. Cumulative emergence curves for adult northern and western corn rootworms occurred as expected, with westerns beginning emergence before northerns, and females of both species emerging later than males (Fig. 1). Western corn rootworm females began emerging 19 DD after western males, and northern corn rootworm females began emerging 33 DD after northern males (Fig. 1). The time between initiation of male and female emergence was somewhat shorter than other studies reported. In laboratory studies, Branson (1987) determined that western corn rootworm females emerged a mean of 4.7 d later than males. In our study, if an average of 12 DD were accumulated per day, western corn rootworm female emergence would begin 1.6 d after western males.

Fig. 1.

Predicted emergence (black line) ± 95% CL (gray lines) fit with Weibull function for adult male and female western (WCR) and northern (NCR) corn rootworms, and total beetle emergence combined across both species (total beetle) in air degree-days accumulated from the biofix determined with emergence cages,

Table 3.

Weibull equations of the form Y = 1 − exp(− [(X − Γ)/ η]^β) where η, β, and Γ are model coefficients ± 95% CL, X is centigrade air degree-days accumulated from the biofix, and Y is cumulative percentage emergence for adult northern (NCR) and western (WCR) corn rootworm males and females, and adults combined across both species (total beetle)

a

Number of observations in each model.

b

Eta (η) = parameter controlling the rate of emergence.

c

Beta (β) = parameter controlling overall shape of the emergence curve.

d

Gamma (Γ) = parameter controlling the lag in the onset of emergence.

The Weibull function explained 80% of the variability between degree-days accumulated from the cage biofix and cumulative percentage emergence for western corn rootworm males (Fig. 1). Western corn rootworm male emergence peaked (50% emergence) 118 DD postbiofix, was 90% complete after 278 DD, and reached 100% 505 DD postbiofix (Table 4).

Table 4.

Degree-day model predictions, ± 95% CL, for 10, 50, and 90% emergence of adult northern (NCR) and western (WCR) males and females, and total beetle emergence combined across both species

a

Centigrade degree-days (CDD) accumulated from the emergence cage biofix.

The Weibull function explained 79% of the variability between degree-days accumulated from the cage biofix and cumulative percentage emergence for northern corn rootworm males (Fig. 1). Northern corn rootworm male emergence peaked (50% emergence) 169 DD postbiofix, was 90% complete after 348 DD, and reached 100% 570 DD postbiofix (Table 4).

The Weibull function explained 83% of the variability between degree-days accumulated from the cage biofix and cumulative percentage emergence for western corn rootworm females (Fig. 1). Western corn rootworm female emergence peaked (50% emergence) 245 DD postbiofix, was 90% complete after 429 DD, and reached 100% 629 DD postbiofix (Table 4).

The Weibull function explained 78% of the variability between degree-days accumulated from the cage biofix and cumulative percentage emergence for northern corn rootworm females (Fig. 1). Northern corn rootworm female emergence peaked (50% emergence) 268 DD postbiofix, was 90% complete after 449 DD, and reached 100% 643 DD postbiofix (Table 4). Small sample sizes for northern corn rootworms probably contributed to greater variability in the models, indicated by lower coefficients of determination (Table 3) and wider confidence intervals (Table 4), compared with the western corn rootworm models.

A model predicting total beetle emergence combined across both species was also developed, because in Iowa, most continuous-corn producers manage both species as one pest complex. The Weibull function explained 85% of the variability between degree-days accumulated from the cage biofix and cumulative percentage emergence for total beetles (Fig. 1). Total beetle emergence peaked (50% emergence) 211 DD postbiofix, was 90% complete after 411 DD, and reached 100% 640 DD postbiofix (Table 4). If an average of 12 DD were accumulated per day, total beetle emergence would peak 18 d (±9 d) postbiofix and would be completed ≈53 d postbiofix. The period for total beetle emergence was very similar to what others have reported in the literature. Fisher (1984) demonstrated that western corn rootworm adult emergence lasted ≈50 d in South Dakota. In a 3-yr study conducted by Ruppel et al. (1978) in Michigan, western corn rootworm emergence was 100% complete 58 d after the first beetle was captured in a field.

Separate models for total northern and total western corn rootworm emergence (each combined across sexes) were developed but are not presented because both curves were nearly identical. The similarity between the two curves most likely resulted from western corn rootworm emergence that was heavily skewed toward females during most years of the study (Table 1).

Validation of Models.

