Errors in Determining lnstar Numbers Through Head Capsule Measurements of a Lepidopteran-a Laboratory Study and Critique1

Transcription

1 Errors in Determining lnstar Numbers Through Head Capsule Measurements of a Lepidopteran-a Laboratory Study and Critique1 FRED H. SCHMIDT, ROBERT K. CAMPBELL, AND STEPHEN]. TROTTER, JR. Pacific Northest Forest and Range Experiment Station, USDA, Forest Service, Corvallis, Oregon Larvae of Choristoneura viridis Freeman (Tortricidae) ere reared individually on an artificial medium in the laboratory under controlled conditions. The larvae exhibited developmental polymorphism, i.e., some larvae had a total of 6 instars, others 7, and still others 8 instars. Except for instar I, all larval head capsules ere recovered and idths measured for each larva in the study population. A frequency distribution curve of capsule idths suggested only 6 instars, hen in fact over 57% of individuals shoed 7 or more instars. Head capsule idth ranges for successive instars, as suggested by the multimodal frequency distribution curve, ere not in agreement ith knon values for the laboratory popula- ABSTRACT tion, nor ere they in agreement hen values for the population ere segregated by sex irrespective of larval instar group. Peaks in the curve could be interpreted only after the head capsule idths in the population ere segregated by instar group. In larval groth regressions for the population as ell as for head capsules broken don by sex and/or instar group, standard error of estimate values appeared to be more sensitive than R 2 values in reflecting relative precisions of equations. Generated frequency distribution curves of head capsule idths support the argument that frequency distribution of head capsule idths cannot be used to assign instar numbers in lepidopterous species ith developmental polymorphism. Ecologists and applied entomologists are concerned ith larval sie, groth rate, age distribution ithin a population, etc. Many of these factors are directly related to the instars of a given species. Questions of interest include, hich instars are preyed upon by given predators, hich are parasitied by given parasites, and hich are most susceptib le to infection by pathogens? Ho much of a given host is the insect capable of consuming, and hich instar ( s) consumes the most and thereby causes the most damage? Are LDris for such agents as insecticides and pathogens the same for different instars of a species (Shepard 1951, Busvine 1957, Stairs 1965, Ahmad and Forgash 1975, Magnoler 1975)? Can the instars of a species be identified, so that seasonal development of populations may be monitored to determine the proper timing of insecticide applications in control programs? Knoledge of the precise number of instars of a species is of fundamental importance to entomologists of varying interests. The number of instars characteristic of a given species may be determined in several ays including direct observation of larvae reared through the entire larval stage and through plots of frequency distributions of head capsule idths taken from larvae representative of the entire larval stage of the species. It has been argued that the latter procedure ill provide a multimodal curve ith each peak being representative of head capsules found ithin one instar. The total number of peaks represents the total number of instars exhibited for the species in the collection of larvae examined. Such multimodal curves, and the resultant classification of the number of instars characteristic of a species, have been in the literature since 1928 (Peterson and Haeussler 1928, Taylor 1931), and are still in current use (Kishi 1971, Fox et al. 1972, Parker and Moyer 1972, Hoxie and Wellso 1974, Vandererker and Kulman 1974, Wilson 1974 ). 1 Received for publication Sept. 24, The model equation lny = a + bx, here Y = head capsule idth and X = instar number, is another method for determining the characteristic number of instars in species. Application of this method to entomological ork as 1st recognied by Dyar ( 189) ho found that "idths of the head of a (lepidopterous) larva in its successive stages follo a regular geometrical progression." Dyar's original purpose in proposing the generaliation as to provide a method of discovering an overlooked instar hen trying to determine the number of molts or instars of certain species (Dyar 189, Richards 1949). The method is not alays applicable (Gaines and Campbell 1935, Fox et al. 1972). Whether any method other than direct observation ill adequately characterie instar number in Choristoncura species (Lepidoptera: Tortricidae) is questionable. In laboratory culture, instead of the "typical" 6 instars that have been reported for the estern spruce budorm, C. occidental1's Freeman (Bean and Bater 1957, Lyon et al. 1972), and the eastern spruce budorm, C. fumiferana (Clemens) (McGugan 1954, Bean and Bater 1957), larvae ith 5, 6, and 7 instars and 6, and 7 instars have been found (Schmidt and Lauer 1977). Larvae exhibiting 6. 7, or 8 instars ere also found in C. 'l1iridis Freeman. Head capsule idth measurements made on larvae of the 1st 2 species ere generally similar to those previously reported, even though the actual number of instars may have been other than the reported 62 The greatest deviation from the expected, typical 6 instars occurred in C. viridis) here more than SO% of the insects had 7 or 8 instars. This study attempts to determine hether or not methods other than the direct observation of developing larvae ill clearly characterie the number of instars in this species. The derivation of the C. viridis stock, the artificial 2 Larval head capsule idths, of last instar larvae, irrespective of the total number of instars exhibited by the species, are ca. the same ithin the species (Schmidt and Lauer 1977). 75

