MONITORING CEREAL LEAF BEETLE LARVAL POPULATIONS Thesis for the Degree of M. S. MICHIGAN STATE UNIVERSITY WINSTON CORDELL FULTON 1975 (‘f-‘\I .1 ABSTRACT (A MONITORING CEREAL LEAF BEETLE LARVAL POPULATIONS By Winston Cordell Fulton Populations in which the age of individuals is distributed with respect to time will have only a portion of there members sub- ject to sampling at any point in time by any sampling method which does not sample all age classes. By considering some of the theor- etical factors affecting the proportion of the whole population which is counted under such circumstances, an estimate of the total seasonal incidence of cereal leaf beetle larvae, Oulema melanopus (L), was developed. Factors which affect the portion of the whole which is counted are the width of the sampleable age class, the timing of the sample with respect to the frequency distribution of the sample- able stages, and the variance of the sampleable age class with respect to time. Age of cereal leaf beetle larvae can be determined by measuring larval head capsules. A sample size of about 50 larvae is required to estimate population maturity. Relationships between population maturity in sweepnet samples, and in foliage samples are established as functions of Winston Cordell Fulton 0D > 48. In addition, the relationships between the population density of eggs, and of each intar, to 0D > 48 are presented. Equations and graphs are presented which allow correction of sweepnet catches for their bias for 4th instar larvae, and to estimate the total seasonal population from single sweepnet or foliage samples. MONITORING CEREAL LEAF BEETLE LARVAL POPULATIONS By Winston Cordell Fulton A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Entomology 1975 ACKNOWLEDGMENTS I wish to express my sincere thanks to Dr. Dean L. Haynes for his advice, support, and friendship during the development of this work. I also wish to acknowledge the strong positive influence which working for Dr. R. F. Morris and Dr. G. L. Baskerville has had in making me into a Biologist. ii TABLE OF CONTENTS LIST OF TABLES . LIST OF FIGURES INTRODUCTION PROBLEM DESCRIPTION METHODS Measuring Maturity by Instar Determination--Visual Examination vs. Headcapsule Measurement. . A Scale on Which to Measure Population Maturity. Sample Size for Estimating Population Maturity A Comparison of Sweepnet_and Foliage Samples for Determining Population Maturity in the Field Estimation of Heat Unit Accumulations Near Survey Sites . . . . . . . Estimating Missing Temperatures . Temperature Prediction Between Weather Stations Times of Observed Peak Larval Densities in Survey Samples Population Maturity in Survey Samples as a Function of 0D > 48 . . Population Maturity in Foliage Samples as a Function of 0D > 48 Population Density of an Age Class as a Function .of 0D > 48 . . . . . . ESTIMATING THE TOTAL POPULATION CONCLUSION BIBLIOGRAPHY APPENDICES Page iv vi I3 I7 20 25 28 38 43 49 55 68 84 86 89 Table 2A. 28. LIST OF TABLES The relationship between the average weighted mean instar based on head capsule measurements, and the instar determined from visual inspection for the 197l cereal leaf beetle survey data . . Correlation matrix for proportion in each instar and weighted mean instar in sweepnet and 2-linear foot cereal leaf beetle samples from the same field. Number of observations = 17. Values of |r| > .482 are significantly different from 0 at the 5% level, and are underlines . Standard statistics for the ll variables considered The average and the standard deviation of the differ- ence of the estimated and the actual daily maximum or daily minimum temperature Change NIthe R2 for maximum temperatures at two weather stations (128 is Allegan sewage plant, 5803 is Newaygo--Hardy Dam) when different numbers of weather statigns are used in the regression equations. are for equations using data for the period of time under "Days of year" column Change in R2 with changes in the number of variables used in the analysis for weather stations at Allegan and Newaygo, maximum and minimum tempera- tures. Variables were maximum and minimum temperatures at other weather stations . Comparison of 0D > 48 accumulations calculated by the sine method from actual daily maximum and minimum temperatures, and from daily temperatures estimated, from multiple regression equations . . . . . Mean and standard deviations for the time (Day, oDI>48) of observed peak larval densities in survey samples from Michigan counties (years l967- l97l). 00 were calculated using temperature data from the airports listed . . . . . . . iv Page 21 23 27 34 36 37 41 Table Page 8. Regression coefficients for the probit of the cumulative density of the life stage listed as a function of 0D > 48. Location, 6 is Gull Lake, C is Collins Road. Crops are wheat, W, and oats, 0 . 57 9. Time of peak density in different fields, as determined by solving the regression equations in Table 8 for 0D > 48, such that 0D > 48= (5- a)/b. The day of the year was found by determining the day of the year on which the given 0D > 48 accumulation occurred at Gull Lake weather station for G or East Lansing weather station for Collins Road, C. W is wheat, 0 is oats . . . 58 lo. Regression statistics for the relationship between the time (PD) of peak occurrence of cereal leaf beetle life stages. E = egg, L = larval instar, n = number of observations . . . . . . . . . . . . . 62 ll. Average number of 0D > 48 between the peak density of different life stages . . . . . . . . . . . 63 12. Relation between some physical factors and the time of peak egg density in wheat at Gull Lake . . . . . 66 13. A comparison of the mean, standard deviation and coefficient of variation for sweepnet population estimates in oats for 4 Michigan counties in l972. Means based on averages of "Number of fields" sampled 3 times . . . . . . . . . . . . . 74 I4. Acomparison of population estimates in Jackson County in oats for the years l967-l972. Population estimates were made using the tabulated data and the algorithm described in the text, and Figure 24. The minimum average estimate is that which can be made by adding a constant to the 0D > 48 values found for sample times. Weather station used was Hillsdale, Michigan . . . . . . . . . . . 77 IS. A comparison of population estimates in Shiawassee County in wheat for the years l967-1972. Population estimates were made using the tabulated data and the ‘ algorithm described in the text, and Figure 24. The minimum average estimate is that which can be made by adding a constant to the 0D > 48 values found for samples times. Weather station used was Owosso Waste Water Plant . . . . . . . . . 78 LIST OF FIGURES Frequency distributions of ages. A is at an initial value of f. B-E are at subsequent values of f as the population ages . . . . . Three factors which affect the proportion of the population counted. The a1 to aj and a'i to a'j are observable ages, p is the mean population age and o is the standard deviation of ages . . . Frequency of occurrence of head capsules of a given Size in samples from the l97l CLB regional survey. Total number measured: 28030. One unit on the horizontal axis - 42.33 micrometers . Value of the weighted mean instar, a population age estimate, computed for each sequential instar determination. Series for two samples are shown The weighted mean instar for sweepnet samples against that for hand-picked 2-linear foot foliage samples for the same field. Numbers are the number of larvae in the foliage samples . . Frequency distribution of the correlation coefficients between maximum temperatures from 26 weather stations in Michigan . Frequency distributions of the correlation coefficients between minimum temperatures from 25 weather stations in Michigan . The coefficient of multiple determination, R2, as a function of the number of airport stations used in the multiple regression. Lines are averages for l2 stations. Top line is for minimum temperatures, bottom for maximums . . . . . . . . 0D > 48 accumulation at one station in relation to that at a nearby station. X, Allegan in relation to Ypsilanti; 0, Newaygo in relation to Lansing vi Page IO I4 I9 24 30 BI 32 39 Figure I0. II. 12. I3. I4. I5. I6. 17. I8. I9. Population densities in oats at different 0D values for 2 Michigan counties for 2 years . Percent 4th instar larvae in samples from several fields in counties Macomb (X) and Gladwin (0) at different 00 > 48 accumulations in l972 . . Percent 4th instar larvae in 1971 survey data. Field collections for each county were pooled for a given sample time. Each letter is a different county . Percent 4th instar larvae in oats in 1972 survey data. Field collections for each county were pooled for a given sample time. Each letter is a different county . . . . . . . . Percent 4th instar larvae in wheat in 1972 survey data. Field collections for each county were pooled for a given sample time. Each letter is a different county. Circled letters were not used in the regression . . . . . . . . Percent 4th instar larvae in foliage samples from several fields in different years from Gull Lake and Collins Road. Letters are different crop- year-location combinations. Where crops are wheat = W, oats = 0, and locations are Gull Lake = G and R, and Collins Road = C. . . Percent 4th instar larvae in 2-1inear foot foliage samples in oats, years 1967-1973, Gull Lake and Collins Road data. Regression lines from 1971 (C) and 1972 (8) survey data are drawn for comparison Percent 4th instar larvae in 2-linear foot foliage samples in wheat. Years 1967-1973, Gull Lake and Collins Road data. The regression line from the 1972 (8) survey data is drawn for comparison. The data for wheat for the 1971 survey were not avail- able . . . . . . . . . Average regression lines for probit transformed data for eggs and each larval instar of the cereal leaf beetle on oats . Average regression lines for probit transformed data from eggs and each larval instar of the cereal leaf beetle for wheat . . . . . . vii Page 40 44 46 47 48 50 53 54 59 60 Figure Page 20. The fraction of the cereal leaf beetle population density which a 15 inch sweepnet sample represents at a given 0D > 48 . . . . . . . . . . . . 70 21. Fraction of the seasonal CLB larval population present in the field as a function of 00 > 48 . . . . . 71 22. Factors which, when multiplied by the sweepnet count at a given value of 0D > 48, correct for the sweepnet bias for 4th instar larvae . . . . . . 73 23. The relationship between the correction to 0D > 48 values to give the minimum population estimate mean for the season for wheat and that for oats. X = Shiawassee County. - = Jackson County. Numbers are the year of observation . . . . . . . . . 80 24. Estimated seasonal larval population at Gull Lake in 1974, based on means and standard errors for 2-linear foot samples from Sawyer (unpublished data) and factors from Figure 22. Means with (-) are for oat fields. Those with (X) are for oats mixed with alfalfa fields and are plotted 10 OD higher than their actual value so the standard error lines do not overlap . . . . . . . . . . . . . . 81 25. Estimated seasonal larval populations at Gull Lake in 1974, based on means and standard errors for 2- linear foot samples from Sawyer (unpublished data) and factors from Figure 22. 0D > 48 have been adjusted for the position of the mean larval density . . . . . . . . . . . . . . . 83 viii INTRODUCTION An important aspect of any study of population dynamics of a species is a consistent estimate of the population level of that species in the study area. In making these population estimates, it is often one spe- cific stage, age, class, or sign which is counted once, or a few times, during the season. Southwood (1966) stated that for animals, the size of rela- tive population estimates may be influenced not only by changes in the actual number of individuals in the population, but also by changes in the life stage and activity of the individuals. Changes in the efficiency of the trapping and searching methods used, and the response of a particular sex and species to the trap stimulus also affect this population estimate. Most of these factors will affect the other four types of population estimate, population intensity, absolute population, basic population, and population index (see Morris, 1955, for a definition and description of terms) as well as the relative estimate. A number of investigators have attempted to correct for changes in the size of relative population estimates which were not due to changes in actual number. These workers employed corrections in the form of empirical factors derived from simple and multiple regression techniques. Ruesink and Hyanes (1973) developed such a method to convert sweepnet catches of adult cereal leaf beetles (CLB), Oulema melanopus (L.), Coleoptera: Chrysomelidae, to abso- lute density. Various workers, including the spruce budworm group (Morris and Miller, 1954) have succeeded in minimizing the effects on popu- lation densities of changes in life stage by concentrating their sampling at one particular point in the life cycle. This was done, not because the inherent Changes in population density with popula- tion maturity was recognized as a problem, but to control the effects of outside agents, such as predators and parasites, or internal agents, such as behavior. Cothran and Summers (1972) recognized that the age of alfalfa weevil larvae had an effect on sweepnet estimates of weevil density because of the effect of age on larval behavior, but they didn't recognize the direct effect of age. The change in the relative population estimate caused by seasonal Changes in life stage has been considered directly by only a few people. Multiple sampling of population through time has been a common feature of all this work, which is adequately reviewed by Southwood (1966), Giles (1971), and Kiritani and Nakasuji (1967). In this work, some of the theoretical aspects of sampling ' from a population in which the age of individuals is distributed with respect to time will be considered. Then methods for appylying this theory to sweepnet gathered samples of cereal leaf beetle larvae will be developed. Finally, these methods will be snythesized into an algorithm for estimating total seasonal incidence of cereal leaf beetle larvae in a field from a single sweepnet sample taken at any point in a broad time span. PROBLEM DESCRIPTION In the context of a sampling problem, any given population is composed of two types of individuals--those which can be observed with a given technique, and those which cannot be observed with this technique. The proportion which can be observed will depend on a number of factors, notably on the technique itself, the timing of the estimate in relation to population development, and intrinsic population parameters relating to the distribution of the individuals with respect to maturity. In general, the term "age" will be used to mean physiologi- cal age, or maturity rather than chronological age. The distinction here is absolutely critical and must be kept in mind throughout the ensuing discussion if confusion is to be avoided. Consider a survey technique where only those individuals in the age interval a. to aj can be observed. Let 2(a) be the relative 1 frequency distribution of age classes. Then for discrete ak, ) k=1 k (1) where n = the total number of age classes. Here, the proportion counted, assuming that all observable individuals are counted, is given by k=i (2) For a continuous and n sufficiently large, we have aj P = f Z(a) da a (3) as an approximation to the proportion counted. Now the age class of an individual is a function of time and often of other environmental factors, such as temperature, relative humidity, food supply, etc. That is, a = f(t,E1,E2,...,Em) (4) where a age, t = time, and E],E2,...Em = environmental factors Operating. A common measure of age for insects is degree-days, imply- ing that t is measured in days and the only environmental factor contributing to the aging of the individual is temperature, measured in degrees. Degree-days are not synonymous with physiological time, however, but only one particular measure of it. While many measures of age might be used, a linear scale will simplify appli- cation of the data to what follows. Substituting the right hand side of eq. 4 for the limits of integration in eq. 3 we have .,E ., .,..., . f(tJ 13 E23 END) P = f Z(a) da (5) f(ti,E E .. E .) I'i.’ 21" ’ m1 As an example, let us assume that the distribution of age ' is normal, ". . . since the normal distribution is used by biologists in belief that it is a mathematical necessity, and by mathematicians because they believe it is a biological reality" (Simpson et a1., 1960). Then 5 becomes ft.,E E E. (J I 'Ij’ 2j30009m‘] 1 e_;5 a- P = I Q—- (6) 0' TI' 0' f(ti,E]i,E2i,...,Emi) 2 where o = pOpulation standard deviation, u = population mean, e = the base of the Naperian logarithm system = 2.7183..., and n = 3.1415... . However, both u and o depend on a and hence on f. It can, therefore, be seen that P depends upon: 1. the 11m1ts of 1ntegrat10n, a, = f (ti’Eli’EZi”°"E .), m1 a. = f(t.,E E a a 11’25’°“’Em:1)‘ 2. u, the population mean at the time of sampling, which in turn depends on f; 3. o, the pOpulation standard deviation at the time of sampling and a measure of dispersion in the population. This, too, depends on f; 4. the form of Z. That is, (l), (2), and (3) apply to distributions other than the normal. In Fig. 1A, the distribution of ages of the population at the initial value of f is shown. The dotted line indicates the position of the mean D. Fig. 18 shows the distribution of ages after some interval Af and so on to Fig. 1E. The two vertical lines from ai and aj represent the limits of integration, i.e., the ages which are observable. The proportion counted, therefore, lies between a1 and aj. It is implicit in Fig. 1 that 0 remains constant with changes in f. That is, that our measure of age is arithmetic. It is also implied that changes in population level are negligible or constant in rate through all ages. These two points are of particular impor- tance since frequently we know the age distribution of the popula- tion at a time such as Fig. 1A would indicate, but the count of individuals is not made until a time such as indicated by Fig. 10. It is therefore implied that the form of the distribution of ages at the point shown in ID is the same as that at the point shown in IA and therefore known. If these assumptions are invalid, an analy- sis more complex than the one considered here is required. In this type of analysis the way in which the distribution changes from point 1A to point 10 would have to be determined, possibly through an application of the generalized diffusion model as presented in Barr et a1. (1972). If the distribution at the time of the count is known, the situation is relatively straightforward and the proportion counted will be determined by the intrinsic population factors indicated in Fig“ 2. Fig. 2A shows the effect on the proportion counted of changes in the observable ages. 8 A 21(f) 1///,I‘\\\\, I3 ZIT+A1TI A > ‘z’ c “J 3 O 1 I” : E Z(Tflszf) /:\\ I) Z(f+A3‘I) / j\ El Z(f‘IAf) /\ OI OI AGE Figure l.--Frequency distributions of ages. A is at an initial value of f. B-E are at subsequent values of f as the population ages (752702-11). Fig. 28 shows the effect of the distance of the population mean age from the age class observed. Fig. 2C shows the effect of different degrees of dispersion in the population. As an example consider a population of spring flowers such as trilliums. If the sampling scheme calls for us to walk through a wooded area and count all of the trilliums which are seen, we will clearly count more on any given day if we can recognize not only those in bloom (Fig. 2A, ai to aj), but also those with only vege- tative parts (Fig. 2A, a; to a3). Again assuming that only those in bloom can be recognized, and if the sampling stroll were taken late in the season, perhaps only a few of the trilliums would remain in bloom (Fig. 28) even though there had been many more to see at an earlier time. If counts were made of two species, say trilliums, which are only in bloom for a short time in the spring (small vari- ance, Fig. 2C), and dandelions, in which some part of the popula— tion is in bloom throughout spring, summer, and fall (large vari- ance, Fig. 2C), a higher portion of the trilliums than of the dandelions could be seen on one day near the peak. It is clear then that the maturity of the population can affect the proportion of that population which is counted at a specific point in time. Not only will it affect the counts of the primary organism, but in cases where parasitized individuals are concerned, it affects the estimates of seasonal parasitism. The effects of maturity on the population density estimate can be mini- mized by choosing a sampling method which collects all age classes present or alternatively by sampling a life stage which is so long IO FREQUENCY, z (a) t——-- Figure 2.--Three factors which affect the proportion of the population counted. The a1 to aj and a'i to a'j are observable ages, p is the mean population age and o is the standard deviation of ages (752702-10). II and stable that essentially all of the individuals in the population are in that life stage at one time. Slightly less effective is to attempt to take the population sample when the mean age of individu- als is near the midpoint of the observable age interval, so that the largest portion of the individuals can be observed. When none of the above alternatives is feasible or desirable, the proportion of the seasonal population which a given sample repre- sents can be estimated. To make this estimate will require three things. First, a good (linear) scale on which to measure age is needed. This will help to minimize changes in the distribution of ages which are due solely to the scale on which age is being measured. Second, an estimate of maturity in the field sample at the sample time is needed. In this way the position of the sampled ages with respect to the position of the mean (Fig. 28) can be determined. Third, the age distribution of the population through the season must be determined. The following sections of this thesis deal with the develop- ment of methods to estimate or measure the above three parameters in larval field popUlations of the cereal leaf beetle as sampled by a sweepnet. After these have been developed, this information then can be used to estimate total season populations of that insect. METHODS To measure maturity requires some means by which different members of the population can be judged more or less mature than other members of the population. In insects, the life stage is commonly used to indicate maturity, and within these categories, other divisions, notably instars, may be recognized. Clearly, for insects with an indeterminate number of moults, observations at one point in time do not indicate the life history or maturity of indi- viduals in the population. In the case of the cereal leaf beetle though, the number of instars was known to be fixed. Therefore, a measure of maturity for the population based on a determination of the instars of individuals in the population was developed and is presented in a subsequent section. In general, the maturity distribution through the season, and the maturity of the field sample could be determined in one of two ways: it could be estimated directly by sampling the population; or second, the physical factors which contribute to aging (eq. 4) could be monitored in the field and the age distribution at the sample time estimated from that data. Both of these approaches will be considered for cereal leaf beetle larval samples. The main data sets available for this study were from the annual cereal leaf beetle larval survey and from intensive sampling I2 I3 plots used to study the within-generation population dynamics at Gull Lake, Kalamazoo Co., Michigan. A description of the CLB survey method is contained in Appendix A. This survey was developed primarily by Dr. D. L. Haynes of Michigan State University. Five years (1967-71) of data were available on magnetic tape when this study began. Only minor procedural changes have taken place in the survey each year. Cur- rently, samples are taken at about 27 sites in five states and Ontario, Canada. For a complete description of the Gull Lake data, see MSU Theses by Helgesen (1969) and Gage (1972, 1974). Initially the survey samples were analyzed to determine their information content with respect to maturity. Measuring Maturity by Instar Determination-- Visual Examination vs. Headcapsule Measurement Head capsule measurements offer an accurate but time consum- ing method of determining instars. Figure 3 shows the frequency of occurrence of head capsules of different sizes in the 1971 CLB survey data. This pattern is similar to that presented in Hoxie and Wellso (1974). All headcaps were measured with an ocular micro- meter in a WildR microscope at 25X. One division of the micrometer represents .042333 millimeters at this magnification. Separate peaks for each of the four instars are evident. Overlap in headcap size between instars is not great, and is least between 2nd and 3rd instars. Based on this data, the relationship between headcap size, X, and instar was determined to be: I4 . .Awmuwonmmmv mgmvmsoguws mm.~¢ u mwxm Fauco~wcog ms“ :o “we: mco .omomN "umgzmmms Lassa: Peach .»m>c:m pmcommmc 30 :3 of 59¢ $353. 5 3.5. :33 a mo 8383 23; mo 3:82:68 hE mucoscmguié 9:5: u~.m ado 049.. on mu 8 em «a on o. o. c. N. o. o .qquuq...qq. .qd.-qu° .00? .1000 CON. l 000. l OOON OO¢N BVAHV'I :IO USSNON 1 Doom ooun coon OOOQ IS lfléiit Divisions Millimeters l X 5_ll X 5_.47 2 11 < X 5_l6 .47 < X 5_.68 3 16 < X 5_21 .68 < X 5_.89 4 x > 21 X > .89 No significant deviation in the pattern represented in Figure 3 was seen in any survey site or for any survey cycle. However, since Hoxie and Wellso (op cit.) showed a significant correlation between prepupa headcapsule width and adult elytra length while Jackman and Haynes (1975) showed a significant inverse correlation between larval density and elytra length of adults, it is quite possible that under some circumstances, not detected in this survey data, variation in the headcap size pattern would occur. The 1971 cereal leaf beetle survey material had been sepa- rated visually into different instars. This material had been saved so that for each county, for each survey cycle, four con- tainers were available which had what workers considered the four separate instars. As many as 125 larvae were sampled from each of these containers, their head capsules measured, and their true instar determined. In Table l the realtionship between the weighted mean instar and the estimated instar is presented. The weighted mean instar was computed as: WMI = (1 N1 + 2 N2 + 3 N + 4 N4)/(N1 + N2 + N3 + N4) (7) 3 I6 TABLE l.--The relationship between the average weighted mean instar based on head capsule measurements, and the instar deter- mined from visual inspection for the 1971 cereal leaf beetle survey data. Visual Number of Average Standard Standard Instar Samples We1ghted Deviation Error Mean Instar l 43 1.60 .46 .07 2 56 2.54 .42 .06 3 59 3.31 .30 .04 4 61 3.92 .14 .02 where WMI is the weighted mean instar and Ni is the number of indi- viduals in the sample in instar i. Clearly fourth instar larvae can be visually separated from all others with a high degree of accuracy, but the other instars cannot be accurately separated from each other with this method. For an accurate direct estimate of the age class head capsules must be measured. A Scale on Which to Measure Population Maturity While the age class of an individual can be determined read- ily by measuring its headcapsule width, more consideration must go into choosing a scale on which to measure population maturity. The number of individuals in each of the age classes could be used, but this would be awkward in making comparisons between populations or samples. A desirable feature of a scale for age of the population would be to use the infOrmation on the number of individuals in each age class to produce a single number. One possible maturity scale I7 is the weighted mean instar. Although strictly speaking all life stages should be included in the equation in order to get a true measure of population age, this could have been accomplished for newer data only with the expenditure of a great deal of resources, and was not available for historical CLB survey data. Therefore weighted mean instar was computed exactly as in eq. 7. In the case of the CLB, each instar is of approximately the same GUration on a OD>48 scale (Helgesen,l969; and Ruesink, 1972). If this were not the case, then weighting factors for the different age Classes would have to be computed based on the relative proportion of time spent in any life stage, compared to the total length of time for all life stages used. That is: suppose life stages m through t are to be used in determining population age and k time units are spent in this interval. Further, suppose that x time units are spent in stage i (m 5_i 5_t). Then the proportion of time spent in the ith stage is Sample Size for Estimating Population Maturity An upper bound on the change in the weighted mean instar for the addition of one more sample can be determined for sample size I8 lgg-to be 3n%. For example, in a sample of 99 head capsules, all of which were firsts, the WMI would be 1.0. If the 100th larvae were a fourth the WMI would be (99+4)/100 = 1.03, a 3% Change. Using the formula 1%9-= sample size, we have that sample size = 100, therefore n=1 and the error is again 3n% - 3%. Actual values for the weighted mean instar computed sequen- tially for two sweepnet samples are shown in Fig. 4. These are representative of the 18 samples which were plotted in this way. The series represented by X's had 100 of 102 larvae measured, while that represented by 0's had 100 of 3130 larvae measured. In each case, the weighted mean instar is reasonably stable after about 50 measurements have been made. While samples with many larvae (>>100) were subsampled by inserting forceps into the vial and drawing out a group of 10 or so, (a procedure which could bias the size sampled) smaller samples, such as those represented by the X's of Fig. 4 were dumped en_ma§§§_onto the counting stage. The pronounced hump at the beginning of the "X" sequence is probably caused by the worker picking out the "easy" large larvae to measure first. In this case as well as in the "0" sequence, however, 50 measurements adequately defined the mean maturity of the sample. Clearly what is "adequate" must be defined in relation to the problem at hand. It is also clear that a bias cannot exist in the sampling method used when the number of larvae in the population sample is less than 100. Also, unless the bias is very great indeed, the value for the WMI determined from 100 head capsule measurements would not change much I9 .AFNLNoummNV :zocm mxm mmpasmm 03» so» mmwxwm .covumcwsxmpmu cmpmcw meucmaamm comm so» umusasou .mpmewumm mam cowumpzaoa m .xmumcw cams umpsmwmz mg» mo ma—m>--.