Initially, each model was visually evaluated by plotting the predicted rate of emergence for 1999 with emergence observed in 1999 (Fig. 2). The models for western corn rootworm males and females and for total beetle emergence effectively explained emergence from the cage biofix in 1999, with cumulative emergence observed in most fields falling within the 95% CL of the models (Fig. 2). Northern corn rootworm male and female emergence from the 1999 cage biofix was more variable than the models predicted. Much of this variability can be attributed to emergence that was significantly delayed at one location (shown with open circles) near Calumet, IA (Fig. 2). The delay is probably not explained by unusual environmental conditions, because western corn rootworm emergence at the same location occurred well within the 95% CL of the models (Fig. 2). Delayed northern corn rootworm emergence at the location near Calumet could be related to individuals in the population exhibiting extended diapause. With extended diapause, a small proportion of the northern corn rootworm population remains dormant as eggs in the soil for two or more seasons before hatching (Krysan et al. 1984). Problems associated with extended diapause have become more frequent over the past 10 yr in northwest Iowa (Rice and Tollefson 1999), and in 1999, northern corn rootworms were collected in emergence cages from a plot of first-year corn directly adjacent to the plot used in this study. This hypothesis is difficult to confirm because no studies comparing emergence rates between normal populations of northern corn rootworms and populations exhibiting extended diapause have been conducted. All five models were nearly as effective at predicting 1999 emergence from the Pherocon CRW Trap biofix compared with the cage biofix (Fig. 2).

Fig. 2.

Predicted adult emergence (black line) ±95% CL (gray lines) and observed emergence (points) for male and female northern (NCR) and western (WCR) corn rootworms, and total beetle emergence combined across both species, for the model validation year of 1999. Observed emergence expressed in air degree-days accumulated from both the emergence cage biofix and the Pherocon CRW Trap biofix. Open circles represent data from one field near Calumet, IA.

Linear regression was used to more closely examine the relationship between predicted and observed emergence in 1999. If the relationship between predicted and observed emergence was perfect, the slope and coefficient of determination for the relationship would equal one and the intercept would equal 0. The models for western corn rootworm males, females, and for total beetle emergence did accurately predict emergence from the cage biofix, because the 95% CL for the slopes and intercepts overlapped 1 and 0, respectively (Table 5). These models also explained between 89% and 91% of the variability in emergence observed in 1999 (Table 5). The models for northern corn rootworm males and females were not as accurate at predicting 1999 emergence from the cage biofix. Both slopes were significantly <1, and the intercepts were significantly different from 0 (Table 5). These models also only explained 76 and 78% of the variability in emergence observed in 1999 (Table 5). If the data from Calumet are removed, the coefficients of determination improve to 0.80 and 0.81 for northern corn rootworm males and females, respectively. Low northern corn rootworm populations in the fields monitored during all years of this study probably increased variability, and reduced the accuracy of the models at predicting northern corn rootworm emergence in 1999.

Table 5.

Linear relationship between predicted and observed adult corn rootworm emergence in 1999 when either 13 emergence cages or a single Pherocon CRW Trap per field were used to determine the biofix

a

Northern corn rootworm (NCR), western corn rootworm (WCR), and total beetle emergence combined across both species (total beetle).

b

Number of observations in each model.

c

Confidence limits (95%).

Very similar validation results were achieved when the Pherocon CRW Trap biofix was used. From this biofix, the models for northern corn rootworm males, western corn rootworm females, and total beetle emergence did accurately predict emergence in 1999, because the 95% CL for the slopes and intercepts overlapped one and 0, respectively (Table 5). The models for northern corn rootworm females and western corn rootworm males were not as accurate at predicting 1999 emergence from the Pherocon CRW Trap biofix. The slope for northern corn rootworm females was significantly <1, and the intercept for western corn rootworm males was significantly different from 0 (Table 5). The amount of variability explained by all five prediction models was also reduced when the Pherocon CRW Trap biofix was used compared with the cage biofix, evident by slightly lower coefficients of determination (Table 5).

The model for total beetle emergence predicted that 10, 50, and 90% emergence should occur at 63, 211, and 411 DD postbiofix, respectively (Table 4). Average total beetle emergence observed in 1999 occurred well within the 95% CL of the model when either biofix was used. From the emergence cage biofix, 10, 50, and 90% of total beetle emergence occurred at 66, 185, and 403 DD postbiofix, respectively. From the Pherocon CRW Trap biofix, 10, 50, and 90% of total beetle emergence occurred at 70, 216, and 432 DD postbiofix, respectively.