2 September 1977] SCHMIDT ET AL.: ERRORS IN lnstar NUMBERS 751 budorm medium, and the rearing methods have been described in Schmidt and Lauer ( 1977). The idths of sloughed head capsules from the 2nd instar to pupation ere measured to the nearest.16 mm using a calibrated, ocular micrometer (Mc Gugan 1954). The instar group to hich each larva belonged as noted at pupation. Only head capsule data from insects that later emerged as adults normal in appearance ere included in the results. The adults ere sexed. A total of 2218 head capsules as measured from a total of 397 larvae. These included head capsules (and larvae) from the folloing groups: 5 (1), 366 (61), and 15 (15) from 6-, 7-, and 8-instar males, or 971 total head capsules (and 176 larvae). In addition, 5 (1), 6 (1), and 147 (21) head capsules from 6-, 7-, and 8-instar females ere measured for a total of 1247 head capsules (and 22 1 larvae). To provide a representative frequency distribution curve of larval head capsule idths, a population of C. viridis as constructed by the random selection (random number table) of 2 of the total 397 larvae. The component groups of this population had proportions equivalent to those observed in previous experiments; i.e.,.5:.5 male to female frequency and an instar frequency of.53,.45, and.2 for 6-, 7-, and 8-instar males, and.32,.63, and.5 for 6-, 7-, and 8-instar females. The curve included idths from all head capsules recovered from all 2 larvae in the constructed population beginning ith instar II. A total of 1122 head capsules as included. These data ere also employed in the calculation of the mean head capsule idth and standard deviation for each instar and in each subsequent breakdon of the population by sex and instar group. The same data ere used to compute larval groth regressions for the population and for the breakdons. The model 1nY = a + bx, here Y = head capsule idth and X = instar number, as used. In these regressions, the standard errors (i.e., v'mse, here MSE =mean square error) are biased slightly donards and R2s upards because all head capsules for each larva in the population, and in each breakdon, ere used in the derivation of the resulting regressions. The data suggested that frequency distribution curves of larval head capsule idths might vary ith the structure of the population being examined. To verify this, curves for populations ith different structures ere simulated. Sample data ere generated by a regression equation that ell predicts head capsule idth (unpublished): 1nY = X X X1 X2 +.7 X2X3, here Y represents head capsule idth, xl head capsule number, x2 instar group, and X3 sex. In a program developed by D. G. Niess, Systems Analyst, Oregon State University, Corvallis, population structures ere varied by changing the proportions of larvae in instar groups, maintaining a 5:5 sex ratio. Head capsule numbers ere obtained ith a random number generator, and 11noise" as introduced into the generated data by incorporating error based on the standard error of estimate of the above equation ( = v'mse =.739). Finally, generated data ere plotted in the characteristic frequency curve for comparison ith distributions based on other population structures. RESULTS AND DISCUSSION Frequency Distribution Curve of Head Capsule Widths and Determination of Number of Instars. Fig. 1 shos the frequency distribution curve of head capsule idths for all head capsules in the larval population. The curve has 5 apparent peaks, ith means at ca..3 (peak a),.433 (peak b), and.633 (peak c), (peak d), and 1.7 mm (peak e), presumably for instars II through VI. This suggests that there are but 6 instars in the population. Hoever, this cannot be the case because a substantial number of head capsules ere included from larvae that had more than 6 instars. Therefore, 11secondary" peaks in the curve, such as those at ca..933 (peak d1) and 1.4 mm (peak d2) may have functional significance. Since the origin of every head capsule in the frequency distribution as knon, it as possible to determine a mean and variation about the mean (i.e., S.D.) for each instar of the population and to calculate a larval groth regression using the means of the successive instars. It as also possible to segregate the capsules into sex and instar classes and to make similar determinations of means and variation about the means for each instar of each subgroup. These calculated mean head capsule idths are shon in Table 1 and also graphically at the top of Fig. 1. There is little apparent difference in head capsule idth beteen head capsules VII and VIII hen the mean head capsule idths of all larvae, irrespective of sex and instar group (i.e., the population), are examined (top line in Fig. 1). Moreover, instars VII and VIII might ordinarily be considered indistinguishable as evidenced by their close mean idths and overlapping standard deviations; and, to a large extent, these instars overlap those of instar VI. When larvae for the respective instar numbers are grouped by sex, irrespective of instar group, this problem is still unresolved. There is little difference in head capsule idths beteen head capsules VII and VIII, and standard deviations overlap for head capsules VI through VIII in both sexes. In females of this sample, the mean idth for instar VIII is less than that for instar VII. Head capsule idth of successive instars are only distinguishable ithout overlap hen animals in the population are broken don by instar group, irrespective of sex. Such a breakdon substantially reduced the variability in head capsule idths from instar V onard. If the population is broken don further by both sex and instar, an even greater reduction in this variability is evident in all subgroups except male 8-instar animals, a subgroup containing but 2 animals.