¢ mxzmwu omcamdmx mum’s: 00. cm on Ob 00 00 cc on ON 0. 6L 0 l o o o x. o 000 000 o o - 000 00800000 0 8030800000080800 oo ofiwnxoBQu x- xx xx xxxxx xxx xx xxx xxxxxxxx xxx xx x .m xx xx I x xx xxx 00. N onfi oh.“ 00.» mm.» 0nd 2km 00.? NVBN 031H9I3M UVLSNI 20 by removing the bias. Thus, while in future work this bias in sampling will be estimated, since its elimination could allow esti- mation of the WMI with fewer samples, it is considered of minor importance here. A Comparison of Sweepnet and Foliage Samples for Determining Population Maturity in the Field In 1974 data were collected to determine the relationship between the instar distribution in standard sweepnet samples and that in the sampled field. The field distribution was taken to be that determined by handpicking all of the larvae from the foliage in several 2-foot sections of a row in the oat field. The number of 2-foot samples taken varied depending on the population density in the field, since the objective was not to compare density esti- mates, but to get a reasonable number of larvae for age comparisons. A summary of the data on which these analyses are based is presented in Appendix B. For the sweepnet samples and for the handpicked samples the proportion, P, in each instar and the weighted mean instar, WMI, were computed. A correlation matrix for these data, along with plant height, HGT, was computed and is presented in Table 2A. For the date of Table 2, we have that n = 17, thus, values for |r| > .482 in the table indicate a regression coefficient between the two variable which is significantly different from 0 at the 5% level (Steel and Torrie, 1960, p. 190). 21 The most notable thing is the general lack of a relationship between values for samples collected by the two different methods. The only exception is the marginal value of r = -.492 between the weighted mean instar from the sweepnet samples and the fraction which was second instar larvae in the 2-foot foliage samples. Con- sidering that there are 15 values for r between the two sets of data, by chance alone, one of these values is expected to exceed the bounds established for the 5% fiducial interval. From Table 28 one can see that in the foliage samples there was a considerably higher proportion of the population in the first two instars (.065 + .130 = .195) than was the case in the sweepnet samples (.007 + .082 = .089). This is reflected in the lowered value of the WMI, 3.21 for foliage vs. 3.48 for sweepnet, and in a larger value of the coefficients of variation for the WMI, 12.2 for foliage vs. 3.8 for sweepnet. In Fig. 5 the WMI for sweepnet samples is plotted over the WMI for foliage samples from the same field. The number beside each point is the number of larvae in the foliage sample. The greater variation in the values for 2-foot samples and the lower mean is evident in Fig. 5. A linear regression on this data (not weighted for sample size) gave a regression equa- tion of y = 3.06 + .13X, where Y is WMI for sweepnet samples and X is WMI fOr foliage samples. The slope is not significantly differ- ent from zero. My conclusion, therefore, is that the proportion of small larvae in sweepnet sampels is essentially constant and not related 22 usmwm; coco u hwxu cmamcm : mg» cw mpaEnm on» eo cowpumxw u cm 5» xmpmcw cams emucmmmz u H23“ coo.F NmP.- Pom. omm.- mmo.- wmo.- mpm.- mmm. Poo. moo. opm.- ha: ooo.~ Pom.- ¢m~.- me.- upw. mmm. omp. mpq.n mm~.- mme. «a S 000.? «cm.. ope. mmm.- mop.- Pmm.- mew. sme. m~¢.- mm m d 80.. 3m. 5.. E... SN. an. MS; so... 2 m ooo.P mom.- ¢F¢.- mmm. mmm. mmF. me~.- Fm ooc.F mmm. NFO.- Nm¢.- Fmo.- cwm. Hz: ooo.— m-.- oom.- mmo.- mmn. ea 2 coo.F mmm.- omm.- mmo.- mg n” U 80; m8... mom; 2 m 80; 88.. E m. 1. coo.F Hz: ohm: «a mm mm Pa Hz: «a mm mm ppm msz mmpgsmm umcnmmZm mmmasmm .um gmmcmg N .cwcppxwuca men use .pm>mp am any as o Eoce ucmgwmmwu appcmo_mwcmwm ago Nae. A _x_ mo mmzpm> .np u mupmwe mo xmnsaz .upo_m «Emu mg» scum mwpqEMm mpummn mmm— mexmu uoow Emmcwp1~ can uwcammzm c? cmumcm come umugmmm: new xmumcw zoom cw cowpgoaoca com chums comumpmcxou-u.<~ u4mgm~ mo conga: mgp mew mgmnszz .Am_-~oxmmxv mSFQEem mane—ct mmdmzdm medic“. 20¢“. m4...mz. 24m! cubic-m; mfin - d 04" Ann I J . a max. n. no. my». nwn . 1 5:. so. so. mm d a . I] a. . ox. not. . —O N mam o1. mwu sum 04“ . .upmww msmm on» com mwpaEmm mmmwpow woo» cmocmpnm umxuwa1ucm; now was“ pmcpmmm mwpaEmm uocammzm Lox enema? cams cmugmwmz mchuu.m og=m_m sum awn fin mun OBLHSIBM W085 SVLSNI NVEW SB'ldWVS lBNdBEMS 25 to the maturity of the field population. This conclusion is based on samples taken over a very narrow range of the field season. While a comparison of maturity in sweepnet samples and foliage samples for a major part of the field season could not be directly made, the maturity measured for each type of sample was positively corre- lated with OD>48, therefore, one could infer that over the whole season a relation between maturity in the two samples could be found, and indeed such a relation was found, and is presented in a later section. Before this could be done, however, values for OD>48 at weather stations near the survey sites were required. Estimation of Heat Unit Accumulations Near Survey Sites Baskerville and Emin (1969) had demonstrated that degree-day values computed from daily maximum and minimum temperatures Closely approximated actual seasonal values when a sine curve was fit to the maximum and minimum and the area under the curve, but above the developmental base temperature, was computed. Therefore, daily maximum and minimum temperatures for all National Weather Stations and affiliates in Michigan were obtained from the National Weather Records Center in Asheville, North Carolina. Estimating Missing Temperatures The maximum and minimum temperatures were read from magnetic tapes, sorted by year and division, and stored on another tape. This left the problem of what to do in the case of missing data which 26 complicates further analysis. Therefore, a program was written which would estimate values for missing data by using data for days before and after the missing day(s). When, for a particular date-location, a missing temperature was encountered by the program, data for 5 days prior to the missing date and for 5 days after were used in a linear regression with corresponding data from each of the other stations in the same geographic division. The station with the highest coefficient of determination, r2, was then used to estimate the missing value. If contiguous data were missing, the whole procedure was repeated. To test the quality of this technique, it was applied to data which was known but which had been removed so that it would be estimated by the procedure. This data is summarized in Table 3. The mean of the estimated values obviously approaches the mean of the samples values, but the value of any particular estimate might be quite seriously in error. This is indicated by the large value of the standard deviations. Note that while the mean error is not significantly different from zero in most cases, there is a tendency for stations with large errors in prediction when one day is missing, to also have large errors when several days are missing. A major problem in this approach could be the passage of weather fronts which would cause a significant change in the tem- perature at a particular station, but that change in temperature would occur on another day at the statiOn to which it is being correlated. Information on hourly wind movement, which was 27 TABLE 3.--The average and the standard deviation of the difference of the estimated and the actual daily maximum or daily minimum temperature. FOR MAXIMUM FOR MINIMUM Station Number Standard Standard Sampled Mean Deviation Mean Deviation a. A single missing datum. Adrian 20 0 2.13 .40 3.41 Boyne Falls 20 .79 4.08 -1.53 2.20 Baldwin 20 - .21 5.77 —1.26 5.90 Bad Axe 20 - .89 1.88 - .16 1.70 Battle Creek 20 .33 2.59 .21 2.94 b. Two contiguous missing data. Adrian 40 .31 2.10 - .54 2.75 Boyne Falls 40 - .19 3.50 -1.32 4.06 Baldw1n 40 -1.68 5.98 - .64 7.13. Bad Axe 40 -1.00 2.25 - .08 2.44 Battle Creek 40 - .17 2.92 .35 2.84 C. Ten contiguous missing data. Adrian 90 .39 2.64 - .24 2.57 Boyne Falls 90 .11 4.25 .07 3.65 Baldw10 90 - .17 6.24 .50 6.62 Bad Axe 90 - .26 3.02 .15 3.40 Battle Creek 90 - .06 3.48 - .40 3.79 28 unavailable in this study, would be necessary to test the validity of this suggestion. This regressing technique was probably adequate for this study, since accumulations in degree days, not the temperatures themselves were being used. However, in future work, other tech- niques should be sought which would yield comparable results from fewer calculations. Temperature Prediction Between Weather Stations While ultimately it is desirable to determine the abiotic conditions existing in the insect environment, in this particular study the objective was to determine the relation of one of these parameters, temperature, between standard weather stations. Future research within the cereal leaf beetle research group at Michigan State University will lead to an improved understanding of the relationship between temperatures at nearby standard weather sta- tions and the insect's environment. Since weather information for the aviation weather network will be available for pest management purposes in the near future, the relationship between a number of airport stations and second- order temperature reporting stations in Michigan was investigated. The approach here is similar to that just discussed under Estimating Missinngemperatures but here the intent is different -—estimates over the whole season are important, and some under- standing of how the stations are related is sought. 29 Daily maximum and minimum temperatures from each station were used. In Fig. 6, the frequency distribution of correlation coefficients for a sample of the data on maximum temperatures is presented. In Fig. 7 a similar distribution is shown for the mini- mum temperatures. Note the high mean and low standard deviation in each case. For each weather station, the data used in the corre- lation were from day 91 to day 210, the CLB growing season, for the years 1964-71. Therefore, about 960 observations were available for each station. At various times and for various purposes, different subsets of these correlations were used. In one case temperature data from 14 airport stations were used in a multiple regression to estimate temperatures at stations near CLB survey sites. The stations Chosen for nearness to survey sites were Dowagiac, Adrian, Hillsdale, Allegan Sewage Plant, Owosso Wastewater Plant, Gull Lake Biological Station, Caro, Newaygo-Hardy Dam, Gladwin, Fife Lake, Petoskey, and Rogers City. Coefficients of multiple determination, R2, with 14 predictors, ranged from .783 for Allegan to .972 for Newaygo. How- ever, as was expected from the high correlation coefficients between all stations, Fig. 6 and Fig. 7, most of the information is in the first station used. In Fig. 8, this is convincingly demonstrated by a plot of the average R2, for the 12 stations considered, over the number of airport stations used in the regression. Choosing the best and the worst from the above stations, Newaygo and Allegan, respectively, a somewhat more extensive analysis 3O 6 4 4 . . . . . 0 N, No no No l >NZN=oNNL N>NN > > > > 7. ON on ON 0. o n .8822 Correlation Coefficient between maximum temperatures from 26 weather stations Michigan (752702-14). 1n Figure 6.--Frequency distribution of the correlation coefficients 31 NF. ON. OF. NO. NO. . u >O2NOONNO N>NN > > > > > m.m Om OO ON ON ON O u .OONO ”u.m ac mm re “wmw ST. DEV. = .0332 = .916 MEAN 300 TOTAL NUMBER ts 1cien Figure 7.--Frequency distributions of the correlation coeff between minimum temperatures from 25 weather stations Michigan (752702-12). 1n 32 .Ammumonmmnv mazewxms Low Eouuon .mmgapmgmaswp Eaewcwe Loy m? m:_p Och .mcowpmpm NF com mmmmxm>m mew mmcw; .cowmmmemmx opawupse ms» cw vwmz mcowpmpm pcoacwm mo Lungs: us» we compucam O NO . m .comumcmEmemn mpawupss mo ucmwuvmemou ogkuu.m mgzmvu N cum: mzo_._.<._.m thou—Ed no mmmiaz N. N. .01. . ._. . m . 