Discussion

Unlike many empirically derived functions, some biological meaning can be attached to each parameter of the Weibull function (Brown and Mayer 1988). Each model has parameters estimating lag in the onset of emergence (γ), emergence rate (η), and overall shape of the emergence curve (β). The influence each parameter has on the overall emergence curve is shown in Fig. 3 using the total beetle emergence model as a template. This model was chosen as a template because it provided the best fit (R² = 0.85) of the five models developed and it was successfully validated with both biofix determination methods. The parameters from the total beetle model are common to each section of Fig. 3 (γ = -31.06, β = 2.00, and η = 290.7). In each section, the value of one of the three parameters has been altered in turn while the values for the other two parameters are held constant. Fig. 3A shows that γ is the intercept on the x-axis and estimates when the first beetle emerges relative to the biofix. The value of γ does not change the overall shape of the emergence curve, unlike β (Fig. 3B). For values of β between 3.25 and 3.61, the Weibull distribution approximates the normal distribution (Brown and Mayer 1988). At higher values of β, the distribution becomes more negatively skewed, and more positively skewed at lower values (Fig. 3B). The rate of beetle emergence is fixed by η, and is the most important parameter for describing differences between male and female northern and western corn rootworm emergence (Fig. 3C). As the value of η increases, the time required to reach 100% emergence increases. An important aspect of the Weibull function is the interaction between β and η. Degree-days between the start of emergence and when cumulative emergence reaches 63% is constant for a given value η, regardless of the value of β, and is proportional to the value of η (Brown and Mayer 1988). This explains why the curves in Fig. 3B converge at a common point. For any value of η, the value of β modifies the shape of the curve and provides the flexibility needed to describe the ‘tail' of late emergence seen with corn rootworms.

Fig. 3.

Influence the values of the Weibull parameters (A) gamma, Γ; (B) beta, β; and (C) eta, η have on the shape of the total corn rootworm beetle emergence curve fit with the Weibull function Y = 1 − exp (− [(X − Γ)/η]β). Except as otherwise indicated on the figure, Γ = −31.06, β = 2.00, and η = 290.7 (parameters for the total beetle model).

These models should offer an improvement over previous models based on air temperature accumulations because of the inclusion of a biofix. Because the models begin predicting emergence on the day the first beetle is captured in a field, much of the variability associated with determination of initial emergence between years and between locations within a year can be eliminated from the models. Factors affecting initiation of emergence that are reduced by using the biofix are the planting date of corn (Musick et al. 1980, Bergman and Turpin 1984) and factors that affect heat-unit accumulation by the immature stages such as soil type, soil moisture, and vertical distribution of the eggs in the soil (Bergman and Turpin 1986). Data for these models also were collected across a large number of environments, and thus should be fairly robust at predicting emergence for most regions of Iowa.

The fits of our models are either improved or comparable to previous air temperature models developed in the literature. Bergman and Turpin (1986) monitored adult corn rootworm emergence at one location over a 4-yr period and developed linear regression models for predicting adult corn rootworm emergence. These models also were based on air temperature degree-day accumulations (base 16°C), but were judged to have little utility, explaining only 35-60% of the variability at 50% emergence for northern and western corn rootworm males and females, respectively. Davis et al. (1996) monitored corn rootworm emergence in a total of 11 fields over 3 yr and used logistic regression to develop curvilinear prediction models. These models used air temperature degree-day accumulations, with lower and upper developmental thresholds of 11 and 18°C, and explained 78 and 82% of the variability in northern and western corn rootworm cumulative emergence, respectively. Although these R² values are comparable to our models, the models developed by Davis et al. (1996) did not accurately predict emergence at locations outside the region from where they were developed without first expressing degree-days as a proportion of the 20- to 30-yr average accumulations from 1 January to 30 September for a given location.

In summary, the Weibull function provided a good description of adult corn rootworm emergence in Iowa. Northern corn rootworm emergence was more variable compared with western corn rootworm emergence; however, northern corn rootworm populations were much lower compared with western corn rootworms in the fields monitored during this study. Incorporation of the biofix improved prediction models based on air temperature degree-day accumulations by reducing much of the variability that determines when emergence begins. Degree-days from the cage biofix explained 85% of the variability in total corn rootworm emergence over 5 yr. The model for total corn rootworm emergence explained 89 and 83% of the variability in total beetle emergence observed from the cage and Pherocon CRW Trap biofixes, respectively, in the validation year of 1999. The peak for total beetle emergence occurred 211 DD postbiofix, which is equivalent to 18 d (±9 d) postbiofix if an average of 12 DD are accumulated per day.

Very similar model validation results were achieved with the Pherocon CRW trap biofix. The Pherocon CRW Trap provides an efficient and cost-effective method for determining the biofix compared with emergence cages. In late June, growers could place sentinel Pherocon CRW Traps in their fields, obtain the biofix, and use the model to help time scouting activities in their fields. These models do not eliminate scouting for adult corn rootworms, but they should improve the scouting efficiency by allowing growers to focus scouting on key periods, such as peak beetle emergence, when populations should be at their maximum in the field.

We thank Anne Warshaw for the use of her corn rootworm emergence data from 1992 and 1993, and Trecé, Inc. (Salinas, CA) for providing the Pherocon CRW Traps and lures. We also thank Joe Russo and Jay Schlegel of ZedX, Inc. (Bellefonte, PA) for compiling the weather data for the models. And finally, thanks to Ed Berry for help with locating field sites and maintenance of the emergence cages during the season. Partial support of this research was provided by USDA-ARS Subcontract 58-5447-6-129. This is Journal Paper 19570 of the Iowa Agriculture and Home Economics Experiment Station, Ames, IA. Project No. 3172.