3 752 ANNALS OF THE ENTOMOLOGICAL SoCIETY OF AMERICA [Vol. 7, no " , I ;;...,....., < "';::""' - -,. _._'_ l. -L 12 r /' / / ---- u... b ""/ > a: """ "' 1 (..) (..) lj. > u :: a:: 6 lj. POPULATION 1% SEX a' 5. Q 5. INSTAR GROUP ;oo; SEX-INSTAR d' lc Q !Q 4 c OB HEAD CAPSULE WIDTH IN mm FIG. I.-Frequency distribution curve of head capsule idths for a larval "population" of C. viridis based on observed idths of all head capsules, except instar I, recovered from individuals in the population. Head capsule idth means. ( ± SD) of each ins tar are given for the population and for each subgroup of the same population. Broken lines connect common instar, or head capsule, numbers ithin a breakdon, and are presented only for the convenience of the reader. As noted above, the frequency distribution exhibited essentially 5 peaks, or 6 instars, and more should have been evident because more than half of the animals in the population had more than 6 instars. When the composition of individual peaks in the distribution is examined, it is obvious that at least some of the peaks in the curve cannot be accounted for hen instar means are computed either for the population or for the subgroups broken don by sex alone (the 3 top lines in Fig. 1). Inconsistencies in this regard occur ith peaks d11 d, and d2 on the curve. Judged by its position in the curve, the d peak, the highest of the 3, probably includes mostly instar V head capsules. When this peak value is compared ith means of instars actually observed for the population and sexed subgroups, the peak contains primarily head capsules from instar VI larvae, hich are mostly males. It is diffcult to account for the d peak because it occurs just ithin one standard deviation from instar VII means of the population and of the male subgroup. It as not even ithin one standard deviation of the mean of instar VI of the female subgroup. The d1 and d2 peaks on the curve are much closer to means actually observed for the instars V and VI, respectively, for the population and for each of the sexed subgroups.

4 September 1977] SCHMIDT ET AL.: ERRORS IN lnstar NUMBERS 753 Table 1.-Mean head capsule idths ( HCW) for all C. viridis instars of the "population" and the respective breakdons of the population in the study. 1. The "population" (Narvae in group= 2) HCW (mm) Instar ±SD II.297±.11 III.413±.27 IV.62±.65 v.966±.156 VI 1.425±.262 VII 1.756±.165 VIII 1.748± Breakdon by sex irrespective of instar group (Narvae) (N=1) (N=1) HCW (mm) HCW (mm) Instar ±SD ±SD II.295± ±.13 IV.41± ±.26 III.61±.68.63±.6 v.948± ±.15 VI 1.42± ±.259 VII 1.684± ±.169 VIII 1.767± ± Breakdon by instar group irrespective of sex Instar group (N,.,.,..) 6 (N=85) 7 (N=18) 8 (N=7) HCW(mm) HCW (mm) HCW (mm) Instar ±SD ±SD ±SD II.3±.9.295±.9.286±.3 III.428±.18.43± ±.4 IV.669± ± ±.5 v 1.116± ±.88.76±.44 VI 1.694± ± ±.66 VII 1.782± ±.49 VIII 1.748± Breakdon by sex and instar group Ins tar group Sex (N,..,..) (N=53) (N=32) (N=45) (N=63) (N=2) (N=5) HCW (mm) HCW (mm) HCW(mm) HCW(mm) HCW (mm) HCW (mm) Ins tar ±SD ±SD ±SD ±SD ±SD ±SD II.298±.9.33± ±.8.298±.9.38± ±.9 III.425± ± ± ± ±.71.37±.18 IV.66± ± ± ± ±.71.51±.22 v 1.77± ±.52.82±.87.93± ± ±.22 VI 1.649± ± ± ± ± ±.49 VII 1.699± ± ± ±.36 VIII 1.767± ±.114 Data shon graphically at the top of Fig. 1. These ere expected to be higher on the curve than the true mean head capsule idths of all instars as found, and certainly higher than the d peak. found in this population. Moreover, a breakdon by If the d peak has little relation to head capsule sex did not appreciably improve this interpretation. idths for instar V, then mm, the maximum When head capsule idths are broken don by for the peak, cannot be assumed to be even a close instar group, irrespective of sex, a more reasonable approximation of the mean head capsule idth either interpretation for the d1, d, and d2 peaks is possible. for that instar in the population or for that instar The d peak is apparently a combination consisting in either sex of C. viridis. Therefore, the frequency largely of the instar V head capsule of the 6-instar distribution curve of head capsule idths did not ac group and the instar VI head capsule of the 7 -instar curately describe either the true number of instars group. These groups represent ca. 96% of the larvae in this population, and presumably this species, or in the population. The broad peak from ca..833-