1.. . N . O... Emu. 1 mm. 33 was performed. First of all, the data were split into 15-day inter- vals, and a separate regression computed for each of these. The R2 values for different numbers of stations are presented in Table 4 for maximum temperatures. The seasonal values are included at the bottom of the table for comparison. The trends are the same for these two-week periods as for the season--that is, one station has most of the information with one or two additional stations being perhaps worthwhile when the initial station is poor. Noteable is the fact that R2 for the season are consistently higher than for the 15-day periods using the same number of airport stations. This is probably caused by the restricted range of temperatures over the two-week period as opposed to the season. The order in which sta- tions were eliminated from the regression changed with the 15-day period being considered but, again, because of the high correlation between all stations, it makes little difference which stations are used. The data for minimums were similar in all respects. Season's data, days 91-210, for Allegan and Newaygo, was computed utilizing one other multiple regression sequence. Minimum temperatures for day n were predicted using minimum temperatures for day n and maximum temperatures for-day (n-l) at a set of several other stations. Maximum temperature on day n at a given station was predicted as a function of both the maximum on day n and the minimun on day (n—l) at the same set of stations. This is in marked con- trast to the earlier set of regressions were only minimums were correlated with minimums and only maximums with maximums. Only 34 mmomn mnomn omomn comm. PFNm. Opum. oFNm. m_Nm. mFNm. m_Nm. mPNm. mpmm. m—Nm. m_Nm. momo moNN Nmom «Nos MMNN. oNNN. omNN. mmNN. moms. m_mN. mFmN. mNmN. Ommn. onN. onN. mNF o_N-_m comaom momm. ”mum. mHmm. cumm. chm. moom. Room. onom. mnom. Hmom. omom. omom. “mom. Nmom. momo omoe. mmmm. nvfio. mmco. «Noo. omoo. mooo. mmoo. mono. mHNo. mNNo. Homo. mono. OONo. mNN oPN1om~ nomm. mmmm. mncm. mmcm. comm. flNom. oNom. omom. Hoom. «com. mvom. «mom. omom. Room. momm moon. mNoc. mono. Homo. mmoo. comm. mNmo. eomm. mumo. Nmmm. ono. NNmo. ommo. ommo. mNH omp-~mp coum. Room. NHHm. omHm. HmHm. qmflm. ofiNm. oHNm. NHNm. mHNm. oNNm. oNNm. HNNm. HNNm. momo nmNo. omoo. Hmom. NNmo. moHo. mNHo. oONo. «Hmo. Nmmo. ommo. ammo. ommo. mmmo. mmmo. mNH omp-oo~ «ohm. mon. momm. mmmm. mflom. Omom. mmem. mmom. ooom. moom. «mom. omom. moom. ooem. momo NmHo. mfimo. Hooo. momo. comm. mNoo. mNHo. moNo. emmo. momo. ommo. NNmo. Nmmo. Nmmo. mNH mop-pop mmom. Nmmm. «com. omem. omom. ocom. Room. momm. moom. Hoom. moom. Noom. moom. momm. momo onco. “mum. vHoo. mNoo. NOHo. memo. ommo. oumo. “Hoo. moqo. NNOo. mNOo. omoo. mmOo. mNH ompaomp NNmm. mon. Hon. HmNm. oon. mon. NmNm. momm. onm. HNmm. ONmm. mNmm. Hmmm. Hmmm. momo moco. Hfimo. memo. mvoo. mooo. vofio. NoHo. moNo. mNNo. oONo. ooNo. ooNo. Homo. Homo. mNH omF-PN_ ooom. Nxflm. oon. HmNm. Nomm. mHmm. onm. Nmmm. Hcmm. oomm. oomm. ommm. oomm. oomm. momo Nomm. oomo. mmoo. oooo. omoo. mHHo. HNHo. HmHo. NmHo. omHo. mmHo. Nomo. ooNo. Nomo. mm“ oN_-oo_ onmm. omom. mmom. omom. Naom. mmom. moom. ooom. NNom. «Nom. «mom. oNom. oNom. oNom. momo oooo. vomo. mHNo. HNNo. mmNo. Nomo. Nomo. mNmo. Nomo. eomo. nomo. momo. momo. momo. mNH oop-pm H N m c o o m m m oH HH NH ma «H cowumum gum» mcowumum mo gmoEaz mo Oxao .:e=_oo Ogmo> mo manoO Loco: were oo nopxmm map Loo moan morn: mcowuoaao coo mEO Na .mcowuuaco cowmmmgmmg on» or own: mom mco_umpm Locummz mo mxonEO: acmgmmmou coo; .Asao huge: N OOO ON OOOOOO--.O NOmox -uommOzmz mm momo .ucmpa mmmzmm como—_< mo mNFV mcomuoum Lmzuuoz oz“ um mmgoumgoosmu Esspxue gov N 35 marginal improvements in predictability over using corresponding temperatures on the same day were observed. In Table 5, the sequence of R2 are listed. Clearly, including information about yesterday's temperature in the form used here is unlikely to be worthwhile. Using the sine method for estimating 0D from maximum and minimum temperatures (Baskerville and Emin, 1969), the day of the year on which the OD accumulations was nearest 700 0D > 48 was determined using actual temperatures for Allegan and for Newaygo for the years 1964-1971. Three sets of maximum and minimum temperatures were then estimated for these two sites, using regression equations with temperatures from 14, (13 for minimum), 2, and l, airports as the independent variables. These 3 sets of maximum and minimum tempera- tures were estimated from regression equations derived from pooled data for the whole growing season. Three similar sets of maximum and minimum temperatures were generated using regression equation which changed every 15 days throughout the growing season. From these maximum and minimum temperatures, and again using the sine method, 00 > 48 accumulated up to and including the day on which the actual value was nearest 700 were computed. Six different estimates of 0D > 48 accumulations to a specific point in time were thus generated for comparison with the actual value. All of these are presented in Table 6. Whether the deviations of estimated 00 > 48 values as pre- sented in Table 6 are serious or significant depends of course on 36 TABLE 5.--Changes in R2 with changes in the number of variables used in the analysis for weather stations at Allegan and Newaygo, maximum and minimum temperatures. Variables were maximum and minimum temperatures at other weather stations. Minimum Maximum Variables Allegan Newaygo Allegan Newaygo 27 .900 .942 .826 .972 26 .900 .942 .826 .972 25 .900 .942 .826 .972 24 .900 .942 .826 .972 23 .900 .942 .826 .972 22 .900 .942 .826 .972 21 .900 .942 .826 .972 20 .900 .942 .825 .972 19 .900 .942 .825 .972 18 .900 .942 .825 .972 17 .900 .941 .825 .972 16 .900 .941 .825 .972 15 .900 .941 .824 .972 14 .900 .941 .824 .972 13 .899 .941 .823 .972 12 .899 .941 .822 .972 11 .898 .941 .821 .972 10 .898 .941 .821 .972 9 .897 .940 .820 .972 8 .896 .940 .819 .972 7 .895 .939 .818 .972 6 .894 .937 .816 .971 5 .893 .936 .814 .971 4 .892 .935 .808 .970 3 .889 .930 .806 .970 2 .881 .928 .795 .968 l .864 .925 .734 .954 37 TABLE 6.--Comparison of 0D 48 accumulations calculated by the sine method from actual daily maximum and minimum temperatures, and from daily temperatures estimated from multiple regression equations. Number of Airports Used to Make Temperature Estimates True Allegan True Newaygo Year Value 14 2 1 Value 14 2 l 1964a 696 586 630 632 708 697 701 703 b 600 595 587 702 715 703 1965a 700 963 1025 1005 704 700 708 725 b 926 956 962 709 710 724 1966a 703 732 735 956 691 687 699 715 711 736 727 692 701 711 1967a 689 647 640 672 700 710 706 697 b 642 652 649 716 705 695 1968a 711 670 680 684 705 695 706 710 b 668 653 642 693 706 703 1969a 701 691 647 635 707 660 650 643 b 660 658 659 673 653 649 1970a 708 655 651 660 698 693 692 715 b 634 640 654 694 696 707 1971a 699 666 640 642 696 666 641 652 b 678 644 655 659 644 640 a One equation for season--day 91-210 b Equations modified every 15 days for days 91-210 38 the precision with which such values must be known. That will be considered in a later section. An approach which was not followed here, but which seems promising, is to just find the degree-day accumulation at stations, then find the regression equation between degree-day values instead of between temperature values. As shown in Fig. 9, values for degree-day accumulations are highly correlated between weather stations. This approach was not pursued further in this study since it was felt that actual temperature values were much more useful for many of our purposes, such as models with temperature dependent functions. Since the weather data was now available, and the magnitude of the error in extrapolating from a weather station to another area had been considered, there remained to be found the relationship between OD>48, and maturity and population density in the field. Times of Observed Peak Larval Densities in Survey Samples Fig. 10 shows that any direct correlation between seasonal population densities and OD accumulations for near-by weather sta- tions is very poor. Differences of about 12500 between the time of occurrence of peak larvae in the same county in two adjacent years, and as many as 20000 between peak larvae in the same year in differ- ent counties can be seen in this figure. Table 7sumnarizes the oat field data for Michigan counties for the years 1967 to 1971. Degree- day values were computed for the nearby airport stations listed. 39 .AoNumooNoNv mommcmo op coppmpmx cw ommazwz .o "Paco—mmo> op compmpmg cw commPF< .x .compmum assume O on page op cowpmmmg cw cowumpm moo um cowummzssuum me A oo--.m «camwm $2525 .0 523.2» .x. 2255 5:2”; ESE: 2 33. 000. 000. 00¢. CON. 000. com com 00¢ CON 0 . A .08 X nu .oocnm .vw x 1 o .08 an nu 1; V mm. x .. com 0W . m N nu 1009mm x VV mu. 0 LooN.mN nu x .02: o .ooo. 4O .AON-NONNONV OLOOx N Low mmwuczoo cmmmzuwz N com mmopm> oo acmsmmwwn pm mama so mmwpwmcmc compmpooom--.o_ mg=m_m 9304: :9. 0550.. 83 mo“. 045438 .2 259. Ohm DNQ ohm ONO nbh huh 050 0N0 ohm man 0%? u c J u d I 1 Lave .Auo .Aum 1 IAXUP moou Nxm.n_ .sm.- ozoo48) of observed peak larval densities in survey samples from Michigan counties (years 1967-71). 0 were calculated using temperature data from the airports listed. 0D > 48 Day of year County Airport Mean SD Mean SD Caro Battle Creek 707 175 164 4.8 Lenawee Jackson 984 203 176 7.2 Jackson Jackson 908 240 172 10.9 Allegan Grand Rapids 698 171 167 8.0 Macomb Detroit City 1005 94 176 4.3 Shiawassee Flint Ap. 728 129 172 6.6 Tuscola Flint 790 147 175 5.6 Newaygo Grand Rapids 791 118 171 6.5 Gladwin Houghton Lake 828 258 185 19.2 Grand Traverse Traverse City 681 94 174 7.5 Emmett Pellston 648 80 183 6.6 Presque Isle Alpena 690 320 184 19.2 OVERALL MEAN 788 175 42 Both the mean 00 value and the mean day were computed. In each case, the precision in using days or 0D is about the same, if you assume that about 2000 per day are accumulated at survey time. That woyld result if daytime temperatures were, for example, 80 while night time temperatures were 56, which are reasonable values for that time of year. Note in Table 7 that the order of the survey sites is more or less southerly to northerly, and that, while on a 00 scale, the time of the observed peak density tends to decrease from south to north, the opposite is true on a day scale, and the time of the . observed peak density tends to increase from southerly to northerly counties. The observed peaks are of course strongly affected by the timing and spacing of the relatively few, usually 3, observations in each county. They might also be affected by insecticide sprays at high population density so that, after treatment, what was an increasing population becomes a decreasing population, with a peak which is displaced with respect to the natural population. This effect cannot be recognized as being caused by a pesticide through an examination of the survey data itself when the number of sampling times is as low as 3 or 4, even when data for individual fields are examined. With so few points on a nonlinear curve we simply don't know which are suspect. Since there appeared to be gross errors in determining the time of peak larval density by using the relationship previously 43 found in the survey between 0D > 48 and peak density, the approach discussed earlier of finding the relationship between 0D > 48 and population maturity, as measured by foliage samples was pursued. First the relationship between maturity estimates from survey samples and 0D > 48 needed to be developed. Population Maturity in Survey Samples as a Function of 0D > 48 Because of the greater ease in determining percent 4th instar larvae in a sample than in determining weighted mean instar, the high correlation between the two measures (Table 2A), and the poor relationship between the sweepnet samples and the actual matur- ity in the field, percent 4th instar larvae was adopted as a measure of maturity rather than continue with weighted mean instar. Since the larvae from all fields within a county had been pooled for the 1971 survey by the time this study started, only one estimate of maturity for the whole county could be made for that years collection. In contrast, maturity estimates by field was available for the 1972 survey material. However, in each of 1971 and 1972, only one estimate of degree-days was available for the county for the sampling date. That is, only data from a nearby standard weather station was available, not information on individual fields and the variation among them. Fig. 11 shows the great variation in the observed maturity for several different fields from two counties. No consistent trend was found for individual fields. This may be partly due to the designed-in narrow range of sample 44 .AeNuNoNNmmv Nmmp cw mcomumpossuum me A no pcmgmwmmv pm on cmzumpu vow “xv osoumz mmwpcoou cm mupwmm Fmgm>mm sogw mm—asmm cm wm>gmp xmumcw one u:mugmm--.- mgzmmm 2. A O. omlo £0 00.0 0405 owh 0mm 040 0mm x 11x 000 0 .. 0. .. cm L on 1 0w 1 on 1 0.0 1 0h .. 00 100 I 00. BVAHV‘I HVLSNI “H? % 45 degree-days, that is, an effort is made to sample at 700 to 800 0D > 48 in the survey. The pooled field data for 1971 is shown in Fig. 12, with a different symbol used for each county. These larvae came predomi- nantly from oats. While counties labeled 8 and C show consistent trends, such is not the case for most of the counties. Even counties 8 and C would show no trend if it weren't for the few samples col- QT lected at 50000 or less. The least squares regression line is given, I partly to show how poorly it fits the data. “II Fig. 13 shows the comparable set of data for oats for the 1972 survey. Again, a few counties show good trends, notably H and L, but the overall relationship is poor. Shown are the least squares regression lines for this set of data and for the 1971 data. The lines are rather similar, however, and predictability is obvi- ously low for either line. The data for wheat for 1972 are shown in Fig. 14. The regression equation is y = 15.3 + .0837x with r2 = .31, and standard error of the estimate S.E. = 14.6. Predictibility was much higher when the two aberrant points, circled in Fig. 15, were deleted. That regression line was y = -20.86 + .138x with an r2 of .72. The higher predictability in wheat than in oats is probably due to the fact that wheat, being a winter grain in Michigan, is available for cereal leaf beetle oviposition as soon as the adults appear in the spring, while oviposition in oats is affected by the vagaries of planting time, and the relative attractiveness of the two crops, wheat and oats, in the early spring. 46 .AomTNoNNonv mpcaou acmgmmm_u a mo xmpump comm .mswu mposmm cm>wm m Lo» ompooo mew: mucoou comm Low mcowuumppou upmmm .Oumo mm>xom pump cw mO>xOF Loumcp sue ucmugmmun.mp meomwm OVAOo OON. 000. 000 000 1 g CON 1 . x N a x 0 q . . O on. I “L ..NN n .m.m xmmmo. + 0n.».: n a EVABV'I UVLSNI "I? % 1 00. 47 ..NO-NONNONV xguczou acmxmmmpo O Ow mmmum. ngm .mswp mpmsOm =m>mm O Lem um_ooq mmmz mpcaou comm Lom mcowuuwppou Upmwm .wumv xw>L3m thp cm mumo cw 00>LMF LMHm2w zuc H:00L0¢11.M~ mgsmvu me A O. 000. 000 00» «I q . 000 000 000 u I . Au 1 .9 Low .0» 2mm 9. a n. A 1.06 x00¢0. + «N6. u a .4 .. 1 cm wz... 5.40 th. o u. 2 1 1%. Oh i! BVAHV'I UVLSNI 9H? 1: m2... 4:3 .5. .. om 100 1.0n: 48 . .ANmumoNNoN. co_mmmgmmm mop c. omma mo: mom: mmmupm. mmpmmmo mpczom pcmmmmmwm O m. Lmupm— ngo .mE.m mFQEOm :m>.m O com umFooo mom: mucsom omOm com mcowpumFFOO upmwm .ONOO mm>xsm Num. c. pOmgz a. mO>cOP gOpmcw cue ucmummm--.m. mmomwm men O. anh 000. 000 000 000 000 .1 J O x . a AV 1.0. 1.9N 1.00 Invc .un .uw G .uh BVAUV'I HVLSNI A"? % om 2.. . N. «.6 . .m.m 39. 