L: w. 3 .0 J. rig», .».- z“ “V I. I}? P AR Y THESIS Pa'llClli-gflfl Stam University This is to certify that ‘the thesis entitled THE WITHIN—GENERATION POPLJLATION DYNAMICS OF THE CEREAL LEAF BEETLE, OULEMA MELANOPUS (L.) presented by Robert Gordon Helgesen has been accepted towards fulfillment of the requirements for Ph.D “Entomology degree in (QM/MM Dean L. Haye Major professor Date October 20, I969 0-169 ABSTRACT THE WITHIN-GENERATION POPULATION DYNAMICS OF THE CEREAL LEAF BEETLE, OULEMA MELANOPUS (L.) BY Robert Gordon Helgesen The cereal leaf beetle, 0u£cma meianopub (L.), has rapidly increased its numbers and range since it was discovered in Michigan in 1962. It was hypothesized that intraspecific density—dependent mortality was the major constraint on the survivorship of this foreign pest as its density increased. In order to quantify fecundity and age specific survivorship the density of three different populations was censused from April to July in 1967, 1968 and 1969. Populations were established in cages where age specific survivorship could be investigated at specific densities. Fecundity was the same at all densities and affected mainly by temperature. Mortality in the first and fourth instar was found to increase with an increase in the log- arithm of density. There was a significant difference in fourth instar mortality between host plants but no difference in first instar mortality between host plants. Two different mortality factors appeared to be involved in the density- dependent mortality of these two instars. Second and Robert Gordon Helgesen third instar mortality, as well as pupal mortality was re— latively constant with respect to density. The cereal leaf beetle has the requisite for population regulation -— a density-dependent feedback system. THE WITHIN-GENERATION POPULATION DYNAMICS OF THE CEREAL LEAF BEETLE, 0u£cma mc£an0pu4 (L.) By Robert Gordon Helgesen A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Entomology 1969 cam-'7 '3 4X70 ACKNOWLEDGEMENTS To Dr. Dean L. Haynes I extend my sincerest appre- ciation for the personal guidance, unselfish contribution of time and wealth of scientific inspirations he offered throughout my prOgram and the preparation of this thesis. Dr. Gordon Guyer's enthusiasm, optimism and support were a constant source of encouragement during my entire program. Some discerning suggestions by Dr. William Cooper enhanced the quantitative aspects of this study. I am grateful for my wife's encouragement, patience and toil in the preparation of this thesis. ii TABLE OF CONTENTS INTRODUCTION REVIEW OF LITERATURE METHODS AND MATERIALS. RESULTS. Fecundity of the Cereal Leaf Beetle Age Specific Developmental Rates. Survival Analysis . . . Variance Analysis of Total Larval Mortality Variance of Age Specific Mortality. . . Model of Within-Generation Survivorship Qualitative Effects of Density. DISCUSSION Survival Analysis . . . . Factors Affecting Within-Generation Survival. Within—Generation Dynamics. . . . . . . . CONCLUSIONS. . . . . . LITERATURE CITED APPENDIX iii Page LIST OF TABLES Table l. Efficiency of yd2 and ft2 sample units in 1967 Galien wheat . . . . . . . . 2. A comparison of fecundity in three populations with different mean elytral length. 3. Developmental time (in days) for instars of the cereal leaf beetle at various temper— atures. A. Survivorship of cereal leaf beetle eggs. 5. Age specific mortality of the cereal leaf beetle in 1967, 1968 and 1969 field studies. . 6. Age specific mortality of the cereal leaf beetle in the 1969 cage density study 7 Two-way analysis of variance of % total larval mortality in 1967-1969 CLB field study. 8. One-way analysis of variance of total larval mortality in the 1969 cage density study. 9. Correlation analysis between instar and total larval mortality in the cereal leaf beetle. 10. The relationship between instar mortality and log1° density in the cereal leaf beetle ll. The relationship between density and age class mortality classified by host plant in the field study. . . . . 12. Mortality of lst instar in Galien wheat, 1967. l3.—l9. Field study data 20.~2l. Cage study data. iv Page 22 33 35 39 “3 AA A6 A7 50 53 5A 63 88 95 Figure 10. ll. l2. 13. IA. LIST OF FIGURES Life cycle of the cereal leaf beetle, Ouicma mc£an0pu4 (L.) . . . The relationship of the mean and variance in square foot samples of Galien wheat in 1967 . . . . . . . . . . . . . One milliacre emergence cages used for summer adult cereal leaf beetle. . . . . . . . . Screening technique used to separate CLB pupae from soil . . . . . . . . . . . . A CLB ovary showing seven ovarioles. The relationship between egg production per female and maximum daily temperature. . . Temperature related developmental curves for the four instars of the cereal leaf beetle. Population curves for egg and larvae of the cereal leaf beetle in Galien wheat, 1967. The distribution of first instar survival calculated by the total incidence method. The relationship between total larval mortality and density in wheat and oats . . . . . . . The relationship between density and mortality in the first and fourth instar of the field study . . . . . . . . . . . . . . . The relationship between percent mortality and log density in each larval instar of the cereal leaf beetle . . . . The relationship of age specific mortality and density in wheat and oats in the field studies . . . . . . . . . . . . . Relationship between observed and calculated total larval mortality using the two- factor model . . . . . . . . . . . . . . V Page 23 29 29 29 32 36 38 Al A9 51 52 55 58 LIST OF FIGURES —— Cont. Figure 15. The relationship between mean weight and mean elytron length of the emerging adult female cereal leaf beetle and its preceding maximum larval population 16. Distribution of survival in a hypothetical instar. . . . . . . . . . . . . . . . . 17. Effect of crowding on fecundity of CLB in 1967 cage study . . . . . . . vi Page 60 66 73 INTRODUCTION Many foreign insects have been introduced into North America. Future technological advances in transportation will intensify this phenonomenon. We can assume that many introduced insects never survived for various reasons, while others, like the Mediterranean fruit fly, Cchatitib capitata (Wiedemann), were eradicated after successful establishment. However, a few insects like the gypsy moth, Ponthetaia dispan (L.), European corn borer, Obtainia nubifiafiib (Hfibner), the codling moth, Canpocapba pomeneflla (L.), and the Japanese beetle, Popi££ia japonica Newman, were able to establish themselves. Once established, they found very little environ- mental resistance and greatly expanded their distribution and abundance. Like other successfully introduced insect pests, the cereal leaf beetle (CLB), Ouficma me£an0pu4 (L.), has rapidly in- creased its number and range. Its preferred host is the suc- culent growth of small grains, and its success threatens the economic production of oats in Michigan. Therefore, popula- tion control is necessary before a certain economic damage threshold is reached. However, before a population control program can be logically designed and evaluated, the dynamics of a population should be quantified. Initial research on the CLB dealt with damage control and biology instead of popu- lation dynamics. Unfortunately, this type of research does not provide very much useful information to design a program in population management. Turnbull and Chant (1961) most aptly suggest that economic entomologists have classically limited their ability to understand the total pest management problem by equating damage control to population control. This study was an investigation of the within-genera— tion population dynamics of the cereal leaf beetle. Natural mortality factors of the population were isolated and quantif— ied in order to construct a mathematical model which would ex— plain natural population changes of the cereal leaf beetle and perhaps expose certain features of the population which are susceptible to control. Castro (1965) and Yun (1967) showed that no parasites or predators significantly affect the cereal leaf beetle in Michigan. Therefore, it was hypo- thesized that most mortality occurring within a generation was a direct cause of intrinsic and climatic or physical mor- tality factors. By accepting the almost axiomatic assumption that a population has an upper limit of density in any given area, certain mortality factors must function through a density- dependent feedback system, at least above certain densities. This density-dependent mortality would tend to hold the popu- lation at some variable and unknown upper limit. Given the somewhat constant planting practices for small grains in Michigan, the most obvious factor which could produce this density-dependent mortality is competition for food which 3 could express itself through direct mortality and qualitative changes in the population. REVIEW OF LITERATURE An excellent account of the history, distribution, general biology and literature of the cereal leaf beetle is given by Yun (1967). From the literature he reviewed, Yun concluded that the cereal leaf beetle has been recorded as a pest of small grains since the mid-eighteenth century. It is presently acknowledged as a general, but sporadic, pest throughout its native range of Europe, parts of north Africa, Turkey, Iran and from central Siberia eastward (Yun 1967). It was first identified in North America from speci- mens collected in Michigan in 1962. However, from our present knowledge of the insect the abundance at that time indicates it was probably introduced at least ten years previously. Castro, ct a£., (1965) described the natural history of the cereal leaf beetle in Michigan. A graphic representa- tion of the life cycle of this insect is diagrammed in Figure 1. MONTH J F M A M J J A S O N D aaaaaaaaaaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaaaaaaaaa eeeeeeeeeee lllllll a = adult pppppp e = egg 1 = larvae P = pupae Figure 1. Life cycle of the cereal leaf beetle, 0u£ema mafianOpuA (L.). 5 They reported that the overwintering adult could be found in forest litter, grass stubble, under tree bark, or any site well protected from temperature extremes. Over- wintering adults become active in March when daytime temper- atures and solar radiation raise their temperature above 55°F. Once active, the beetles are arbitrarily called spring adults. Spring adults feed on winter grains as well as native and cultivated grasses for a few days after emergence. Fe— males generally mate very soon after leaving the overwinter- ing site and continue to mate throughout the ovipositional period. Oviposition occurs from mid—April to June in Mich— igan, usually on the basal one third of the upper leaf sur- face. Generally, smaller more succulent grain plants are preferred for food and oviposition. Larvae feed on the upper surface of the leaf and chew through to the lower cuticle. When development is com- plete the prepupa drops or crawls from the plant and enters the soil to pupate. Merritt (1967) reported that mortality in the pupal stage ranged from 4% to 2A% with mortality prob- ably related to soil moisture. Adults, arbitrarily called summer adults, emerge in July and feed on grasses and corn for two to four weeks. Responding to some environmental or physiological cue the summer adults seek an overwintering site and enter a state of reduced activity until the follow- ing spring. Under laboratory conditions Yun (1967) reported 65% mortality for overwintering adults held at A3°F for 90 days. POPULATION THEORY Several theories have been constructed to render numerical population change understandable (Nicholson 1933; Thompson 1939; Andrewartha and Birch 195A; Milne 1957). Under- lying all of these theories is an almost axiomatic assumption that population size cannot increase indefinitely without some upper limit. Exactly how and why populations change numerically is the major source of controversy among these theories. This review will deal with those features of each theory which contribute most to understanding the population dynamics of the cereal leaf beetle. Nicholson (1933) was the first to construct a logical and detailed theory of population dynamics and it is the basis for most subsequent theories. Nicholson (1954) used the observations of Howard and Fiske (1911) and Chapman (1928), and the mathematical models of Lotka (1925) and Volterra (1926) to postulate that a population and its environment exist in a state of dynamic balance because of density-related resist- ance to infinite population growth. The following quotation summarizes his point of view: "Populations are self—governing systems. They regulate their densities in relation to their own properties and those of their environ- ment. This they do by depleting and impairing essential things to the threshold of favorability, or by maintaining reactive inimical factors, such as the attack of enemies, at the level of toler— ance. The mechanism of density governance is al- most always intraspecific competition, either amongst the animals for a critically important requisite, or amongst natural enemies for which the animals concerned are requisites. 7 Far from being a stationary state, balance is commonly a state of oscillation about the level of equilibrium density which is forever changing with environmental conditions. Although pOpulation densities can be governed only by factors which react to density change, factors which are uninfluenced by density may produce profound effects upon the density.” Nicholson's theory can be summarized quite accurately by an oversimplified mathematical model (after Cole 1957): Nx+1 = NXROg(x), where the present population density (N ) is equal to the x+l product of the previous generation density (Nx) times the net reproductive rate (R0) times a "governing" factor (g(x)). This model is restricted to populations with non- overlapping generations. Since Nicholson concluded that pop— ulation change was a result of both the density of the exist- ing population (N) and the environment (E), then, the "govern- ing" factor, g, must be a function of both density and en- vironment and since the full effect of the environment de- termines the carrying capacity or mean density (N ), then: max g = f(N,E) = [ l — N ] , O < g < l Nicholson's theory has received both widespread acceptance and criticism among population ecologists. For example, Thompson (1956) accurately points out that the factor of chance plays a much greater part in population dynamics than Nicholson eludes. In a statistical sense, Nicholson has used a deterministic model where a stochastic model would be most accurate to describe population dynamics. Following this stochastic argument, Thompson (1956) suggests that environmental conditions met by any population vary tremend- ously in both time and space, and the mean density defined for a population is not a single event, but a distribution of events in a probability set. Andrewartha and Birch (195A) observed frequent and extreme fluctuations in Australian grasshoppers and concluded that these insects as well as many others were regulated by environment and not by density-dependent factors. However, in a Nicholsonian sense they have simply stressed the import- ance of the environment as the determinant of the carrying Icapacity (Nmax)' Milne (1957) emphasizes the incomplete nature of Nicholson's dogmatic classification of density-dependent and density-independent factors responsible for numerical popu- lation change. Milne (1957) concluded that natural enemies of a population are imperfectly density—dependent and can only control increase of a population in combined action with density-independent factors. EFFECT OF DENSITY ON POPULATION CHANGE Andersen (1957) compiled a review of the effects of density on the birth and death rate of a population. He in- vestigated Kostitzin's (1939) assumption that the birth and death rates of a population are linear functions of its 9 density. Boggild and Keiding (1958) clarified Kostitzin's (1939) assumption on the relationship of mortality and den- sity by stating that above a certain density the fraction of the population dying between birth and the adult is a linear function of its initial density. That is, X = a - bx, x where x is the initial density, y is the number surviving and a and b are constants of the equation. By a simple algebraic manipulation of this equation Beggild and Keiding (1958) showed that the survival process may be divided into two components: y = ax - bx2 where the number surviving is equal to some constant mortality factor (ax), such as genetic or intrinsic death, and a para- bolic component (bxz) showing that mortality is due to mutual influence of individuals proportional to the second power of density (x). The square of density expresses mutual influence because at a specific density (x) each individual is affected by x—l individuals so mutual influence in the total population is x(x-l). However, as x + w, x(x—l) + x2, or the second power of density. Andersen (1957) concluded from several important laboratory findings that the assumption of Kostitzin (1939) and B¢ggild and Keiding (1958) was correct. In the labor- atory, Yun (1967) showed that there was a linear relationship between the logarithm of the number of larvae placed on a 10 grain plant and the survival of those larvae. Unlike mortality, fecundity is not a linear function of density as Kostitzin (1939) assumed. Andersen (1957) con- cluded, after an exhaustive review of literature, that: "Above a certain limit of density the fecundity (n) is a linear function of the reciprocal of the density (N)." Mathematically, that is: n = a + b/N where a and b are constants. Biologically, the reciprocal of density (b/N) can be interpreted as (from Andersen 1957): 1) amount of food available per female 2) the number of oviposition sites per female 3) amount of space per female Yun (1967) showed a similar relationship between adult density and fecundity, but the densities were so unnaturally high that unrealistic interference must have occurred. Most studies reviewed by Andersen (1957) were from homogeneous laboratory conditions and populations with uniform age dis— tributions. Aside from the quantitative changes in response to density a population can also exhibit certain density—related qualitative changes. Ullyett (1950) showed that the size of adult Chtybomia chKOAOpyga decreased with increasing initial larval density and that fecundity increased with increasing female size. Greenbank (1956) showed that fecundity increased linearly with increasing size of the female pupal spruce bud- WOI‘ITI . ll EFFECT OF CLIMATE AND WEATHER ON POPULATION CHANGE The influence of climate and weather on animal popula- tions is considered by Andrewartha and Birch (1954), Birch (1957), Greenbank (1956), Klomp (1962) and Wellington (195“). In these studies climatic factors are considered as they affect insect fecundity, growth and survival. Yun (1967) showed that a day length in excess of 8 hours is necessary for cereal leaf beetle oviposition and a maximum rate is obtained at 16 hours., Oviposition also increases with increased temperature. However, his data did not support his conclusion that fluctuating temperatures had an adverse effect on oviposition because he compared a con- stant temperature treatment of 80°F to a day-night temperature of 70°F to 50°F. Under these conditions a comparison is not possible. Yun (1967) also showed that developmental time of the cereal leaf beetle decreased with increasing temperature according to Davidson's (19AM) logistic equation: developmental time temperature a, b & K = constants l+ea+bx Y Y = ———K——— where, x The effect of this relationship is such that under a con— stant temperature of 58°F larval development is complete after 27 days while at 90°F only 8 days is required. In the field situation direct solar radiation can raise the body temperature of some insects 10° to 15°F (Wigglesworth 1965). This makes application of laboratory results to 12 natural conditions somewhat difficult. Dickler (unpublished) and Yun (1967) showed how ex- treme temperatures affect survival of eggs, larvae and pupae of the cereal leaf beetle. Their data showed that survival in these age classes is little affected by the temperature regimes found in lower Michigan from April through July. Greenbank (1956) points out that a decrease in temperature not only prolongs developmental time, but it increases the amount of time the immature insect is exposed to mortality factors, or increases the probability of death. The desiccating action of low humidity and wind must have some effects on survival of young larvae at the time of eclosion and ecdysis, but this can only be inferred from the literature (Wigglesworth 1965). DESIGN OF FIELD STUDIES Sampling efficiency seems to be a universal problem in population studies and has received considerable attention by Embree (1965), Harcourt (1961a, 1961b, 1962, 1963, 196A), Hughes (1963), LeRoux, ct afi. (1963), Lyons (1964), Morris (1960, 1963). In all cases the objective of the design was to accurately and efficiently estimate absolute field densities in time and space. Most of the concern in estimating absolute density has been in determining the optimal sample unit size, number of samples, and sample frequency and efficiency. The problem is to define the universe to be sampled and select an appropriate sample unit (Morris 1960, 1955; Southwood 1966). Morris (1955) offered the following six l3 considerations for the selection of a sample unit: "1. In order for the sample to be representative of the universe, the sample unit should be of such a nature that all units in the universe have an equal chance of selection. 2. The sample must have stability. That is, the number of units available to the in- sect pepulation must not be affected by changes in growth habit of the plant caused either by intrinsic factors or by repeated insect damage. 3. The prOportion of the insect population using the sample unit as a habitat must remain con— stant. 4. The sample unit should be reasonably small so that enough units can be examined on a given plot and date to provide an adequate estimate of variance. 5. In absolute pOpulation work, where estimates of population per acre are required, the sampling unit must lend itself to quantita— tive assessments of the number of units per acre. 6. An important practical consideration is the facility with which the sample unit can be delineated in the field and collected without serious loss of disturbance of the insect pOpulation." Methods for determining the most efficient sample units are suggested by Southwood (1966), Lewis and Taylor (1967), Lyons (196A) and Morris (1963). Lyons (196A) used precision, and the time required to collect one sample unit, as the most important criteria in designing an efficient sample plan. Most of the authors mentioned above agree that the standard error should be maintained around 10% of the mean, because at this level variance of the mean estimate due to sampling is minimal. The sample size required to lower the standard error below 10% of the mean often becomes so large that efficient sampling is no longer feasible. Embree (1965) showed that the estimation of sample size, N, can easily be IA obtained by the formula: = 2 _2 N s /sX where s; is 10% of f. However, he showed that if there is a relationship between the magnitude of the mean and the variance, a table or graph of sample size (which maintains SE = 10% of E) versus the mean is helpful in determining the adequate sample size for a certain sample area. Harcourt (1961, 1963, 1964) and Richards (1961) con— sidered the problem of sample frequency in terms of the in— sect's developmental rates. They made effective use of the insect's developmental curve to predict the optimal sampling frequencies of a certain insect species. ANALYTICAL TECHNIQUES FOR POPULATION DATA In addition to estimating population size, changes in density through time must be evaluated in such a way that survival probabilities may be assigned to specific age class— es. Life tables conveniently summarize these survival prob- abilities. Various population parameters which can be cal- culated from one life table (Birch 19A8) can be compared to those of other life tables by variance analysis and/or re- gression analysis. Numerical change within one instar can be accounted for by recruitment from the preceding instar, moulting and age specific mortality. This is complicated in the cereal leaf beetle because eggs are laid over an eight week period 15 and all age classes occur simultaneously. It is not possible to follow one uniformly aged cohort in a natural field sit— uation. Separating age specific mortality in such a popu- lation is an analytical problem studied by the following authors: Dempster 1961; Kiritani and Najasuji 1967; Richards and Waloff 1954; Richards, Waloff and Spadbury 1960; South- wood 1966. Southwood (1966) explains a simple, yet very basic, method to calculate age specific mortality from this type of data. If the population is censused frequently enough an occurrence curve of each instar can be established. The area under each curve is the total number of instar—days. From this the actual number of individuals entering the instar (NI) can be calculated by dividing average developmental time (d) into instar—days (NT): N =.N_T. I d Southwood (1966) showed that the method is most accurate when the distribution of mortality is light at the beginning and heavy toward the end of the instar. If this procedure is repeated for each instar the number entering each instar can be compared to determine age specific survivorship. Richards and Waloff (195A) used the "Y"—intercept of a regression line fitted to the negative slope of the total instar occurrence curve to approximate the number of individuals entering that instar. This assumes a constant mortality and developmental time and requires a well defined peak in the total instar occurrence curve. l6 Richards, Waloff and Spadbury (1960) offer another method for analysis of instar survival. They reasoned that the total incidence (instar—days) of an instar (N) is expressed as: a a - t _ n(K —1) N‘nf K Cit-m— o where n = total entering the instar daily survival rate duration of stage time c+m x II II II And, if the observed N could be compared with what should have occurred, a-n, this difference would reflect mortality within that age class: In contrast to the previous method this method assumes much of the mortality takes place early in the instar development. The method, however, is very sensitive to accurate estimation of developmental time. Dempster (1961) treats census data as a series of simultaneous equations (Io+It) (Ado+Adt) AN = Pa - ———§——— tuI.... - 2 tua d where population change (AN) during a certain sample interval (t) is equal to the fraction of the total eggs hatching during that interval (Pa) minus the average occurrence of each age class (IO+It/2) times the mortality during the 17 sample interval (tui). The only unknowns in this equation are uI.. the age specific daily mortality rates. If 'uad’ there are more samples than unknowns the unknowns can be solved by a system of simultaneous equations. Unlike the other methods age specific developmental time is not required. However, to be most efficient the sample interval should be close to the average age specific developmental time. The method appears to be the most efficient of all the methods reviewed. DEVELOPMENT OF THE LIFE TABLE Pearl, ct aZ. (19Al) were the first to seriously apply life table analysis to the study of insect populations. As early as 1947 Deevey (19A7) criticizes ecologists for leaving the construction and analysis of life tables to stat- isticians and laboratory ecologists. He gives a comprehensive discussion of the various types of life tables and the meaning of the various parameters which may be calculated. Deevey (19A7) described a life table in the following way: "A life table is a concise summary of vital statistics of a pOpulation. Beginning with a cohort, real or imaginary, whose members start life together, the life table states for every interval of age the number of deaths, the sur- vivors remaining, the rate of mortality, and the expectation of further life. These columns are symbolized by d , lx’ q , and ex, respectively, where x stands for age. " ' Birch (19A8) improves the versatility of the life table by integrating the age specific life table and the age Specific fecundity table of the female population. From this table additional parameters, such as net reproduction rate 18 (R0) and mean length of the generation can be calculated. Morris and Miller (1954) suggest several modificatiions of the life table to make it more applicable to insect popu- lation studies. They suggest that the age column (x) should emphasize that stage where mortality occurs rather than strict adherence to chronological age. The age column might then have unequal age intervals. They also suggest another column dxF that summarizes the factors causing the mortality in that age interval. They also noted that there is little use for a summarization of life expectation (ex) in insect populations of one generation per year. Ives (196A) discusses the problems encountered in the development of life tables for insect populations. He accur- ately concludes that the single most important problem in developing life tables for insect populations is sampling. Yun (1967) constructed life tables for a laboratory population and a field population of the cereal leaf beetle. These tables were important in indicating where high mortality could be expected in the cereal leaf beetle, and what age classes needed the most detailed study. However, one life table for one generation of an insect in one environment hardly describes its population dynamics. Morris and Miller (195A) conclude that, "More valuable information can be shown, ... by continuous life tables for many generations and for different environments." It is interesting to note that in the spruce budworm study, Morris (1963) used 81 life tables to establish population trends and Embree (1965) developed 35 life tables to study the population dynamics of the winter moth. l9 POPULATION MODELS Morris (1963) states that population models "... reveal in a quantitative way exactly how much is understood about the population dynamics of a species..." The popula— tion model quantitatively explains the dynamic processes of population change. The equation reviewed in population theory NX+l = NXROg(x) is such a model. However, there are many other models which describe population phenomena. Watt (1961) proposes an approach to modeling the within-generation sur- vivorship of an insect population. He begins with a series of submodels developing the probability of survival for each age class being studied. These are constructed by serially adding the percentage values of the most important mortality factors, in the age class being considered, until the majority of the mortality in an age class is accounted for. This value is subtracted from one to give the survival of the age class. A typical submodel is: Segg = (M1 + M2 + M3...Mn) - 1; where Mn equals a mortality factor percentage. In the case of a life table the survival for the age class can be computed directly, but how mortality came about will not be understood. Then he explains total generation survival as the product of the series of probabilities of survival for each of these submodels: SGen = SEgg SI 811 SIII SIV SP SA PF; 20 where generation survivorship (SG) equals the product of age specific survival (S ) times the proportion of adults that are females (P) times the mean fecundity (F). Morris (1963) and Embree (1965) use this model to explain the dynamics of the populatons they studied. Watt's basic model, of course, must be modified according to the various interactions and properties of a particular population which might affect total survival. In developing this model, Watt (1961) discusses the history, philosophy and techniques of building inductive and mixed inductive-deductive population models. METHODS AND MATERIALS FIELD STUDY The object of the field study was to quantify numer- ical population change of the cereal leaf beetle within a generation. One method of measuring this change is by fre- quent estimation of the absolute population density through- out a generation. Accuracy of the absolute density estimate can be optimized with the selection of an appropriate sample universe, sample unit and sample size (n). Sample universe. A one acre sample universe was se— lected because an acre of oats or winter wheat was small enough to be reasonably sampled, but large enough to reflect the variance inherent in most grain fields. Since within field variance was also of interest the one acre plot was systematically subdivided into ten equal subplots from which random samples were taken. Sample unit. The sample unit could have been a por— tion of the grain plant, the whole plant or an area unit of several plants. However, wheat and oats are relatively small plants and at lower beetle densities (e.g., one egg/A00 plants) a large number of plants would have to be collected to obtain a reasonable estimate of the absolute density. The most efficient method of sampling large numbers of plants was using an area sample unit that included several plants. A 21 22 sample unit of one square yard was arbitrarily selected for the 1967 field study. During this study square foot samples were also used to estimate densities in the field. Table 1 shows a comparison of the efficiency of the two sample units. TABLE 1. EFFICIENCY OF YD2 AND FT2 SAMPLE UNITS IN 1967 GALIEN WHEAT Sample Statistics 2 ft2 needed in Unit E 52 N* sample rt2 iuo 3,432 17 17 yd2 980 52,900 5.5 A9.5 *for SE = 0.1 f When the square yard sample unit is used three times as much plant material is required to maintain the same ef- ficiency (SE = .l E) as the square foot sample unit. Sample size: Figure 2 shows that the variance in- creased proportionately with the mean in the 1967 field study. In order to maintain the standard error at 10% of the mean the sample size (N) had to be adjusted to cover most of the means expected in the field. Using the variances from Figure 2, at a mean of 10 CLB/ftz, the sample size is 25 units and at a mean of 160 CLB/ft2 the sample size is 19 units. There- fore it was adequate to remove two to three square foot samples at random from each subplot, or a total of twenty to thirty square foot samples per plot. 23 VVAHIAHEE Figure 2. l 1 1 I 1 1 l n 20 40 60 80 100 120 140 160 MM The relationship of the mean and variance in square foot samples of Galien wheat in 1967. 2A Sample frequency. In 1967 a sample frequency of one sample per week was selected. However, it was found that too much development had occurred in the population to accurately develop age specific population curves. Therefore, a sample frequency of three days was chosen for the 1968 and 1969 stud- ies because this frequency was close to the average develop- mental time of one larval instar. Field procedures. The 1967 field study included a low density area at Gull Lake and a high density area at Galien, Michigan. One acre plots in larger fields of oat and winter wheat were established at each location. The sample unit consisted of the grain plants in one square yard. Each sample unit was randomly located and removed from each of the 10 subplots once a week during the egg and larval stages. The samples were returned to the laboratory in plastic bags for counting. The pupal stage was sampled by taking a one half square yard soil sample 2 1/2" - 3" deep from each subplot. The soil was washed through 1/8" screen which separated the soil from the pupal cells (see Figure A). Before the summer adults began to emerge, 3 one—milliacre cages were placed at random throughout the plot (Figure 3). Newly emerged adults were removed from these cages at two to three day intervals. The 1968 and 1969 field studies were similar to the 1967 study but included three Michigan locations; a low den- sity area at East Lansing, a medium density area at Gull Lake, and a high density area at Galien. Three 1 ft2 samples in 1968, and two in 1969, were randomly selected from each 25 subplot at 3—A day intervals during the egg and larval stages. The processing of these samples and sampling for pupae and emerging summer adults was the same as in 1967. An additional sample was added to the 1968 and 1969 field studies. In order to obtain an independent estimate of oviposition during the sample interval, a series of plants consisting of 2 linear row-feet were marked off in each sub- plot in both oats and wheat. At each sampling the eggs were counted and pinched so that no eggs remained after counting. This was continued until oviposition had ceased. Temperature and humidity were recorded on hygrothermo- graphs placed on one of the plots at each location. Solar radiation was recorded on pyroheliographs at Gull Lake and Galien in 1968 and 1969. EXPERIMENTAL CAGE STUDY To quantify the effect of density on age specific survival, a gradient of very low to very high density popu— lations was established in 6—milliacre cages. These cages were placed in oats at the MSU Entomology Research Facility in East Lansing. This study was very similar to the field study except many more densities could be studied at one time and place with a minimum of variance. The cages ex— closed any native predators and moderated the influence of meteorological events on survival. In 1967, populations of 100, 200, 500, 1000, 2000 and 4000 spring adults were established in each of six 6-milliacre cages. In 1968 and 1969 twelve cages were 26 available so four different densities (100, 500, 2000, 5000 spring adults/cage) were replicated three times. In each cage twenty—five one row-foot sample Sites (15 in 1969) were staked out at random. Unlike the field study, the phenology of the organism in each sample site was followed in time. Therefore, the number occurring in each age class at the sample sites was recorded at a sample frequency of four days. No plants or larvae were removed. After pupation all plants were removed from the cage. The soil from each row-foot sample site was removed, and processed in the same way as the field study. Emerging adults were collected from the cage three times during the emergence period. In 1969 three oviposition sites, Similar to the 1968 and 1969 field studies, were established in each cage in order to estimate oviposition during the sample interval. QUALITATIVE EFFECTS OF DENSITY Size; To test the qualitative effects of density on the cereal leaf beetle the sizes and weights of newly emerged adults from all the field studies and the 1967 caged density studies were compared. Thirty individuals from each population were placed in a laboratory oven at 106°C for A8 hours and the dry weight of each individual was measured on a Kahn Electrobalance with an accuracy of i0.0005 mg. The elytral length was measured with an optical micrometer. Fecundity. Several studies have shown a relationship between the size of female pupae, or resultant adults and the number of eggs they are capable of laying. To test this 27 relationship in the cereal leaf beetle two experiments were performed. Newly emerged spring adults from a high density area (Galien), a medium density area (Gull Lake), and a low density area (East Lansing) were placed in cages in the lab- oratory and in the field under natural conditions. The num— ber of eggs laid by each female was followed at various in- tervals throughout the life of the female. In the laboratory, twenty pairs of beetles from each of the three density areas were placed in cellulose acetate cylinders atop a 2-inch pot of small barley plants. The pots were replaced every three days, when the eggs were counted. Counts continued until the female died. Males were not re- placed if they died before the female. The laboratory was maintained at a constant temperature of 78°F and 50% R.H. with a 16 hour day. In the field fecundity study, sixty pairs of beetles taken from the same areas as those in the previous experiment were placed in separate sleeve cages. The cage enclosed an individual wheat plant. This test was set up during the last week of April, 1969. Egg counts were made at weekly intervals. The evaluation procedure was the same as in the laboratory study. The sleeve cages were constructed of nylon screen formed in an eight inch cylinder, 32" tall. The top was formed by an 8" embroidery hoop which was attached to the top of a 36" stake. The seam of the cylinder was stapled to this stake for support. Destruction of the eggs after counting was accomplished with a long dissecting needle. 28 One plant provided sufficient food for the entire life of the adult female. DEVELOPMENTAL RATE Larval developmental rates were observed at differ- ent temperatures so temperature-dependent developmental curves could be constructed for each instar over the range of temperatures studied. Twenty—four larvae on individual A-inch pots, were placed in each of three Sherer—Gillette table top growth chambers maintained at 60°F, 70°F, and 80°F (with 70 to 80% R.H. and 16 hour day). The age class status of larva, established by exuviae and head capsule width, was recorded daily. 29 Figure 3. One milliacre emergence cages used for summer adult cereal leaf beetles. Figure A. Left. Screening technique used to separate CLB pupae from soil. Figure 5. Right. A CLB ovary showing seven ovarioles. RESULTS FECUNDITY OF THE CEREAL LEAF BEETLE Southwood (1966) defines fecundity as the total egg production and fertility as the number of viable eggs laid by a female. Since fecundity is the numerical input of a population system, the factors which determine this input are of a major importance in population studies. It was hy- pothesized that ovarian composition, temperature and adult size were the most important factors influencing egg produc- tion in the cereal leaf beetle. Ovarian composition. The insect ovary is composed of a number of ovarioles responsible for egg production. Since the number of ovarioles can directly determine fecundity, ovaries were dissected from sixty spring adults from Galien, Gull Lake and East Lansing to determine variation in numbers of ovaries and ovarioles. The size of these spring adults varied considerably. Figure 5 shows a dissected ovary with seven ovarioles and eggs in various stages of development. All females had two ovaries each containing seven ovarioles or a total of fourteen ovarioles per female. Temperature. Yun (1967) showed large differences in egg production at two different temperature regimes in the laboratory. However, more information was needed to estab— lish the influence of temperature on egg production in the 30 31 field. The ovipositional activity of the cereal leaf beetle was measured every three days at oviposition sites in the field. When these results were plotted against the mean maximum daily temperature measured during the three day sam— ple interval a definite linear trend was observed. Figure 6 shows that the rate of oviposition increased linearly from 50° to 75°F. The linear relationship cannot be extrapolated beyond the endpoints of this range because the rate of ovi- position quickly becomes non-linear at low and high temper- atures. The cereal leaf beetle does not oviposit during the night, so maximum daily temperature was used as an indicator of daily temperature influence. Other meteorological events, such as solar radiation and wind influence body temperature and, hence, oviposition. However, the strong relationship between temperature and rate of oviposition shows that these factors are relatively minor. Table 2 shows the fecundity of three differert populations reared in the laboratory and field. In the laboratory, at a constant temperature of 78°F, mean fecundity ranged from 205 to 360 eggs per female. In the field mean fecundity ranged from 53 to 61 eggs per female. The mean daily temperature during the field experiment was 62°F. The relationship between temperature and oviposition rate suggest that this suboptimal temperature regime was probably responsible for the large difference in mean fecundity between the laboratory and field experiment. The fecundity of the field fecundity experiment was lower than that which expected in natural populations, as in Figure 6, because the cages modified the warming effect of direct solar radiation. 32 lfi —- l4 ” 12 ” Eglfl " 2E"- 5.. 4.. 2 50 BB 10 80 MEAN MAXIMHM flAllY TEMPEHAIHHETPT Figure 6. The relationship between egg production per female and maximum daily temperature in the cereal leaf beetle. 33 Logan ppmpcmpm a+ Eopmmnm mo moonwopx Amhmpv *mma.m mm.ms sm.mn mm.mA seam mesa oa.xa so.H mw.sm om.am mm.om sees Amwwov *mmH.m mm. A ma.mn ma.ma Assesseme oa.xa mm.o ss.om Hm.:m so.mm sees oqum Anamov *Hm.m 33.3“ mm.:A 00.:A seam mesa oa.xm sm.m me.mm om.sm oa.sm sets Amwwmv *Hm.m ma.mmA :m.msA 00.0:A Assesseeu oa.xa mm.m mm.mom ms.mom om.mmm same mmoeamomlm wcfimsmq ummm oxmq HHSU cofiamu memzmq Qpma pew ado pom mo>a36 COHpmasoom .m opsmfim mwdn 00 OH zli/lflfiflnfi .OOH ONH 39 Egg; The total number of eggs laid in a sample unit was measured directly in 1968 and 1969, but in 1967 the num— ber laid per sample unit was calculated by the total inci- dence method. Survivorship of the egg was measured directly in the laboratory. As samples came into the laboratory for counting, eggs were placed in petri dishes on moist filter paper and incubated at 80°F. Results are presented in Table A. Some mortality in Table A was unnatural because of dessi— cation and fungal growth in a few petri dishes. TABLE A. SURVIVORSHIP OF CEREAL LEAF BEETLE EGGS Locality Year Host Plant % Survival Galien 1967 oats 85 wheat 69 Gull Lake 1967 oats 93 wheat 100 ' Galien 1968 both 91 Gull Lake 1968 both 78 Dickler (unpublished) found similar values for eggs laid in the laboratory. ' For purposes of the survival analysis egg survival was accepted aspa 90 percent constant. First instar. The number of individuals entering instar I was calculated by multiplying the total number of eggs laid by egg survivorship or 0.90. The number entering instar I could also be calculated by dividing the total incidence of the first instar by its developmental time. However, using the formula below, the survival values of AD the first instar were plotted according to their egg densities along the median development line in Figure 9. A hypothetical distribution line was drawn from the origin of development to total development through the cluster of high density points on the median developmental line. This distribution suggests that mortality is high early in the instar at high densities. Therefore, the actual number entering the first instar (100% level in Figure 9) would be considerably under- estimated using the total incidence method. S = total incidence II/dev. time II I # eggs x 0.90 Second and third instar. Survivorship for these in- stars was calculated by the total incidence method outlined above: = total incidence III/dev. time III II total incidence II/dev. time II S = total incidence IV/dev. time IV III total incidence III/dev. time III Fourth instar. Survival of the fourth instar was cal- culated by dividing the total number to pupate by the total number to enter the fourth instar: S = absolute density of pupae IV total incidence IV/dev. time IV ‘ll 7. SllflVlVAl Figure 9. Al ---——---aunlu. I ' lllllil] ilSlll Ill tensity g3 1 high I Isily El 1 letipu lav. lllll luv. llll flfVElflPMENTAl 1le (HAYS) The distribution of first instar survival (solid line) calculated by the total incidence method. The survival values corresponding to egg density are plotted along the median line. A2 Pupae. Pupae were sampled after all larvae had entered the ground. Adult emergence prior to pupal sampling did not affect the estimate of absolute density because emp- ty as well as full pupal cases were recovered. Pupal surviv- orship was calculated by dividing the absolute density of pupae into the absolute density of adults recovered in the emergence cages . S _ absolute density of summer adults P absolute density of pupae Total larval survival. Total larval survival was calculated by dividing the absolute density of pupae by the absolute density of eggs: S = absolute density of pupae L absolute density of eggs Egg survival, included in this calculation, was defined as part of total larval survival. Although within—generation survival (i.e., the fraction surviving from egg to adult) is easily calculated by dividing the egg density by the adult density, the total larval mortality was of most in- terest in development of the population model. The results of the preceding calculations for age specific survival in the field and cage studies are tabu- lated in Tables 5 and 6. Mortality, rather than survivor- ship, was used in these tables, but the transformation back to survivorship is simple (SX = l-MX). There were two causes for the negative mortalities seen in these tables: A3 poo“ osmsvm hoo* m.© a.» 3.: 30336 ppmpcmpm m.om H.03 m.mm can: 00 ms 3 mm 3 ma ma 0 I- as om: mm mm 03 a 3 me am mm mm AH- :m was m: o seaweed In ms me as om om- om 3 mo ensm so we mm 3m mm as: 33 0 mm mm mm: 3s mm m3 H 3 3m am mm mm em H. mm 033 0 mm mm mm om mm mm 3 3 mm mm :m mm m: mm or 3mm 0 :3 am om Hm mm m: mam 3 mo exam Haze m3 mm H m3 mm s: mm o 00 mm mm: Hm om om omm 3 am we mm me am mm mm cam o 00 mm mm H: mm a: mmm 3 we H: mm Ho mm m: 3m mooa o 00 am mm H: mm mm omfi 3 mo seaanu Hausa Hs>an 3H HHH HH H HBBOB *mmmm HmpoB mono snow coameoq Aooaxv NBHQ¢BmOZ mm¢do mw< WMHQDBW quHm mood Qz< wood .wmmfi ZH MAEmmm mnnq 3H HHH HH H Hapoe mmwwm mpa5p< Lonesz mcfinam Hmpoe amassz owmo AOOHXV MBHA.10 Density l7.A7 P<.01 Interaction 1.A7 P>.10 *with 1,1A degrees of freedom Therefore, it was hypothesized that the variability of total larval mortality was attributable to the differences in density from cage to cage. The densities in the 1969 cage density study were designated by the initial number of adults placed in each cage as described earlier: 5000 adults, 2000 adults, 500 adults, 100 adults. The total larval mortality values from Table 6 were classified according to these lettered densities in Table 8. A one-way A? analysis of variance of these values showed that there was a significant difference in total larval mortality amongst these four densities. TABLE 8. ONE-WAY ANALYSIS OF VARIANCE OF TOTAL LARVAL MORTALITY IN THE 1969 CAGE DENSITY STUDY Density rep H K M L 96 90 57 A8 98 56 A8 A9 92 56 ‘ 60 23 F3,8 = 10.1A: P<.01 Relationship of host specific mortality to density. The results of the field and cage studies indicated that much of the variance in total larval mortality could be explained if the relationship between larval mortality and density was understood. Because of the convincing laboratory studies discussed earlier it was hypothesized that there was a linear relationship between density and larval mortality. In order to investigate this hypothesis total larval mortality in wheat and oats was plotted against the total number of eggs laid per square foot for each population in the field and cage studies (Figure 10). It was shown earlier that there was no significant difference in total larval mortality between oats and wheat. However, upon closer in- spection of Table 7 it was discovered that the broad classif— ication of high and low density used to test this hypothesis A8 masked a real survival difference in the two host plants. Figure 10 shows that total larval mortality in wheat was higher than in oats over all densities. Figure 10 also shows that total larval mortality is a linear function of the logarithm of density and increases with increasing den- sity. VARIANCE OF AGE SPECIFIC MORTALITY The components of total larval mortality must be an- alyzed separately in order to understand the relationship between mortality and density. The simplest hypothesis to explain this relationship is that mortality in each instar increases as density increases. It is important to first investigate the relative importance of mortality in each instar to the variance of total larval mortality. Table 9 shows the correlation analysis between total larval mortality and instar mortality from the field study (Table 5) and cage study (Table 6). The negative mortality values of the fourth instar in Table 5 were adjusted to zero for this analysis because the total incidence of the fourth instar, in these cases, was under— estimated due to very long sample intervals (in the 1967 field study, Table 5). From this analysis it appears that the first and fourth instar were the most highly correlated with and account for 29% to 68% of the variance of total larval mortality in the field and cage studies (Table 9). A9 .sesen camfie an» as muamcmc can mafiampmoe Hm>nma ofimaoodmnpmon no ofinmcoaumHoa one .oH onswam 11:331. :23 ===_ ==_ =— _qd.qd _‘ L _ ...... .o. _. _ _____ q _ _ _ O \ - as "1" I 1.. l V H I" v n == 1.. mu" 0 l H II; v ] l l . an .in O a:— 50 TABLE 9. CORRELATION ANALYSIS BETWEEN INSTAR AND TOTAL LARVAL MORTALITY IN THE CEREAL LEAF BEETLE Age Class r2 Significance Field I 0.29 .01.10 III 0.003 P>.10 IV 0.20 .05hma comm CH 3pflmcop moa new mafiamupoe pzoomoa consume QHQmCOHpmHoL 039 .NH onswfim 3:23: :32.— ..._ ... : .... ... : :04 q a — dqdqqdd d 1 dun-dd d 1 A dH-dfl - 4 uddfifid — - — qq-qqfifiq H . - 1: 1 .u 3. ...._n-_ ..._—_»-_ d_~q_ u u - ___d14- - — —_—-dq u d - qqqddd q 4 -««-<__ q d qqqdd—q d d O \ \ I 0 a 1 .. J l.- __.-_”-_ ...—a-_ llllllllfll X 53 The relationship between log density and mortality was also graphed for the second and third instar in Figure 12. How- ever, data from the cage study was not used because sampling problems caused gross underestimation of second and third instar mortality. These lines show that second instar mor— tality is relatively constant over all densities in the field. There is a slight tendency for third instar mor- tality to decrease with increasing density. Regression and correlation statistics of the graphs in Figure 12 are listed in Table 10. TABLE 10. THE RELATIONSHIP BETWEEN INSTAR MORTALITY AND LOGIO DENSITY IN THE CEREAL LEAF BEETLE Age Class a b r2 Sign. Field I -lA.l 19.A 0.51 P<.0l II 30.8 0.00A 0.008 P>.10 III 58.8 —10.0 0.11 P>.10 IV 6.7 21.3 0.26 .05.10 Only the correlations between log density and first instar mortality were significant at greater than the 1% probability level in the field and cage studies. In the cage study as much as 65% of the variance in first instar mortality could be accounted for by density. Figure 13 shows that relationship between age specific mortality and density in cats and in wheat. There is no 5A difference between host plants and no significant correlation between density and mortality in the second and third instar (Table 11). However, Table 11 shows that 67% and 56% of the variance in first instar mortality could be explained by density in oats and wheat respectively. In the fourth in- star the slopes were similar in both host plants but mortality was 30% higher in wheat than in oats. Table 11 shows that A8% and 66% of the variance in fourth instar mortality could be explained by density in oats and wheat respectively. TABLE 11. THE RELATIONSHIP BETWEEN DENSITY AND AGE CLASS MORTALITY CLASSIFIED BY HOST PLANT IN THE FIELD STUDY Instar coefficient of determination (r2) Oats Wheat I 0.67* 0.56* II 0.01 0.10 III 0.16 0.09 IV V 0.A8* 1 0.66* *Significant correlation: P<.01 MODEL OF WITHIN-GENERATION SURVIVORSHIP Within-generation survival (SWG) was defined as that fraction of the population which survived from the egg to the adult. In the cereal leaf beetle this includes survival within the egg, four larval instars and pupa. The equation proposed earlier serves as a generalized model for within- 55 .oapoon mmofi Hmoaoo on» mo nmpmcfi Hm>pma some CH mafimCop woa pew mafiamppoe pcoonoa coozpon afinmCOHumHop one .ma opzwfim 3:33: :33. .... ... ._ .... .._ ._ quad: . q q «dqqdq—_ q -qqq4_ u q q -_u-—— _ qqqqq— u— — qqqq-q— u — llll'lllfll % ...._n-_ ....—_a.. dud-u — a q dfid—qd q — u du—q-qq d q qua. d u - fidu~14 q q - udqdddfi- a L O o L - o |\\\ 00 o o b I 1 t 1 : ""Il"'bl"h I. 1 """""" . . ........ 1 1 1 : . I. l O n . ... 1 a 1 1 .. 1 1 .. -._u-_ _ .._”—_ 56 generation survivorship of the cereal leaf beetle: ch = SE' SI' 311' S111' SIV' SP Survivorship in the egg, second instar and pupa were constant, with random variance, over all densities: SE 0.90 (from laboratory results) S 0.68 (from field study) II S 0.70 (from field study) P However, survivorship in the first, and fourth instars varied predictably with density and host plant. The regression statistics from Figure 13 were used to form regression equa- I’ SIV at the densities and host crops studied. The regression statis- tions for field populations which would predict S tics were divided by 100 to transform them from percent to fractional values: sl = l-(-.31 + .26 log x) S = l-( .02 + .15 log x) where, x is the den- I(wheat) sity in total eggs SIV(oats) = l—(-.ll + .26 log x) per square foot SIV(wheat) = SIV(oats) + '30 when these components are combined the two-factor model takes this form: SG(oats) (0.9)-(l-(-.3l+.26 log x))-(0.70)-(0.60)- (l-(-.11+.26 log x))-(.70) SG(wheat) (0.9)-(l-(.02+.15 log x))-(0.70)-(0.60)- (SIV(oatsP +°30)'('70) Although total within-generation survivorship was 57 measured, only the accuracy of the most dynamic portion of the model, total larval mortality, need to be tested. Pupal mortality was considered constant. Using the same type of model, total larval survivorship (including egg survival) was calculated as: Figure 1A shows the predictive value of this model. The observed values are from the field study and the calcu— lated values from the above model, using the density values in Table 5. A regression analysis of the observed and calculated total larval mortality showed that host plant and density accounted for 63% of the variance in total larval mortality. Climate, locality, time of planting and sample error probably account for most of the remaining 37%. However, only sixteen populations were studied over the three year study period and any attempt to factor mortality beyond host plant and density would lead to very tenuous results. The analysis of the relationship between observed and calculated total larval mortality, in Figure 1A, indi- cates that the calculated mortality at low densities is con- sistently overestimating the observed, but approaches reality at higher densities. Another weakness of the model is that it lacks the feature of time. It treats age classes as total entities and does not explain the interactions of age classes as 58 ZlABVAl MINIMUM-ABS. l l l l l l l as 75 15 95 '/.lAHVAl MflHIAlIIY "CANE. Figure 1A. Comparison of observed and calculated total larval mortality of the cereal leaf beetle using the two—factor model. 59 they progress in time. At high densities the probability of survival for a first instar is greater early in May when there are no other instars present than in June when second, third and fourth instar larvae are feeding. Also, develop- mental time is faster at warmer temperaturres later in the generation so exposure to physical mortality factors is less than for instars occurring early in the generation. QUALITATIVE EFFECTS OF DENSITY Aside from the strictly numerical relationship of mortality and density, qualitative changes in the pOpula- tion can result from the effects of density. The mean elytral length and dry weight of 30 emerging female adults from different populations was plotted against the logarithm of density of that population, in Figure 15. The graph shows that the mean elytral length and dry weight of the female cereal leaf beetle decreases as log density increased. The same results were seen in emerging male adults, although males were generally smaller. These results account for the difference in mean elytral length for beetles from Galien, Gull Lake and East Lansing in the fecundity experiment. The relationship between density and larval head capsule size and dry weight was also investigated. Unfortunately, the larval samples were taken from each study area at one point in time and were not representative of the total instar pop- ulation. 60 MEAN FEMAIE WEIGHT (ME! “I ~L _~ mg mg 6N ==_ .cofiumasmoo Hm>ama Eseflxme wcfipooopo new pew oapoon mama Hwopoo madame wasps wcHwLoEo on» mo cpwcofi conpzfio cmme pew pnwfimz :mms soozpoo aficncofipmaon one _N_= »__M.L= La>=o 0c>0 0<>0 D 888 888 888 888 888 Ln+mc_ 0N.0 O O 383 8 888 888 000 000 000 000 000 00 888 888 930w 00.0 230 930 0n.0 no; mm; mv.o 05.N nN.n mn.0 no; n_.N 070 50; no.0 888 838 888 00° GOO 000 N Lo+mc. Ln+mc_ mpmo 5N.0 av; 50.. 5N.0 on; 00; Nn.o n5; 50.N 5m.0 n0.N 50.N 0_.0 .0; no.0 Lo+mc_ 9900. 00.0 00 00.0 00 00.0 .m 00.0 00 00.. 00 00.. _m 5~.0 00 00.. 00 55.0 m 0~.0 00 00.. 00 00.. .m 00.0 00 .0.~ 00 00.~ .m 00.0 00 .0.~ 00 50.0 .m .5.0 00 .0.0 00 50.0 .m .5.0 00 00.0 00 00.0 .m 00.0 00 00.0 00 00.0 .m No.0 00 00.0 00 0..« m 0000 +000 wocoumuc_ _n+05 _3000 :30 o_ :30n_ :30 o. :30 50 :30 v0 >3..0 >0: em >0: .0 5oz 5. 0+mo 92 000 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.00 00.0 00.0 05.00 00.5 00.00 00.00 00.0 00.0 00.0 050:0 000 .00.0 50.0 00.0 00.0 00.0 00.0 00.0 00.0 00.00 00.0 00.00 00.00 50.0 00.00 05.00 00.0 00.00 00.00 50.0 00.00 05.00 00.0 00.00 00.00 ”cyan 00.0 50.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 0.030 000 00.0 00.0 00.0 «0.0 00.0 00.0 05.0 00.0 00.0 00.0 00.0 00.0 05.0 00.0 00.0 50.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 nun-:0 000 00.0 00.0 00.0 00.0 05.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 05.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 0 nun-am 0000 Ho oo.n on.n 00.” an.o “0.“ 50.“ no.” no.” so.“ n~.o o... n~.n on.o~ 00.0 o~.c on.“ 5a.." «0.00 50.nn ....“ no.0 no.0 00.nd n~.o« nn.cn “...“ 5s." a... on.- oo.o~ oa.0~ oo.~. .5.0 0n.~ a... o..o0 nn.n~ oo.o~ oo.~ no.0 «... 0n.- o~.o~ o..«n an.o no.“ .0.~ ~o.. on." ou.a~ an.o o~.o an." an.n no.o a... o~.o nn.° n..o «0.0 90.9 as.“ oo.o o0.o oo.o do.“ oo.e co.” n u nus-nu «cu-a0 pmmnz 005 00.0 00.0 00.0 00.0 05.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.5 50.0 00.0 00.0 00.0 00.00 00.00 05.0 00.00 00.00 00.0 00.00 00.00 00.0 00.00 00.00 00.0 00.5 00.00 05.0 00.0 00.0 00.0 00.0 00.00 0 u-uo-u 5000 00.0 00.0 00.0 00.0 00.0 00.0 05.0 00.0 00.0 00.0 00.0 00.0 00.0 50.0 00.0 00.0 50.0 05.0 00.0 00.50 05.00 00.0 00.00 00.00 00.0 00.00 05.00 05.0 00.00 00.000 00.00 05.05 00.000 00.00 00.00 00.000 00.0 00.00 00.000 con-000:0 0nuoh an an m 050 N am an m can a“ an a» m can nu an an m can .0 mm a» M can 00 um an m can «0 an an M £50 0 an an M can 0 mm an M can 0 an an N ha: o~ mm a» m an: o~ an an m 5.: 0“ an an m 5.: an 9.00 00.0 94 n0 00.0 00.0 00.0 00.0 00.0 00.0 ou.0 ~n.0 00.0 50.0 00.0 on.~ n0.o n0.~ 0n.N ~5.o nN.~ oo.~ 05.0 -.N 00.n no.0 50.5 00.N« 05.0 ~0.~ 0n.n no.0 00.n 00.0 no.« n~.n 00.0 00.0 00.0 00.0 00.0 00.N o0.“ usaau nun no.~ 50.0 ~5.0H on.o on.H an.~ 05.0 an.n on.n No.0 o~.0 oo.- 5o.n No.0 00.nu 00.n No.5n nn.0~ no.“ 0n.o 00.0N 0~.N 00.0 n«.a~ oo.n n5.5~ 00.nn 5n.N 00.0u nu.0~ Ho.N on.NH n0.on 0H.~ no.0 no.0 Ho.u o~.n 00.0 Acuou no.a 5~.0 u5.0u 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 oo.o 00.0 00.0 oo.o 0.090 50 00.0 00.0 00.0 on.o Nn.~ Ho.~ on.0 0N.N no.~ on.o Ho.~ n5.o 50.0 oo.n n5.n 00.0 no.n n0.H NH.0 Nn.o 0N.0 no.0 NN.o no.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 0 HCUIBH 05 and 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 an.o oo.o on.o 00.0 00.0 n~.o 05.4 00.0 05.n on.0 an.o nn.0 an.d on.n nn.~ n0.~ 55.0 no.0 ~0.n 00." nn.0 no.0 00.0 on.n n0.n 0~.n n0.n nn.0 «0.0 00.0 on." 50.0 00.N 00.0 nu.0 no.0 00.0 0o.n no.0 05.n no.0 05.0 N~.0 0n.n no.0 n0.~ 00.0 0H.o 00.0 no.0 00.0 0n.0 00.0 nn.0 00.0 00.0 00.0 on.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 n « uuuoan qu-nn mpmo 00 00.0 00.0 00.0 00.0 00.0 00.0 no.0 NN.0 no.0 nn.o 50.0 nn.o 00.0 0H.“ 00.N No.0 AH.0 oo.n nn.0 5n.u n0.n 0H.H oo.n n~.0 00.0 0H.N n0.n n0.o no.d no." 00.0 00.N n0.~ 00.0 00.0 00.0 on.o «n.o 0H.o a unu-an nnn 00.0 00.0 00.0 00.0 an.o nu.0 50.0 an.o 0H.o 50.0 00." 00.u 00.0 «0.0 no.0 ~0.N no.0 05.nu 50.H nn.o no.uu on.u 50.5 00.nd oo.n no.5u no.0N 0~.N no.0H 00.A~ oo.~ 0~.nu on.5N 0~.H no.0 no.0 50.0 5o.n oo.n OUQO‘uOflH Huuoa an an m ”an «a an an m Han ou an an m «an 5 um an m «an N a» an m can o~ an an m can n~ an an m can on an an m can 0‘ an an m can «a an an M flan 0 an an M can n an an m 5.: o~ an an M 5a: on alum Quin 00.~ o~.n 0H.n 05.0 00.n on.0 nn.~ u5.n «H.o no.d 5N.5 no.nd 55.N on.un on.n« 50.~ on.0 nn.5n oo.~ 05.0 nd.n~ o5." 05.0H 00.0n no." NN.0 nu.ou no.“ on.n~ nn.0n flinch 00.« o~.n on.n 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 cunnn 5n 00.0 00.0 00.0 00.0 on.d oo.~ No.0 no.0 0N.n No.~ on.0 05.n "0.0 N0.H 00.n 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 00.0 oo.o 00.0 00.0 00.0 0 nan-an wQDBm quHm Uszz