5 754 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA [Vol. 7, no mm ( d1) consisted largely of the instar V head capsule of the 7 -instar group, a group representing ca. 54% of the population. The peak at ca mm (d2) consisted largely of the instar VII head capsule of the 8-instar group, a group representing ca. 3% of the population. When head capsule idths are broken don by both sex and instar group, interpretation becomes easier. It is probable that the d peak consisted largely of the instar VI head capsule of the male, 7 -instar group and of tl].e instar V head capsule of the female, 6-instar group, ith a relatively minor contribution of the instar VI head capsule of the male, 8-instar group. These combined groups represent ca. 4% of the larvae in the population. The d1 peak of the curve consisted mainly of instar V head capsules of the female 7 -instar group, representing ca. 32% of the population. The d2 peak probably consisted mostly of the instar VII head capsule of both males and females in the 8-instar group, representing ca. 4% of the population. A similar analysis of the probable composition of most other peaks, both major and minor, of this population, and presumably the species, can be made. Gaines and Campbell (I935) used a modified approach to that described above in interpreting a complex frequency distribution curve of head capsule idth of the black cutorm, Agrotis ipsilon ( = ypsilon) (Hufnagel), from the data of Satterthait ( I933). Satterthait found that 6-, 7-, and 8-instar "classes" occurred normally in that species. Gaines and Campbell concluded that the frequency distribution curve method of determining the number of instars of a species "ill give clear results only hen the insects being measured are fairly homogeneous in rate of development and number of instars. If the population being studied is a mixture of individuals having n and n+ I instars and a corresponding difference in rate of development, it might be diffcult or impossible to interpret the frequency distributions." This statement ould accurately apply to C. viridis, as ell as to other lepidopterous species that exhibit postembryonic, developmental polymorphism (Schmidt and Lauer I977). Larval Groth Regressions from Head Capsule Width Measurements of Successive Instars.-Table 2 shos larval groth regressions of head capsule idth measurements of successive instars for the population and for the various subgroup breakdons shon in both Table I and Fig. 1. In a comparison of R2 values for regressions of the population and subsequent breakdons of that population, a relatively poor fit of the head capsule idth data to the regression as expected, and R2 values for the regressions of the various subgroup breakdons ould reflect substantially better fits. This as not the case. The R2 value as surprisingly high, in vie of the knon heterogeneity in the population. Little or no additional variability in the data could be accounted for, as judged from R2 values, hen the data as broken don into subgroups based on sex, irrespective of instar group. An improvement in R2 values as not evident until the population as broken don by instar group, irrespective of sex. When head capsule idths in the population ere broken don by both sex and instar group, R2 values of regressions for the male subgroups shoed little or no change from the pre- Table 2.-Larval groth regressions based on head capsule idth measurements of all head capsules recovered from 2 C. viridis larvae. N in group analyed Larval groth regression S.E. '% (animals) of (head capsule idth (Y) esti- Head group or subgroup Breakdon X in star number (X)) R2* mate** Larvae capsules analyed 1. Population In? = X.953 O.I IOO 2. Sex IOO Male In? = X I IOO 5. Female In? = 1.4SI+.372 X.957 O.I32 IOO lnstar Group IOO 6-instar In? = I X instar In? = X instar In? = X :5 Sex-Instar Group IOO Male instar In? = X instar In? = X.983.8I instar In? = X Female IOO 5. 6-instar 1n? = X I instar In? = X instar In? = 1.49I+.3I4 X * Coeffcient of determination. ** Standard error of estimate = V MSE.