4 8.8- . x .00 nXu. 49 The relationships established above do have a considerable variance associated with them. It seems likely that much of that variance could be explained by having a better estimate of 0D > 48 for each county, but such information is not currently available. Therefore, for this study, the above established relationships between maturity in the survey samples and 0D > 48 in the field, as estimated at a nearby standard weather station, will be used. Having accepted these relationships between maturity in survey samples and 0D > 48, there remains to be found an estimate of the acutal maturity in the field as a function of 0D > 48. Population Maturity in Foliage Samples as a Function of 0D > 48 A large amount of data forthe intensive population study at Gull Lake and Collins Road were available from theses (Helgesen, 1969; Gage, 1972, 1974). Precise descriptions of how samples were collected for this data are given in these theses, but, in general, a number of 2-linear foot row samples were collected and all of the eggs and larvae on the stems were picked off and counted. The instar of a larvae was determined by visual inspection, and this data is therefore suspect, but again, 4th instar larvae can be separated from the others with a high degree of accuracy. (See page 16.) Using that intensive sampling data, in Fig. 15, the portion of the population which was in the 4th instar at the time of samp- ling is plotted over the degree-day value at that time. Each 50 ll ILIJZ .AOO-NONNONV .2 .N .OOO. "O .O .OOO. 2 NO .O .OOO. n O ”o .m .onm. u x .3 .m .mom. u o .2 .o .oum. mo .o .onm. < cmsh .o u OOom mcwppou ccO .m ucm m u meo ..:m mLO mcoquuo. ucO .o n mpOo .3 u uOmzz mLO moose msmcz .mcomeconom coprmopucOmmTQOOO ucmxmmmwm mmO mmmmumo .uOom mcwppoo ucO meo ppoo seem mmOmm acmmmmmwu :. mOmem .Omm>mm seem mmFoEOm mmOmpom cw mO>mOp mOumcw cum mcmmxmm--.o. mmamwm mvo.. 00». 00: 00a . 00h D00 00m 1|: . q u q .n— q. - O 4 .. O.. O .o. c. u x 4 O .8 c. .2 22v. nonwb x .c 9. 9.5 m mw 4 1.9? ux .4 mm a a O O m m .00 av x U x c‘ o .3 now _.I w < ._oN. mm m a: 1.00 ..um .nXu. 51 letter represents a different field-year-crop combination. To ease defining the letters in Fig. 15, let the crops be wheat = N, oats = 0; locations be Gull Lake, two different fields = G and R, and Collins Road = C. Then in Fig. 15, A = 1970, G, 0; E = 1970 G, N; J = 1969, R. W; K = 1970, R, 0; L = 1968, C. 0; M = 1968, R, 0; N = 1968, R, W. Many more fields are available but this sample is typical of the whole. Clearly the relationship is remarkably good ”‘j“1I_ for any one field in any one year. For example, for an oat field at Gull Lake in 1970, labeled "A" in Fig. 16, Y = -52.0 + .093x, r2 = .95, S.E. = 6.3. It is also clear that values in oats (A, K, L, M) tend to occur later than similar values in wheat (E, J, N) and that differences between Gull Lake and Collins Road (L) are no greater than differences between two fields at Gull Lake in the same year (A and K). Note in several fields that the last sample shows , a lower percent 4ths than does the previous sample. This could quite easily happen when the peak density of larvae passes through the 4th instar and pupates, leaving a considerably smaller popula- tion of younger larvae on which to base the maturity estimate. In any case this normally occurs late enough in the season (>1000°D) that it is mostly of academic interest, since surveys would not normally be conducted so late. It would, however, affect the slope of any linear regression used to characterize the data. For example, for J, y = -92.4 + .173x, r2 = .83, S.E. = 10.2 with all points (x > 500); but, y = -126.9 + .226x, r2 = .96, S.E. = 5.2 with the last point (895, 45.5) deleted. 52 Figures 16 and 17 are the complete data set for both Gull Lake and Collins Road for the years 1967 - 1973. Fig. 16 is for oats, Fig. 17 is for wheat. For each of these, linear regression lines were calculated using data pairs with x > 500 to avoid the initial zeroes. For oats the regression equation is: ‘< II -47.5 + .0918x, (r2 = .54, S.E. = 16.9) (9) T1 for wheat: -69.6 + .141x. (r2 = .45, S.E. = 22.0) (10) '~< 11 These are labelled "A" on their respective figures. It is clear that a great deal of precision is lost in directly pooling the data, and in the future an attempt must be made to determine the causes of the changes between fields and years in the slopes and intercepts. In Figures 16 and 17, the regression lines for maturity in survey samples as functions of 0D >~48 are drawn for comparison to the regression lines for these data (maturity in foliage samples) as a function of 0D > 48. The line for the 1972 survey data is labelled "8" while that for the 1971 data in oats is labelled "C." Maturity data for wheat were not available for the 1971 survey. Using the regression lines and reading from the vertical axis in Fig. 17, at any given value of 0D > 48, the difference between maturity in survey samples and in foliage samples is about 45 percent 4th instar larvae. Similarly in Fig. 16, the difference in percent 4th instar larvae between the regression line for the 53 .AmmuNoNNoN. commemsom mom :szc mLO OpOo mm>mom Am. mum. ucO .o. .nm. onm mmcw. cowmmmmmmm .OuOo OOom we...oo ncO mem ppoo .mNmFTNom. mLOmm .mpOo c. mmpoEOm mmOw—om poo» xOmOPFTN cm mO>xOF LOumcw gum mcmmmmm--.o. mmammm me A 0.. con. 00: com 02. can com I x d x u N 0414711 J O . .. o. 0 O O 1 ON 0 O 0 525m .5. . on m. o 0 K .7 C . Iu'v . 00 .000 A x. of m , . >m>mom N5. 8 *0. I N5 0 0 I 00 w 0.0. n .U.m ’00 a x050. +oficux> o .00 .I . V . H . 2. m a; O my a 4.0m 4 “whuanv L.90 U o o .. OO. 54 ..OO-NONNONV mFOOpr>O mo: mmmz >m>mom an. mom mow uOmzz mom OpOu mop .OOOWLOOeou mo. czOLv m. . ONOO xm>m=m Am. Num. ms» so». mow. cowmmmmmmm mob .OpOO uOom OOPPFOo OOO meO _.:o mumpuuom. mmOm> .pOmgz c. mmPOsOm mmprom poop LOmcwpuN c. mO>xOP mOme. sow ucmommmuu.m. mczmwm men O. 00». 00. . 000 00.. 000 00m GI u d I d u d 4 . d . d o 1.0. ..ON . O» O. .v I. .:o.. “x mm . O. 5 av . 8 .ooonx. .00 W 9.. .. N. 525m «3. M 0.3 . .mm . 2. W. x.o;..o comm..n_> 1.00 5mm; . O. ,< o o o m o .. OO. 55 maturity data from the 1971 survey data (line C) and the maturity from foliage samples (line A) is about 30%. The line of oats for the 1972 survey data (Fig. 16, line B) gradually approaches the foliage sample line, so the difference between them, in terms of percent 4th instar larvae gradually decreases. However, over the range of 0D > 48 at which the survey normally takes place, 600-800 0D > 48, the vertical distance between line A and line B is approxi- mately the same as that between line A and line C. These differ- ences in the vertical positions of the regression lines for the survey data and for foliage samples are further strong evidence of the strong bias of the sweepnet for large larvae. In equations 9 and 10 are the means for estimating the maturity in the field at the time the samples are taken, given that the 0D > 48 value for the sample time can be determined. This was the second of three requirements mentioned in the introduction as necessary in order to estimate the total seasonal population. The third requirement, an estimate of the maturity distribution through the season will be developed in the next section. Population Density of an Age Class as a Function ofTOD > 48 Data on the density of cereal leaf beetle eggs and larvae were available in the intensive sampling work from Gull Lake and Collins Road as mentioned earlier. Individual observations of density of a particular life stage were converted to cumulative percent occurrence of that life stage in a particular field, as a 56 function of 0D > 48. These gave sigmoid shaped curves, which were then transformed to probits, and linear regression lines fit to the data for each field. The slope, b, and intercept, a, of each of these regression equations is listed in Table 8. Solving each of these equations for Y = 5, we have the mean or 50% point for each set of data, the degree-day value at which peak occurrence of the life stage occurred (Table 9). Going one step further, we have, also in Table 9, the day of the year on which the peak occurred. For Table 8, means and standard deviations for wheat and for oats are at the bottom of the table. From these the 1% occurrence point and 50% point were computed for each of the life stages. The lines are shown in Fig. 18 for oats and Fig. 19 for wheat. Gage (1974, p. 79), using mostly the same data, computed 50% points for oats of 467 i 35 for eggs, and 723 i 25 OD>48 for total larvae based on observed peaks. These compare very well with the 503 for eggs and 695 0D > 48 for total larvae found here. Using Fig. 18 for oats or Fig. 19 for wheat, the pOpulation of eggs, any instar, or total larvae which has already occurred in the field by any given value of 0D > 48 can be read directly from the vertical axis. The portion of the population which is still in the field at a given 00 > 48 value, 00, can be estimated by the difference of the proportion of the population which had developed at 00 minus one half the length of the life stage, and 00 plus one half the length of the life stage. This is because at the time of 57 TABLE 8.--Regression coefficients for the probit of the cumulative density of the life stage listed as a fenction of °D>'48. For Location, 6 - 6011 Lake. C - Collins Road. Crops are I - Hheat and 0 - Oats. Instar Total Egg lst 2nd 3rd 4th Larvae Place Field Year Crop a b' a b a b a b a b 4 P G 67 H 3.84 .428 - G 67 0 .12 .631 1.06 .583 1.54 .520 - .31 .658 -7.69 1.302 .74 .581 G 68 H 2.37 .598 - .84 1.228 - .44 .994 - .39 .984 G 68 0 2.32 .616 .15 .815 - .86 .875 -3.45 1.111 -4.42 1.197 - .68 .831 C 68 O 1.89 .751 -1.92 1.205 -2.73 1.241 C 68 H 1.73 .950 .09 1.218 -l.32 1.387 G 69 H 1.88 .788 .17 .914 -I.38 1.064 -2.86 1.205 -4.88 1.407 - .73 .947 G 69 0 1.95 .593 - .54 .887 -1.24 .887 -2.20 .942 -3.16 1.071 -1.17 .853 C 69 H 2.88 .626 - .47 1.002 -1.49 1.050 -5.10 1.489 -9.24 1.984 -1.52 1.042 C 69 0 .76 .691 -1.00 .897 -2.38 .987 -3.47 1.002 -5.49 1.148 -1.07 .779 G R 70 W 2.22 .817 - .18 1.093 -1 18 1.172 -l.35 1.016 -1.85 1.034 .09 .849 G R 70 0 2.55 .524 .57 .713 .14 .686 .02 .664 -2.27 .870 .07 .677 C 70 H 2.17 .683 -I.69 1 190 - .06 .789 -2.20 .985 -6.49 1.417 - .87 .839 C 70 O 1.13 .674 - .47 .754 -1.17 .800 -3.02 .966 -6.27 1.299 -I.78 .854 G T 70 W 2.93 .618 .86 .933 .73 .822 - .13 .883 -l.06 .948 .77 .779 I G T 70 0 2.48 .516 — .36 .820 .18 .689 - .28 .694 - .83 .723 - .31 .717 G R 70 H 2.61 .649 G T 70 H 2.23 .797 G T 70 0 .90 .751 G R 70 0 1.51 .652 C 70 H 1.83 .698 C 70 0 .42 .784 G T 71 0 4.22 .350 .04 1.058 1.50 .721 1.63 .671 3.26 .412 1.36 .639 G R 71 0 -2.98 .585 .55 .830 .41 .816 -2.12 1.069 - .41 .873 - .53 .914 G B 71 0 3.02 .536 1.28 .720 - .80 .927 -1.95 1.028 -2.04 1.023 - .21 .842 G T 71 H 3.96 .422 2.90 .508 -2.78 1.680 -1.92 1.227 1.74 .663 G R 71 W 3.03 .671 -2.32 1.289 .25 .960 G B 71 H 3.28 .571 .17 .995 -1.34 1.137 .06 .921 G A 72 O 3.91 .403 1.16 .706 .71 .746 - .32 .824 -2.57 1.054 .18 .778 G B 72 0 2.70 .440 1.52 .569 1.11 .584 .64 .617 - .76 .765 .40 .657 G A 72 H 2.16 .815 -1.24 1.296 -2.83 1.470 G 8 72 W 2.27 .830 1.47 .759 - .65 1.068 - .72 .938 -2.07 1.081 .48 .805 G C 72 H 2.69 .625 1.54 .725 .54 .829 - .48 .894 - .91 .868 1.10 .693 G E 73 0 1.28 .529 -I.59 1.022 -2.10 .912 -2.79 .959 -I.52 .831 G L 73 0 .13 .593 -2.12 .920 G E 73 H - .13 1.283 -I.92 1.248 -5.34 1.679 G L 73 H 1.85 .823 - .01 1.005 -3.51 1.118 - .39 .827 G L9 74 W - .56 1.533 Hheat Mean 2.37 .741 - .48 1.012 - .90 1.076 -I.85 1.080 -3.48 1.228 - .60 .990 S .87 .201' 1.23 .234 1.15 .266 1.61 .211 2.86 .339 1.73 .295 50% Point = Z = 355 541 549 634 691 565 11 Point = 42 311 332 364 498 330 Oats Mean 2.12 .573 .03 .826 - .23 .781 -1.35 .862 -2.72 .978 - .62 .808 S 1.21 .122‘ 1.07 .178 1.30 .139 1.67 .179 2.93 .262 1.12 .158 501 Point = Z = 503 601 669 737 789 695 1% Point = 97 320 371 467 552 408 .All values of b must be multiplied by 10'2. 5553 TABLE 9.--Time of occurrence of peak density in different fields. as determined by solving the regression equations in Table 8 for 00 > 48. such that 00 > 48 = (5-A)/b. The day of the year was found by determining the day of the year on which the given 00 > 48 accumulation occurred at Gull Lake weather station for G or East Lansing weather station for C. Collins Road. H - wheat; 0 = Oats. °o > 48 for 50: Day of the year for 50: Place Field Year Crop Egg lst 2nd 3rd 4th Total Egg lst 2nd 3rd 4th Total G 67 H 271 G 67 0 774 676 665 806 974 734 6/15 6/12 6/12 6/16 6/24 6/14 G 68 H 439 476 547 688 699 548 5/25 5/29 6/ 4 6/ 9 6/10 6/ 4 G 68 0 435 596 669 760 787 683 5/25 6/ 6 6/ 9 6/12 6/14 6/10 C 68 0 414 574 647 701 724 623 5/26 6/ 7 6/10 6/12 6/14 6/ 9 C 68 H 344 403 502 456 5/16 5/25 6/ 4 6/ 1 G 69 H 395 529 600 652 702 605 5/23 6/ 1 6/ 7 6/12 6/14 6/ 8 G 69 ‘0 514 625 703 764 814 7]] 5/31 6/10 6/14 6/19 6/22 6/16 C 69 H 338 546 618 679 718 626 5/17 6/ 6 6/12 6/15 6/18 6/12 C 69 0 614 682 748 845 914 780 6/12 6/15 6/20 6/26 6/29 6/23 G R 70 H 341 474 527 625 663 578 5/18 5/25 5/29 6/ 2 6/ 6 5/31 G R 70 0 468 622 708 737 835 728 5/24 6/ 2 6/ 8 6/ 9 6/13 6/ 9 C 70 H 414 562 642 731 811 700 5/21 5/30 6/ 4 6/ 9 6/12 6/ 8 C 70 0 574 726 771 830 868 794 5/31 6/ 9 6/10 6/13 6/15 6/11 G T 70 H 335 444 519 581 639 543 5/18 5/23 5/28 5/31 6/ 4 5/29 G T 70 0 488 653 699 760 807 741 5/25 6/ 5 6/ 8 6/10 6/12 6/ 9 G R 70 u 369 5/20 G T 70 H 348 5719 G T 70 0 545 5/30 G R 70 0 535 5/29 C 70 H 454 5/22 C 70 0 585 5/31 G T 71 0 223 468 486 502 422 569 5/14 6/ 3 6/ 4 6/ 4 5/30 6/ 7 G R 71 0 345 537 563 666 620 604 5/23 6/ 6 6/ 7 6/12 6/ 9 6/ 8 G B 71 0 370 516 625 676 688 619 5/24 6/ 5 6/10 6/12 6/13 6/ 5 G T 71 H 246 414 463 564 492 5/16 5/30 6/12 6/ 7 6/ 4 G R 71 H 294 568 495 5/18 6/ 7 6/ 4 G B 71 H 301 485 557 537 5/19 6/ 4 6/ 6 6/ 5 G A 72 0 271 543 575 646 719 620 5/19 6/ 2 6/ 4 6/ 7 6/12 6/ 6 G 8 72 0 522 611 666 707 753 700 6/ l 6/ 6 6/ 9 6/12 6/14 6/11 G A 72 H 348 482 532 5/22 5/29 6/ 2 G B 72 H 329 465 529 609 654 562 5/21 5/28 6/ 2 6/ 6 6/ 8 6/ 3 G C 72 H 370 478 538 613 681 564 5/23 5/28 6/ 2 6/ 6 6/ 9 6/ 3 G E 73 0 704 645 778 812 913 784 6/10 6/ 8 6/12 6/14 6/18 6/13 G L 73 0 820 774 6/14 6/12 G E 73 H 400 555 616 5/23 6/ 3 6/ 6 G L 73 H 382 498 761 652 5/21 6/ 1 6/ 8 G L9 74 H 363 -—‘.= " 'Ah-A . . 59 .AmmumonmmnV mumo co mpumma emm_ Pomcmu 8:8 4o cmgmcw Fw>LaF zoom new mama com upon umsgommcmcp pmnoca gee mmcwp commmmcmmc mmmgm>¢<4 JcmF comm vcm mama Lo; mama vmecowmcmcp “Paoca com mmcmp covmmmcmmc mamgm> 48 (Helgesen, 1969; Ruesink, 1972). Sampling oats at 700°D > 48, we have that the proportion of the population between 580 and 820°D > 48 is in the larval stage at that time. From Fig. 18 the proportion at 58000 > 48 is 18, at 820°D > 48 it is 84%. The difference between the two is 66%. That is, 66% of the whole population is in the field to be observed at 700°D > 48. Using the data of Table 9, linear regression analyses were performed using (1) eggs as the independent variable, and lst, 2nd, 3rd, 4th, and total larvae as dependent variables; and, (2) stage n as the independent variable with stage n + 1 as the dependent vari- able, n = instar l, 2, 3. These regression equations are presented in Table 10. This table is farily significant since it shows that a very large part of the variance in the time of occurrence of peaks is contributed by the initial variance in the time of peak eggs. Note for example that in cats, 80% of the variance in the time of peak occurrence of fourth instar larvae is explained by the time of occurrence of peak egg density. Another interesting point brought out by Table 10 is that the slopes of a few of the regression lines computed are significantly different from 1, for example egg to first instar larvae in oats with a slope of b = .390. This indi- cates that for any given year, the later that peak density is observed on a 00 scale, the shorter will be the time on a 0D scale to the occurrence of peak first instar larval density. This could result from the insects growing faster at higher temperatures than If? 62 TABLE 10.--Regression statistics for the relationship between the time (OD) of peak occurrence of cereal leaf beetle life stages. E = egg. L = larval instar, n = number of observations. X Y a b r2 Hozb = 0 Hozb = 1 n 0 TS E L1 418 .390 .72 * * 14 E L2 463 .420 .63 * * 14 E L3 489 .502 .74 * * 14 E L4 387 .807 .80 * N 11 E Total 429 .579 .86 * * 12 L1 L2 45 1.02 .79 * N 14 L2 L3 51 1.02 .86 * N 14 L3 L4 -302 1.48 .90 * * l4 WHEAT E L1 284 .570 .33 * N 14 E L2 356 .555 .32 N N 11 E L3 371 .749 .62 * N 9 E L4 266 1.16 .52 * N 10 E Total 196 1.08 .54 * N 14 L1 L2 66 1.01 .93 * N 10 L2 L3 168 .848 .77 * N 9 L3 L4 52.9 .993 .85 * N 8 *Reject the null hypothesis, 5% level. NDo not reject the null hypothesis. 63 is predicted by using 0D, or conversely, that it grows slower at low temperatures than expected. When the slopes of these equations in Table 10 are signifi- cantly different from 1, it is not strictly speaking correct to say that the time, in OD, between peak occurrence of two life stages is any one value, since the difference will depend on the time, in 0D, at which the peak density of the earlier life stage occurred. Given these reservations, the average time in 0D between the peaks of each life stage are presented in Table 11. For example, on the average from the data set used, it is 250 0D > 48 between peak egg density in oats and peak density of 3rd instar larvae in oats. The comparable value for wheat is 282 0D > 48. Asterisks indicate slopes of regression equations significantly different from 1 as presented in Table 10. TABLE 11.--Average number of 0D > 48 between the peak density of different life stages. 0D > 48 Between Peaks Earlier Later Stage Stage Oats Wheat ”€13239" 33%; E99 lst instar 126* 131 114 124 Egg 2nd instar 185* 201 Egg 3rd instar 250* 282 Egg 4th instar 294 324 Egg Total larvae 245* 223 lst instar 2nd instar 59 64 59 2nd instar 3rd instar 65 84 51 3rd instar 4th instar 45* 49 46 *Slope of the regression line for time of peak significantly different from 1. 64 It is easy to see that each of the time intervals given in Table 11 represents 1/2 of the development time of the earlier life stage, plus 1/2 the development time of the later life stage in that row. Now these data are derived from the differences between points on probit regression lines. They are in remarkable agreement, par- ticularly in oats, with the directly comparable laboratory data from Helgesen (1969) and Yun (1967). The values for wheat are consis- tently higher than for oats or for Helgesen's data. This could indi- cate longer development time on wheat. Given an initial starting point—-the time of peak egg density --the probit transformation of the maturity data produces results which would be expected based on results from other sources. Returning to the data of Tables 8 and 9, the reciprocal of the slope, 1/b was plotted over the value of 0D > 48 at the peak. These values are the standard deviation and the mean, respectively, of the normal distribution represented by the probit equation (Finney, 1971). While all stages on both crops were tested, only 5 = 268 -.169 (Peak OD)S.E. = 33.9, r2 = .46} 282-.231 (Peak 01)), S.E. = 36.0, r2 = .48} had eggs in oats {l/b and L4 in oats {S slopes which were significantly different from 0. There are numer- ous possible explanations for this relationship, one of which is a nonlinear response to temperature so that at higher temperatures, which normally occur later in the year, developmental velocities are greater than would be expected from degree-days alone. A large pr0portion of the variance in the time of peak density of any given instar can be explained by the time of peak egg 65 density (Table 10). Therefore, if the variation in the time of peak egg density could be explained in terms of variation in some physical factor, this physical factor could be monitored and used to minimize the unexplained variance in the time of peak density of age classes. Using the data for wheat in Tables 8 and 9, the average times of the peaks was computed for each year at Gull Lake. The day of the year on which that value occurred was also found. These are the values given in Table 12, together with values for some physical factors which were considered. Note that there is slightly more variation in the time of the peak measured on a 0D > 48 scale than on a day scale. The standard deviation on a 0D scale is 57, on a day scale, 2.9, but there are 13.9 0D/day being accumulated at that time of the year, so we have 57 compared with 40. Again referring to Table 12, none of the "independent" fac- tors, day with last snow or the rates of accumulation of OD, either singly or in combination, were significant in predicting the peak egg density. The rest of the factors in Table 12, in terms of the rate at which they are changing at peak egg density, are no more constant than is peak egg density itself in 0D > 48, and therefore, offer no advantage oVer 0D > 48. These attempts to explain some of the variance in the time of peak egg density in terms of some physical factors have failed. It is an area worth investigating further however, and such will be done in future work. Information that would be very useful in such 66 4.. E .I . . 111. 1 1.11.4 .xmmeo mpaumm um cczocm mg» co zocm umup soc; N .vmccauoo mucmnwucv mom xmmn gown: mo mpuers mg» cw wowcma ameum on» com fl ma. wH. ma. NH. mH. efl. oH. so. 0H. .x\m eHH HmH new woe oHH mmH mmm mem m.m e.m m.~ mm m mom Hum mmma mmem max meNH mmem Name . om. m.ma flea mmm .M mme nooH mme mmnm mes ommH Hmmm Name m.w H.m m.n Ho om n.0m HeH mom en NNm omm Hmmfl moHN omm mnefl mfimm comm m.m N.w N.m Ne ooH n.NH Nee Ham ms mum mem mneH NmNN mmo mafia mmmm name m.NH m.0H w.m me mm o.- meH men mm mom eom mama Hmom mom mHoH eNHN meme 5.“ m.o e.m me em o.eH umH omm Hm eum «mm emNH coma 0mm HmoH mmfim mmme o.mH m.mH m.HH Rm HoH m.mH mmH mem cu mmm mnofi monfi mmmm mkn mHmH mHmN meme H.m m.w o.o mm km . m.e NeH mom mo mam mNNH mccm eNHm mmm mneH NHNN mofim H.m e.“ o.m No em N.HH oeH mme we cum Hmm moofi mflmm NHm moHH oemw mome m.m o.m o.m om Hm m.NH mmH Hum so cexa mmxo cmxo mac cexa meo omxoo Nxoo oce1coH oom-ooH oom-om Nzocm Nzocm moo\o men xmma Luw> o o o o o 0 “we; “mm; H com: com: ” 3ocm “we; sock “mommm amcpo ”cmmZHmm amo\me a so»; new: N o mama xoo .mxo4 szu an poms: cw xpwmcme mam xmma we we?“ we“ use weapon» Feuwnga meow :mngma cowuapmmur.up mgm<~ 67 a study would be the time of peak egg density in wheat at several different locations in one year. In that way, hypothesis about photoperiodic effects could be tested. While variance is quite high, Figs. 18 and 19, and Table 8 do provide infOrmation on the distribution of maturity in the field through the season, again assuming that the 00 value at the sampling time is known. This completes the third and last requirement needed to make an estimate of the total seasonal population from a single sweepnet count. 1 ESTIMATING THE TOTAL POPULATION Since all of the information needed to make an estimate of the proportion of the total population counted at any time during the CLB larval development period has been developed, an estimate of the total population can be made. The following algorithm is employed for sweepnet counts: 1. Then Find the mean number of larval per 100 sweeps for the county. = Y. Find the mean percent 4th instar larval in the samples. = Z. Find the 0D > 48 accumulation on the sample date. = DD. Find, from Fig. 18, the proportion of the whole population that is present between 00-120 and DD + 120, = l/k. Estimate the true proportion of 4ths in the field at DD from eq. (9) or eq. (10). = b. Assuming that the sweepnet collects all of the 4th instars in its path, let total seasonal population number of 4ths observed true total population present true proportion of 4ths = X/A observed population observed proportion 4ths = X/Y N> 48 and the pro- portion of 4ths in the sweepnet samples can be used in place of actual percent 4ths (Step 2). For oats, the average of the 1971 and 1972 linear regression coefficients was used giving: Percent 4ths = 1.46 + .0696 x 0D > 48 (15) Using this approach, one can find the fraction of the whole population that is observed at any value of 0D > 48 throughout the season for wheat and for oats. This was done in Fig. 20 which is convenient for making estimates for a small number of samples. Figure 20 gives the fraction of the total seasonal popula- tion which is observed with a sweepnet. This can be divided into the fraction that is present at a given value of OD:>48 (Fig. 21), and a correction factor which, when multiplied by the sweepnet count, corrects for the fact that sweepnets sample predominantly 4th instar larvae. This factor is computed as: 70 .Ampumofimmnv me A no co>wm o no mucommcooc oposom umcoooZm soc? mp o cows: »u_mcoo :owpo_:ooo mFaooa moop Fomcou on» 4o cowpoocm och--.om oczmwn oehoo o8. oo: ooo. ooo ooo o2. com con 8.. 1 I 1 d - O J.— ud w .3. ...... O N o J -o. m. u 1 .2. d o d n" W .8. u o n" o n.0mW an 33 I no. 5:: .1... M a: -on. a 71 .Amp-~oammmv me A co co cowaucom o mo opowe on» cw pcmmoco comqu=ooo Po>cop moo Focomomm oz» mo compumcu--._m ocommm ocAOo com. 000. com com 00¢ com u q d u d a 11 ‘ o a. 1 .. “a V O nu a. m 0 a. . I. an m V 1 a . d .v nu d n w 10. II... nu N ..0. MW 8 q: 8 .mhazo uxlnlon ..~. mm 0 ham—.3 Tie 80. 72 For oats: 1.46 + .o 696 0D > 48 .o -47.5 + .0918 on > 48. F0 0 > 48 > 518 (16) undefined, 0D > 48,: 518 And for wheat: -20.86 + .138 CD > 48 -69.6 + .141 0D > 48 FW ’OD > 48 > 494 (17) N undefined, on > 48 5_494 These follow directly from eq. 13, and are graphed in Fig. 22. It is clear that at the time of peak larval density a very significant part of the whole population is present in the field. Counts within 50 0D of the peak would represent greater than 65% of the population in oats and greater than 75% in wheat. For sweepnet estimates, the high rate of change of the func- tions shown in Fig. 22 with respect to CD in the 500 to 600 08 range suggest that where possible, sweepnet samples should be made at 0D values higher than about 600 to 650, where the slope is fairly con- stant and low. Using the previously defined algorithm, seasonal population estimates for four Michigan counties for 1972 in oats, and one county in wheat were made. The results are presented in Table 13 where, for each county, the number of fields samples, the mean, standard deviation, and coefficient of variation of the population are pre- sented. Sweepnet counts were made three times in each county, means 73 .Ammumonmmnv mo>cop copmcw cpe com mown umcomozm mg» cow uuoccoo .me A no mo mopo> co>mm m po pcoou umcooozm ocu an owwpowppae cos: .gumsz mcouuou--.mm ocomwu me A a. com 02. can con 1 q a q d d d O L o. W n .. on n d .11. w - on u 0 e N J.— . 0.. W I. O 8 9:6 I L n on 2m; I L oo 74 TABLE 13.--A comparison of the mean, standard deviation, and coefficient of variation for sweepnet population estimates in oats for 4 Michigan counties in l972. Means based on averages of "number of fields" sampled 3 times. Number Estimated Coefficient of Mean Standard of County Fields Population Deviation Variation, % Lenawee a 50 93 65 70 b 50 1061 1083 102 c 50 944 969 103 Macomb a 11 142 122 86 b 11 473 382 81 c 11 913 517 57 Newaygo a 25 303 83 27 b 25 2647 1776 67 c 25 2486 1208 49 Cladwin a 14 741 490 66 b 14 4003 3958 99 c 14 3967 2708 68 Newaygo a 9 10 9 90 (Wheat) b 9 381 503 132 c 9 65 47 72 g_. Actual mean count per 100 sweeps b_.... Estimated total population using mean % 4ths in sample g_. . Estimated total pOpulation using established relationship between 0D > 48 and % 4ths in sweepnet sample 75 here are the means of the means for each sampling date, and the standard deviation and coefficient of variation apply to these means of three sample means. Compare these statistics in Table 13 for the three differ- ent population figures on which they are based: A. Direct sweepnet counts B. Direct sweepnet counts and observed proportion of fourth instar larvae in the sample used in the algorithm and C. Direct sweepnet counts used in the second version of the algorithm which estimates the proportion of fourths in the sample given the 0D > 48 value at which it was taken. These samples are not optimum for testing the algorithm since they were a small part of the data set used in constructing it. However, they were used here since they are several levels of complexity away from being direct comparisons with themselves. The striking thing is the large increase in the estimated over the counted population for both versions of the algorithm. But there is a strong tendency for the estimated percent fourths to give a lower popu1ation estimate than the observed percent fourths in the samples. Low standard deviations and coefficients of variation for the population estimates b and c would have indicated that the algorithm was working properly and predicting constant populations from samples taken at any time during the season. These values are not low and are lower than the value for the direct sweepnet counts 76 only about as often as they are higher. The variation tends to be lower using estimated percent fourths, rows c instead of actuals, row b. In Table 14 for oats and 15 for wheat the average number of CLB larvae per 100 sweeps collected at the 0D > 48 values listed are shown for the years 1967-1972. The population data for oats is from the Jackson County survey, the 0D values are based on maximum and minimum temperatures from the Hillsdale weather station. The population data for wheat is from Shiawassee County while the 0D values are based on temperature data from the Owosso Waste Water plant. These counties show typical results of application of the seasonal estimate algorithm. A general tendency is for over- compensation in the early and late season, giving unbelievably high estimates and not a great degree of constancy. Reasoning that part of t’herproblem might be in the value for 0D > 48 computed for the date, new population estimates were com- puted by adding :5 0D > 48 to each of the original values. This process was repeated until a minimum value for the means of popu- lation estimates for the year was found. The number of 0D which must be added to the original to give this minimum is listed in the rightmost column of Tables 14 and 15. This process had Some rather significant effects on the individual estimates within a year. Unusually high estimates in early and late season still occur, however, This whole process of adding or subtracting 0D to effectively shift the population in time had to be rather arbitrary at this time 77 TABLE 14.--A comparison of population estimates in Jackson County in oats for the years 1967-1972. made using the tabulated data and the algorithm described in the text, and Figure 24. Population estimates were The minimum average estimate is that which can be made by adding a constant to the Weather station used 0D>48 values found for sample times. was Hillsdale, Michigan. Sampled Minimum 00 > 48 Added at Mean Per Estimated Average to Give Year 0D > 48 100 Sweeps Population Estimate Minimum 1967 533 25.73 2030 1483 722 249.30 991 981 899 56.00 444 467 1939 93.91 5035 5504 X 2125 2109 5 1968 451 8.80 21135 65619 706 323.90 1347 2506 1168 453.51 386853 56019 X 136445 41382 -85 1969 554 239.20 6989 555555 775 289.98 1134 1395 1025 1136.12 47821 11806 1282 190.15 4245304 232279 X 1075312 200259 -100 1970 572 155.30 2747 9477 861 1627.37 9389 7557 1009 769.43 24904 14761 1103 195.86 36695 17974 X 18434 12442 -35 1971 701 362.73 1536 1446 863 28.90 169 198 1065 .77 66 99 X 591 581 20 1972 586 61.53 809 366 687 76.63 345 295 817 62.10 276 405 X 477 356 60 78 TABLE 15.--A comparison of population estimates in Shiawassee County in wheat for the years 1967-1972. Population estimates were made using the tabulated data and the algorithm described in the text, and Figure 24. The minimum average estimate is that which can be made by adding a constant to the °D>48 values found for sample times. Weather station used was Owosso Waste water plant. Sampled Minimum 00 > 48 Added 0 at Mean Per Estimated Average to Give Year 0 > 48 100 Sweeps P0pu1ation Estimate Minimum 1967 773 12.02 151 75 998 2.24 5315 43 X 2733 59 -195 1968 802 35.88 691 216 1905 8.13 24402 121 X 12546 168 -220 1969 570 78.55 529 1043 711 129.19 844 671 837 115.42 4101 2056 X 1825 1257 -40 1970 731 109.17 848 1923 998 85.98 204047 1336 1104 13.37 1790321 1619 X 665072 1626 -210 1971 644 100.69 494 512 820 6.93 180 111 X 337 312 ~30 1972 535 43.33 519 234 673 15.33 80 127 X 300 181 65 79 since the relationship between OD's at standard weather stations and in crops is not known. One indication that the corrective values found in Tables 14 and 15 may have some real existence is the fact that values for wheat and for oats are highly correlated (Fig. 23). One further test of the technique for estimating seasonal population from single samples was conducted, that was based on A two linear foot samples from Gull Lake in 1974. Applying factors from Fig. 21 to Sawyer's 1974 data (unpub- lished), which were not used in development of the models, pro- ! duced the population estimates shown in Fig. 24. Two classes of I fields were used: oats and oats mixed with alfalfa. Means for oats alone have a dot. The standard error of Sawyer's population estimate was converted directly to standard error bars in this graph. The data for oats/alfalfa are plotted 10 00 higher than their actual value so that the lines wouldn't overlap those of the oats alone. Clearly there is a tendency for the population estimates to get larger as the season progresses. For oats alone, Sawyer had computed the total incidence of CLB larval to be 6.42 and the total incidence iniflmaoats and alfalfa to be 10.23. The average value of the predicted means on the range 50039D31100 was 11.7 for oats alone and 13.2 for oats/alfalfa. Consideration of Fig. 24 suggests that the correction factors are displaced in time, under-correcting early in the season and over-correcting late in the season. Therefore, the standard 80 .Anp-~onmmuv cowpo>eomno eo coo» asp mew memosoz .xucooo comxuoe n . .Aucooo oommozowsm n x .mpoo com page one pomcz cow comoom «so go» some oposwpmo cowpo_oooo Eoewcws ocu o>wm op moopo> me A so on corpoocsoo mgp :mmzuoo nwgmcowuopmc ochlu.mm mcommm ham-‘3 z. otAOo O... 29.59300 00. on o or... 00.. On.- OON. Oman «1 q q a u q q CON- so. . 2. 8 O O 8 .8... an... O .1... O . on - N l. O 5&0- w > 0 . 1 . 0 0 No.0 I M F08 O N :x 8 ”0. I Nb III .82.. + 6.5 n a . on N w I. S . oo. L OD. 81 481r 3.1 11 301p 25 - 201P ESTIMATED TOTAL POPULATION .6. _ I 2+5; 00 500 660 $0 800 900 1000 1100 ’0 > 48 Figure 24.--Estimated seasonal larval popu1ation at Gull Lake in 1974, based on means and standard errors for 2-linear foot samples from Sawyer (unpublished data) and factors from Figure 22. Means with (-) are for oat fields. Those with (X) are for oats mixed with alfalfa fields and are plotted 10 00 higher than their actual value so the standard error lines do not overlap (752702-16). 82 probit transformation described earlier was applied to the oat and oat/alfalfa data. It was found that the mean for the oat data occurred at 780 0D > 48, and the mean for the oat/alfalfa data occurred at 740 0D > 48. Physically these differences from the 700 00 mean of Fig. 25 are rather small and difficult to predict in field locations. For example, on June 3 accumulated 0D > 48 at the research trailer, on which the data in Fig. 24 are based, was 567. Less than 1/4 mile away at the Gull Lake Biological station accumulated OD > 48 was 535 on June 3. Yet the effect of these small changes in 0D > 48 can be quite significant as is demonstrated in Fig. 25, where the same data used to generate Fig. 24 was used, but the time axis was shifted so that the observed means corresponded to the mean of the correction factor in Fig. 21. This was done by subtracting 80 from each of the values for 0D for oats alone, and 40 from those values for the oats/alfalfa combination. The population estimates are now remarkably stable over a very wide range of 0D. After making these adjustments of 80°D for oats alone and‘ 400D for the oats/alfalfa combination, Sawyer's total incidence estimates (see above) were again compared with the means for all estimates on the range 4803?D:1120. Sawyer's estimate for oats was 6.42 vs 6.55 from this approach. For the oats/alfalfa combination, his estimate was 10.23 which is compared to 10.24 for this approach. 83 .Ampimommmuv oe . me n no Axv oppompo muoo com mom 1 no u no A.v memo so» ..o.w .xpwmcoo Fo>cop some on» mo cowu_moo mew cow ooumomoo coon m>og me A no .NN ocomwu Eosm mcouooe oco Aouoo oozmwpnoocov Loxzom Eocm mmposom uooe Loocwplm com msoccm ocoocoum oco memos co comma .enmp cw oxoo Fpom an mcomuopaooo ~o>cop pocomomm oouoswpmu--.mm mczmwu 24w: m1... m0... owhmnao< ovAOo 00W. 00: 000. com 000 00» com 000 00¢ con New 6 m .6 m a: Z NOLLV'IncIOd '1V.|.O.l. OBLVWILSB CONCLUSION Early in the development of this work it became clear that survey sweepnet samples of cereal leaf beetle larvae were unlikely to be the optimum type of sample for estimating the total seasonal population from a single sample. That is, they were not likely to lend themselves easily to a demonstration of the usefulness of the theory developed in the introduction. The reasons for the survey data being sub—optinal lie primarily in the strong bias of the sweepnet for large larvae, the few samples taken through the season in the normal survey situation, and a lack of adequate information on weather, notably temperature, in the insects microhabitat. But conditions such as these are the rule rather than the exception in insect surveys and the attempts made here to cope with these limi- tations of the sweepnet survey may be more important than the actual demonstration of the effect of age distribution on the pro- portion of the population sampled, or in the technique for estimat- ing total population. The final synthesis in this thesis after all applies to only sweepnet samples of cereal leaf beetle larvae, while the methods for reaching that point apply to a broad range of sampling techniques and organisms. As to future work in this area, the following should be considered: (1) Application of the generalized diffusion model 84 85 (Barr et a1. 1972) to estimating the distribution of population density and maturity at a given time. (2) Application of simula- tion models to determining maturity distribution. (3) Development of different sampling methods for CLB larval surveys which will give more precise information on maturity, possibly a combination of sweepnet and foliage samples of some type. (4) Development of microhabitat models so that maturity of field population can be inferred from the value of monitored abiotic factors. (5) For agricultural crops, the effect of pesticide application on the maturity distribution should be investigated. (6) In general, factors which affect the maturity distribution in the field, the between field and between year variation, should be further investi- gated. (7) Because of the importance of the time of peak egg density in wheat, an explanation of some of the variance in that value should be sought. (8) Some of the variance in the maturity distribution in oats might be explained by further knowledge of the factors involved in the movement of cereal leaf beetel adults from wheat and grasses to oats. This knowledge should be pursued. Throughout this study, the thing which repeatedly was limit- ing was a knowledge of the accumulated degree-day value in the different fields. Considering the general usefulness of such infor- mation and the knowledge of how to acquire it, it is probably the most important area of research for further development of single sampling techniques for estimating total seasonal population. LITERATURE CITED 86 LITERATURE CITED Barr, R. 0., A. N. Kharkar and K. Y. Lee. 1972. Population balance models in ecological systems. Presented at the 65th Annual Meeting of the AICHE in the Symposium on Modeling of Ecological Systems. Baskerville, G. L. and P. Emin. 1969. Rapid estimation of heat I accumulation from maximum and minimum temperatures. Ecology. 50: 514-7. Cothran, W. R., and C. G. Summers. 1972. Sampling for the Egyptian Alfalfa weevil: A comment on the sweep-net method. J. Econ. Entomol. 65: 689-91. Finney, D. J. 1971. Probit analysis. Cambridge University Press. 333 pp. Gage, S. H. 1972. The cereal leaf beetle and its interaction with two primary hosts: winter wheat and spring oats. M.S. Thesis, Michigan State University. Gage, S. H. 1974. Ecological investigations on the cereal leaf beetle, Oulema melanopus (L.), and the principal larval parasite, tetrastichus julis (Walker). Ph.D. Thesis, Michigan State University. Giles, R. H., ed. 1971. Wildlife management techniques, 3rd ed. The Wildlife Society, Washington, 0.6. 633 pp. Helgesen, R. G. 1969. The within-generation dynamics of the cereal leaf beetle, Oulema melanopus (L.). Ph.D. Thesis, Michigan State University. Hoxie, R. P. and S. G. Wellso. 1974. Cereal leaf beetle instars and sex, defined by larval head capsule widths. Ann. Entomol. Soc. Amer. 67: 183-6. Jackman, J. A. and D. L. Haynes. 1975. An inverse relationship between individual size and population density in the cereal leaf beetle, Oulema melanopus. Environ. Entomol. 4: 235-7. Kiritani, K. and F. Nakasuji. 1967. Estimation of the stage-- specific survival rate in the insect population with over- lapping stages. Res. Popul. Ecol. 9: 145-52. 87 88 Morris, R. F. 1955. The development of sampling techniques for forest insect defoliators, with particular reference to the spruce budworm. Can. J. Zool. 33: 225-94. Morris, R. F., and C. A. Miller. 1954. The development of life tables for the spruce budworm. Can. J. Zool. 32: 283-301. Ruesink, W. G. and D. L. Haynes. 1973. Sweepnet sampling for the cereal leaf beetle, Oulema melanopus (L.). Environ. Entomol. 2(2): 161-72. Simpson, G. G., A. Roe and R. C. Lewontin. 1960. Quantitative Zoology. Rev. Ed. Harcourt, Brace and Co., New York. 440 pp. Southwood, T. R. E. 1966. Ecological Methods. Methuen and Co., London. 391 pp. Steel, R. G. D. and J. H. Torrie. 1960. Principles and Procedures of Statistics. McGraw-Hill Co., New York. 481 pp. APPENDICES 89 APPENDIX A 90 APPENDIX A CEREAL LEAF BEETLE SWEEPNET SURVEY In each survey site, proceed along the chosen route until the required number of fields have been swept with the desig— nated number of sweeps per field. Never survey more than 3 oat (or 3 wheat) fields per mile of driving; hence, if you are doing both crops, you may do 6 in one mile. If 30 wheat and 30 oat fields have not been obtained when you get all the way through the designated area, go back and sweep any unsampled fields you can find, disregarding the above condition of 3 maximum per mile. Mark on the survey map an O for each oat field and a W for each wheat field you see that borders within 100 feet of the road. Circle the letter for the fields that you sweep, so that their exact location is known. On the data form, record temperatures, winds, sky conditions. Record crop height in inches. NOTE: If water accumulates in the bag when sweeping, add alcohol as usual and then freeze the sample at the end of the day. Keep it frozen. If water is no problem, you may either freeze or merely cool the samples. MICHIGAN ONLY Each township has 49 stations: each station is divided into 4 segments. Segments are numbered: N Stations are numbered: 5 4 %3 2 1 2 fun-l H" (A) ,3; 71 ’- do q> ha i (o 1 1» to m (n |li-ul N 19 20 21 2 23 24 47 38 3 21¢ 49 10 =- 2 30 29 28 31 32 33 34 35 36 ,p 4h I I I 10 (I) \I L < N 01 N D———Lem 1h 9 A) ()0 U1 0 10 92 CEREAL LEAF BEETLE SURVEY Wind Conditions _£2gg_ Slight 1 Moderate 2 Heavy 3 Sky Conditions Code O-25% Cloud 1 26-50% Cloud 2 _51-95% Cloud 3 96-100% Cloud 6 Wet or Raining 5 93 1972 CEREAL LEAF BEETLE HEAD CAPSULE ANALYSIS On a single field basis, visually separate the larvae into 4th instar and "other" instars. Count the number of 4th instar larvae. Then measure the head capsule of 10 of these 4ths (if there is that many). Fourth instar larvae will have head capsule widths greater than 21.0 divisions on the ocular micrometer when the Ocular is 10X and the objective is set at 25. Record this information on the coding sheet and follow the last 4th instar measurement with a 999 value so these can automatically be separated. Pro— ceed to measure the head capsules of up to 100 of the larvae in the "other" category, and record this information. Coding from Format: Card Column Information Stored 1-2 County 3-5 Township 6-7 Month 8-9 Day lO—ll Year 12-13 Station 14-15 Segment 16—17 Host 18-21 Number of 4ths 22-78 3 digit head cap measurements up to 10 fourths followed by 999 followed by up to 100 other measurements. Some samples did not have the fourths separated and hence do not have columns 18—21 used, nor the 999 used. These usually had 150 head cap measurements, if there were that many to be measured by keypunching subsequent lines of headcap values for the same field were started in column 3. COUNTY Berrien Cass Lenawee Jackson Allegan Barry Macomb Shiawassee Lapeer Tuscola Newaygo Kalamazoo Gladwin Lake Grand Traverse Oscoda Emmet Presque Isle Chippewa lngham Huron CODE 14 15 20 25 27 33 36 46 ‘13 55 58 66 68 82 534 KEY TO: COUNTIES, TOWNSHIPS 6 STATES MICHIGAN TOWNSHIP EQEE STATE CODE Weesaw 23 1 Howard 21 1 Riga 15 " Pulaski 11 " Dorr 45 " Thornapple 41 Ray 42 " New Haven 43 " Lapeer 22 Elmwood 54 " Ensley 14 " Ross (Gull Lk) 44 ” Grout 21 " Chase 14 " Hayfield 12 " Comins 23 " Resort 11 “ Posen 15 " Rudyard 14 “ Lansing 41 " (Collins Rd) OHIO Townsend 10 3 86° 86° 83° 84° 85° 82° 84° 83° 85° 85° 84° 85° 85° 84° 85° 83° 84° 84° 82° LONGITUDE 28' 00' 49' 39' 43' 55' 07' 17' 37' 22' 33' 38' 38' 03' 04' 42' 32' 30' 30' 41° 41° 41° 42° 42° 42° 43° 43° 43° 42° 43° 44° 44° 45° 45° 46° 42° 41° LATITUDE 51' 51' 46' 07' 44' 45' 05' 38' 20' 24' 67' 52' 33° 44' 20' 15' 12' 42' 15' Mi Whitley Pulaski Armstrong Whitechurch(York) Hildmay (Bruce) Fergus (Wellington) Merlin (Kent) Amherstburg (Essex) Glencoe (Hiddlesex) Ontario (Brock) Renfrew (Bromley) Columbia (Leeds) Walworth (Bloomfield) Addison (Washington) Lima (Grant) CODE 10 90 90 91 92 93 95 96 98 71 72 73 TOWNSHIP INDIANA Jefferson Salem PENNSYLVANIA CANADA 95 CODE 20 30 00 00 00 00 00 STATE CODE 85° 86° 81° 80° 82° 82° 81° 79° 76° 89° 88° 88° LONGITUDE 20' 54' 35' 25' 09' 25' IO' 03' 42' 07' 57' 15' 18' 17' 29' 41° 40° 40° 49° 44° 43° 42° 42° 42° 44° 45° 43° 42° 43° 42° LATITUDE 03' 57' 45' 00' 06' 42' 16' 06' 47' 17' 34' 18' 33' 24' 47' 96 CEREAL LEAF BEETLE SURVEY CARD FORMAT Card Information Columns Stored 1-2 Month 3-4 Day 5-6 Year 7'9 Collector Number 10-12 CollectOr Number 13-14 County Code 15-16 Township Code 17-18 Station Code 19-20 Blank 21 Segment or field 22-23 Host Code 24-27 Time of day (24 hr. Clock) 28-30 Temperature, °F 31 Wind Code 32 Sky Code 33 Blank 34-37 Number of Sweeps 33'49 Blank 50-53 Number of Adults 1971 and earlier 1972 and later 54-57 Number of lst instar Total larvae larvae 58-61 “ 2nd “ Blank 62-65 ' “ 3rd " Blank 66-69 ” 4th ” Blank Card Columns 70-72 73’75 76-77 78-79 80 97 Information Stored Lost larvae Lost adults Lab workers initials 01 signifying regular survey. State Code. 98 KEY TO HOST PLANTS HOST 'EQQE None 00 Grass 01 Oats 02 Wheat 03 Corn 04 Barley 05 Rye 06 Speltz . 07 Timothy 08 Quack Grass 09 Forbs 10 Sudan Grass li Oat Stubble l2 Wheat Stubble l3 Corn Stubble l4 Barley Stubble l5 Rye Stubble 16 Speltz Stubble 17 Other 18 Mixed Grains 19 For survey summaries, “oats” has included Spring grains, while ”wheat” should be restricted to winter wheat. All other categories were grouped in ”other” and assigned the number 18 for purposes of the summary. VDQQO‘M‘NNH ADAIR APPLETON BEHMLAND BEHNKE BENNETT BLACK BLANDFOR BRITTON BRUNNER BUDINGER CARLSON CARR CAULKINS CLEMENS CLAUCHER COOK COOPER DAMICO DAVID DURREN EICKMEYE EMMERT ESSIG FARLARDE FARISS GAINES GARDNER GERSTENS GRAEBER HANNA HERNANDE HOVER HOWARD JEWETT JOHNSON JOSLIN KEEM KELMER KESLER KILMER KIMSEL KING KIRCHER KRUSE KUNDE LINDY LININGER LOREE MARKOVIC MAXWELL 99 COLLECTOR CODE LIST FOR CEREAL LEAF BEETLE SURVEY MCAVOY MCCLINIC MCMANAS MESECHER MILLARD MILLER MOORE MORGAN MURREY NEIDLING OLSON PEACH PIRKLE RANN REDDING REEDER ROGERS SCHULTZ SHERMAN SMITH, F. STEPHENS STOUT TEED TERRILL THIEWES THOMPSON TIEMANN VANDERVE VANVORST WARE WENZLOFF WEST WHITTAKE WOLSCHON WORTH ZEANDER ZUMBROCK WARNKE BLOOMER RING REMINGTO ALLARD HICKS HARTMAN SHEETS YASINSK SCHNEIDE SANNE MOLOUGHN 100 ATKINS 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 GATHOLD BOLLINGER HASKELL BASSLER SNODGRAS KARDEZ BONCZAR SAPP PILARSKI SAYLER MURREY RUESINK DRAPER DEDITCH KETNER SCHOTT NAGY WEIKEL NYBERG GOSCHKE TRAUB LAHTI TAYLOR CLINE NELSON ROWE WOOD BROWN NASLANIC LIST BUTLER CLARK WENDLETO HENDERSO CLOVER WARREN SCHAEFFE KOMANETS VOORHESS CULY TERESINS CUMMINGS POREDA HAYNES MATTESON CHAMBERL ZINK WILLE HAGGART DOLLHOPF 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 MARLOWE DERKS REINOEHL SPARKS PHILLIPS MAIER BIRTWIST GOODNER MARTELL POSSLER PELA FRAZIER LEACH ROBINSON HARDY WESCOTT PANGBORN ANDRUS BROZOVIC SHERMAN COLE LOGAN SHAW, J. SMITH, B. GREENE HERRINGTON KOENIG BOURDO HENRY PASICHNY K UPFOLD CARROLL FISH BROOK NOP J. SCOTT SCOTT KEEPER ROSS HILL PORTER RICHARDSON PICKARD ADAM ZETTLER LA FLECHE 201 202 203 204 205 206 207 208 209 210 211 212 213 214 21-5 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 APPLEGATE FULTON BUYSSE BOSSLER BEKEY LAPP BALABAN MRS. HERRINGTON DOTY CLARK STURER FERGUSON SIMMENS HELGERSON GRUMANN MARCHANE UNDERLY BAKER LAMPERT AREND LOOMIS RAULIN JOCKINEN LARSON DIXON LEHLER ADAMS HAVEN MASSEY HORSLEY SCHLEIHAUF POULTOR RIDGETOWN LANG WALKER MARTIN SINCLAIR CONRAD ESAU APPENDIX B 100 101 ucosmmm u u < xFucmaa< :F cmcwmmu mm aflmwwmwm H M .aucsoo u a Fan. Nov. mcF. Fmo. em mm mF m mF.m no N mN FF on on oN o one. nmN. oFN. moo. om RN MN m oF.m moF N eF NF mm on FN o mme. va. Fmo. oFo. 6N mN m F om.m mm e oN me ¢F em ON 0 oom. ooN. oom. ooo. m N m o oo.N oF F eF «F om em oN o Fae. mMN. onF. NFF. w e m N oo.m NF N FF me ¢F ¢n oN 0 com. Non. FmF. moF. om oo FN oF No.m mmF N on FF mo ¢F oN o mom. mmN. an. ooo. RN FF w o F¢.m .oe e om mo mN ¢F NF o ooN. ooo. ooo. ooo. F w o o oN.m m m mm Ne oN on NF 0 omm. mom. omo. ooo. oN vF N o om.m on F oN mF oo on FN o omF. omN. ooo. ooo. m F o o mn.m e F me em mm vs NF 0 ooo.F ooo. ooo. ooo. NF o o o oo.¢ NF e no «F om en oN 0 mos. FMN. ooo. ooo. oF m o o NF.m mF e. ON cw on em oN o ooo. ooo.F ooo. ooo. o F o o oo.m F N oN mo ¢F on oN o mmm. NNN. mmm. FFF. m N m F NF.N o F om me mN #5 NF 0 oom. ooo. ooo. oom. F o o F om.N N N mN me mN on NF o FNv. mom. me. mmo. ¢N FN o m oF.m mm c MN mF no em FN o mom. moN. moN. moF. m m m N mm.N mF m cm mF oo on FN o o m N F e m N F U o a w E: 28:52 38 :oFuLoaoca Lanszz V :oFumooo .mpmu anEmm umxomgiucmguummFaEmm umxowauucm: an vmcwssmumo mm uFmFm asp cF was» on moFaEem pmcnmmzm mcp cw conzaFLumFu empmcw mo comFLmano go» some ean--.Fm xHozmam< 102 One On OOOFELOOOO mm Fmawmzm OOF\ mm>LmF mo LOOEOOV OuchmO :oFumFOOOO Ocm OOOFm; Oogu t. .mszchm Om>gsm umcnmozm nemucmum FON NN OOO. ONO. ONO. OFO. OO NO N F O0.0 OOF N ON FF OO ON ON O NOOF ON OOO. ONO. OFF. OFO. OO OO ON O O0.0 OON N OF NF OO ON NN O OOFO ON FOO. ONO. FOO. OFO. OO NO O F O0.0 OO O ON OO OF ON ON O OOF FN OOO. OOO. NOF. ONO. NO OO OF N O0.0 OO F OF OF OO ON ON O OOF OF NNO. NNO. FOO. OOO. FO NO O O N0.0 OO N FF OO OF ON ON O NON ON OOO. OOO. OOO. ONO. OOF ON NF O O0.0 OON N OO FF OO ON ON O OFNF NN OOO. OOO. OFF. OFO. OO OO FF F O0.0 OOF O OO OO ON ON OF O OO OF OON. OFF. NOF. ONO. ON O O F O0.0 OO O OO NO ON ON OF O OOO OO OFO. OOO. OOO. OOO. FO OO O O O0.0 OOF F ON OF OO ON FN O OF OF OON. OOF. FNO. OOO. FF N F O FN.O OF F OO OO OO ON FN O OOO OF OFO. OOO. OOO. OOO. FO OO O O N0.0 OOF O NO OF OO. ON ON O ONF ON ONO. OONO OOF1 OOO. NO ON OF O O0.0 OOF OO ON OF OO ON ON O OF ON OOO. OOO. NNO. OOO. N O F O O0.0 OF N ON OO OF ON ON O NFO OF OOO. ONO. OOF. OFO. OO NO OF F ON.O OOF F OO OO ON ON OF O O FN OOO. OOO. OOO. OOO. N N O O O0.0 O N ON OO ON ON FN O NOO ON OON. OOF. OOO. OOO. ON OF O O ON.O OOF OO ON OF OO ON FN O NOF ON OOO. OON. OOF. OOO. OO ON OF O O0.0 OOF O OO OF OO ON FN O 4.8“. :95: O m N F O m N F O u a 8 .1: Hz: Longs: mumO 4no.5 OOFOLOOOOO Omnsaz coFumoog .mpmu mFasem umcaomzmiummFOEmm Omonnivcas an Omchgmme we OFoFm mgp OF pmsa op mmFOEmm pmcammzm mg» OF OOFOOOFEOOFO empmcF Oo comFLmOEoo Low mamu ONOF--.NO xHOzOOO< APPENDIX C 103 104 fim flm coma «H mm OH 0v ZQmH Ho NNmH HZ m mAQ4h Aflammmv NO NO ONN No OO ON OO Nmzzm No ONON OOONNN> mmomo NN NN OOO ON NO NO OO ONNNN No OOON 3O3 O NNmzmoo NO NO ONO NO NO ON NO zmzmmzmx No OONN OzMNNz 9N NOON NNNNoo ON: ON: OON 00 OO NO NO OUZONN No ONON Noomom mamam NOBOzoNoo NO NO OOO ON ON NN OO zOUOommmo Oo NNON ONO NOONN mm ZOONOONNU NO NO ONO OO OO NN OO NNONO No OOON ENON NNO zOOaOmo NO NO NNO OO OO NN NO ZOOON No ONON meaogmamo NO NO OOON NO NO NN OO maamOaNOz No NNON ON NNNNN 2O> zoNNZOOo NO NO ONO ON NO NN NO ONoomOe No .NNNN omOu NON NON ONNN ON OO ON OO NOONxmz No ONNN oONNNNOu NO NO ONN OO OO ON OO xNo>mNNONo No ONNo ONNON OZNON ON ON ONN OO OO NN NO zmmam zm> Oo OOOO ONNOozNzooNO ON: ON: ONN NN OO NO NO «swoon: Oo NNNO mxmozmmamz ONNNON ONN NO NO ONO NO NO OO OO usemoamgz No ONNO mm N NON ONO NO NO OONN NN NO ON OO zoszoezo No ONNO EON OZONONNN NO ON NOO ON OO Oo NO szNmmm Oo ONNO NNONNNO NONNOO zoezmm6 NO NO OOON NO NO NN OO zONN No. NOOo 32: N Noozmommm ON: ON: ONO NO NO NN NO NOO No OOOO NaNo NOO NN NN ONO NN OO ON NO zoomNOo No NOOO Mummo mNaaOm« NO NO OON ON OO NO OO OOONNO No OOOO 323 O OOONNN NO NO ONO NO OO OO NO NOON Oo OOOO zNzoqmm NO NO ONN oo NO NO NO zomam NO NNOO NNO NON NO NO OOO NO OO oo OO Nozmmozazoz Oo NONo OazONNO NO NO NNO OO NO NN NO zOzmeOOOz ON ONNO OoNz No >st momma 224 ON: ON: OOO ON NO Oo OO OZNNNO Oo NONO NZNNN Oomzmm OszNO ON: ONE NOO ON NO Oo OO OZNNNO Oo OONO NO om: O2NNNN. ON NO OON OO OO NN NO NONNONO Oo OONO OzNO NO NO NNO NO OO NN NO zOONNNN Oo ONNO NZONN Nomzmm ZOONNNN NO NO NO NNNN NN OO NN OO OOONON No NOOO N920 NON NOON ONNONNN ON: ON: OON No OO OO NO ONzszN ON NNoo sz N ZONNNN d N m T. N N m wmqmfifi O24 mSHB ZOHB¢>mmmmO NOILVAHTS HGDIIONOT NOISIAIO °ON XHGNI HGOLILVT $92500 ZOHBflBm mezH ZOHBdBm mmmfidm3 ZdOHmUHE 1.14)" (III) I‘lutlll . i .1 mm mm mhm mm mm NV mw Sfifimoo mo mmmm .522 NO NO OOO ON NO OO OO OzooNO Oo ONON 2 O ONNN>ONNNOO mm mm ONO NO NO NO NO zomam No OOON 32 O OOONN NOONON OO NN ONO NO NO ON OO oomoN OO NNON ZOO NOON ONON NO NO ONO ON OO ON NO QONOzONOx OO OOON .Oam .Non NOON NNOo NO NO NNO OO NO ON NO szOz ON NNON mszN NazNoN mmmomw NO NO NOO ON ON NN OO zNOoazoz OO NNON mzz N ONNN>zmmNo zNN zNN OONN NO ON OO OO NNONzONo Oo NNON ozNNNONo ON: ON: OON NO OO NO NO 92mm OO NNON N NO omz ONNNON onNoO NO NO OON NO OO OO OO NOONO NO NNON OOO N ONONOz onNo NO NO NNO ON OO OO OO Oanao Oo ONNO ONON OONN zm>Om OZONO zNN zNN ONN ON OO NO NO zN3NONo OO ONNN zNONONo NO NO NOON NO OO NO OO oommao Oo ONON NON zoo NNONNOO NO NO OOO NN OO NO OO ONszm No OONN Oz N NNONMZONN ON: ON: ONN OO NO NO NO mmmwmzmo ON OOON NO 0m3 ezNNNO ON ON OOON ON ON ON OO OONN>ONN mo . No OONN 3m O OOON ONNN NO NO OON NO OO NO OO ONNNN No NNNN NNNONON ON ON NNN ON OO OO NO ONomomo Oo NNON BNO>N % ON: ON: ONO NO NO OO OO ONNNN No ONON N OOOZOoOm 1. NO NO ONO ON OO NO NO szNNmm OO OOON Oz O NNNONo OON OO OO OOO oN NO NN OO ooOoN Oo NNON OOan OOON NO NO OOO ON OO NO NO ZOOozN No ONON NO N ozNOZON NOON NO NO ONO ON ON No OO xNo>mNNOmo No NOON ZONNon NOON NO NO OOO ON OO NN OO OamNNNNo No ONNN ONO Nxm NOONON NOsto NO NO ONN OO OO NO NO OmOo Oo OONN z N OONOOzoo ON: ON: NNO ON NO ON NO Nszz ON OONN NO ON: ONON: NNONNONO ON: ON: ONO No NO ON NO szOz ON NONN N NNONNNO NaNo NNONNNNO zNN zNN ONO OO NO NO OO OszNNOo No ONON z N macs NO NO NO ONO ON NO ON NO NZNOz ON ONON zmomNOmo .d M m m 3 m G I M O O m N w M m . v. m m m m OONOON NZO m m m m N OzNa zoNaO>Nmmmo N m 3 Nezsoo N p zoNNONO XMQZH ZOHBfiBm mmxe¢m3 ZfiOHEUHE ON: ON: NNO ON NO ON OO ONNOOONOZ NO ONNO O NO OO3 ONNOOONOS NO NO ONN NO OO NO OO OOzONOmN NO NNOO NNNO ONNO: NN NN ONO NN OO OO OO NNONONOOOOO NO NNOO 323 N OOONNONzOz NO NO OOO ON OO ON OO OONONzOz NO OOOO OONONzOz OO OO OON NO OO ON OO 3Oszo OO NONO O N zoeNON NO NO ONO ON OO OO NO zoOOz OO OONO OO O zoaszOON NO NO ONO ON NO NO NO NOONON ON OOOO NOONON ON: ON: NOO ON OO NO NO 2OO02N NO NOOO O NO OO3 O2NO2ONO NO NO ONNN NN OO ON OO mmeOOONz NO NOOO ONON NNO NNNO OOON ON: ON: OON ON OO ON OO O3mNNNOO NO ONNO NOOO N O OONOOOZNM NN NN NONN NO OO NN OO zoamOOOO NO ONNO NOONON O 2 2092mm NO NO OON ON OO NN NO OONOZONOM OO OONO NOOO ONOBO OONOzONOM ON: ON: ONN ON OO ON NO zoOMuOO NO OONO NO OON zoOMOOOO NO NO ONON NN NO NN OO ONNOOONOS NO NNNO OzNzONmON ON ON ONON NN ON ON OO ONOOOOO NO OONO OOO320NN % NN NN OONN . OO OO OO OO 20O2NOONO NO ONOO OVE N93 229282 29: 1: NO NO OON NO OO NO NO ONzoN NO ONOO Ozm N ONzoN ON: ON: NONN NO OO NN OO zozzOOOON OO ONNN NO OO3 OOON zoNOOOOO« OO OO ONNN NO OO NN OO OOMOOOONz OO NNNN 3O3 O OOON zoaOOOOO ON: ON: NOON ON OO ON NO zoaOOOOO NO OONN NO OON zonOOOOO zNN ONN ONO NO OO OO NO O3ONNO OO OOON O2ONNOO NN NN OONN ON OO NO NO ONOOONNNO NO NNON ONOOONNNO NO NO ONNN OO OO NN OO ONON3ONO OO OONN OOON O2NOONO NN NN OON OO OO ON NO O2OOOO OO NONN 323 O ONNONOOO NO NO OONN ON OO OO OO OOONNOO NO OONN 202NOO NO NO OON 3 ON OO NN NO NNNOO NO NOON szNaOOO m L M 3 m 3 m m I O a O m N m m m . . V I I S X OONOON O2O m m m m N OONN zoNeO>NmOOO m m m Nazpou N w zoNNONO xmazH ZOHBdBm mmm8¢m3 zameUHE NN NN OON NN OO NO NO zoazNNu NO OONN O2NOO azNOO NO NO ONO ON OO OO OO NNO>NNNONO NO NNNN O2ONON NN>ONO OszO NO NO NO OOO NO OO NO NO OOzNNOOz NO ONNN NO OO2NNOO21NOO20N NO NO NO ONO OO OO ON NO 3O2NOOO NO NONN ONNNONO azNOO ON: ON: NOO OO OO NN NO 3O2NOOO NO NNNN NO OON 3O2NOOO« NO NO NOO OO NO NN NO 3O2NOOO NO NNNN OO N3N SOOzou 3O2NOOO NO NO ONN OO NO NN OO NNON OOOONNN OO NOON O N NNNO ONOOON ON: ON: OOO ON NO NO NO NNONO NO ON OOOO NN NN3ONOO3 zoNON NNON OO NO ONN ON NO NN NO onNNOO ON OOOO NOBNNOON NNONO OONezoN NO NO ONO NO OO NN OO amzzm NO NOOO NNNOONON ON: ON: ONN OO OO ON OO amzzm NO ONOO NO OON 209ONNNN« NN NN ONN NN OO NO NO NOOOO3ONNO NO OONO NN NONO3NNOO3 OOOO3O OO OO NON NN NO NO OO zoOOzoazo NO ONNO zoOOzoazo NO NO ONO ON ON ON OO ONON OOOOONN OO OONO NNON ONONO NO3O20 ON: ON: OOO ON OO ON OO NOON NO ONOO NOON NNONO NNNNO3N2 mm ON: ON: NON ON OO NN NO OONO3N2 OO NOOO OOO NONON OONO3N2 1. ON: ON: NNO ON OO ON NO zoOmNOOz OO NNNO NO OO3 zoONNOOz« ON ON OOO NN OO ON OO NNONO NO ONOO OzNOstz NO NO ONN OO OO ON NO ONNNOOON OO NOOO NNNONN>N22 NzOOOmNN 92 ON ON: NNO NO NO ON NO OSOOOz ON OOOO NOOO NO OzmzNNO 9220: NO NO NOO NN OO OO OO 3O2ON3NN NO NNOO NONON NNON O2ONON 9902 NO NO OOO NN OO ON NO zoOmNOO2 OO NOOO OOOOazoz NN NN NNO ON NO. OO NO NONzoz ON OOOO NONzoz NN NN NON OO OO OO OO OOOOOO OO NNOO azONN ONONN ON: ON: ON: ONN NO NO ON NO zoaOO2N>NN ON NOOO 2NO O2N>ONN so ONONNNz NO NO NOO NN OO NN NO O2ONON2 OO ONOO O2ONON2 ON: ON: ONON ON NO NN OO NNNNOONON NO OONO NO OON ONNOOONO2« d M m L 3 N... m m N O m m m m m m . V I I I ONNOON O2O m m m m w ONNN zoNNONNNOOO m m m Nezooo . zoNNONO xmozN zoNNONO NNNNON3 2OONNON2 OcoNumuO Hmzumos HMOQHNOO NO NO NNN NN NO ON NO 3O2ONNOO3 ON ONNN O NON: NOOO NezONNONN NO NO OOO ON NO OO NO 3O2OBNOO3 ON ONON 3OO O ONNNN3 NO NO ONO OO OO OO OO O3NNNNNO NO ONNO azNON OONNONNN3 ON NO OOON ON OO ON OO 3O2NOO OO OOOO 2 N NonNO NOO3 NO NO ONON NN NO ON OO ONONOOO NO OOOO NNNSONONO3 MW NO ON: ONN NN OO ON OO OONONO OO NNOO Ozm NN NNNONNO2O> ON: ON: ONO ON OO OO OO NONNNONB NO NO NONO NO OON NNNO NONN>ON9« NO NO ONO ON OO OO NO NNNOOO NO NO OONO ONN>NN ONNNN NO NO OON ON NO ON OO O3NNNNNO NO NOOO NNON ONNON zozszoaNOe NN NN ONN NO NO ON OO OmzNzozmz NO NOON 3 O 20O2NNNNNO ON ON NNO NO NO NO NO OOszO NO ONON OONO2ONO NN NN OOON ON OO OO OO zoNN NO NNON O N NOOOszNO NO NO NO ONO NN OO ON NO 2mNOO 2O> OO ONON NNON NNO 2O>ON NNOoO NO NO NO ONN NO OO NN OO NNONONOONOO NO ONON NN3 NNOz Nmsz ON: ON: NNN NN OO ON OO O3NNNNNO NO OONN N NO OO3 ONNO: .O NNOOO« NO NO ONN OO NO ON NO OONNzOO NO OONN NNOOO2OO d m _ O O O a N m m N O m m N m m a . ONNOON OzO . m m u m x NzNa zoNNO>NmOOO m m m Nezsou m m zoNNONO N H 3 .