6 September 1977] SCHMIDT ET AL.: ERRORS IN lnstar NUMBERS 755 vious breakdon by instar group, irrespective of sex. R:: values of female subgroup regressions, on the other hand, did appear to be slightly higher than a breakdon by instar group alone. Standard error of estimate (SEE) proved to be more sensitive than R2 values in reflecting the increase in precision of regression equations associated ith successive breakdons of the population into sex and instar group (Table 2). Regressions in subgroups broken don by sex, irrespective of instar group, had SEE values of.141 and.132 for males and females, respectively. These values ere little different from the SEE value of. 136 for the population. In populations broken don by instar group, irrespective of sex, the SEE values ere.7,.82, and.77 for regressions of the 6-, 7-, and 8-instar subgroups, or almost a 2-fold reduction from the SEE value obtained for the population regression. vvhen the population as broken don by both sex and instar group, a still further reduction in SEE values usually occurred. This reduction as substantial for the female-instar subgroups, resulting in SEE values less than half those for the population. Obviously in this sample, most of the variation in the head capsule idths of C. viridis has been accounted for by regression hen the SEE value is ca..1 or less. Computer Simulations of Frequency Distribution Curves of Larval Head Capsule Widths.-The computer simulations of larval head capsule idths ere plotted in a frequency distribution curve format. Four populations of head capsules ere generated. Of the 4, 3 ere generated ith sex ratios of.5:.5 each, males to females. All the simulated head capsule idths of each simulated population came from larvae of only one instar group. In one population, all the represented larvae exhibited a total of 6 instars (i.e.. 6-instar group). In another group, all the represented larvae exhibited a total of 7 instars. and in still another, all the represented larvae exhibited a total of 8 instars. The resulting frequency distribution curves are the top 3 in Fig. 2. Since head capsule data for instar I larvae ere omitted in the derivation of the simulation model, the number of peaks in the multimodal frequency distribution curves should be one less than the number in the respective instar group. This proved to be the case. ith 5. 6, and 7 peaks being evident for curves of the 6-, 7-, and 8-instar groups. A 4th simulated population as structured by sex and instar group according to classes shon in the loest curve of Fig. 2. Because this population had the sex and instar group makeup of a natural population, its frequency distribution curve is expected to he similar to that found for the population shon in Fig. 1. vvhile the curve (bottom Fig. 2) shoed a great deal of "noise." probably due to the error term used ith the model. similarities beteen the 2 curves are apparent. Peak maxima occurred at ca..3(a)..4 ( b). and.6 mm(c) and peak minima at ca..33 and.5 mm in both curves. The peak located from ca mm (e) as evident in l % 8 INSTAR u NCAPS = 22 1% 7 I NSTAR 4 NCAPS = 12 :: :::> 8 2o >-8 u :::>6 1% 6 INSTAR :: NCAPS = l % IN INSTAR CLASS SEX d HEAD CAPSULE WIDTH IN m m FIG. 2.-Computer simulation of frequency distribution curves of head capsule idths using the model : In?= X X2 - o.62 X1X2 + o.oo7 x2x3 Y represents head capsule idth, X1 head capsule number, X2 ins tar group, and X3 sex. An expected peak for instar I is omitted in each curve. The sex ratio in each of the populations is.5:.5, males: females.

7 756 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA [Vol. 7, no. 5 both curves. The simulated curve as much "noisier," and finding a precise location for the peak maximum, or the mean head capsule idth for the instar that as presumed to be represented by it as, therefore, precluded. The complex and diffcult to interpret region of the curve of the natural laboratory population (i.e. beteen.75 and 1.5 mm, Fig. 1 ), as equally complex in the simulated curve. If the population represented in the latter had been composed exclusively or even predominately of 6-instar larvae, a minimum ould have been expected at ca mm. But this as not the case. The curves for the homogeneous groups suggested that this peak is composed of simulated head capsules V and VI of. the 7- and 8-instar groups. But since the 8-instar complement to the population as only 3.5%, and therefore negligible, this peak consisted largely of instar V head capsules of the 7-instar group. It is probable that the peak beteen 1. and ca. 1.3 mm consisted largely of simulated head capsules of instars V and VI of the 6- and 7-instar groups. The findings by computer simulations are in general agreement ith the conclusions made by examining the frequency distribution curve of the actual population presented in Fig. 1. CONCLUSIONS The use of frequency distribution curves to determine the characteristic number of instars of a species or population may not alays be reliable. As pointed out by Gaines and Campbell ( 1935), such curves "ill give clear results only hen the insects being measured are fairly homogeneous in rate of development and number of instars." In lepidopterous species hich sho developmental polymorphism (e.g., C. 'l!iridis). a complex frequency distribution curve can result. Conversely, it can be surmised that complex frequency distributions for head capsule idths in other species may indicate developmental polymorphism. ACKNOWLEDGMENT We thank R. B. Ryan and M. E. Martignoni, Pacific Northest Forest and Range Experiment Station, and N. H. Anderson, Oregon State University for reading the manuscript, and P. Kanarek, Oregon State University, for revieing the statistical presentation in the manuscript. REFEREN CES CITED Ahmad, S., and A. J. Forgash Toxicity of carbaryl and diainon to gypsy moth larvae: changes in relation to larval groth. J. Econ. Entomol. 68: Bean, J. L., and H.. Bater Mean head idths for spruce budorm larval instars in Minnesota and associated data. Ibid. 5: 499. Busvine, J. R A Critical Revie of the Techniques for Testing Insecticides, Commonealth Institute of Entomology, London. 28 pp. Dyar, H. G The number of molts of Lepidopterous larvae. Psyche 5 : Fox, R. C., N. H. Anderson, S. C. Garner, and A. I. Walker Larval head-capsules of the Nantucket pine tip moth. Ann. Entomol. Soc. Am. 65 : Gaines, }. C., and F. L. Campbell Dyar's rule as related to the number of instars of the corn ear orm, Heliothis obsoleta (Fab.), collected in the field. Ibid. 28 : Hoxie, R. P., and S. G. Wellso Cereal leaf beetle instars and sex, defined by larval head capsule idths. Ibid. 67: Kishi, Y Reconsideration of the method to measure the larval instars by use of the frequency distribution of head-capsule idths or lengths. Can. Entomol. 13: Lyon, R. L, C. E. Richmond, J. L. Robertson, and B. A. Lucas Rearing diapause and diapause-free estern spruce budorm (Choristoneura occidentalis) (Lepidoptera : Tortricidae) on an artificial diet. Ibid. 14: Magnoler, A Bioassay of nucleopolyhedrosis virus against larval instars of Malacosom. neustria. ]. Invertebr. Pathol. 25 : McGugan, B. M Needle-mining habits and larval instars of the spruce budorm. Can. Entomol. 86: Parker, D. L., and M. W. Moyer Biology of a leafroller, Archips negundamts, in Utah (Lepidoptera: Tortricidae). Ann. Entomol. Soc. Am. 65: Peterson, A., and G. }. Haeussler Some observations on the number of larval instars of the oriental peach moth, Laspeyresia molesta Busck. ]. Econ. Entomol. 21 : Richards,. W The relation beteen measurements of the successive instars of insects. Proc. R. Entomol. Soc. London (A). 24: 8-1. Satterthait, A. F Larval instars and feeding of the black cutorm, Agrotis ypsilon Rott. J. Agric. Res. 46: Schmidt, F. H., and W. L. Lauer Developmental polymorphism in Choristoneura spp. (Lepidoptera: Tortricidae). Ann. Entomol. Soc. Am. 7: Shepard, H. H The Chemistry and Action of Insecticides, McGra-Hill, Ne York. 54 pp. Stairs, G. R Quantitative differences in susceptibility to nuclear-polyhedrosis virus among larval instars of the forest tent caterpillar, Malacosoma disstria (Hubner). ]. Invertebr. Pathol. 7: Taylor, R. L On 'Dyar's Rule' and its application to safly larvae. Ann. Entomol. Soc. Am. 24: Vandererker, G. K., and H. M. Kulman Stadium and sex determination of yelloheaded spruce safly larvae, Pikou.ma alaskensis. Ibid. 67: Wilson, L. F Life history and habits of a leaf tier, Aroga argutiola (Lepidoptera: Gelechiidae), on seet fern in Michigan. Can. Entomol. 16: