Historic, Archive Document Do not assume content reflects current scientific knowledge, _ policies, or practices. ees Topics in Forest Entomology , elected Papers from the XVth International Congress of Entomology Current Topics in Forest Entomology Selected Papers from the XVth International Congress of Entomology Washington, D.C, August 1976 W.E. Waters, Editor (University of California, Berkeley, California) United States Department of Agriculture Forest Service General Technical Report WO-8 February 1979 a Eee ee ET, Se NE ee ee For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402 This publication reports research involving pesticides. It does not contain recommendations for their use, nor does it imply that the uses discussed have been registered. All uses of pesti- cides must be registered by appropriate State and/or Federal agen- cies before they can be recommended. Certain pesticides can be injurious to humans, domestic ani- mals, desirable plants, and fish or other wildlife -- if they are not handled properly. Use all pesticides selectively and care- fully. Follow recommended practices for the disposal of surplus | pesticide containers. PREF ACE The papers in this volume represent a special selection of those presented in the technical sessions of the Forest Ento- mology Section of the XV International Congress of Entomology in Washington, D.C., August 19 to 27, 1976. These sessions, a major component of the Forest Entomology program, were organ- ized under the eight topics by W.E. Waters, Section Chairman and Organizer, with the assistance of Harold 0. Batzer, H.C. Coppel, C.J. DeMars, A.T. Drooz, W.J. Mattson, M.L. McManus, J.B. Simeone, and D.L. Wood. Six related topics comprise the subject matter of this volume. With the participants responsible for each, these are: I. Detection, Evaluation, and Prediction of Forest Pest Insect Populations -- C.J. DeMars, Forest Service, U.S. Department of Agriculture, Berkeley, California. II. Assessment and prediction of Pest Insect Impacts on Forest Uses and Values -- H.O. Batzer, Forest Service, U.S. Department of Agriculture, St. Paul, Minnesota. III. Concept and Practice of Integrated Pest Management in Forestry -- D.L. Wood, Department of Entomological Sciences, University of California, Berkeley, Cali- fornia. IV. Biological Control of Forest Insects -- A.T. Drooz, Forest Service, U.S. Department of Agriculture, Re- search Triangle Park, North Carolina. V. Novel Approaches to Forest Insect Control -- H.C. Coppel, Department of Entomology, University of Wisconsin, Madison, Wisconsin. VI. Forest Insect Ecology and Control -- M.L. McManus, Forest Service, U.S. Department of Agriculture, Hamden, Connecticut. At the Congress, the sessions on Topics V and VI met jointly with the Biological Control Section. The papers of a session entitled The Role of Arthropods in Forest Ecosystems were considered to be of broader interest to forest ecologists and resource managers and are to be published separately by Springer-Verlag, New York, under the editorship sk ok ak of W.J. Mattson. The subject matter of another session, Insect Pests of Wood in Structures (co-sponsored with the Stored-Prod- ucts and Structural Insects Section) also justified separate publication of its major papers; these will appear in the journal Organismen und Holz (Duncker und Humblot, Berlin), edited by Professor Giinther Becker. A number of papers evoking interest and discussion at the Congress unfortunately were not suitable for publication because of special use of visual materials, an audience re- sponse format, or other technical reasons. The support of K.R. Shea, Director of Forest Insect and Disease Research, and M.E. McKnight, Staff Entomologist, Forest Service, U.S. Department of Agriculture, Washington, D.C., is gratefully acknowledged. W.E. Waters., Editor The statements of the contributors from outside the U.S. Department of Agriculture may not necessarily reflect the policy of the Department. 1166 ILIOILS CONTENTS Detection, Evaluation, and Prediction of Forest Pest Insect Populations Systematic Variable Probability Sampling for Estimating A Douglas-Fir Cone Population Trend Prediction in Forest Insects Detecting Forest Insect Pests in Canada Survey of the Nun Moth, Lymantria monacha L., in the Swiss Alps by Means of Disparlure Assessment and Prediction of Pest Insect Impacts on Forest Uses and Values One Aspect of Damage to the Forest Caused by Insects Measuring Damage to Lodgepole Pine Caused by the Mountain Pine Beetle Economic Impact of Mountain Pine Beetle on Outdoor Recreation Impact Analysis, Interpretation, and Modeling Concept and Practice of Integrated Pest Management in Forestry Night Application of Aerial Sprays Using Multi-Engine Aircraft, Inertial Guidance Equipment, and Incremental Application Technology for Insect Control, with Special Reference to the Spruce Budworm Page ity) he) 53 33 39 43 50 55 55 IV. CONTENTS Organization and Implementation of Comprehen- sive Research and Development Programs on the Gypsy Moth, Douglas-Fir Tussock Moth, and Southern Pine Beetle in the United States Discontinuous Stability in a Sawfly Life System and Its Relevance to Pest Management Strategies Quantitative Evaluation of Pest Management Options: The Spruce Budworm Case Study Integrated Management of Pine Weevil (Hylobtus abtetts L.) Populations in Sweden The Concept of Impact in Integrated Pest Management Biological Control of Forest Insects Some Recent Approaches to the Biological Con- trol of Forest Insect Pests in Japan Operational Use of Bactllus thuringtensts against the Spruce Budworm Biological Control Prospects of an Egg Parasite, Telenomus alsophtlae Viereck Insect Parasites as Regulators of the Gypsy Moth Population at Hawk Mountain, Pennsylvania Novel Approaches to Forest Insect Control Exploiting Olfactory Interactions between Spe- cies of Scolytidae Development and Evaluation of Synthetic Inhib- itors for Use in Southern Pine Beetle Pest Management 64 68 82 103 110 117 in L7/ 120 123 128 135 135 CONTENTS Page Manipulating the Entomophagous—Mycetophagous Nematode, Deladenus Siricidicola, for Bio- | logical Control of the Woodwasp Strex Noc- : tilto in Australia 144 Reproductive Incompatibility in Prtyogenes | chalcographus (L.) 148 | The Potential of Insect Growth Regulators as i Ecologically Acceptable Agents for Con- | trolling Forest Insect Pests 151 Using Behavior Modifying Chemicals to Reduce Western Pine Beetle-Caused Tree Mortality | and Protect Trees 159 VI. Forest Insect Ecology and Control 165 | Several Types of the Mycetangia Found in (Scolytid and) Platypodid Ambrosia Beetles 165 | Genetic Features of Douglas-Fir Tussock Moth Populations 167 | vii tt So Sry stu?’ — a a a ” Fe ia ‘ rk ey gre 2 i’ Pre ia AS siya? an iad), oe “ ~ae@ er ee | Prt sary’ Ate. Cert ty fei tieres of — —. wt ee ere% ets "ves : tour cs x wre ad h i qn a ay o Por) ~ Fer a . as nA | > Re? Geach ext) 0 ie. boi Gone ae he "WA y Cary a P Yee e.. - ia ‘ hh eel ee i) ae | a ‘= ; is \ ¢ fo ae 7 & ~ os, hs i Oi ap Se p ‘i yi "4 oe 4 7 . oe ; = j fe 1 4. ‘ : . o> ‘ ; @ i vee - 2 j : pptme wy . ¢ foe aes si be p> Br’, » ieee nt be FFieder: | ot os sy gh as ¢ = ‘pears Wee a wy as ae a oft — . I. Detection, EvaLUATION, AND PREDICTION OF Forest Pest INSECT PoPULATIONS Systematic Variable Probability Sampling for Estimating a Douglas-Fir Cone Population Introductton In studying the dynamics of a Douglas-fir cone moth (Barbara colfaxtana (Kft.)) popula- ‘tion on Douglas-fir (Pseudotsuga menztesit (Mirb.) Franco) (Nebeker 1974), a means of estimating the total numbers of cones was re- quired to estimate absolute population den- sities of the various immature stages of B. colfaxiana on a given area. In addition to being useful as a research tool, such a tech- ‘nique might also be useful in the management of seed production areas and orchards if it allowed advance planning of manpower and logistic requirements. The procedure dis- cussed will be exemplified by results obtained during 1971 on the Buckhead Seed Production Area, Lowell Ranger District, Willamette Na- tional Forest, Oreg. Estimation of Cone Population A systematic sampling design with vari- able probability was used to estimate the total number of cones on the Buckhead Seed Production Area. This technique uses pre- Based in part on research conducted by the senior author in partial fulfillment of the degree Doctor of Philosophy at Oregon State University, Corvallis, Oreg. This work was supported by a National Science Foundation Training Grant, GZ-1372, in "Pest Population Ecology.” Respectively, Department of Entomo- _ logy, Mississippi State University, Missis— _Sippi State, Miss. and Department of Statis-— tics, Oregon State University, Corvallis, Oreg, T. Evan Nebeker and W. Scott Gueccens scribed probability proportional to an index variable (ppx). The general method was used for many years in sample surveys prior to the series of papers by Hartley and associates (c.f., Hartley and Rao 1962 and Hartley 1966) which outlined the theory. In general, the tech- nique is very useful in many cases of practi- cal sampling, and provides the most convenient variable probability sampling scheme with fixed sample size. Recent references in the forestry area are Stage (1971) and Overton et al. (1974). The procedure requires that each tree in the universe (or sample) be characterized by an index variable that is related to the variable of.interest for that unit. If a list of the index variable can be obtained cheaply, then ppx systematic sampling can be more ef- ficient than other sampling methods. How much more efficient depends on the character of the functional relation between the index and the variable of interest. Generally, a positive correlation is an indication of gain, and a negative correlation can lead to loss in efficiency. In the present application, the index variable was defined as the relative number of cones on a tree, with the number of cones The notation ppx is used to desig- nate probability proportional to a variable, x. The notation pps sometimes used in this context is better reserved for probability proportional to (sample) sum, a technique useful in similar circumstances. Variable probability sampling can be applied at any level in a sampling scheme. Our application to alZ trees in the universe of interest is a special case. ee ues) ine eo ea ee EE being the variable of interest. The index was then assigned in visual relation to the tree having approximately the greatest number of cones. Inspection was from the ground. Each tree in each of 10 plots was given an Index I according to the criteria outlined in table 1. Trees were ranked in order of index, from the highest to the lowest in each plot, and the accumulative total T,; was cal- culated for all 10 plots. Table 1.--Summary of indexing system and criteria for each index Index Range Criteria (1) 1.00 .75-1.00 Heavy -- cones throughout .90 the crown, dense .80 .J0 .50- .74 Medium -- cones throughout . 60 the crown, sparse; or mid- .50 and upper-crown only, dense -40 .25- .49 Light -- cones in mid- and 30 upper-crown, sparse to dense .20 .00- .24 Very light -- cones only in .10 upper crown or scattered .00 very sparsely With the time available for sampling, it was decided that samples of n = 12 of the N = 365 cone bearing trees were to be selected as sample units to estimate the total number of cones on the area- To select a sample of n = 12, a sample interval s had to be computed: s = T,/n and a random number r between o and s se- lected. Sample selection then follows the rule: a tree t, is in the sample if a number from the set see rts, rt2s, .. .-, r+(n-l)s] is contained in the interval (Ty) T,)> af where > Te I,, To=o i j af If s > Max I, then this sample will con- tain exactly n trees, and the probability of inclusion in the sample will be x, = nI,/T, fo each j. If some I’s are greater than 3! a interval must be modified appropriately. r After the sample trees were selected, the same procedure was again followed in selecting the whorls to be sampled. ated and indexed. Within each selected whorl, one } branch was selected at random, with equal probability for total cone enumeration. sample selection can be summarized in the following steps: selected. Step l. Step 2. Step 4. Step 5. Step 6. Step 7. Step 8. The sample unit estimations for whorl, tree, plot, and area can be summarized as follows: Step l. K i Whorls were enumer=) Four whorls per tree were The Assign an index I to tree j in the area. € Arrange the indices from highest to lowest in each plot and sum indices over all trees in the area. Select n trees of the total N cone bearing trees systemati- cally with probability propor- tional to Ty: | ) | Assign an index W, to each | | whorl i on a selected tree. Arrange the whorl indices W from highest to lowest for this | tree and sum the indices. =] Select k whorls of the total K 1 whorls per tree systematically (| with probability proportional to W,- 7 From each of the selected whorls, select one branch at random (numbering the m branches clockwise with the one | pointing most northerly as num- I ber one) with equal probabil iia || | Count all live cones y on the i branch selected. cones on the i whorl is: A ] Estimation of Fhe number of | Cyd = mn, - y where: m, = pumber of branches : a He in the i whorl. she: Step 2. Estimation of Epe number of clusion probabilities for these cones on the j__ tree is: plots. k nN is Ne is A t - = 4 mae Dis ie a ey ae i=1 ri 4=1 “caj where: where: Tr. ax kW. , i we ue nt. caj Th4 The inclusion probability of the i whorl on the j__ tree. the conditional pnclusion proba-— Step 3. Estimation of the number of bility of the 3° tree in plot cones on the area is: a being sampled. Considering the entire area, an estimate , B - of the variance of T._ is conservatively ap- ge SD mayan proximated (overestimated) by: of : Ty as Ne uaVNN | 82 2 AWA . 49) A A where: a We = n( = m=)(D ee - (at ) ) mj = j=l : Ty. Table 2 summarizes the estimates for each and n = number of sample trees. of the 12 trees sampled and then projects an estimate of the total cones on the area. As Step 4 EKA GS Gis eile oF mentioned before, the precision of the esti- ; cones per plot requires identi- mates depends on how well correlated the in- filedttontokthe icondittonaldn= dex was with the actual number of cones. Table 2.-- Area (T.), Plot (T alee nee (t..) and the contribution (t /1j) of each tree to the total (Tt) t8ne estimated and indices for the Buckhtad’ Seed Y Production Area, September 1971 Plot No. of cone~ Mean Tree no. ‘ : no. bearing trees Index sampled 7 Index ty ‘Va ty 4/5 I 36 -60 24 -07088 85 523 34,744 Usaue) 31 -02502 30 791 31,615 II 22 43 ~- => tee => => III 43 245 14 -07505 -90 3209 43,877 42,758 42 -04170 -50 615 14,748 IV 34 44 29 -05837 -/0 1355 (AU MT) 23,214 V 57 34 3 -06671 80 2977 45,570 44,626 5 -01668 -20 252 15,108 VI OL 29 6 -03336 -40 385 16,245 11,541 VII 47 sol 36 - 05837 ./70 4778 49,354 81,857 29 -00417 505 20 4,796 VIII 19 39 ted => me ao oe IX 11 39 => == tee => ~= X 35 »45 8 07505 -90 3505 90,290 46,702 11 02919 B35 2924 100,171 Vy = 424,523 \ VT) = 164,567 Zty 5/15 = 424,515 Theoretically, if the index was absolute, bution of within-tree estimation error. In there would be perfect positive linear rela- general, various estimates for this sort of |} tionship between the index and the actual num- sampling do not accurately reflect the gain i ber of cones, and the variance of the estim- precision from the technique, and this is one> ator would be zero, except for the contri- area in which work needs to be done, l € cl Table 3.-~ Summary of the tree (ty) and the whorl (ty 4) cone estimates for the Buckhead yer ve Seed Production Area in September 1971 [. 0. St ee ee BNE OO eee b Plot Tree Tree Whorl Whorl Rn A no. no. index no. index m. y te T. I 31 30 15 G 3 12 36 32000791 14 9 6 17 102 .48000 12 9 5 15 75 .48000 ; 10 8 6 22 132 42667 I 24 .85 15 8 5 19 95 31068 523 it i 9 n 19 76 sages | 7 8 5 0 0 .31068 | 4 4 4 0 0 5534 III 14 .90 15 9 5 39 195 .29032 «3209 12 9 5 67 335 29032 10 8 7 51 357 ~~. 25807 7 Nea 5 0 0 .19355 | 42 50 10 a 6 9 54 .47458 618 9 4 5 16 80 .27119 | 8 4 4 14 ee Sahl : 7 5 6 0 0 .33898 IV 29 .70 17 9 3 19 57 39130 135mm 15 9 5 28 140 .39130 14 8 4 26 104. 34783 | ll 1c 4 30 120. -.21739 V 5 20 8 8 3 9 27. +-.50000 ~—Ss 252 7 9 4 16 64. 56250 6 9 4 6 24 56250 4 | 6 3 18 .43750 3 -80 12 9 6 76 456 .26471 ~+«-2977:~*| 9 8 5 37 185 23529 6 8 5 22 110 23529 1 62 6 0 0 .05882 VI 6 -40 13 i 6 7 42 .29474 385 1 ll 9 7 4 28 «37895 9 9 4 16 64 37895 | 6 B 6 0 0 4.21053 \| VII 36 .70 13 9 4 173 692 134952 4778 || 10 9 6 152 912 .34952 6 a 6 0 0 27185 4 6 4 ll 44 23310 29 .05 7 i 3 2 3.62500 20 6 1G 5 1 5 62500 5 i 3 2 6 .87500 3 iG 3 0 0 .75000 X 8 .90 18 a ya ail 339 22764 + ~—«-3505 16 9 5 73 365 —-. 29268 13 9 9 3 27 «29268 oe 4 7 22 154 22765 ll ee 10 9 6 22 132 .43374 2924 _~—| 7 8 9 4 36 38554 L] 9 a 4 19 96» 2ag735 6 4 8 54 432. .19277 DCSCUSSIZOH Hartley, H.O. and J.N.K. Rao. 1962. Sampling with unequal probabilities i and without replacement. Ann. Math. 8 It can be seen in table 3 that a number Stat. 33: 350-374. of whorls with high indices had low cone esti- mates and vice versa. This apparently re- Nebeker, T.E. flects the variation in cones per branch at a 1974. Population dynamics of the Douglas- given whorl. It is indicated that, to im- fir cone moth, Barbara colfaxiana (Kft.) prove the estimates, the branches in each (Lepidoptera: Olethreutidae). Un- selected whorl should also be indexed and published Ph.D. Dissertation, Oregon _selected in the same manner as the whorls and State University. 84 p. trees. Overton, W.S., Lavender, D.P. and Herman, R.K. 1974, Estimation of biomass and nutrient capital in stands of old growth Douglas- Literature Cited fir. IUFRO Biomass Studies, College of Life Sciences and Agriculture, Univ. of Maine. Orono, Maine. p. 89-104. Hartley, H.O. | 1966. Systematic sampling with unequal pro- Stage, A.R. , bability and without replacement. J. 1971. Sampling with probability propor- Amer. Stat. Assoc. 61: 739-748. tional to size from a sorted list. U.S. Dept. Agr. For. Ser. Res. Pap. INT-88. 16 p. Trend Prediction in Forest Insects H. Klomp Introductton In predicting the trend of forest in- sects, we try to forecast the future popula- tion density of an insect with some precision from observed facts and experience. The aim of the predictions is to provide a basis for deciding whether or not to control injurious insects. The purpose of predictions for control actions follows from the object of a control strategy, maximization of the benefit/cost ratio (Conway 1973): cost of damage controlled cost of damag uncontrolled 3 cost of control This objective function being accepted, the control worker is confronted with the problems of: 1. forecasting and evaluating damage when the population remains unchecked; 2. estimating in advance the reduction of damage under different control programs; and 3. estimating the cost of the control pro- gram adopted. The estimation of (3) is relatively easy, because most of the costs are labor and materials. To solve the more complcated problems raised under (1) and (2), it is necessary to have an expectation of the damage caused by the insect in the future. Plant damage is related to insect density, and because trend prediction refers to trends in insect abundance, we shall consider insect density instead of damage. Research Institute for Nature Management, Arnhem, Netherlands 1 The main features of trend prediction fo practical purposes are: 1. formulating a prediction of the density at time t_(x>0), based on demographic x data determined at time to; 2. stating the probability that the pre- dicted density will be correct; and 3. recurrent surveying in the field of the demographic data referred to under (1). The formulation of a predictive function requires information on the relationship be- tween population data at ty (which may refer to the density, the reproductive rate, and/or the mortality rate) and the density at time t_- It is valuable to know how detailed ) this information should be in order to make | a prediction that will be correct with a given probability, P. It is demonstrated later in this paper that the amount of infor- mation required is dependent on the pattern of fluctuation and on the nature of the key factor(s), which primarily contribute to population change. ' These features will be elucidated briefly. The interval, to-t, should be long enough to enable the control worker to cal- culate expenses and profits, and to make the practical preparations necessary for the con- trol program adopted. According to the predictive function men- tioned above, economic infestation levels can be expected to occur at t_, when the demo- graphic data at t, reach critical values. For the control worker, it is important to | know the probability that this expectation is correct. This probability can only be derived from historical population data, the data which are also needed to derive the predictive function. In general, the probability of a correct prediction will increase with a re- duction in the interval t,-t_, because the more variable factors have been excluded. It will be necessary for the control worker to know whether or not the demographic data reach critical values at t,. Therefore, a method is required to recurrently estimate the size of the population variables, which are incorporated in the predictive equation. The practical side of this aspect of trend prediction will not be considered in this paper. Predictions Based on the Denstty of the Preceding Generation In univoltine insects, the simplest re- lationship between the densities (N) of two successive generations can be expressed by the difference equation (1) where R, is the net rate of reproduction over generation t. In some species, the variations in Net are mainly due to variations in Ni» because R_ is relatively constant. In such species, the overall range of fluctuation is considerably greater than the range of change between successive generations. As a result, predictions can be based on the serial corre- lation between successive generations. An example of this type of fluctuation is presented by the larch bud moth (Zetraphera dinitana Gn.) in the Upper Engadin in Switzer- land. Baltensweiler (1971) reports for 1950- 1969 the highest larval density of this species roughly 50,000 times greater than the lowest density. The range in the rate of increase or decrease between successive gener- ations varies between 6 and 14 times, with exceptional decreases of up to 100 times. The correlation between successive larval den- sities is presented in figure l. Baltensweiler (1971) also reported that there was general defoliation in the area when log N > 5.8. This level can be reached only when the log density of the previous genera- tion (log N_) is higher than 4.7, thus es- tablishing Fhe critical level. Economic damage demanding large scale control action in 6 LOG N, 47 t__cRITICAL LEVEL Figure 1,--Correlation between successive larval densities in a population of larch bud moth, Upper Engadin, Switzerland, 1950-69. (See text.) generation ttl is predicted when the critical level in generation t is surpassed, but the prediction will only be correct in 50 percent of the cases (fig. 1). The predictive value of a relationship of this type is low, even though it is based on a considerable body of data: vig. the density of larvae over a period of 20 successive years, data which is unavail- able for the great majority of other harmful insects, However, important additional information . is available for the larch bud moth which can be used to improve the probability of a cor- rect prediction. Auer (1961) and Balten- sweiler (1964) have shown that the pattern of fluctuation of this moth in the Engadin is cyclic with periodic defoliation occurring every 8-9 years. This implies that, from the prediction point of view, the descending part of the cycle can be left out of consideration, The density figures for this part of the cycle are located below the 45° line in figure l, and those higher than the critical level ace underlined. When these are discarded, pre- dictions can be expected to be correct in 75 percent of the cases, In general, even this is far from a firm basis for decisionmaking with respect to con- trol actions. In this cyclic species, how- ever, a wrong decision does not have serious consequences, because, first of all, in the ascending part of the cycle, the density of generation ttl will at least be very close to the level of general defoliation, once the critical level has been surpassed in genera- tion t (fig. 1, point marked with a dagger). Secondly, if the defoliation level is not yet reached in ttl, it will certainly be reached one generation later (t+2) (fig. 1, point marked with an asterisk), and this will be prevented by successful control in generation tiles The prevention of defoliation touches up- on another important aspect of trend predic— tion in this species, The cyclic behavior of the population results from a time lag in the reaction of the food plant to defoliation. The quality of the larch needles is reduced for 3 to 4 years after partial or complete defoliation. This affects both the reproduc- tion and the mortality of the population re- sulting in the regression phase of the cycle (Benz 1974). Therefore, preventing defolia- tion will disturb the cycle, and this may have far-reaching consequences for the whole system, and hence for trend prediction. It has been suggested by Van den Bos & Rabbinge (1976), based upon a simulation model of the dynamic behavior of the larch bud moth, that excluding the effects of food quality, while leaving food quantity and parasitoids as effective mortality factors, results in population den sities fluctuating at an extremely high level | which would probably cause mass tree mortality after some years. This suggests that, in the - field, parasitoids will be unable to stabilize host density at a low level when defoliation is prevented by chemical control. If this is so, the host will rapidly build up its nunm- bers, and will reach defoliation levels once more, However, the number of years after whist this will happen is not fixed. To monitor this would require an intensification of the field survey program, which is already rather extensive due to the asynchronous occurrence 0 peak densities in different parts of the species range. In the larch bud moth, the predictability, of infestation levels one generation ahead is relatively high, due to the regular cyclic be havior of the population. When there is less regular periodicity and population declines | can also occur without preceding defoliation, \| the chance for a correct prediction is con- | siderably lower. Using the population data of) the pine looper (Bupalus pintartus L.) pub- lished by Schwerdtfeger (1935, 1941) and re- | plotted on a logarithmic scale by Varley (1949), a graph was prepared of the serial correlation between successive pupal densities (figs. 2A and B). Varley reports that outa were recorded in 1888, 1894, 1900, 1905, 1917 1928, and 1936, but Sehwecderenee (1941) has | i pointed out enae the relatively low peaks of 1894 and 1905 did not reach significant infes- tation levels. Consequently, large scale defoliation by the larval stage does occur ris the preceding log pupal density ore ) is equal to or higher than 0.5. ayaa in figure 2B, this density level is reached only when fhe log pupal density in the preceding generation (log N_) is equal to or higher than -0.6. However, predictions based on this critical level ERIS a low probability of being correct, because outbreaks occurred in only 8 out of 23 years (P= 0.35). i As was the case with the larch bud moth, the years after an outbreak (underlined in fig. 2) are immaterial to the argument. When | | these years and the year 1881/82 (squared in | fig. 2B), which had an unknown prehistory, are discarded, the predictability of outbreaks improves substantially (P= 0.58), but is still) too low for practical use. Predictions Based on Denstty and the Net_Rate of Reproduction \ When the range of overall fluctuation of i a population is only slightly greater than the| + PEPE Tere nr Ss elena Ta cal on a alinGar ae eee 1890 1900 1910 1920 1930 1940 | LOG N,,, 7? - 6 [ 1 ee ‘ e [el @ = 4 ° | 2 e@ © ele ® ool @ z O 1 LOG Ny oo nol 0.6 Figure 2,.--Fluctuations of the pupal density of the pine looper (log numbers per =f) at Letz- lingen, Germany, 1881-1940 (A) and correlation between successive pupal densities of the same population (See text.) range of change between successive generations, variations in N. ) are due both to variations in N, and in R, entation 1). Im these cases, predictions of the density of the defoliating stage in generation t+l must be based on esti- mations of both N, and R- R is the product of the survival rates in the successive stages or age intervals of the insect and its reproductive rate. Thus, according to figure 3: N, = N, x R) = N).S3.S,.F.S)-S, (2) It has been found, in some species, that the variations in R are essentially determined LOG DENSITY 2 Figure 3.--Intra and inter-generation fluctuations of an imaginary population of an univoltine | insect over nearly three generations, to show the relation between two successive generations || of defoliating larvae: N F, (Morris 1959, Watt 1963). This component ha been called the key factor of population fluc tuation (Varley & Gradwell 1960). When the factor operates during one of the later stag of the life cycle of the insect (pupa, moth) the inter-generation variation of the compo= | nents affecting the earlier stages will be / relatively small. This gives rise to a cor- relation between the density of the stage itt occurring after the operation of the key fac- tor and the density of the defoliating stage following. For instance, if in figure 3 the population fluctuations are mainly due to | variations of S, or F (or, when both are key | factors), egg density and the density of the defoliating larvae will be correlated, and I I by the variations of one of these components ; | ] ' j EGGS F SMALL LARVAE DEFOLIATING | LARVAE » where Ss) to S, indicate survival | = Gg isc Si. rates in different age intervats, aid F is the reproductive rate of the moths, this relation may be used to predict the in- ha cidence of outbreaks. This method was used for the white-fir sawfly, Neodtprion abietis Harr., by Struble * (1959) to show that control may be necessary whenever the egg density exceeds 20 clutches in a 200-twig sample. However, he was unable to indicate the probability of this prediction being correct, because the correlation between ~ egg and subsequent larval densities was based on only a 4-year census. Another example of the use of the corre- lation between egg density and larval density to predict the latter was reported by Shepherd & Brown (1971) for the forest tent caterpil- lar, Malacosoma disstrta Hbn., which defoli- ates trembling aspen in North America. Larval ‘ densities were expressed as defoliating classes (light, moderate, and severe), and the incidence of these classes was predicted on the basis of the number of egg-bands per ’ sample. However, in testing the accuracy of the predictions in the field, only 65 percent of the larval populations proved to be in the correct class. Locally, trees usually are defoliated 4 years in succession. This is possible be- cause the trees recover their foliage during the first 2 years of an outbreak, but recovery is much less in later years. Owing to this declining recovery, the egg hatch rate and the larval survival rate decrease as the outbreak progresses. In consequence, at the same high egg density, the damage by larval feeding proves to be severe at the beginning of an infestation, but light to moderate at the end. In compensating for this effect by using “sliding class-boundaries,"” the accuracy _ of the prediction could be increased to 75 percent. However, this still leaves too great a possibility for a needless, and hence unjustifiable, application of insecticides. In conclusion, prediction of the numbers of the defoliating stage can be based on the density of an earlier stage of the same gen- eration when there is a high correlation be- tween these densities. This is so, when the key factor(s) operate before the stage on which the prediction is based. Although this fore- cast is apparently based on density alone, in contrast to also using net reproduction rate, this is only seemingly so. The variable effects of the previous density of the defoli- ating stage (N_) and of the key factor(s) are included in the variation of the density of the stage used for prediction. For population fluctuations which are mainly due to variations in S, or S, (fig. 3), the correlation between egg denSity and the 11 numbers of the defoliating stage will be poor. Thus, for the pine looper in Holland, as shown for two generations in figure 4, the density of large larvae is primarily determined by the amount of juvenile mortality, most of which occurs during the first few days after hatching (Klomp 1966). As a result of variations in juvenile survival, the mortality from egg to defoliating larvae varied from 27 to 96.5 per- cent. This gives rise to a poor correlation between the densities of the two stages of the life cycle (fig. 5A). In contrast, the density of small larvae in the beginning of August, occurring after the operation of the key factor, is fairly well correlated with larval numbers later in the season (fig. 5B). If the latter relation is used for the prediction of defoliation in September-October, it is necessary to census larval density in August every 2 or 3 years because in this species, epidemic levels are normally reached within 3 to 4 years (fig. 2A). This frequency of estimating density is perhaps practicable if large forested areas with virtually the same larval density are involved. However, as shown for Germany by Engel (1942), and for England by Bevan & Brown (1961) and Davies (1962), there May be considerable differences in density be- tween the compartments of a pine forest, some- times only 500 meters apart. This situation requires such an extensive census program that the available time interval (t, - t_) of roughly 45 days could be limiting. *This im- plies that, for species with fluctuations governed by key factors operating shortly be- fore the defoliating stage, and having a highly patchy distribution pattern, predicting out- breaks with much precision seems to be an un- attainable goal. Predictions Based on the Stze of the Key Factor Instead of using measured densities of preceding stages of the same generation, esti- mates of the size of the key factors themselves may be used to predict the density level of the defoliating large-larval stage. This method was used by Morris (1959, 1963a, 1963b) in developing predictive equations for the black- headed budworm, Acleris vartana Fern., the European spruce sawfly, Diprion hercyntae Htg., and the spruce budworm, Chortstoneura fumtferana Clem. In the first species, a univoltine defo- liator of fir and spruce in Canada, larval density in generation t+l is predicted from the larval density and the rate of larval par- asitism in generation t. Parasitism is easily measured in this species by dissecting or LOG DENSITY EGGS | | LARVAE | SMALL DEFOLIATING | «~LARVAE | Figure 4.--Survivorship curves of the pine looper in Holland, showing the predictive value of juvenile mortality for the density of the defoliating larval stage. rearing the larvae obtained in the population For this procedure, Morris used population sampling. data on N and p from a coniferous stand in New Brunswick collected over 12 successive years (fig. 6). He concluded, from the consid- erable improvement of the correlation coeffi- cient, that parasitism was a key factor for this population during the period that the population was studied. This makes sense when larval parasitism is a key factor, i.e., when larval density (N +) in generation t+l is essentially determined by the rate of parasitism (p,) in generation t, in addition to the contribution of larval density (N.) in generation t. This was tested by Morris by comparing the correla- tion between log N and log N, with the nd d correlation between log N.,, an log {N.Ci-p, and appraising the effect of the assumed key factor by measuring the improve- ment it made in the correlation coefficient. This method differs from those described above, in which densities of an earlier stage | of the same generation have been used, pecauall| {N, C1=p dt is not a real density of larvae | surviving parasitism; it is only an index. There is considerable mortality of both (°3*82 99S) °7961-O0S61T “PUPTIOH ut tedooT outd oy uz (gq) 2eA -1e[T umoi3 AT[N} pue umo13 JTey Jo SaTITsuep sy useMjJoq UOT}eTeII0D YSTY oy pue ‘(V) eeAIeT umoi3 AT[NJ jo AjTSuep ay} pue AjTSuep 38a uaemjeq uoT}eTeIIOD 100d sy BuyMous sudeig--*c¢ aan3Ty ysn6ne ALISNAG WAYVT SOT ALISNAG D553 901 GL Ol 0) O72 SI Vv i= O G) | > < > rr 0 m = ” =| ~< Oo 2) =r fe) Oo 2 49q0}90 ALISNAG IWAYVT SOT 13 “6S6T SFAIOW JO BJep ASzIFY *3xX9 By UT UsATS ST YOTYM Jo uoTQenbse ayq ‘auTT UOFSSe91391 APAUTT 9Yy} SqUeseider g UT BUTT ey ‘*UOTIJeIeUaS auO Oz SuTyYOeRT vie -ejep aqtseied asneoeq ‘fy i REL qutod Few sey g yderg *(g) (€6‘0=1) potaed awes ay} azaao uoyTje[ndod sawes ay ut (74-1) N 307 pue ae N 30T usanjeq UOT eT e1109 ua pue (y¥) Wiompnq pepesy—-y2eTq 9yQ Jo uoTjetndod yoTMsuNaAg MeN e& UT (/9‘Q=1) N 30] pue I+ N 30[ uveMJOq UOTJeTeIIOD |aYyI--*g 2aNn3Ty Ag-1) tn} 501 . € q parasitized and unparasitized larvae due to points (years) from the regression of log N other causes, both during and after the period on log N_ were related to an easily measured of activity of the adult parasites. Morris expression of weather: the mean maximum daily developed a method to correct larval survival temperature during the main period of larval for this mortality, but this correction will development (roughly June 1-July 13). Morris be omitted here, because the adjustment did simply added the regression of deviations on not significantly improve the correlation. temperature to the relation between successive larval densities, giving the predictive equa- Predictions of log N can now be based tion : ftl By on the regression of log , on log we Pm f giving the equation (fig. BY: log Nea] = 0.98 + 0.76 log Ne + 0.18(T,-T) log Ney = 0.53 + 0.93 log Ne (1-p, (3) (4) However, this equation seems to be of where no use for the prediction of defoliation. Morris has reported that, in New Brunswick, T, = mean maximum daily temperature populations of the black-headed budworm June 1-July 13 in year t. decline before food shortage becomes an important factor. This implies that equation T = the average value of T. over a (3) is based on data from an area where long period of years, being 66.5 F. outbreaks of this species do not occur. Therefore, the question may be raised as to Morris tested the predictability of this equa- ‘the validity of the predictive value of the tion by calculating the regression of observed | equation for populations in other stands. In values of N on those calculated from areas where the budworm periodically causes equation (a§t this regression had a standard -defoliation, the parasite's numerical response error of estimate of 0.41, which is too high to increases in the host population is prob- to permit good predictions. ably insufficient to bring about a downward trend in the numbers of the host. Under such Equation (4) was based on data from var- conditions, food quantity and quality, rather ious plots in Area I, including those that than parasitism, are likely to be key factors, suffered from severe defoliation and others as is the case with the larch budmoth dis-— that did not. By restricting the analysis cussed earlier. to the latter, thus eliminating effects of food shortage and possibly the mass dispersal Another objection to the predictive value of moths induced by overpopulation, Morris was of equation (3) is that it is based on only 12 able to improve the predictability of the generations. Over this series of years, para- equation considerably. The regression of sitism indeed proved to be a key factor, but observed values on calculated values in this there is some evidence from other species stud- case explained 79 percent of the variance ied over longer sequences of years, showing and showed a standard error of estimate of that fluctuations of numbers may be governed by 0.29. ‘different factors over successive intervals. However, the forest entomologist in- Another species considered by Morris terested in predicting larval densities (1963a), the spruce budworm, was studied on causing economic damage in situations where several plots of an extensively forested area overpopulation phenomena may be expected to (Area I) in northwestern New Brunswick during play a role before heavy defoliation occurs the outbreak period of 1945-1960. Because the will have to await the development of predic- overall range of fluctuation in this period is tive equations which incorporate these fac- considerably larger than the range of annual tors. Where some factors are largely weather Imcreases or decreases, there is a high cor- dependent, such as moth migration, reliable relation (r= 0.67) between log Ni, and log N,. prediction of economic damage for definite stands of forest will remain a wish for a long In this species, unlike the black-headed time. budworm, there is hardly any improvement of the ‘correlation coefficient between log Ne4] and log NC1-p ) over that between log N and log Nee is shows that parasitism Is not a The Change of the Key Factor over key factor in this budworm, at least not ~ Long Intervals during the outbreak period studied. There was Some evidence suggesting an influence of weather on population development, and Morris For the prediction of population change Showed that the deviations of individual with some precision, knowledge concerning the 15 key factor(s) causing this change is essen- tial, unless the larval densities of succes- sive generations are highly correlated and fluctuations are cyclic. In some species, prediction can be based on the density of a life stage occurring earlier in the life cycle than the defoliating stage, provided the key factor operates before that earlier stage. In all other cases, predictions have to be based on the size of the key factor. Thus, predic- tions of population changes of the winter moth in Wytham Wood, England are as yet impossible, because the main cause of the year to year changes is largély unknown (Varley & Gradwell 1971). The changes are thought to arise from variations in the degree of synchronization between bud burst and egg hatch, factors which are probably weather dependent. Therefore, identification of the key fac- tors is essential, and should be based on a detailed life table study of the species in- volved. With some exceptions, case studies of this kind have been limited to 10 to 15 gener- ations or less. There is, however, some evi- dence that population change may be governed by different key factors over successive in- tervals of that length. This is illustrated by the long term population study of the pine looper in Holland. LOG LARVAL DENSITY 0.8 We have studied the fluctuations of this moth in a pine plantation over a period of 26 consecutive years (fig. 7). Detailed life table analysis as described by Klomp (1966) and key factor analysis according to Varley & Gradwell (1960) have shown that fluctations over the years 1950-1963 were mainly due to variation in juvenile mortality, most of i affects the tiny larvae just after hatching. After 1963, there was a sudden | of an ichneumon fly, Poectlostietus cothur- [| natus, killing an increasing proportion of th pupae up to the 1967-1968 generation, and causing a tremendous decline in host numbers. The proportion infected in 1968 is unknown, but from 1969 onward the parasite’s numbers rapidly decreased, giving rise to a recupera- tion of the host population within 4 to 5 years. system during the 1950-1963 period, during which it killed a small, slightly variable proportion of pupae each year. It emerges from the host pupae in June, but does not attack pine looper larvae before their devel- opment is well advanced in September-October. Oviposition on earlier stages of the pine looper is refused, and the parasite has at ) This parasite was also present in the | | 1950 - 1963 Figure 7.--Long-term fluctuations of a pine looper population in Holland. The year to year changes of density are mainly due to first instar larval mortality over 1950-1963, to a non- specific ichneumon fly over 1964-1974, while the crash 1975-1976 is caused by a polyphagous egg-parasite. ise": Umea a Jiraasl 1 T 1970 1964 -1975 least two alternate hosts on birch which are infected in July. The new generation emerges from these alternate hosts in September. Birch trees do not occur in our study plot, but are present in large numbers at distances of 500 meters or more. Large numbers of the parasite probably invaded the study plot in the late summers of 1965, 1966, and 1967, because the low numbers emerging there in June could not have been responsible for the high proportion of larvae subsequently attacked in the autumn. Variations in juvenile mortality still played a part over this period, as shown by the variation in the steepness of the decline in looper numbers from 1963 to 1967. The much more dominant effect of pupal parasitism is, so to speak, superimposed on the variable effects of the juvenile mortality. Then, finally, there was a sharp decline of larval numbers in the 1976 generation. This was caused by a 98 percent kill of the eggs, due mainly to the polyphagous egg parasite, Trichogramna embryophagum Htg. With some exceptions, this parasite killed from 20 to 40 percent of the eggs in previous years. It is highly polyphagous and may have invaded the study plot from the surrounding deciduous forests; further, its host searching may have been favourably influenced by the extremely high temperatures (30-35° C) over a consider- able part of June. Whatever the cause may be, the egg parasite acted as a key factor in that year and affected population change to an ex- tent never before observed. For trend prediction, an occasional switch of key factors over time has far-reach- ing consequences, For the pine looper, it implies that in outbreak areas, larval density in August should be intensively censused annually. All this means that trend prediction in forest insects is of limited applicability, and this implies that as yet there is no good basis for decisionmaking with respect to con- trol actions, because the principal data needed for maximization of the benefit/cost Yratio given at the start are not available. Literature Cited Auer, C. 1961. Ergebnisse zw6lfjahriger quantitativer Untersuchungen Populations—bewegung des Grauen Larchenwicklers (Zetraphera griseana Hb.) im Oberengadin, 1949- 1958, Mitt. Schweiz. Anst. forstl. 17 Versuchswesen 37: 175-263. Baltensweiler, W. 1964, The case of Zetraphera griseana Hb. in the European Alps. A contribution to the problem of cycles. Can. Ent. 96: 790-800. Baltensweiler, W. 1971. The relevance of changes in the com- position of larch bud moth populations for the dynamics of its numbers. Proc. Adv. Study Inst. Dynamics Numbers Popul. 19/70: 208-219. Benz, G. 1974, Negative Rtickkoppelung durch Raum und Nahrungskonkurrenz sowie zyklische Verdnderung der Nahrungsgrundlage als Regelprinzip in der Populationsdynamik des Grauen Larchenwicklers, Zetraphera diniana (Guenée). Z. ang. Ent. 76: 196-228, Bevan, D. & R.M. Brown 1961. The pine looper moth Bupalus pintarius in Rendlesham and Sherwood forests 1959. Rep. For. Res., For. Comm. 1959-1960: NOMS) Bos, J. van den & R. Rabbinge 1976. Simulation of the fluctuations of the grey larch bud moth. Simulation Mono- graphs, Pudoc, Wageningen: 1-83. Conway, G.R. 1973. Experience in insect pest modelling: a review of models, uses and future direc-— tions. In Insects: studies in popula-- tion management. P.W. Geier et al. eds. Ecol. Soc. Austr. (memoirs 1), Canberra: 103-130. Davies, J.M. 1962. The pine looper moth, Bupalus ptntartus, at Cannock Chase in 1960. Rep. For. Res., For. Comm., 1960-1961: 176-182. Engel, H. 1942. Populationsbewegung des Kiefernspan- ners (Bupalus pintartus L.) in verschie- denen Bestandstypen. Z. ang. Ent. 29: 116-163. Klomp, H. 1966. The dynamics of a field population of the pine looper, Bupalus piniarius L. Adv. Ecol. Res. 3: 207-305. Morris, R.F. 1959. Single-factor analysis in population dynamics. Ecology 40: 580-588. rey ee ama era rer a a me a a | a a a mt NE as aE Morris, R.F. 1963a. The development of predictive equa- tions for the spruce budworm based on key factor analysis. Memoirs Ent. Soc. Canty St: etl6=129) Morris, R.F. 1963b. Predictive population equations based on key factors. Memoirs Ent. Soc. Can. S22) 6=—2)10 Schwerdtfeger, F. 1935. Studien uber den Massenwechsel einiger Forstschadlinge II. Uber die Populations- dichte von Bupalus piniarius L., Panolis flammea Schiff, Dendrolimus pint L., Sphinx pinastrt L. und ihren zeitlichen Wechsel. Z. Forst- und Jagdwesen 67: 449-482, 513-540. Schwerdtfeger, F. 1941. Uber die Ursachen des Massenwechsels der Insekten. Z. ang. Ent. 28: 254-303. Shepherd, R.F. & C.E. Brown 1971, Sequential egg-band sampling and prob- ability methods of predicting defoliation 18 by Malacosoma disstria Hbn. Can. Ent. 103: 1371-1379. ! Struble, G.R. 1959, Egg sampling reveals trend in white- fir sawfly abundance, J. Forestry 57: [? 510-511. 1949, Population changes in German forest | Varley, G.C. | 117-122. | pests, J. Anim. Ecol. 18: Varley, G.C. & G, R. Gradwell 1960. Key factors in population studies Jin Anima "Ecol 29:9 1399-400. Varley, G.C. & G.R. Gradwell 1971. The use of models and life tables in| assessing the role of natural enemies. In: Biological Control, C.B. Huffaker (ed.). Plenum Press, New York: 93-112. Watt, K.E.F. 1963. Mathemetical population models for H five agricultural crop pests. Memoirs | Ent. Soc.) Gant 525) 6S—9r | | The Forest Insect and Disease Survey is an operational unit within the Canadian Fores- try Service charged with the responsibility of detecting and evaluating forest pest outbreaks on a nationwide basis throughout Canada. In 1956, Blair M. McGugan, who at the time was the Survey Coordinator attached to Divisional Headquarters at Ottawa, presented an excellent paper describing the origin and development of the Canadian Forest Insect Survey at the Tenth International Congress of Entomology in Mon- treal. The intent of this article is to re- view McGugan's paper briefly, to describe the present capability of detecting forest insect pests in Canada, and finally, to offer a few comments about the future. Canada is a large country. It is the second largest country in the world, with an area of about 10 million km”. It extends 9248 km from coast to coast, and shares a common border with the United States for 6400 km. Much of Canada is covered by forests which stretch across the continent in an un- broken belt 960 to 2080 km wide. These forests are among our greatest renewable re- sources and provide raw material for the lum- ber, pulp and paper, plywood, and other wood- using industries that are so vital to our economy. In addition, the forests of Canada control water runoff and prevent erosion, shelter and sustain wildlife, and offer un- Matched opportunities for human recreation and enjoyment. Productive forests cover nearly 2.6 million km”. The total volume of wood in these forests is estimated at nearly 22 bil- lion m. Four-fifths of this wood is coni- ferous and one-fifth is deciduous. The nature of the forests in any country is influenced by many factors, including cli- Mate, geology, and topography. In Canada, different combinations of these fators and Canadian Forestry Service, Forest Insect and Disease Survey Unit, Great Lakes Forest Research Centre, Sault Ste. Marie, Ontario, Canada. 19 Detecting Forest Insect Pests in Canada Gordon M. Howse the post-glacial migration of tree species from the south have resulted in eight dis- tinct forest regions (fig. 1), which may be subdivided further into 90 sections, each with its own ecological characteristics. The largest of these regions, occupying three- quarters of Canada's productive forest area, is known as the Boreal Forest, stretching in a broad belt from the Atlantic Coast westward and then northwest to Alaska. The forests of this region are predominantly coniferous, with spruces, balsam fir, and pines the most common species. Many deciduous trees are also found in the Boreal Forest, with poplar and white birch being the most widespread. The Great Lakes-St. Lawrence and Acadian regions are found in eastern Canada south of the Boreal Forest. Here the forests are mixed, with many species represented. The principal conifers are white and red pine, hemlock, spruce, cedar, and fir. The main deciduous trees are yellow birch, maple, oak, and basswood, Entirely different in character is the Coastal Region of British Columbia. Here the forests are coniferous, and because the cli- mate is mild and humid and rainfall is heavy, the trees are very large in comparison with those in the east. The principal species are cedar, hemlock, spruce, fir, and Douglas fir. In the Subalpine, Montane, and Columbia re- gions of western Alberta and the interior of British Columbia, lodgepole pine, Englemann spruce, and several true firs are found along. with Douglas fir and other coast species. The only true deciduous forests in Canada occupy a relatively small area in the southernmost part of Ontario, which is predominantly an agri- cultural district. The British North America Act (1867) is the core of the Canadian constitution. It specifies the major arrangements governing the organization of the government and the divi- sion of powers between the central (or federal) government and the provinces. Canada today consists of ten provinces and two huge north- ern territories (fig. 2). The provinces from east to west are Newfoundland, Nova Scotia, Prince Edward Island, New Brunswick, Quebec, Ontario, Manitoba, Saskatchewan, Alberta, and 06 ‘suot~3e1 Jsez0zZ QYsTe 24. UTYITM pue uof3e1 yse8ieT ay ST JSei10q Teer10g oy] } ( as Q | wih é 44 il ily ya ) qT ae VGYNVD 30 SNOIOSSY LSIWO4 *epeuey ut MOIS sotToeds 9819 QT eWOS *pejyeustsep uveq eAey SUOTIDaS jsSaI10F *pueT 3Se10F aATONpoad s,epeueg Fo JZusoied ¢/ Jnoqe siaaod *suoT3e1 Jsa10zF JOUTISTp WYUSTe sey epeueg--*] 9IiNsTy (°*seiqued yoreaseaz jsei0j queseider seaenbs yIeT_) °*seizjueD yoaeesez 4se1OJ TeuoTSe1 xTs Jo uot zoUNZ ay jo jied se Adaang eseesTq pue ZOeSUT JSet0q ay Aq peteaod st eaae a8ny SsTYyL *SeTIORTAIE} UTSYIIOU OMI pue seoUTAOAd QT SesTidwod epeueD—-°Z ain3sty WV S25 as Lins \ \ NOLNOWQ] \ \ \ 2 f | 1 i 1 \ \ ' \ S< 9°71 British Columbia. In addition, there are the Yukon and the Northwest Territories. Each province has its own governing structure whereas the territories still come under the jurisdiction of the federal government. Eighty percent of Canada’s productive forest land is publicly owned. This condi- tion was established by the British North America Act, which assigned to the various provincial governments the exclusive right to enact laws regarding management and sale of public lands within their boundaries, includ- ing the timber and wood on those lands. In the northern territories, which contain only about 8 percent of the country’s productive forest land, the forests are administered by the federal government. For many years, the policy of both the federal and provincial governments has been to retain in public ownership all lands not re- quired for agricultural purposes. In some of the older settled areas of Canada, however, the proportion of privately-owned land is high. This is especially so in the three Maritime (Atlantic) provinces where nearly two-thirds of the productive forest area is privately owned. Thus, the administration and protection of most of Canada’s productive forest area is vested in the various pro- vincial governments, which make the forests available to private industry through long- term leasing and other arrangements. With 1l different forest authorities in Canada, a great diversity of policies and forest laws might have been expected to develop; in fact, the similarities are far more significant than the differences. And furthermore, although most of the productive forest land in Canada is publicly owned, virtually all timber har- vesting is carried out by private industry, an arrangement believed to be unique among the major wood-producing countries. It is estimated that 99 million ae of wood are cut each year from Canada’s forests. About 90 percent of the wood harvested is con- iferous while the remaining 10 percent is de- ciduous- In addition to the wood used by in- dustry, if is estimated that more than 56.6 million m are lost to insects, diseases, and forest fires on an annual basis. Much of the effort to prevent losses by these major natural enemies is carried out through more intensive forest management, reforestation, basic and applied research, forest pest sur- veys, and contre! operations. In Canada, the federal government concen- trates much of its forestry effort on research, which is carried out by the Canadian Forestry Service of the federal Department of the Envir- onment. Forest research is also conducted by provincial governments and universities. laboratories, field stations, and experimenta areas in six regions across the country, as well as a number of institutes in Ottawa and elsewhere. Extensive research, both basic ané applied, is carried out on forest management, forest fire control, forest insects and dis- eases, and forest products. The Canadian Fo estry Service is also responsible for conduct- ing surveys of forest insect and disease condi tions across Canada and for providing informa tional and educational material for the public on the wise use of the nation’s forests. The Canadian Forestry Service operates With the foregoing summary serving as both an introduction to the topic and an over- all framework, it is evident that the functio of detecting forest insect pests, at least in the Canadian context, is not a simple task. The sheer physical size of the area presents formidable challenge. A well-organized team of field and laboratory personnel possessing considerable knowledge, skill, and ingenuity is required to monitor pest situations effec-\ tively under conditions such as we face in Canada. | In 1931, the Division of Forest Insects of the Entomological Branch organized a ser- vice known as "Forest Insect Intelligence Service.” A number of special circulars ont principal forest insects was published giving popular accounts of the life history and habit of the pests as well as colored illustrations showing the insects themselves and the nature and appearance of the damage caused by them. | These circulars, together with printed ques- tionnaires, were sent to the forest industry and provincial forestry services with the request that they be distributed to the men working in the forest and that these men be asked to make reports on the occurrence of insects in their respective districts. The plan met with general approval and valuable information was received. | In 1936, a regional survey was started is | eastern Canada for the specific purpose of assessing the status and spread of the intro- duced European sawfly. The first receiving center for forest insect reports and ai | tions was established in Ottawa in 1936. | national Forest Insect Survey was proposed “a | 1937 to include forest insects generally and | other receiving centers were set up in 1937 at Fredericton, New Brunswick and Vernon, | British Columbia, at Duschenay, Quebec in | 1938, at Indian Head, Saskatchewan in 1940, | and at Winnipeg, Manitoba in 194l. In these early years, the Forest Insect Survey relied primarily upon observations and collections provided by cooperators in the field. These cooperating personnel were highly fire-oriented, with little training or interest in biology or entomology, and the overall quality of the work was relatively poor -- though certainly better than nothing. After World War II, Forest Insect Survey Groups, including a ranger service (field) of subprofessional staff, were established at the following centres: Fredericton, New Bruns-— wick, Sault Ste. Marie and Ottawa, Ontario, Indian Head, Saskatchewan, and Vernon, British Columbia in 1945; Calgary, Alberta in 1948; Victoria, British Columbia in 1949; Truro, Nova Scotia in 1953; Winnipeg, Manitoba in 1954; and Corner Brook, Newfoundland in 1955. In 1953, the Sault Ste. Marie lab- oratory accepted responsibility for all of Ontario, and the original Ottawa Survey Centre was disbanded. Survey activities of the Indian Head laboratory, which had dealt with prairie shelterbelt, were discontinued in 1954, and responsibilities were divided among the Winnipeg and Calgary laboratories. The province of Quebec operated an independent survey, cooperating closely with the federal laboratory in Quebec until 1973, when the Laurentian Forest Research Centre took over full responsibility for all aspects of forest insect and disease survey there. Forest disease surveys were gradually incorporated into the organization throughout the 1950s, although development in this area was slow and most rangers required training in forest pathology. A processing system for field data was inaugurated in the early 1950s using Reming- ton-Rand equipment which was later converted to a fully automatic IBM system in the late 1960s. The original Remington system proved quite useful: it was employed, for example, in the preparation of regional summaries for Major compilations such as the Forest Lepidop- tera of Canada series initiated in 1958. Although the general aims of the Forest Insect Survey have changed little over the years, the methods and emphasis have undergone considerable adjustment. The primary purpose of the Survey was, and still is, the detection of insect outbreaks and an annual census of important forest insect conditions. Initial- ly, this was accomplished almost exclusively through the use of cooperators but results, while encouraging, allowed only broad qualita- tive interpretation. Following 1945, a small staff of forest biology rangers was hired for the purpose of instructing and working with cooperators. Before long, the rangers were providing most of the collections and observa- tions, and these were of consistently good quality. As a consequence, ranger staffs were 23 increased, coverage and detection became more comprehensive and complete, and the use of cooperators was de-emphasized. As McGugan stated in 1956, “there was a trend from what might be termed the ‘taxonomic phase" of the Survey to the ‘ecological phase'." The ecolo- gical phase lasted from the mid-1950s until the late 1960s. We are now in an “economic phase" -- economic in the sense that our atten- tion, in terms of detection and evaluation, is focused on the most important, or potentially important, insect pests known, and on their hosts. In retrospect, the change in emphasis from the "ecological phase” -- which was really a broad, faunistic survey -- to the "economic phase" was probably natural and realistic, although it was hastened considerably by events of the winter of 1969-1970. In 1969, severe budget constraints required a major reorgani- zation and retrenchment, including the closing of the Winnipeg laboratory and the amalgamation of the three prairie provinces into the Prairie Region, with headquarters in the Northern For- est Research Centre in Edmonton. The various Survey Units across the country sustained major cutbacks in staffing, which required consider- able reorganization and a consequent shift in emphasis, All regions were forced to reduce the number of forest biology rangers (technicians) on field assignment, thereby increasing travel time and the size of the survey districts under surveillance. At present, the operational responsibili- ties of the Forest Insect and Disease Survey cover the whole of Canada, as part of the function of the six regional forest research centres (fig. 2). For survey purposes, the country is subdivided as follows: the pro- vince of British Columbia and the Yukon Ter- ritory are assigned to the Pacific Forest Research Centre in Victoria, British Columbia; the three prairie provinces and the Northwest Territories to the Northern Forest Research Centre in Edmonton, Alberta; the province of Ontario to the Great Lakes Forest Research Centre in Sault Ste. Marie, Ontario; the province of Quebec to the Lauren- tian Forest Research Centre in Quebec City, Quebec; the three maritime provinces (New Brunswick, Nova Scotia, and Prince Edward Island) to the Maritimes Forest Research Centre in Fredericton, New Brunswick; and New- foundland-Labrador to the Newfoundland Forest Research Centre in Saint John's, Newfoundland. The staff organization and resource re- quired to conduct Survey functions within each region may be summarized as follows, although it must be borne in mind that there is some variation from region to region (see figure 3 for Ontario's organization structure). In each forest research centre, Survey operates under the general direction of a program man- DISEASE SURVEY OFFICER (LAB) IDENTIFICATION TECHNICIAN DISEASE SURVEY OFFICER (FIELD SENIOR PATHOLOGY HEAD FIDS UNIT CHIEF TECHNICIAN FIELD SERVICE INSECT SURVEY OFFICER (FIELD SENIOR ENTOMOLOGY INSECT SURVEY OFFICER (LAB) LABORATORY SUPERVISOR TECHNICIAN CULTURING TECHNICIAN TECHNICIANS TECHNICIAN COLLECT1ON Figure 3.--Staff organization of a regional Forest Insect and Disease Survey Unit. cular chart illustrates the organization of the Survey Unit in Ontario at the Great Lakes Forest Research Centre, Sault Ste. Marie. ager in charge of forest protection. Survey is the largest single unit or project in each region and is headed by a professional with Many years of experience in pest surveys. At the professional level, he is supported by an Insect and a Disease Survey Officer, an en- tomologist, and a pathologist, respectively. Professional officers also head up the insect and disease diagnostic and technical services in most centres. The professionals advise the Survey head in their particular disciplines and participate in planning and executing the program. The support professionals also carry out survey-oriented research as time permits. Appropriate technical support is provided in all units. This support varies according to regional program priorities, but covers certain core areas. The backbone of the detection function is provided by the dis- trict-assigned ranger who also carries out major liaison and technical service duties and takes part in appraisals, usually in the off-season. Normally, the activities of one or more district rangers are supervised by a senior ranger who also covers a survey dis- SIX REGIONAL SUPERVISORS EIGHT AREA INSECT 24 TECHNICIAN TECHNICIAN INSECT IDENTIFICATION This parti- trict. A chief ranger is responsible for the overall supervision of the rangers. One or more technicians support each of the insect and disease diagnostic and taxonomic service groups, appraisal or research functions. In all, the six regions have been alloca ted approximately 24 man years of professional staff and more than 90 man years of technical staff. In addition, varying amounts of stu- dent and casual support are allotted to Sur- vey depending on their availability and on Survey requirements, Approximately $2 million is required annually for salaries, travel, and supplies. A fundamental question one might ask is: “Why do we want to detect forest insect pests?" In the author's opinion, the purpose of detection surveys is summed up in the fol- lowing three objectives: 1. To determine the distribution of native insect organisms and their damage. Pest distribution data have been gathered and recorded for some 40 years or more. Much of the information has been published or otherwise made avail- able but much still remains in files or on punch cards awaiting analysis. The second, and perhaps most tmportant, objective ts to detect changes tn popula- tion or damage levels of selected organ- isms knoum to be potentially damaging, espectally to discover rising, and pos- stbly threatening, populattons and out- breaks. The deltneatton of outbreaks may be constdered part of thts objective as well as being the first step tn the ap- pratsal process. Most of Survey's current efforts and resources are directed toward meeting this objective. Control or management actions usually result from this type of activity. For example, aerial control operations have been mounted against some 24 different forest insect pests in Canada in recent years. In the Ontario Region, forest insect pests have been classified according to a historical analysis of outbreaks and damage so that priorities have been established. For example, there are nine species in Category 1 (highest priority), which in- cludes insects of major importance to forestry in Ontario because of their continuous or recurring abundance and their severe direct or indirect impact on trees. Among these insects are such well known pests as spruce budworm, jack pine budworm, larch sawfly, forest tent caterpillar, European pine sawfly, white pine weevil, native elm bark beetle, smaller European elm bark beetle, and saddled prominent. In Category 2 are insects or complexes of species causing appreciable damage to natural stands but usually over a period of several years or in a limited area, mostly with unknown impact. This category contains some 24 species such as large aspen tortrix, pop- lar leafroller, birch skeletonizer, oak leaf miner, etc. In Category 3 are in- sects causing appreciable damage pri- marily to plantings or nurseries and somewhat less damage to natural forests. Some 29 species are considered, including Leconte's sawfly, red pine sawfly, root and shoot weevils, bud and shoot miners, cutworms, aphids, spittlebugs, and June bugs. In Category 4 are insects known to have caused appreciable forest losses elsewhere in North America and considered to be potentially serious forest pests of Ontario. Included here are some 16 species, such as gypsy moth, European spruce sawfly, balsam woolly aphid, and hemlock looper. Category 5 contains 16 ; 25 species that are important primarily because of damage done to shade trees; Category 6, species or groups causing damage to finished or partially processed wood (six families); Category 7, insects known to have caused limited damage in small areas (71 species); Category 8, species found commonly each year but only in low numbers in the history of Survey (105 species); Category 9, species causing damage to unimportant tree species such as alder, willow, cherry, etc. (46 spec- ies); and Category 10, species collected infrequently (2770 species). The rangers concentrate their atten- tion on those species in the first four categories, Emphasis is placed on the economically important tree species that occur within the Survey Region, and it is damage to those species that is of primary concern. When Survey was reor- ganized in 1969-1970, it was felt that since detection must now be carried on in larger areas than had been the case pre- viously, field staff would not normally be able to detect changes in insect popu- lations at low levels. Therefore, rather than look for insects (general collect- ing), they were instructed to look for damage. These, of course, are gene.al instructions to the field staff, and, in some cases, special surveys designed to detect the presence, change in abundance, or spread of a particular insect are frequently organized. 3, The third objective ts to detect the tntroductton of tnseets exotte to the regton. While vigilance is maintained continu- ally for introduced pests because they pose a special danger, specific attention to particular exotic species by specially designed surveys is the method most like- ly to succeed. These objectives, therefore, represent the major reasons for detecting forest insect pests. The next question that needs to be dealt with involves our procedures and methods, or, in other words, our technical capability, for meeting our detection objec- tives. How are forest insect pests detected? The main reliance for detection is placed on the Survey rangers who carry out a series of planned or impromptu observations in their districts. These are frequently supported by quantitative sampling procedures designed to monitor, at their appropriate life stages, the populations of particular insects. Ground vantage point and aerial observation (fig. 4) are employed by the rangers seeking Figure 4,--Prompt detection of forest pest problems in Canada's forests requires frequent and comprehensive aerial observations, visible indication of the presence of pests. manent sampling plots, while not as useful a Sampling involves a variety of methods suit- detection procedure as "blind" or random samp- able to the particular pest stage sampled. ling, do provide good information on year-to Foliage may be removed from tree crowns year population trends; however, the high costs (with pole clippers or a line thrower) and of permanent sampling plots require a close examined (fig. 5), or beating samples may be look at their cost/benefit status. taken. Light traps and pheromone traps are used, although interpretation of numbers of It is difficult to state just where re- insects trapped is uncertain. However, traps mote sensing stands in relation to forest of this type are useful for detecting the insect pest detection. There seems to have presence of a particular species. They are been some oversell concerning satellite cap- _relatively inexpensive and some limited con- ability for detecting forest insect outbreaks, _clusions may be drawn regarding numbers. Per- and methods have not yet been developed for ‘ VE aman jh Par wey vei AAS i i ‘|Figure 5.--A forest research technician with the Forest Insect and Disease Survey is shown using a pole clipper and basket to collect spruce budworm infested foliage from the mid- crown of a balsam fir tree. A basket is used to prevent loss of active, later instar larvae (when they are in the later stages of growth). 27 operational application. In Canada, ERTS or Landsat imagery and high altitude photography are still in the experimental stage, at least as far as detection is concerned. It is to be hoped that this is primarily a question of technology and will eventually be resolved. Lower altitude aerial photography is fairly commonly used for purposes of delineating out- breaks or assessing damage. The detection aspect would not seem to be particularly significant. What will the future bring to the func-— tion of forest insect pest detection? It is difficult to foresee how we can improve sig- nificantly upon the insect ranger with a good pair of eyes, an inquisitive nature, and a thorough knowledge of his district. Undoub- tedly, technological advances will occur that will be useful or less expensive than existing methods. Satellite imagery and high altitude photography are examples that come to mind. But there is no replacement for the field man. Apart from technological advances or standardized nationwide approaches to detec-— tion, prediction could become the principal tool of Survey personnel. The basic elements of prediction are available, although our knowledge of many is limited at the present time. Examples of these elements are know- ledge of the insect (i.e., its life history, behavior, etc.), its abundance as determined by its environment and other factors, popula- tion behavior in relation to climate, cyclic patterns, knowledge of migratory patterns and behavior and monitoring thereof, knowledge of our forests, and so on. In other words, if we possess enough knowledge about the systems involved and can monitor or measure the appropriate parameters, it might be possible to predict when and where outbreaks will occur. This brings us back to detection. Appropriate management action dealing with 28 a pre-outbreak situation would probably be much more effective and much less costly. At the present time, we depend largely on, and respond to, visible symptoms such as damage or large numbers of insects.Manage- ment actions in this situation are costly and probably will not change the course of events. Literature Cited Anon. 1967. Canada: One hundred 1867-1967. Can. Year Book, Handb. Libr. Div., Dom. Bur. Stat., Ottawa, Ont. Anon. 1969. Canada 1970: The official handbook of present conditions and recent progress. Year Book Div. Dom. Bur. Stat., Ottawa, Ont. Anon. 1976. Canada's forests 1976. Dep. Fish. Environ., Can. For. Serv., Ottawa, Ont. de Gryse, J.J. 1951. Forest entomology. 94-109. Sci. News 21(98): McGugan, B.M. 1958. The Canadian Forest Insect Survey. Proc. Tenth Internat. Cong. Entomol. 4: 219-232. Rowe, J.S. 1972. Forest regions of Canada. Dep. Environ., Can. For. Serv., Ottawa, Ont. Publ. 1300. 172 p. Survey of the Nun Moth, Lymantria monacha L., in the Swiss Alps by Means of Disparlure © J. Ke WiAleeymony” The nun moth, Lymantria monacha L., is considered to be one of the most destructive pests existing today in the coniferous forests of Europe- Outbreaks have occurred periodi- cally in Germany, Austria, Czechoslovakia, Poland, Rumania, the USSR, and Spain. In recent years, outbreaks have been reported from Sweden and Denmark, also. In Switzerland, outbreaks occur only in the Alps. The last outbreak 11 years ago in the southwest Swiss Alps affected 300 ha (740 acres) at an altitude of 1100 to 1600 m (3600 to 4250 feet). The forestry service was taken by surprise at this turn of events, for the population buildup had remained unnoticed. For this reason, we are interested in an effective early survey method. The attraction of nun moth males by fe- males has been known for a long time; this behavior was put to practical use for the first time in 1931 (Ambros 1937). In 1949, the reciprocal attraction of the nun and gypsy moth males by the females of the other species was discovered (Gornitz 1949), and in 1955, this was confirmed by means of additional experiments (Schwinck 1955). Following the synthesis of disparlure in 1970 (Bierl et al. 1970), the first field experiments were carried out in Germany, Czechoslovakia, and Denmark to test the effectiveness of this product against the nun moth (Schoenherr 1972; Skuhravy et al. 1974; Schroeter and Lange 1975). Recent investigations have shown that the natural sexual attractant of the nun moth chemically is identical with disparlure (Bierl et al. 1975). The optical identity of these two substances has not been established, however. : Swiss Federal Institute of Forestry Research, 8903 Birmensdorf/Zurich, Switzerland. Results of field experiments in Germany support in the meantime the opinion that L. dispar secretes (+) -disparlure, 29 Tests with disparlure-baited traps were first carried out in the Swiss Alps in 1973, and systematic experiments have been conducted since 1974. Table 1 shows the sites and number of traps. One site is in the Rhone Valley (Albenwald) while all others are in the Zer- matt Valley. The type of trap used was the Pherocon 1 C Trap, manufactured by Zoecon of Palo Alto, Calif. In 1974, three traps were placed at each of the eight sites in the Zermatt Valley and six were placed at Alben- wald. The following year, the number was doubled at the Zermatt Valley sites and nine were used at Albenwald. The traps were placed in a horizontal line along the slope of the mountain, 2 to 3 meters (6 to 10 feet) up on the trunks of trees, and 20 to 40 meters (65 to 131 feet) apart from one another. In the 1974 experiments, the traps at each site were baited with three different concentrations of disparlure: 0.1, 0.2, and 0.4 mg. As no significant difference in the number of males attracted resulted from the use of different: concentrations, only two concentrations were used in 1975, 0.1 mg (A) and 1 mg (B). The lower disparlure concentrations in 1974 and 1975 were applied in cotton rolls -- much like those used by dentists -- along with 3 mg of Trioctanoin in order to retard the release of the attractant. The higher concentration in 1975 was placed in plastic caps with no addi- tional ingredient, with the caps themselves taking over the function of the carrier. Table 2 shows that in 1975 the traps con- taining the stronger concentration attracted significantly more moths than those with weaker concentrations in all sites. This is whereas 1. monacha produces both (+)- and (-)-disparlure [Klimetzek, D., Loskant, G., Vite JP, Mori, K-, 1976:) Disparlure: differences in pheromone perception between Gypsy Moth and Nun Moth. - Natur- wiss. 63, 581-582]. Table L.--Distribution of disparlure-baited traps in the Albenwald and the Zermatt Valley, 1974/75 Altitude 3 3 Site Tree species Aspect m 1974 STEAM WES) Albenwald Fir,Spr,La, Pic N 1150 6 6 + 3 St. Niklaus Spr ,La,APi NW 1400 3 3 tp 3 Herbriggen R La,Spr W 1350 3 3 ar 3 Herbriggen L La,Spr E 1300 3 3 P 3 Randa R La,Spr W 1400 3 3 + 3 Randa te La,Spr E 1400 3 3 + 3 Tasch R La,Spr W 1450-1500 3 3 + 3 Tasch L La,Spr E 1450-1500 3 3 + 3 27 27 24 ; R and L designate opposite sides of the valley Fir = Silver fir Spr = Norway Spruce La = European larch Pi = Scotch pine 3 APi = Alpine pine A = 0.1 mg and B = 1 mg concentrations of disparlure Table 2.--Number of LZ. monacha males in disparlure-baited traps in the Albenwald and the Zermatt Valley, 1974/75 Disparlure 0.1 mg Disparlure 1 mg Site Number of males 1974 __ Number of males 1975A Number of males 1975B Total Per trap Total Per trap Total Per trap Albenwald 293 48.8 183 30-5 237 79 St. Niklaus 100 S3e6 69 23 198 66 Herbriggen R 99 33 52 1753 217 123 Herbriggen L 65 sled) 81 27 161 BNS6 Randa R 99 33 69 23 96 32 Randa L 66 22 14 4.7 63 eal Tasch R 56 ish 7 i) Soi 30 10 Tasch E 35 bead 8 ell 29 O71; also evident from figure 1, showing the catches at Albenwald in 1975. Upon looking at the male catches by the lower disparlure concentrations in 1974 and 1975, it is noted that with one exception the catches in 1975 were lower than those of 1974 (table 3). This may be due to the competitive effect of the traps with the higher concentra- tion. In the case of the one exception (a gain of 24 percent per trap in Herbriggen L), this might be pure chance, but a local increase in population might have been the cause. If one similarly compares the results of 1975 using 1 mg disparlure with those of 1974 using 0.1 to 0.4 mg, one notices that at only four sites were significantly better catches obtained using the stronger concentra- tion. At four other sites in the upper part of the valley, the number of males caught per trap using the stronger concentration 30 remained below that of the previous year. This might be explained by the fact that there were approximately the same number of moths in this area as in the previous year, and the males simply were distributed over more traps. Table 4 is an attempt to correlate the total number of males attracted in each of the two years with the estimated forested area below the line of traps, whereby the following assumption was made: The males react to the disparlure only toward evening (Schroeter and Lange 1975). At this time of day, the wind is moving downward along the slope of the moun- tain. Since the source of scent can only be detected downwind, this makes the forested ‘area below the traps a potential source of moths which can be attracted. The area assumed to be affected by the attractant was increased at all sites in 1975 due to the increase in number of traps used that year as compared with 1974. The percent increase in = ae oO =) LIVE CROWN RATIO *SO’x 100’ HABITAT PLOT “STAND STRUCTURE (GAF 10) “REPRODUCTION “TREE HEIGHT * CROWN DENSITY - INCREMENT BORING ° MISTLETOE RATING * GROUND PHOTO Figure 1.--The basic plot sampling design. trees were permanently tagged for future reference. Additional data was taken on tree height, live crown ratio, mistletoe occurrence, growth rates, crown density, reproduction, habitat type, and herbage yield. Each plot was permanently marked and photographed on the ground for future comparisons and remeasurements. 36 Results Lodgepole pine mortality was estimated two ways: by aerial photography interpreta- tion, and by the variable plot method. Esti- mates of the number of lodgepole pine per acre killed by the mountain pine beetle by both of these methods follow: METHODS >5 inches dbh >/ inches dbh Aerial photos 113.2 Pbos' Variable plot Wall 79.6 Considering the high level of mortality, there was generally good agreement between the two methods For the photo interpretation por- tion, the R values for all trees and those 7 inches dbh and greater were 0.80 and 0.94, respectively. The photo method is believed to best describe the overall tree losses as well as diameter distribution, and although there is a disparity between the two methods in the small diameter classes, they are closely in accord with trees greater than 10 inches dbh. w a ) < in w a o w w a Fr 3 SS) iO =I (12 13 14 DIAMETER CLASS (INCHES) Figure 2 depicts the stand structure, living and dead, following the outbreak in 1972-73 while figure 3 shows the 1972-73 live stand component compared to that measured in 1975. Initial stand mortality of merchantable lodgepole pine 9 inches dbh and greater amounts to 62 percent of the trees and 65 per- cent of the volume for the lodgepole component and 56 percent of the trees and 55 percent of the volume for the stand (tables 1, 2). There is a very little overall difference between the two sampling periods. Approxi- mately, six lodgepole pines per acre died, mostly from windthrow (tables 3, 4). How- ever, ingrowth practically nullified this difference. The decrease of approximately eight trees per acre in the 6-inch class may be partially due to ingrowth into the 7-inch class. iY COP , LITT) LODGEPOLE PINE LIVE LODGEPOLE PINE DEAD OTHER SPECIES LIVE Figure 2.--Stand structure of lodgepole pine forest following an outbreak of the mountain pine NOWASH/S3e beetle, 37 w a VY < [a4 w a ” wl wW a E iy I9TS ae ca, species DIAMETER CLASS (INCHES) Piisigy 7230 Yj a LY Figure 3.--Comparison of stand structure of mountain pine beetle depleted lodgepole pine forest following an outbreak, The difference between the two cruises is greater when the data are converted to board feet volume 5 inches dbh and more: Board feet volume in trees >5-inches dbh Year 1975 1972 Diff. Year 1975 1972 Diitifi. Year 1975 1972 Diff. Lodgepole pine 1ve Dea ota 5103 7050 12153 4894 6110 11004 +209 +940 +1149 Other species Live Dead Total 1797 73 1970 1705 835 2540 +92 -662 -570 Tota! ive Dea ota 6900 YES) 14123 6599 6945 13544 +301 +278 +579 38 1972-73, and three years later, 1975. The important factor here is that al- though some trees, particularly lodgepole pine, continued to die (mostly windthrow and secondary insects), there was a net increase of 301 bd. ft., with lodgepole volume increas- ing by more than 200 bd. ft. Now that the stand has stabilized somewhat, and barring a natural catastrophe, gains should continue to outstrip the losses. The implications of this exceptionally heavy tree mortality on forest fire hazard and Management is evident. These data eventually will be converted to tons of fuel per acre as criteria for on-the-ground fuel loading and other fire hazard classifications. Table 5 shows the reproduction component as estimated during the 1972-73 and 1975 surveys. The overall level of stocking for all tree species is more than adequate, but personal observations indicate large varia- tions in both distribution and abundance between areas. Some form of supplemental stand examination may be required to specifi- cally identify those areas inadequately stocked or in need of conversion. Table 1.--Stand structure Nepees per acre) of a lodgepole pine forest following a mountain pine beetle outbreak, Targhee National Forest, 1972-1973 Trees per acre Area Diam. Lodgepole pine Other spp.¢ Total Class Live 1 Dead l Live 1 All spp. (In.) (No.) Percent (No.) Percent’ (No.) Percent” (No.) Percent Porcupine 5 20.8 8.2 12.2 4.8 8.6 SoS} 41.6 16.3 Targhee N.F. 6 20.4 8.0 5.9 228 6.8 2.7 Sout 13.0 i/ 19.9 7.8 8.7 3.4 6.9 Asi B50 13.9 Basal Area/acre 8 18.6 Vo8 10.0 3.9 DS Ze 33.9 Wsh53) 9 U7 oa/ 6.9 Ala ll 8.3 2.6 1.0 41.4 16.2 Live Lpp. 40.0 10 62 3.6 9.2 3310 2.7 eal 21k: 873 Dead Lpp. 51.2 ll D)g il 2.0 10.8 4.2 es 0.5 2 6.7 Other Spp. 15.7 12 58 0.9 Vo 2.8 eS 0.6 11.0 4.3 13 1.6 0.6 ae3 Wed 0.9 0.4 6.8 6) Total 106.9 14 0.8 Or SL eZ 0.5 0.2 4.4 Ike 7 15 0.3 Op 1.8 0.7 eal 0.4 SEC Nee 16 0.1 0.1 ies) 0.5 0.7 0.2 Zésdh 0.8 >16 O53 0.1 deal 0.8 16 OF 4.0 ¥6 TOTAL in Gy/eal 45.9 97.7 38.2 40.5 15.9 255.3 100.0 Percent of total stand Douglas-Fir, Subalpine Fir, Englemann Spruce, Limber Pine, and Quaking Aspen Table 2.--Stand structure (volume per acre) of a lodgepole pine forest following a mountain pine beetle outbreak, Targhee National Forest, 1972-1973 Volume per acre scribner Area Diam. Lodgepole pine Other spp. Total Class Live 1 Dead 1 Live 1 All spp. 1 (In.) (Bd.Ft.) Percent” (Bd.Ft.) Percent” (Bd.Ft.) Percent” (Bd.Ft.) Percent Porcupine 5 499 3.9 252 1.9 63 0.6 814 6.4 6 480 3.8 140 ipa 36 0.3 656 5.2 7 455 3.6 168 WES) 60 ORD 683 5.4 8 669 ge: 272 all 77 0.6 1018 8.0 9 925 Usd 873 6.9 61 0.5 1859 14.7 10 643 5.1 588 4.6 126 1.0 1357 10.7 11 419 358} 791 6.2 66 0.5 1276 10.0 12 311 2.4 788 6.2 106 0.8 1205 9.4 13 213 17, 528 4.1 85 0.7 826 6.5 14 124 1.0 496 3.9 68 0.5 688 5.4 15 60 0.5 296 Ze 162 153 518 4.1 16 14 OFT 257 2.0 135 Thea 406 Sue >16 82 0.6 661 af 660 bie 1403 LO TOTAL 4894 38.6 6110 47.8 1705 13.6 12709 100.00 Percent of total stand Douglas-Fir, Subalpine Fir, Englemann Spruce, Limber Pine, and Quaking Aspen 39 Table 3.--Stand structure (trees per acre) of a lodgepole pine forest following a mountain pine beetle outbreak, Targhee National Forest, 1975 Trees per acre Area Diam. Lodgepole pine Other spp.¢ Total Class Live 1 Dead 1 Live 1 All spp. 1 (In.) (No.) Percent (No.) Percent (No.) Percent’ (No.) Percent Porcupine 5 22.0 8.5 14.7 5.6 6.1 oS} 42.8 16.4 Targhee N.F. 6 WRT 4.9 6.8 2.6 MO) 4.2 30.5 nb 7/ 7 23a 8.9 8.7 SoS} Saul Ne 34.9 13.4 Basal Area/Acre 8 19.6 Vos 11.0 4.2 6.2 Doli 36.8 14.1 9 16.6 6.4 21.5 823 23 0.8 40.4 15.35 Live Lpp. 41.3 10 OS7/ 4.1 9.2 S25 Pell 1.0 22.6 8.6 Dead Lpp. 52.8 11 5.8 ee 11.6 4.5 1.8 (5 7/ 19.2 7.4 Other Spp. 15.8 12 a4 123 Wot 2.8 Boil 0.8 WaT 4.9 13 1.8 G7 4.3 56 0.4 0.2 6.5 25 Total 109.9 14 0.5 0.2 Soul eZ 0.9 0.3 Aa5 Nod) 15 0.4 0.2 1.8 Oe? 0 0.4 BZ es 16 Onl 0.1 3 0.5 0.7 0.3 Zell 0.9 >16 0.2 Watt Oe. _0.8 ey. 0.7 (Noi 1.6 TOTAL 116.9 45.1 103.4 39.6 40.0 153 260.3 100.0 l Percent of total stand Douglas-Fir, Subalpine Fir, Englemann Spruce, Limber Pine, and Quaking Aspen Table 4.~--Stand structure (volume per acre) of a lodgepole pine forest following a mountain pine beetle outbreak, Targhee National Forest, 1975 Volume per acre scribner Area Diam. Lodgepole pine Other spp.< Total Class Live l Dead 1 Live 1 All spp. 1 (In.) (Bd.Ft.) Percent (Bd.Ft.) Percent (Bd.Ft.) Percent (Bd.Ft.) Percent Porcupine 5 560 4.0 308 2.2 18 0.1 886 6.3 6 300 Ze 160 eZ 80 0.6 540 4.0 7 531 3.8 210 155 29 0.2 770 555 8 720 LG 352 25 68 0.5 1140 8.2 9 854 6a 1142 8.2 57 0.4 2053 14.7 10 762 5 630 4.5 144 1.0 1536 PSO ll 477 3.4 1009 URC 109 0.8 1595 11.4 12 394 2.8 816 5.9 160 Ns 1370 9.9 13 241 7 576 4.1 39 0.3 856 6.1 14 76 0.6 503 3.6 142 NEO 721 35/4 15 85 0.6 325 a3 144 1.0 554 3.9 16 38 0.3 286 2.0 135 1.0 459 Sans >16 65 0.5 733 5.2 672 4.8 1470 10.5 TOTAL 5103 36.7 7050 50.4 1797 12.9 13950 100.00 1 Percent of total stand Douglas-Fir, Subalpine Fir, Englemann Spruce, Limber Pine, and Quaking Aspen 40 Table 5.--Summary of reproduction survey, Targhee National Forest, 1972~73 and 1975 Seedlings Saplings Live Dead Year Species Trees/acre Percent Trees/acre Percent Trees/acre Percent 1972-73 Lodgepole pine 425.0 49. 104.0 69.4 35.0 66.0 Subalpine fir 67.0 Ue 8.0 BGS} 0.0 0.0 Douglas-fir 125.0 14, 18.0 12.0 8.0 Syed Englemann spruce 10.0 Ihe 2.0 a3} 0.0 0.0 Quaking aspen 205.0 24. 7.0 4.7 10.0 18.9 Limber pine 21.0 Zo 11.0 Lod 0.0 0.0 All Species 853.0 100. 150.0 100.0 53.0 100.0 1975 Lodgepole pine 398.0 42. 75.0 56.4 47.0 75.8 Subalpine fir 136.0 14. 21.0 15.8 1.0 1.6 Douglas-fir 78.0 8. 20.0 15.0 6.0 9.7 Englemann spruce 18.0 Ihe 3.0 Zoe 0.0 0.0 Quaking aspen 294.0 Siibe 3.0 2583 8.0 1229 Limber pine 70 ils LO 8.2 0.0 0.0 All Species 941.0 100. 133.0 100.0 62.0 100.0 There was relatively high mortality of lodgepole pine saplings between the two surveys (47 per acre) which is primarily due to segondary bark beetles, particularly Jps spp.’ which characteristically follow in the aftermath of a mountain pine beetle outbreak (Evenden and Gibson 1940). The plant association or habitat type analysis is incomplete at this time, but a preliminary scan of the data indicates that more than half of the site study plots were classified as being in the Psuedotsuga men- ztestt/Colamogrostis rubescens (Douglas-fir/ pine grass) habitat type. By far the most severe lodgepole mortality occurred in this lower elevation vegetational zone, which differs from that reported by Roe and Amman (1970) who measured the most intense beetle activity in the Abtes lastocarpa/pachtsttma mystnites habitat type or middle elevational zone. It is now known, however, that the severity of mountain pine beetle outbreaks in lodgepole pine is inversely related to eleva- tion (Amman and Stipe 1972). The plant classification and abundance data, coupled - Gibson, Archie L., Status and Effect of a Mountain Pine Beetle Infestation on Lodgepole Pine Stands. Forest Insect Laboratory, Forest Service, U.S. Depart- ment of Agriculture, Coeur d' Alene, Idaho. (Unpublished report, 1943.) 41 with crown density, may provide some insight into change in forage capacity. These results are preliminary and stem from only a portion of the data collected during the survey. They are being reported to show the drastic and obvious changes in a lodgepole pine forest caused by the mountain pine beetle. The immediate changes are in stand structure, species abundance and com- position, volume, reproduction, and other readily measurable factors that will aid in the identification of major impact areas. Subsequent compilation and analysis of other data such as tree growth rates, mistletoe intensity, habitat relationship, fuel loading, and the collection of supplemental, corrobora- tive data including recreation use trends, streamflow records, changes in road, trail, and fence construction costs, and their analysis and interpretation will be incor- porated into a subsequent report. In the meantime, however, it is hoped that this preliminary information will provide the base upon which the final analysis can be built. Literature Cited Amman, Gene D., and Lawrence E. Stipe. 1973. Lodgepole Pine Losses to Mountain Pine Beetle Related to Elevation. USDA Forest Survey Research Note INT-171, 8) p.> Lllus). Evenden, James C., and A.L. Gibson. Roe, A.L., and G.D. Amman. 1940. A Destructive Infestation in Lodge- 1970. The Mountain Pine Beetle in Lodgepole pole Stands by the Mountain Pine Beetle. Pine Forests. USDA Forest Service Re- Jour. For. 383) ) 27275 search Paper, INT-/1, 23 p. 42 ; ; ; ; pen Economic Impact of Mountain Pine Beetle on Outdoor Recreation Eee ehalisontandu sitll siniceritas = This study estimates the economic impact of outdoor recreation as a contributor to total value of forest resources in an area which has been heavily infested by mountain pine beetle. The area of study was the Island Park area in the Targhee National Forest in eastern Idaho. This National Forest includes a popular recreation area west of Yellowstone and Grand Teton National Parks. Recreation-—- ists use this area for both destination and nondestination purposes. The recreation opportunities include water sports, hiking, and related outdoor activities. It is classi- fied as one of the major recreation areas of Idaho. The Island Park area has been heavily infested with mountain pine beetle since 1960. The major tree species involved is lodgepole pine, which at present is utilized for poles, fence posts, roundwood, cordwood, and pulpwood. The recreation resources of the area are directly impacted by the mountain pine beetle as was indicated by the large number of dead trees observed in infested campgrounds. The question which is uppermost in the minds of the resource managers is: To what extent is the mountain pine beetle affecting recreational and other resource values in the Targhee National Forest? Secondly, how did recreationists react to the large number of dead trees in the infested areas? : The work herein reported was funded in part by an IPM sponsored project en- titled, “The Principles, Strategies, and Tactics of Pest Population Regulation and Control in Major Crop Ecosystems.” Journal No. 7513, Idaho Agricultural Experiment Station. This paper was published previously using estimated rather than actual campground use data in the Southern Journal of Agricul- tural Economics, December 1975, p. 43-50. 43 Objectives The purpose of this study is to estimate the economic impact of the mountain pine beetle infestation on the recreational re- sources of the area and project this economic impact on the future recreational use of the area. The specific objectives were to: 1. Survey recreational users in selected campgrounds in the Targhee National Forest to obtain information on re- creational patterns and uses 2. Develop recreational demand models to estimate the economic impact of mountain pine beetle on recreational use in the Targhee National Forest. Data The basic data used in this study were obtained from personal interviews with approxi- mately 500 recreational users in six camp- grounds in the Targhee National Forest during July and August of 1973. The campgrounds were selected for this study based upon the degree of mountain pine beetle infestation evident. Three of them were defined as infested (over 50 percent of the trees affected by mountain pine beetle), and three as noninfested (under 50 percent of the trees infested). All areas of the Targhee National Forest exhibit some degree of mountain pine beetle infestation. Respectively, professor and research associate, Department of Agricultural Eco- nomics, University of Idaho, Moscow, Idaho. Rivas, A., 1974. “Economic Evaluation of Mountain Pine Beetle Control on the Targhee National Forest," unpublished report presented at Twenty-fifth Annual Western City, Utah. The interview procedure was to visit the campgrounds in the evening, leaving question- naires with the recreationists and allowing them to fill them out overnight. In order to obtain a representative, unbiased sample of the user population, a systematic method of stratified sampling was used. The sample was stratified by campground, and users of each campground were sampled systematically. Questionnaires were collected at different campsites in each campground each day so as to assure a representative sample. Of the 500 questionnaires handed out, 90 percent were returned and 30/7 were useful in the analysis. The information obtained from the ques-— tionnaires consisted of a profile of recrea- tional users of the area, a catalog of the activities in which they participated, origin- destination data, and the transfer costs of the recreation trip. The transfer costs included the cost of transportation and costs directly related to participating in the recreational experiences. The tabulated questionnaires indicated that approximately 86 percent of the rec- reationists were repeat visitors and only 14 percent were first time visitors. Recreation was the major purpose of the trip for the majority (53 percent), and it was a vacation trip for most of them (49 percent). Only 30 percent of these recreationists visited areas besides the Targhee compared to 60 percent who did not, and 90 percent indicated that ‘they planned to return to the area in the future. The activities most frequently participated in were fishing, camping, sightseeing, canoeing or rafting, hiking, photography, swimming, and waterskiing. The average length of stay in these campgrounds was 6.4 days and the average size of group was 3.8 persons. Information on the user's average mileage, travel time, and trip cost is shown in tables 1 to 3. The differences in the average mileage traveled, average travel time, and average costs of recreating between residents and nonresidents were not very large. The reasons for this were related to the fact that many of the residents visiting the area came from western and northern Idaho which is 500 to 800 plus miles distant by highway. Secondly, when out-of-State people indicate that the major purpose of their recreation trip was to visit some other area, the mileage charged to their visit to the Targhee was computed from the last stop prior to their Targhee visit and on to their next stop. This was done in order to allocate travel costs in a reasonable manner between destination and nondestination recrea-— tion. A third factor was that the majority of the out-of-State recreationists who use this area come from northern Utah (approximately a 300-mile trip). Table 1.--Average miles traveled to and from the Targhee National Forest, 1973 Average Average User group mileage to mileage from Idaho residents 517 465 Nonresidents 612 642 Total sample 550 527 Table 2.--Average travel time in hours to and from the Targhee National Forest, 1973 Average Average User group hours to hours from Idaho residents 47.2 5a Nonresidents 57.6 63.9 Total sample 50.7 60.0 Table 3.--Average trip expenditures made to recreate in the Targhee National Forest, 1973 Average Average expend- User group total cost iture in Idaho Idaho residents $188 $151 Nonresidents 191 141 Total sample 188 147 Methodology The procedures used to evaluate economic impact would logically compare two situations. The first step would be to hypothesize what the situation in the Targhee National Forest would be without the mountain pine beetle and compare this with the situation where the beetle infestation exists. The difference measures the economic impact of the mountain pine beetle on outdoor recreation. Equation (1) below indicates a simplified model: R w/o mpb — R w/mpb = E.I. (1) where: R w/o mpb = economic value of recreation without the pres- ence of mountain pine beetle R w/mpb = economic value of recreation with mountain pine beetle infestation E.[. = economic impact of the mountain pine beetle The evaluation procedure relies upon separa- tion of campgrounds in order to compare those infested with those not infested. Where the economic impact of recreation was estimated, it was done by interviewing recreationists camping in infested and noninfested camp- grounds. The evaluation technique involved de- veloping a statistical demand model which estimates the number of visitor-days of outdoor recreation as a function of round trip mileage, estimated travel time, and cost per visitor-day. Once an equation was devel- oped, it was then possible to determine the average transfer costs and the average con- sumer surplus per visitor-day. The general form of the demand curves developed is shown in Equation (2) below: YS Chae [8 Ok as eae fe aes (2) where: Y = number of visitor-days per group X, = round trip mullllearen X, = estimated travel imeg X, = costs per visitor-day per person a+B = constants € = error term The above general model utilizes a mul- tiple regression least squares analysis. The ‘usual assumptions of this estimating technique are made. Analysts The demand equations developed in the analysis are shown in the tabulation below. 2 For a more detailed discussion see Clawson, M. and Knetsch, J.L., “Economics of Outdoor Recreation.” Resources for the Future, Johns Hopkins Press, 1966, and Nawis, F., “The Oregon Big Game Resource: An Economic Evaluation.” Unpublished Ph.D. Thesis, Oregon State University, Corvallis, Oregon, 1972. Ec Both round trip mileage and esti- mated travel time variables were included so as to improve the predictability of the model. The inclusion of both variables simul- taneously in the model inevitably results in some degree of multicollinearity, but both variables were deemed necessary to the model, and, therefore, both were included. 45 The R statistics in the three equations varied from 0.435 to 0.564. The estimated economic values are shown in table 4. Demand relationships were estimated for: (1) All campgrounds, (2) campgrounds which were heavily infested with mountain pine beetle, and (3) campgrounds which were lightly in- fested with mountain pine beetle. For pur- poses of convenience, the terms infested and noninfested are used to describe (2) and (3). Table 5 indicates average number of visitor- days per. group, average cost per visitor-day, and average consumer surplus per visitor-day values. The average consumer surplus per visitor-day was obtained by integrating the equation between the average cost per visi- tor-day and the highest reported cost per visitor-day estimate. The consumer surplus is defined in the usual sense as the benefit which consumers receive but do not pay for. It can be interpreted as a net resource value for publicly owned properties if the assump- tions are made that the marginal utility equals the marginal cost at each point on the curve above the average cost per visitor-day, and that the government is a discriminating monopolist. Equations used to estimate the demand for outdoor recreation in the Targhee National Forest, 1973: 1. All campgrounds: N = 181°, p> = 0.499, F = 58.54 ¥" = 13.732 + 0.005X,* + 0.632X,* - 1,142K,*° (1.0617)(0.0012) (0.0140)*(0.2701) 2. Infested campgrounds: N = 113, p> = 0.435, F = 28.00 Yy" = 13.920 + 0.004X,* + 0.732X,* - 1.083X,** (1.2592)(0.0017)! (0.2125) (6.3703) Y = number of visitor-days per trip, X) = round trip mileage, x = hours traveled to recreation in area, and X, = cost per visitor-day per person. b Although 307 questionnaires were useful in analyzing the socioeconomic charac-— teristics of the users, only 181 were useful in statistically estimating demand. Only complete questionnaires in regard to variables, ape 6 a DCA ME Waral NC. CGleBay no missing data) were included in the statistical analysis. * Coefficient significant at the 5 percent level, and estimates of the standard errors of the coefficients are given in paren- theses, b 3. Noninfested campgrounds: N = 68, R = 0.54, F = 27.60 Y" = 12.869 + 0.006X,* + 0.555X,* - 1.083X,%*°* (1. 9803) (0.0018) {0. 2027) (0. 4224) 2 It is evident that there were differences between the estimates of value derived for the three demand equations. The average cost per visitor-day was $2.95 per day in all camp- grounds; $2.85 per visitor-day in infested campgrounds, and $3.10 per visitor-day in noninfested campgrounds. The average consumer surplus values were $7.80 per visitor-day in all campgrounds, $7.75 per visitor-day in infested campgrounds, and $8.95 per visitor- day in noninfested campgrounds. Table 4.--Estimated economic values for out- door recreation in selected campgrounds in the Targhee National Forest, 1973 Average Average Average consumer Campground visitor- cost surplus categories days per per per group visitor- visitor- day day All 16.8 $2.95 $7.80 Infested 1558 2.85 Hes) Noninfested 183 Se 895 The above results indicate that there was a difference in recreationist responses between infested and noninfested campgrounds. It is assumed that this response measures the desirability of recreating in campgrounds which do not have a large proportion of dead trees. The demand curves are measuring the response of recreationists to the environment by the length of their stay, and by the amount of money they spend. Estimation of Losses The losses were determined by calculating the differences between the estimated average consumer surplus and recreation costs for the infested and noninfested campgrounds. The average consumer surplus values were estimated by holding the other variables in the estimat- ing equations at average levels; and the recreation costs were the estimated average cost per visitor-day. The method used to develop the loss values is indicated in table 5. The calculation is to subtract the con- sumer surplus value of the infested camp- grounds from that estimated for the nonin- fested campgrounds ($8.95 - $7.75 = $1.20/ 46 Table 5.--Estimated losses of recreational values resulting from mountain pine beetle infestation in the Targhee National Forest, 1973 Average Average number Average consumer Item of cost surplus visitor- per per days per visitor- visitor- group day day Noninfested campgrounds 18.3 $3.10 $8.95 Infested campgrounds 15.8 2.85 Hod!'S Net difference 2.5 0.25 20 visitor-day). A similar calculation was made for the cost per visitor-day expenditures ($3.10 - $2.85 = $0.25/visitor-day). These residuals were then summed to determine the total value (marginal value per visitor-day) of $1.45 per visitor-day. This value is an estimate of the economic cost of the mountain pine beetle infestation in terms of its impact on recreational values. The values generated above were aggre- gated to determine the magnitude of the total losses caused by the mountain pine beetle in the Targhee National Forest. This was done first for the campgrounds which were studied, then for all campgrounds in the forest, and, finally, for all the campgrounds in the forest assuming the average level of infestation which currently exists in the Targhee National Forest. In the case of the campgrounds which were studied, the estimated losses reflected the existing situation with regard to the infesta- tion levels of mountain pine beetle in the six campgrounds studied. The loss estimates were based on U.S. Forest Service estimates of recreational use in these campgrounds. This estimated use was 177,600 visitor-days. The estimated losses were $257,520 based on the average loss per visitor-day of $1.45 esti- mated from the demand equations developed previously. This value can be allocated as follows, $44,400 in reduced expenditures, and $213,120 in reduced consumer surplus (table 6). Additional data on campground use were available from U.S. Forest Service records. These data indicated that the total number of visitor-days in all campgrounds in the Targhee National Forest during 1973 was 252,000 visitor-days. In making loss projections, it was assumed that the level or degree of mountain pine beetle infestation would be the same as that observed in the campgrounds Table 6.--Estimated potential economic losses in outdoor recreation values assuming that the campgrounds studied were infested with mountain pine beetle, 1973 tem Value All campgrounds studied (177,600 visitor-days) 1. Not infested a. Expenditures $ 550,560 b. Consumer surplus 1,589,520 Total $2, 140,080 2. Infested a. Expenditures $ 506,160 b. Consumer surplus 1,376,400 Total $1,882,560 3. Not infested - infested (difference of above figures) a. Expenditures $ 44,400 b. Consumer surplus Zi 20 Total $ 257,520 previously studied. This assumed an infesta- tion level of 0 to 50 percent in the nonin- fested campgrounds and a 50 to 100 percent infestation level in the infested campgrounds. The first projection was made by assuming that all 19 campgrounds in the Targhee Na- tional Forest were infested and the economic losses were calculated as shown in table 7. Table 7.--Estimated potential economic losses in outdoor recreation values assuming that al] campgrounds in the Targhee National Forest were infested by mountain pine beetle, 1973 Item Value Total campground use (252,000 visitor-days) 1. Not infested a. Expenditures $ 781,200 b. Consumer surplus 2,255,400 Total $3,036,600 2. Infested a. Expenditures $ 718,200 b. Consumer surplus 1,953,000 Total $2,671,200 3. Not infested - infested (difference of above figures) a. Expenditures $ 63,000 b. Consumer surplus 302,400 Total $ 365,400 47 The total losses were $365,400. This was determined as follows: a) Expenditure losses were $0. 25/visitor-day x 252,000 visitor days = $63,000, and b) consumer surplus or net resource benefit losses were $1.20 x 252,000 visitor-days = $302,400. A second estimate was made which assumed that only half of all the campgrounds in the Targhee National Forest would be infested at any one time. This relationship was assumed because there were no empirical data available to verify a greater or smaller level of campground infestation. What this assumption does is to introduce an aspect of marginality into the analysis in a gross way. The loss values estimated were half the value of those estimated for the previous estimate (table 8). The calculations were: a) ($0.25/visitor-day x 252,000 visitor-days)/2 = $31,500 loss of expenditures, and b) ($1.20/visitor-day x 252,000 visitor-days)/2 = $151,200 loss of consumer surplus. These values sum to $182,700 in terms of annual economic losses of recreation values in the Targhee National Forest. Table 8.--Estimated potential economic losses in outdoor recreation values assuming that half of all campgrounds in the Targhee Na- tional Forest would be infested by mountain pine beetle at any one time, 1973 Item Value Half of the total campground use (126,000 visitor-days) 1. Not infested a. Expenditures $ 390,600 b. Consumer surplus W277 00 Total $1,518, 300 2. Infested a. Expenditures $ 359,100 b. Consumer surplus 976,500 Total $1,335,600 3. Not infested - infested (difference of above figures) a. Expenditures Soo 500 b. Consumer surplus 151,200 Total $ 182,700 Investment tn Control Program If a recreational management agency is interested in developing a control program, an obvious question is, given the estimated losses caused by mountain pine beetle, how much money can the agency justify spending on control measures? This question can be thought of as how present losses are evaluated in terms of future losses. The present value of all future losses needs to be determined. This can be done by assuming that the esti- mated losses are an opportunity cost and discounting them at an appropriate discount rate. The rate used in this analysis was 7 percent. The formula used to develop these capitalized values was: = capitalized value L = aggregate annual economic losses discount rate of 10.0 percent = iT The present value of the economic losses for each of the three examples discussed above is shown in table 9. The total capitalized value for each example was: $2,575,200 for the campgrounds studied; $3,654,000 for the potential loss due to infestation by the moun- tain pine beetle in all campgrounds in the Targhee; and $1,827,000 value for loss of half of the campgrounds or visitor-day use in the Targhee National Forest. These values may seem high until compared to present investment in controlling infestation in the Targhee National Forest. "The present infestation began in 1960 and large scale control efforts started in 1962. Control efforts were undertaken to protect multiple use values which were never quantified. Control costs_through 1970 were 10.3 million dollars." The capitalized values are the present value of the recreational losses resulting from the mountain pine beetle infestation in the Targhee National Forest. These values can also be used to determine the upper limit on the amount of investment which could be justified for a pest management control program. The difficulty is that the decision- maker needs to know what his potential losses may be before he can determine the amount of investment he should be using, or if he should be concerned with a control program at all. Summary This study used recreational demand models to estimate the economic impact of mountain pine beetles on recreational use in Riviasi,) vAc), Ops) (CLL, pian 40s 48 Table 9.--Present values of economic losses in outdoor recreation values in the Targhee National Forest, 1973 Item Capitalized value 1. Studied campgrounds, 177,600 visitor-days a. Expenditures b. Consumer surplus Total $ 444,000 ZS Z00 $2,575,200 2. All campgrounds, potential losses, 252,000 visitor- days a. Expenditures b. Consumer surplus Total $ 630,000 3,024 ,000 $3,654,000 3. Half of all campgrounds, potential losses, 126,000 visitor-days a. Expenditures b. Consumer surplus Total $ 315,000 1,512,000 1,827,000 the Targhee National Forest. The procedure followed was to estimate the demand for both infested and noninfested campgrounds and compare the consumer surplus and transfer cost estimated derived from the demand models. These estimates were then used to simulate various infestation conditions to determine the magnitude of average annual losses from mountain pine beetle infestation. These losses were then capitalized to determine the total value of damages, and this value was interpreted as the upper limit for investment in control measures for the mountain pine beetle pest control program in the Targhee. There are several concerns which should be recognized when using transfer costs as a surrogate for prices in estimating consumer surplus values for outdoor recreation. First, an adjustment was made to account for nondes- tination use, because in some cases the hours and mileage traveled were incidental to the Targhee visit. A second factor affecting the estimation of consumer surplus was that the data used for this study were obtained during the summer of 1973, a period of rapidly rising gasoline prices. These price increases had the effect of dramatically increasing the average cost per visitor-day compared to | earlier years. A third factor was that the consumer surplus values are point estimates. These point estimates are assumed to have wide and unknown bounds, and computation of these bounds is difficult if not impossible. This paper is an attempt to use demand models to measure the economic impact of mountain pine beetle on outdoor recreation in the Targhee National Forest. The results imply that such measurement is possible, and that the loss estimates developed in this analysis may be compared to losses of other resources in the forest. More research is needed to develop models for other resources to obtain loss values related to mountain pine beetle damage. Literature Cited Clawson, M. and Knetsch, J.-L. 1966. "Economics of Outdoor Recreation." Resources for the Future, Johns Hopkins Press. 49 Nawis, F. 1972. "The Oregon Big Game Resource: An Economic Evaluation." Unpublished Ph.D. Thesis, Oregon State University, Corvallis, Oregon. Rivas, A. 1974. “Economic Evaluation of Mountain Pine Beetle Control on the Targhee National Forest." Unpublished report presented at Western Forest Insect Work Conference, Salt Lake City, Utah. Impact Analysis, Interpretation, and Modeling William A. Living in a world of scarce resources which have competing uses should lead the forest resource manager to question those actions which use these resources, including his when he acts as a pest manager. There- fore, one may ask: (1) Why should resource managers make impact analyses? (2) How can they usefully interpret them once they are made? and (3) Why bother with impact models? Why Make Analyses We have just had an excellent discussion by Dr. Stark on the concept and definition of impact. The major thrust of his and most defi- nitions is that impact is the net effect of an insect upon the physical ability to produce goods and services which are desired by man. The key phrase in this sentence is"... which are desired by man.” Insect impact has meaning only to the extent it affects some- thing which man desires. Let me more fully explain what I mean by "something desired by man" lest I be accused, as many economists are, of knowing the price of everything and the value of nothing. There is a very broad spectrum of goods and services produced from our forest resources.e These include not only the more tangible and easily quantified items, such as timber and grazing, but also those less easily quantified, such as recreational and esthetic services. All of these goods and services I Associate Professor of Forest Econo- mics, Department of Forestry and Forest Pro- ducts, Virginia Polytechnic Institute and State University, Blacksburg, Virginia. The author is currently Integration Coordinator for the U.S. Department of Agriculture program entitled The Expanded Southern Pine Beetle Research and Applications Program. The find- ings, opinions, and recommendations expressed herein are those of the author and not necessar- ily those of the U.S.Department of Agriculture. Teuse hiner which are affected by an insect or insect complex should, to the extent possible, be included in an impact analysis. I would like to clarify another point and that is that "net effect" of impact may be either positive or negative. The term "pest management" is negative because the word "pest" by definition means something detri- mental to mane However, we should recognize, at least philosophically, that the net effect of an insect or insect complex could be beneficial to man. I will be speaking in the negative context because that is why most forest control activities take place. The reason, then, for making an impact analysis is to discover whether an impact is positive or negative and also to discover or predict the amount of the positive or negative effects. Now, to be realistic, few impact analyses will be made until a resource manager first perceives that there is an interaction between an insect and what man wants from the forest. Further, the impact must be perceived to be large enough to make some kind of difference. The reason for this is obvious -- impact analyses cost money; they take time and effort away from other desirable activities and there should be a good reason for incurring these costs. The reason for incurring the cost is to provide the forest manager with information needed for him to decide upon which course of action to take. He must decide whether to do nothing or to initiate a control activity, and, if the latter, he must decide upon which activity or combination of activities to use and how large a program to initiate. But this brings us to the second question, how are impact analyses interpreted to help the mana- ger make these decisions? Interpreting the Analysts After impact is first perceived, the forest manager -- sometimes almost intui- tively -- decides whether it is small enough to be acceptable. This in itself is an interpretation which may be based on casual observations, experience or the results of some extensive surveys. The conclusion may be that: (1) The impact is acceptable and will be ignored henceforth, therefore, no further interpretation is needed; (2) the impact is currently acceptable but that it should be monitored frequently, therefore, further interpretation and decision is postponed to a later date; or (3) the impact is presently unacceptable, therefore further interpretation and decision is needed at once. A second judgment must usually be made if the impact level is umacceptable about whether or not a control activity is available or can be developed. Lack of control activity can force acceptance of otherwise unacceptable impacts. However, in most cases, a control activity exists even if that activity is to fund research to find better control activi- ties. If control activities are desired, I would let my economic bias come to bear and suggest that impact and other data be used in a benefit-cost analysis to help interpret the impact analysis and to provide a more complete guideline for decisionmaking. A benefit-cost analysis framework is one of the more useful methods of interpreting the results of an impact analysis. Its framework is quite general and the basic principles are appli- cable to many cases. The general framework is this: The benefits of a control program are the damages which that program prevents from occurring. Estimating benefits usually requires esti- mating the damages without the program, and then either estimating the proportion of these damages which the program will eliminate or estimating a second damage level with the program and subtracting it from the estimated damages without the program to obtain damages prevented. Valuing damages may be very difficult because either the quantities are unknown, such as the impact on an elk herd caused by the mountain pine beetle, or because the impact is not valued in the market place, such as the value of one head of elk. In some cases, we may just have to be satisfied with establishing that an impact exists and indi- cating whether it is positive or negative instead of putting the value of the impact in a benefit-cost analysis. However, to be useful for decisionmaking, a great deal more information is needed. It 51 is not only desirable to know the amount of damage reduction which a new control program will cause, but it would also be desirable to know how much more the damages would be reduced if the program size were increased or how much the damages would be increased if program size were reduced. This information is needed to provide guidelines for deciding on the most efficient program size. Finally, an analyst must estimate the costs of a particular control program. The cost analysis is usually easier than the benefit analysis because most costs are labor or materials for which estimates are readily available. Of course, nonmarket costs which do occur, such as adverse environmental effects of insecticides, should also be included just as nonmarket benefits should be included. The interpretation of the estimated benefits and costs is fairly straightforward. One simply adopts the control activity which has the greatest positive difference between benefits and costs, and a program is not adopted if the benefits are less than costs. This means that if the damage reduction is less than the control costs, the indicated course of action is to allow the insect to continue unchecked for the time being. Benefit-cost analyses are not, however, a complete guideline to forest management decisions. First, there are the obvious imprecisions in benefit-cost estimates. Second, there are those items omitted from the analysis because of our inability to quantify them, for example, the impact of insects on esthetic values. Finally, there are some areas which benefit-cost analysis is not designed to cover, most notably the impacts of insect damage on income and wealth distribu- tion. This latter area is concerned with shifts of income and wealth from one group to another. A recent example which came to my attention is the possible shift in the county tax burden in the western United States due to mountain pine beetle attacks on residential properties. Here the homeowners obtained reduced personal property assessments because mountain pine beetle killed trees on their house lots and hence lowered the market value. This meant that the tax was shifted to prop- erty owners who had no attacked trees and who consequently were paying relatively more tax than before the attack, assuming the total amount of tax collected in the county re- mained the same. This type of shift may be relevant to pest management decisions but is not included in a benefit-cost analysis. Why Model Impact The remaining question is: Why model impact? The reason is to predict the impact (or damages) for use in making control ac- tivity decisions. Note that there are other reasons for modeling, for example, population dynamics models which may lead to better understanding of life processes and hence allow improved control tactics. However, we are discussing impact and its analysis, and I would maintain the major reason for modeling impact is to provide information for control activity decisions. But why are impact models needed? Don’t we know what the impact is? The answer in many cases is no. Not only do we lack accu- rate estimates of the past impact of many forest insects but we certainly do not have estimates on what the future impact will be. It is this latter information that is needed because decisionmaking is for control activi- ties taking place in the future. For example, we would not plan control activities for the coming field season if we thought an insect population was going to collapse under its own weight. Similarly, the benefits of many control activities, particularly those that prevent rather than suppress impacts, take place over a number of years in the future. Assessing the benefits of these prevention programs, therefore, requires predicting the damage reduction in future years which, in turn, requires predicting future impacts. An understanding of risk and uncertainty and how to treat them will help do this. The management team for the Expanded Southern Pine Beetle Program has outlined four major types of impact models which they feel are important, but which are not yet devel- oped. These models are defined by the geographic area covered and the use to which the results are put. I would like to present them to you here because I believe they help to define the context of impact modeling (table 1). The first type of model is used to pre- dict impact over wide geographical areas for the next 5 to possibly 50 years. Here, wide geographical area is defined as an area which covers several counties or a forest survey unit sized area and which may contain hundreds of thousands or millions of acres (or hectares) of forest land. Impact information is desired to predict the benefits of long-term preven- tion programs for use in benefit-cost analy- ses. A second type of impact model would also predict impacts over wide areas but only for l year in the future. Hopefully, this model would give more precise estimates of damage and be better able to account for annual fluctuations than a model which predicts decades into the future. It would be used to plan the next field season’s activity and to make budget requests for submission to legis-— latures. The third model also predicts wide area impact but only for a month or so into the future. This model would be used to plan field operations during the season, for example, intensification of detection or con- trol efforts. We would also hope to have a hazard index available for use with this model to indicate the general areas on the ground where increased activity might occur. The fourth and final type of model would predict the likelihood of a severe outbreak on a specific area. It would seek to predict how large a specific spot will grow if left uncon- trolled. The purpose of this type of impact model is to guide operational decisions for suppression activities. It comes closest to Table 1.--Types of impact models Model number Area covered I Wide area years II Wide area LE Wide area IV Specific spot Information predicted Impact over next 5-50 Impact next year Impact next month Growth if uncontrolled D2 Use of prediction Estimate benefits of long-term pre- vention programs Plan budget request for coming field season Plan field activities Plan suppression activities on outbreak predicting what is commonly called the "econo- mic threshold" of control activity. We have also found it useful to think in terms of two different impact modeling techniques. The first may be called a popula- tion dynamics, or life table technique and the second a probability, stochastic, or actuarial technique. Note that, as with many categori- zations, these are not mutually exclusive, and that some models may have elements of both ‘categories. The population technique models, which may contain probability elements, simulate insect population numbers throughout their life cycle and translate these numbers to trees attacked and hence to impacts. The emphasis is on modeling the insect. The actual models may be much more complex than this, for example, they may model tree phy- siology in order to determine the next tree attacked. Population modeling has the advan- tage of usually including basic physiological processes which cause attacks and hence can cope better with changes in exogenous causal variables. One would speculate that the popu- lation technique would be most useful for Type IV models and least useful for Type I models, in descending order, because of difficulties in aggregating results over wide areas and long time periods. Probability technique models take a more empirical approach and seek to relate the probabilities of attack and spread to obser- vable abiotic and biotic factors affecting the host material. The emphasis is on modeling the host. Probability modeling has the ad- vantage of requiring less detailed knowledge of basic biological processes and the disad- vantage of being less able to cope with changes in exogenous causal variables. One 53 would speculate that the probability technique would be most useful for Type I models and least useful for Type IV models, in descend- ing order, because it is based on probabili- ties of attack and spread rather than vari- ables affecting biological processes which can be measured for an individual spot. Summary and Conelustons To summarize: (1) Impact analyses are made to supply information for deciding which course of action to take; (2) the analyses can be usefully interpreted within a benefit-cost analysis framework; and (3) models are needed to better predict future impacts and hence potential control program benefits. The need for better impact analyses and prediction models is becoming quite evident to many forest and pest managers. I detect, at least in the U.S., increased scrutiny of pest Management expenditures and requests to demon- strate that the benefits of proposed control activities exceed their costs. There appears to be less willingness to accept control pro- posals on the basis of: “bugs are bad because they kill trees" or “we better get them now before they eat the whole forest,” and an increased desire for quantified costs and results. I personally endorse this trend because it can lead to better decisions and more efficient use of resources, whether in the public or the private sector. However, these estimates require more knowledge, which, in turn, will require better records of past impacts and the development of more models to predict future impacts. III. \ \ ‘ CoNCEPT AND PRACTICE OF INTEGRATED PEST MANAGEMENT IN FORESTRY > Night Application of Aerial Sprays Using Multi-Engine Aircraft, Inertial Guidance Equipment, and Incremental Application Technology for Insect Control, with Special Reference to the Spruce Budworm ¢ | , A. P. Randall and R. Desaulederss 2 The relationship of aerial spray deposits on a coniferous forest and the degree of spruce budworm, Chortstoneura fumiferana, (Clem.), control based on ground sampling devices was” established in the early 1950s by Hurtig et al. (1953) at 10+ drops/cm for a coarse (MMD-250 ») DDT spray. These values were modified by Fettes (1962) gnd Randall (1969) to include 20-40 drops/ cm for ultra low volume concentrate sprays in insecticides. Research by Himel and Moore (1967 and 1969) showed that the highest degree of mortality to the western spruce budworm Choristoneura oecetdentalis (Freeman) was achieved by contact with drops of less than 50 w diameter, thus confirming the effectiveness of a fine droplet spectrum. Researchers over the years have used the above ranges of drop deposit para- u Respectively, Fisheries and Environ- ment Canada, Canadian Forestry Service, Chem- ical Control Research Institute, Ottawa, Ontario; and Quebec Department of Lands and Forests, Ste. Foy, Quebec. The authors gratefully acknowledge the support of the aircrews of Conair Avia- tion Ltd., Abbotsford, B.C., Canada and Chrysler Flying Services, Thermopolye, Wyoming, (U.S.A.), for their support and services during the trials. A very very special note of thanks to Mr. L. Pollock (C.C.R.-I Experimental Pilot) for aerial monitoring and engineering support throughout the project. Lastly, the authors wish to thank Messrs B. Zylstra and J. McFarlane for their devotion to duty and support during the trying phases of plot establishment and deposit monitoring of the experimental blocks. 55 — meters as an index for equipment calibration prior to use on operational budworm sprays. Calibration trials undertaken on a Douglas DC-7B spray aircraft (Randall and Zylstra 1972) indicated the potentiality of multi- engine aircraft for incremental spray tech- nology using swath intervals of 3000 feet based on a deposit coverage of 40 drops/cm . The basic concept of multi-engine use was predicated on an economical but effective swath width, large payload capability, elec- tronic navigation for swath lane alignment, and effective spray coverage for optimum insect mortality. The calibration trials were conducted using the Porton Method of crosswind spray application whereby the height/wind product determined the degree of downwind displacement of the drop spectrum to ensure the desired degree of spray coverage (Gunn et al. 1948). From these early trials, the spray dispersal (and hence, proposed droplet distri- bution) on the target site was based on the concept of incremental spray application (Randall and Zylstra 1972). The first semi- operational trials using the DC-7B spray aircraft equipped with the Litton LTN-51 inertial guidance system for spray block location and swath lane offset of 3000 feet confirmed the reliability of multi-engine aircraft, electronic swath guidance naviga- tion, incremental application technology, and efficacy of deposit distribution for spruce budworm control (Randall 1974 a,b; 1975). The success of these early incremental spray applications trials prompted the Quebec Department of Lands and Forests to embark on a program of incremental drift application in 1973, using multi-engine spray aircraft operating under electronic guidance control for swath lane location and automatic swath lane offset of 3000-foot intervals, thus eliminating the spray block system of visual terrain boundaries that constituted hazard zones for multiple spray application (Desaul- niers 1973 and 1975; Paquet and Desaulniers 1974). During the period of 1972 to 1976, an active cooperative program of research and development on application technology, swath lane navigation, pesticide formulation, and spray efficacy was undertaken to improve the Quebec spray operation and maintain an effec- tive spray program in the face of rising costs. The successful adaptation of a high capacity aircraft to produce a ULV droplet spectrum of MMD of 90 to 100 yu at volume output of 100 to 200 gallons (US)/minute and the subsequent improvement of the Litton LIN-51 navigation system to maintain a high degree of performance (Boivin and DeCamp 1975) opened up the potentialities for night appli- cation of pesticides for large scale spraying against the spruce budworm. Aertal Application Techniques As early as 1930, it was recognized that the most suitable weather for applying dust formulations by aircraft occurred during calm days when upward thermal air currents were minimal (Swain 1930). These findings were adapted and used with the introduction of oil formulations for the control of forest defol- iators (Stewart 1949) and led to the practice of aerial application during the early morning and late evening hours of the day when wind velocities, air temperatures, and inversion conditions were conducive to maximum spray deposition. The findings that the earliest sprays in the morning and the latest sprays in the evening usually provided the best deposits and hence insect control results, led to speculation that night application of pesticides would be extremely advantageous. This concept has been successfully adopted in agricultural spray programs in the United States using high intensity light beams for visual guidance (anonymous, 1966) but is hardly acceptable for forest spraying over unchartered terrain quite unlike agricultural fields. Furthermore, Federal regulations would require 700-foot altitude clearance above the highest point of land over the spray area. The introduction of the multi-engine aircraft and electronic guidance system, however, provided the means to test the concept. 56 Early Expertmental Night Trtals During the course of equipment develop- ment and modification of spray systems, numerous calibration trials were conducted at remote airports and late in the evening after sunset, thus providing the occasions to undertake "ad hoc" night trials. The first attempt to obtain deposit data during a night application occurred late in September 1973 at Val D’Or, Quebec, using a CL-215 converted water-bomber aircraft. Two crosswind spray runs were made at 1000 feet above a 20-mile layout of card samples. Meteorological conditions were extremely unfavorable for spray deposition with a high pressure region over the test area and wind velocities of up to 18 knots at spray height. Results were very disappointing with a deposit recovery of 1 large drop across the 18 miles of deposit cards. This was attributed to a malfunction of a nozzle. The results, how- ever, indicated the unsuitability of #2 fuel oil for high level spraying, and led to the development of a less volatile oil formulation (unpublished data) comprised of number two and number four bunker fuels. The trail also confirmed the findings of Mount et al. (1970) that wind velocity increases with increased altitude. The second opportunity occurred later in 1973 during a series of calibration trials with a Douglas DC-6B spray aircraft to deter- mine ULV droplet spectrum characteristics and effectiveness of low volume output on spray droplet deposition. Meteorological conditions were ideal (cool with winds at 0 to 0.5 mph at ground level and 0.5 to 1 above tree tops). Because of delays in equipment adjustment and rate of flow calibrations on the final cali- bration trial and fog conditions on the run- way, the available daylight hours were exceeded. Rather than abort the last trial, the pilot elected to make the final run under night conditions at 250 feet above the layout with the aid of ground flares and radio communication to ensure proper track orienta-— tion. During spray emission, wind speeds of 15 knots were encountered by the aircraft above the layout whereas ground and treetop wind conditions as indicated by smoke flares were 0 to 0.5 mph at 180° to the upper air stream. Maximum drop size (Dmax) of the spray deposits were displaced 700 yards downwind from the point of spray emission and covered the entire 3 miles of the sample layout. Results from this trial under night conditions illustrated the misconception of so-called ideal spray weather based on ground and treetop observations and stressed the need for meteorological equipment (Pibal or Radiosonde) to sample the weather profile from ground to spray height emission. The data from the day (FTX=14) and night (FTX=19) calibration runs when plotted on a theoretical incremental deposit basis for a swath interval of 3000 feet, illustrate the effect of wind speed and altitude upon spray deposition (fig. 1). By the spring of 1974, the new low vola- tile oil formulation (a 50/50 ratio of non- volatile to low volatile liquid fractions) was in operational use on daylight spray applica- tion programs against the spruce budworm. Simultaneously, a program of pilot training and equipment use of the LTN-51 was instigated to eliminate in-flight errors of navigation. Early in the spring of 1974, a preli- minary series of night trails was established DC-6 CALIBRATION 1973 at Lac des Loups, Quebec to ascertain the practicability of night operations. A 50-mile layout was established in order to provide two flight swaths aross the area for comparison of spray deposition of dissimilar spray formula- tions (fig. 2). Low level (200 feet) and high level (1500 feet) flight lines were estab- lished at right angles to the layout to provide crosswind deposit characteristics. The program, unfortunately, was plagued by adverse meteorological conditions associated with late spring snowstorms and further compounded by installation problems of one of the Litton LTN-51 guidance systems in one of the experimental aircraft. By the time all problems were resolved, the aircraft were required on the operational spray program and the night trial was shelved until after completion of the spray project. At midnight, 23/10/73| 18:03] 0-1! 2-3] 41! 46 200 400 600 800 !000 !200 !400 !600 IOmiles yards N = °o <4 w a Ww a (2) a a Date Time Wind Temp No.of | Pressure] Dosage ==> j | W | D |Nozzles} P.S.I. oz/acd gal /mi 118 iG Lal \ AN AA Cx et Silo nA vy \ (FEET) HEIGHT OF AIRCRAFT 200 400 600 800 /000 !200 !400 !600 |.Omiles yords { SWATH INTERVAL INSEE Figure 1.--Douglas DC-6b calibration trials (1973) under ideal day emission (FTX-14) and night emission conditions (FTX-19) showing a theoretical incremental cumulative spray deposit over 4-5 swath intervals. required. 57 For maximum effectiveness a total of 8 to 12 swath intervals are / / Grand Loke ‘fl ! & | ff / 1 / 4 Q Poulter ae Loke Loc des 4 > Lac des Loups Airport 7 4 te Embarrass , Ny 6 Loc ' Kove Kondiaronk Lake ' Lake ae ru ' a Delaney Hee }e---------- Gale Loke yA [a] Doolittle 5 uw Loke Nautical Miles Figure 2.--Map of the lower end of the 1974 experimental night spray layout showing the location of the low level spray pass (mile 4) and high level spray pass (mile 13). June 12, 1974, a DC-6B aircraft was released to provide a single spray run across the layout. Results of a single-line emission run at 1500 feet above and across the layout plus an aborted run are presented in figure 3. Spray deposits were recorded from both passes across a 6 mile area of the layout with a maximum deposit of 6 drops/cm occurring at a position 1.5 miles from the point of emis- sion (fig. 3). Drop sizes as small as 20 microns (u) were collected on the kromekote cards and glass plates, thus confirming the relative low-volatility of the #2/#4 oil formulation. Projected cumulative deposits from a series of successive spray swaths at 3000-foot intervals indicated that a potential toxic deposit of 20 spray droplets/cm could be deposited on the target site of spruce and fir foliage to kill early instar budworm larvae or adult moths (Kettela 1975). Plans were, therefore, established to undertake a night spray application of a pesticide against early instar larvae of the spruce budworm in 1975. In the spring of 1975, an experimental site of 35,000 acres was selected at Maniwaki, Quebec, a distance of 60 miles from the air field. Two sampling lines were established across the area to provide spray deposit and biological data on transects at 1/20 mile intervals. A cross section of the experi- DEPOSIT DENSITY (DROPS/CM2) 220 210 | I SAMPLE POSITIONS ACROSS no sample cards =U 200 190 180 170 BLOCK 11 | + 10 9 MILE POSITION Figure 3.--Transect of spray deposit (drops/cm*) collected downwind from a single high level pass (1500 ft) across the layout plus an aborted run are presented for lines A and B. 58 — LEVEL — SEA re) > ° a we =_ w C. road ELEVATION PROFILE -%,y,-road crosses same position twice € |-inight spray i] Tote ek 5 ~=-,e da teccieas, stecoat! tat Sy COR Gay, ‘0 NN eae 4, Profile ACROSS BLOCK; 1975 ; MANIWAKI. Figure 4.--Transect of experimental night spray block showing the profile of road elevation, day and night land profile and spray emission altitude. mental block is shown in figure 4 in terms of the road profile, the day land profile (based on the highest point of land 2 miles on either side of the road), and the night land profile (based on the highest point of land on the swath lanes within the block). Spray emission altitude was set at 700 feet above the highest point of forest terrain bounding a 5-mile zone surrounding the spray block. All profiles are shown with reference to elevation above sea level. The first attempt to spray the experi- mental block on June 4 under ideal weather conditons was aborted due to improper program- ming of the LTN-51 and a malfunction of the spray system on the Constellation 749 spray aircraft. Meteorological conditions during the next 4 days were extreme, with a low centered over the experimental site. Ground winds were 3 to 5 kilometers/hr. Since the spray aircraft was scheduled to depart on the morning of June 10, it was decided to spray the experimental site on June 9 taking into consideration the high altitude winds. Pibal readings were taken before, during, and after spraying to provide a profile of wind speed and direction across the experimental block 59 (fig. 5). Because of the high winds at emission altitude, the aircraft swath lanes were offset 6000 feet upwind of the block. oil/fenitrothion formulation containing 9.25 percent active ingredient, 25.82 percent arotex cosolvent, 24.98 percent #2 fuel oil and 40.15 percent # 4 fuel oil was applied at a dosage rate of 15.36 oz. (U.S.)/acre under starlight conditions using a Constellation 749 spray aircraft. The abovewing mounted boom and open nozzle system was operated at 14.5 PSI at a spray emission rate of 167 gallons (U.S.) per minute. All sampling units were collected 4 hours after spray completion to provide adequate time for deposition of the small drops within the target area. A tran- sect of the deposit data in terms of drops/cm and volume (oz/ac) is shown in figure 6. An Spray deposit data indicated that the Dmax spray droplets were carried 3 miles downwind before encountering the forest canopy- Maximum deposits over one drop/cm were recorded 2 miles downwind beyond the block boundary thus indicating the bulk of the spray was deposited well downwind of the block. Preliminary assessment of deposit data WIND SPEED (KNOTS/ HOUR) 10 15 20 June 9, 1975 drection 2135 hrs EDT 2145 hrs EDT — = Ww rl} uJ => uJ =} aq uJ ie) uJ > [e) faa) o~ 4 | ol ac © lJ ac WIND DIRECTION (AZIMUTH ) Figure 5.--Pibal recording of wind speed and direction during night spray trial, 1975. LINE- A DROPS/ CM2 FLIGHT LANES Mitts lula cle ai Alpi yas Te LY I\ I \ alr fy \ A Kivi eq \ \ FLUID OZ/AC __~_ / \aA \ \ / y Woes y v LINE -B FLIGHT LANES DROPS / CM2 each ee ae FLUID OZ./AC. _ __ BEUIDFOZ BERSAC N = oO or Lu ae op) (ats e) or Q 140 120 100 80 60 40 SAMPLE STATIONS 1720 OF A MILE APART Figure 6.--Transect of spray deposit density across the experimental layout. 60 indicated a droplet spectrum with a maximum drop size of over 300 microns (yu) and mass median diameter of approximately 130 uw. A value for the number median diameter of 105 yu was recorded for the spray within the block; this value, however, would not be representa- tive of the NMD for the entire application. Tentative results of the efficacy of this single application of 2 oz. fenitrothion indicated a larval mortality of spruce budworm approximating 25 percent-30 percent with 0 percent foliage protection. Results and Discusston The night spray trials were exploratory trials designed to provide some parameters of spray deposit in an attempt to extend the favorable spray weather period on spruce budworm projects. It was recognized that with night operations, spray emission altitudes would have to be raised to 700 feet above points of highest terrain in the spray blocks and in the turning area at the ends of the block. The majority of the trials were under- taken under adverse conditions of timing and meteorological conditions of high altitude winds. Meteorological conditions at ground level, however, were acceptable for spray application. In spite of adverse conditions a good deposit in terms of drop numbers and volume/cm was recorded that was not unlike those obtained during daylight hours. These results, therefore, could well represent the upper end of spray application parameters for high altitude spray emission. The success of spray operations under daylight conditions depends to a large extent on the meteorological conditions during the critical period of spray release and subse- quent deposition of the spray droplets onto the target site and beyond. Failure to consider the wind component, temperature gradient, and heat turbulence during spray release seriously jeopardizes the success of the operation, particularly when the target site is a small woodlot or spray block (Hurtig et al. 1953). This effect is well illustrated on a grand scale in figure 6 where the win- nowing effect of the crosswind component and spray altitude has resulted in the deposition of the largest (Dmax) drop sizes onto the target area in low numbers but in relatively high volume deposits per unit area. It,is well recognized that coverage (drops/cm ) and not volume (oz/ac) is responsible for maximum larval mortality (Hurtig et al. 1953). 61 It is therefore not surprising that, under these conditions, a mortality of only 25 percent was recorded against the late instars of the spruce budworm. One could expect a higher mortality, however, beyond the downwind boundary of the block where spray coverage is higher as shown by the increasing slope of the drop deposit graph. Of greater significance, however, is the potential use of incremental application techniques based on the findings of the uniformity of deposit across a wide area and deposition of droplets as small as 20 uw from spray emission heights of 1500 feet. Because of the nature of the night trials, scientific equipment to record droplets below 20 uw were unavailable. Mount et al- (1970), however, using bioassay techniques, reported excellent control of caged mosquitoes 3 to 5 miles downwind from spray emission altitudes of 1000 feet thus indicating the presence of droplets below 20 yu. These findings are in agreement with those of Yule and Cole (1969) who re- ported airborne spray particles miles downwind of operational sprays. Conelustons During the course of studies of night meteorology, it was found that wind velocity increased with increase in altitude and that the altitude of calm conditions above the forest canopy was influenced by the surround- ing terrain. Droplet deposition is directly affected by high altitude winds which in turn determine the degree of spray drift according to drop size. Results from these trials suggest that for operational use of high altitude (700 to 1500 feet) night spraying, a rapid system of meteorological data retrieval would have to be developed in order to utilize the optimum weather conditions for spray deposition. The present system is not sufficiently advanced for the control of spruce budworm larvae, but could be readily adapted for large-scale spraying of flying insects that are suscep- tible to uniform deposits of insecticides at low dosage rate on a cumulative deposit basis. Literature Cited Anonymous, 1966. They fly by night. Nov. 1966. Chemagro Courier, Boivin, G. and DeCamp, S.T. 1975. Operational aspects of inertial guid- ance in the 1975 Province of Quebec spruce budworm control program. Pro- ceedings of the Fifth International Agricultural Aviation Congress, National Agricultural Centre, Warwickshire, England, p- 350-363. Desaulniers, R. 1973. Aerial spraying against the spruce budworm in Quebec in 1973 and plans for 1974. Report presented at the Annual Forest Pest Control Forum, Department of Forestry, Environment Canada, Nov. 14, 1973, Ottawa. Desaulniers, R. 1975. Quebec protects its forests against the spruce budworm with four-engine air- craft. Fifth International Agricultural Aviation Congress, National Agricultural Centre, Warwickshire, England, p. 99. Fettes, J.J. 1962. Forest Aerial Spraying - Dosage con- cepts and avoidance of hazard to fish and wildlife. Proc. Fifth World Forestry Congress. Vol. 2, 924-929. Gunn, D.L., Lea, H-A.F., Botha, D.H., Calloway, S., Clackson, J.R-, Immelman, A., Taljaard, J.J., and Ward, J. 1948. Locust control by aircraft in Tangan- yika.- Published by the Anti-Locust Research Centre, London and the Locust Control and Research Section. Dept. of Agric., Pretoria, South Africa. 153 p. Himel, C.M. and Moore, A.D. 1967. Spruce budworm mortality as a func- tion of aerial spray droplet size. Science 156: 1250-1251. Himel, C.H. 1969. The optimum size for insecticide spray droplets. J. Econ. Entomol. 62 (4): 919-925. Hurtig, H., Fettes, J.J., Randall, A.P., and Hopewell, W.W. 1953. A field inves- tigation of the relationship between the amount of DDT spray deposited, the physical properties of the spray and its toxicity to larvae of the spruce bud- worm. Suffield Report No. 176, Part 6, 141 p. S.E.S. Defence Research Board, Canada. Ketalla, E.G., 1975. Aerial spraying against the spruce budworm in 1975 and a forecast of condi- tions in the Maritimes Region in 1976. Report presented at the Annual Forest Pest Control Forum, Dept. of Forestry, Environment Canada. Nov. 14, 1975, Ottawa. Mount, G.A., Adams, C.T., Loftgren, C. S., and Weidhaas, D.E., 1970. Altitude and drift of ultralow volume aerial sprays of technical malathion for mosquito control. Mosquito News. 30(4): 580-584. Paquet, G-, and Desaulniers, R. 1974. Aerial spraying against the spruce budworm in Quebec in 1974 and forecast for 1975. Report presented at the Annual Forest Pest Control Forum, Dept. of Forestry, Environment Canada, Nov. 19, 1974. Randall, A.P. 1969. Some aspects of aerial spray experi- mentation for forest insect control. Pro- ceedings of the Fourth International Agricultural Aviation Congress, Kingston, Canada, p- 308-315. Randall, A.P., and Zylstra, B. 1972. Evaluation of a modified Douglas DC- 7b aircraft and spray system for forest insect control. Information Report CC- X-23. Canadian Forestry Service, Envir- onment Canada, 50 p. Randall, A.P. 1974a. Insecticides, formulations, and aerial application technology for spruce budworm control. Proceedings of a symposium on the spruce budworm. pe. //- 90. Alexandria, Virginia. (Miscellaneous publication No. 1327, U.S. Dept. of Agric., For. Service.) Randall, A.P. 1974b. Changing concepts and technology for the control of the spruce budworm in Canadian forests following the intro- duction of ULV treatments. British Crop Protection Council. Monogram No. ll, p- 152-165. Randall, A.P. 1975. Field evaluation of the Litton LTN-51l inertial guidance system, multi-engine spray aircraft and incremental spraying for the dispersal of liquid insecticides for the control of the spruce budworn, Choristoneura fumiferana (Clem.). Proceedings of the Fifth International Agricultural Aviation Congress, National Agricultural Centre, Warwickshire, England, p- 336-349. Stewart, K.E. 1949. Application of DDT sprays by aircraft in Canada for the control of the spruce budworm. Jn Forest spraying and some eg a SS lel el ets ise at = Ee effects of DDT. Biol. Bull. #2. Div. of Res. Ont. Dept. Lands and Forests. Swaine, J.M. 1930. Airplane dusting operations for the control of defoliating insects. App. A. p- 72-87. In Rept. on Civil Aviation and Civil Gov’t. Air Opera- 63 tions for the year 1929. Nat. Defence, Ottawa. Can. Dept. Yule, W.N., and Cole, A.F.W. 1969. Measurement of insecticide drift forestry operations Proc. 4th Int. Agric. Aviat. Congr. Kingston, Ontario, Canada, p- 346-351. in Organization and Implementation of Comprehensive Research and Development Programs on the Gypsy Moth, Douglas-Fir Tussock Moth, and Southern Pine Beetle,in the United States {is > " S bor §,2 Rey Ge TRAteheES The Problem The Douglas-fir tussock moth (Orgyta pseudotsugata\McD.), the gypsy moth (Porthe- tria dispar|L.), and the southern pin beetle (Dendroctonus frontalis)Zimm.) are among the most destructive forest pests in the northwestern, northeastern, and southern United States, respectively. Major outbreaks of the tussock moth occur every 7 to 10 years and cause severe defoliation and tree mor- tality in Douglas-fir, grand fir, and white fir forests over extensive areas for periods of about 3 years. In 1974, over 2.3 million acres of forest were infested in Washington, Oregon, and Idaho. Over 400,000 acres were treated with DDT to avoid extensive tree mortality. Gypsy moth populations reached a peak in the northeast in 1973 when defoliation of hardwood forests occurred on about 1.75 million acres. Oaks, birch, and maple suf- fered significant growth loss and, under intense or repeated defoliation, large numbers of trees, principally white oaks, died. During the 3-year period 1972-74, an outbreak of the southern pine beetle in 10 Southeastern States caused losses exceeding 1.5 billion bd. ft. of pine sawtimber and 350 million cu. ft. of pulpwood. Less than 25 percent of the dead timber was salvaged. Dectston to Act In response to strong expressions of concern over continuing losses from land managers, owners, and other interested groups, Program Manager, USDA Southern Pine Beetle Program, Forest Research Center, Pineville, Louisiana. 64 Robert W. Long, then Assistant Secretary for Conservation, Research and Education, U.S. Department of Agriculture, called upon four Federal agencies -- Agricultural Research Ser- vice (ARS), Animal and Plant Health Inspection Service (APHIS), Cooperative State Research Service (CSRS), and Forest Service (FS) -- in August 1973 to pool their resources and develop a coordinated program to provide means for reducing losses from the three insects within a relatively short time period. Continuing damage from the three pests empha- sized the urgency of developing more effective control methods. Planning, Funding, and Implementing the Program Field and Washington Office agency staff members, working closely with the Department, organized research and development programs of 3- to 5-years duration for the tussock moth, the gypsy moth, and the south- ern pine beetle. Because of the complexity of the problems and the need to accomplish specific tasks within definite time periods, a new approach to program planning was used. This planning technique, called Adapted Convergence Technique for Agricultural Re- search (ACTAR), provided: 1) Clearly stated, well defined program objectives, 2) a logically ordered plan to reach those objectives, 3) flexibility which allowed for the inherent uncertainty of accomplishment associated with research and development, and 4) a chrono- logical scheduling of planned activities. ACTAR also provided bases for developing benefit/cost analyses, appraising the most effecive way to organize the overall programs, and monitoring progress. Each of the three programs was designed to accomplish two common objectives: 1) To more fully utilize available knowledge to reduce damage caused by the insects and 2) to develop improved or new methods and materials for preventing or suppressing damaging outbreaks. Although the three programs were to differ slightly in structure and in the manner in which they were to be handled, three major areas of investigation were to be pursued: 1) Measurement and prediction of pest popula- tion trends and their impacts on forest ecostystems; 2) development of methods for preventing or suppressing pest population outbreaks; and 3) development of integrated pest management strategies. The ultimate goal was to provide an integrated pest management system within which the damage caused by each insect would be prevented or reduced to tolerable levels. An array of tactics and strategies would be developed which would be compatible with forest management objectives and environmental conditions and could be used by forest land Managers and pest control specialists. Planning was completed by March 1974, re- viewed within the Department of Agriculture, and approved by the Office of Management and Budget. Testimony before the House and Senate Appro- priations Committees by Federal agency and key organizational representatives from the prob- lem areas resulted in a Budget Amendment which was signed by the President in August 1974. Research and development efforts in what was to be collectively referred to as the Combined Forest Pest R&D Program was activated in the Department of Agriculture in FY 1975 with an appropriation of $6 million. This, plus regular (base) funds available to the Federal agencies, brought total funding in that fiscal year to $9.3 million -- $2.0 million for Douglas-fir tussock moth, $4.8 million for gypsy moth, and $2.5 million dollars for southern pine beetle. Comparable funding was subsequently provided in FY 1976 and is anticipated in FY 1977. The Douglas- fir tussock moth and gypsy moth programs were scheduled for 3 years and the southern pine beetle program for 5 years. Though each program is expected to make major advances in providing means for coping with each of the pests, it is also anticipated that there will be a need for followup work to complete, report on, and implement the new technology. Organization and Administration of the Program The Combined Forest Pest R&D Program is administered from the Office of the Secretary, U.S. Department of Agriculture. This departs from the usual pattern in which Federal agencies conduct research or action programs. 65 A Staff Officer assigned to the Secretary’s Office provides staff support and coordinates activities among the three insect programs. Each program is directed by a Program Manager, also assigned to the Secretary’s Office. All four individuals report directly to a nine- member Program Board chaired by the Deputy Assistant Secretary for Conservation, Research, and Education. Each Program Manager is located in a headquarters office in the field and is assisted by a Research Coordinator and an Applications Coordinator. The Program Manager is responsible for selecting, scheduling, and approving funding for activities described in an Annual Plan of Work and Budget. Research and development proposals are solicited both on a competitive and non-competitive basis from a broad array of Federal, State, univer- sity, and private research and action organi- zations. Much of the work currently underway was initiated in the first year of each program, but new studies have or will be undertaken in subsequent years. Each proposal is reviewed by a peer group of specialists (¢d hoc Technical Review Panel) and recommendations made to the Program Manager regarding acceptability based on scientific merit and likelihood of contribut- ing to program goals. Publication and ac- tivity records of the proposed investigators are also reviewed to determine the potential productivity of these individuals. In other words, every effort is made to select investi- gators with good "track records." Each year, Program Management submits Annual Plans of Work and Budgets for review by the Program Board. The Board provides a national overview, guidance, and advice. During each review, accomplishments during the past year are assessed, progress toward objectives evaluated, and any needed program adjustments recommended to the Program Mana- gers. With the recommendation of the Board, the Annual Plans are approved by the Assistant Secretary for Conservation, Research, and Education, and serve as guiding documents for each insect progam. In the field, once projects are accepted, detailed study plans may be requested from the investigators if their original or revised proposals do not contain sufficient detail to serve as operating documents. Projects may be of one to several years duration, may be funded entirely out of one fiscal year’s money (entire agreements) or on a year-to-year basis (open-ended agreements), and may take the form of cooperative agree- ments, grants, or contracts. Funds are provided through CSRS and FS for the Douglas-— fir Tussock Moth and Southern Pine Beetle progams, through APHIS, ARS, CSRS, and FS for the Gypsy Moth Program. Allocations to these agencies or their field units are made based on recommendations from the Program Managers. Final negotiations on the details of work and funding are made with the performing organiza- tions -- federal forest experiment station, state forestry organization, state agricul- tural experiment station, university, indus- trial firm -- by Program management working closely with the funded Federal agencies and appropriate Federal or State representatives. Continuing control over approval and assess- ment of work and funding is maintained by each Program Manager. To establish and maintain a coordinated effort within each program, investigators in closely related studies (e.g., sampling, population dynamics, behavioral chemicals, site/stand attributes) have been grouped into working or subject area groups, each with a leader. Examples of these working groups are as follows: Douglas-fir Tussoek Moth Program Insect Dynamics Effect on Ecosystem Components Site, Tree, and Stand Dynamics Socio-Economic Evaluation Direct Control Methods Indirect Control Methods Survey Methods Integration Technology Transfer Southern Pine Beetle Program Impact Insect Sampling and Population Dynamics Mortality and Competition Factors Site, Stand, and Climatic Character- istics Host, Insect, and Symbiotic Factors Behavioral Chemicals Toxicants Forest Manipulative Practices Integration These working (or subject area) groups period- ically meet with Program management to review status of accomplishments and any work in progress or planned. Study plans and progress reports are exchanged between investigators in each group each year. In addition to coordination efforts within the programs, closer ties and coordina- tion have been developed through periodic meetings of the Program Managers or entire Program management staff with the Staff 66 Officer. In addition, interprogram working meetings are also held within various special- ty areas in which there is involvement in similar work in two or more programs (e.g., nucleopolyhedral viruses, Bacillus thurtngtensis chemical toxicants, behavioral chemicals, pesticide application techniques). A wide range of current or anticipated problems is discussed; accomplishments, experience, plans and/or ideas shared; outside technical input solicited; and reponsibilities for continuing work assigned where appropriate. Program management continuously monitors and evaluates progress and needs for redirec-— tion, additional or new effort, and termina- tion of unproductive or lower priority work. This is accomplished through review of semi- annual and/or annual progress reports, attend- ance at working group meetings, on-site visits to projects, solicitation of reviews, com- ments and criticisms from outside experts or special interest groups, annual program workshops, and other contacts with organiza- tional administrators and investigators. Proposed administrative actions relating to strengthening, reducing level of effort, and redirection or termination are reviewed with technical peer groups prior to implementation. Similarly, continued support for acceptable work, broad reviews of accomplishment, and needs for changes in program priorities or direction are discussed with an administra- tive advisory group. Although a large number of projects are funded for more than a year, all organiza- tional administrators and investigators have been informed that continued support for mul- tiple year studies or the initiation of followup studies is dependent upon accomplish- ment, availability of funds, and needs of the programs to maintain flexibility in making shifts in emphasis. Support for more funda- mental work varies between the programs based on the state of knowledge, length of program, and estimated probability of success but, in all cases, overall program emphasis is sched- uled to shift from a research to an applica- tions orientation over time. This tends to be both a source of frustration and a chal- lenge to many scientists in the life sciences who have not been accustomed to accomplishing results within a time limit nor considering ultimate practical use in their total plan- ning. Where appropriate, procedures have been established for data sharing and analysis in closely related lines of work. Consideration has also been given to the release of interim or final results through various outlets (e.g., in-house publications, referred jour- nals). These actions are intended to focus attention on the selection of critical para- meters for measurement, the need for standard- ization of data capture and analysis pro- cedures where appropriate, the early and continuing analysis of results and refinement of study approaches, the timely release of information, and protection of author rights. As might be expected in any large-scale program, adjustments in level of support and emphasis have changed with time. Initially, full funding was not allowed for each of the programs and certain tasks were specifically deleted from funding consideration. This necessitated some internal adjustments in level of support to accommodate the activities to be funded in each program. Later on, as research and development results were ob- tained, along with more experience in managing large, interdisciplinary programs, the 3 to 5 year activity schedules were revised one or more times to accommodate changes in R&D opportunities, reflect progress in developing new knowledge, better utilize available expertise, and improve the organizational structure of the respective programs. Simi- larly, levels of pest activity have varied in the respective regions necessitating shifts to new areas for some tests while, at the same time, permitting initiation of studies of endemic populations in other areas. The Douglas-fir tussock moth and gypsy moth programs are scheduled to be completed in FY 1978. Plans are being made to conduct final field and pilot tests, summarize program accomplishments in the form of updated "state of the art" compendiums and user guides, and to identify resources -- manpower, dollars, facilities -- no longer needed in the programs and tasks that remain to be completed by the Federal agencies and others. Similar plans will be developed in approximately 2 years for the southern pine beetle program. Finally, it would seem appropriate to make some early observations on the apparent effectiveness of conducting programs in the manner I have described to solve major forest pest problems. First, I would note that the approach permits a concerted effort with sufficient resources to accomplish significant results. No single organization could expect to accomplish similar results with more limited resources except over a very long time period and, possibly, at greater total cost. Secondly, the use of a competitive approach has generally fostered a healthy atmosphere and led to a much more concentrated R&D effort by individuals and groups both within and between the Federal, State, and private sectors. Thirdly, investigators and many of their administrators have shown keen interest and enthusiasm in participating in such a 67 program to meet the needs of a larger user public. Individuals are able to grow profes- sionally, to produce results that are accept- able both to their parent organizations and in the broader scientific community and to mingle, compete, and interact with other disciplines. Finally, many investigators have come to a fuller realization of the contribu- tion that their individual work can make in achieving a larger goal, the development of an integrated pest management solution to a major forest pest problem. Summary I have attempted in these few minutes to describe three major forest pest problems which have led to a U.S. Department of Agri- culture action program. Four of the Depart- ment’s agencies are working closely with state agricultural experiment stations, univer- sities, colleges, state forestry organiza- tions, and industry in an accelerated research and development effort. The Program is administered from the Office of the Secretary, USDA, and is directed in the field by program managers for each of the pest problems -- Douglas-fir tussock moth, gypsy moth, and southern pine beetle. Common objectives and goals have been established and the three programs have been underway since FY 1975. Studies have been undertaken which will ultimately improve pest control specialist and land manager abilities to understand and predict pest population trends and their impacts. Site, tree, stand, and climatic characteristics influencing outbreak incidence and severity are under intensive investigation. Field and laboratory work is underway which will provide new toxicants, behavioral chemicals, and forest manipulative practices as well as increased understanding of mortality and competition factors, and insect/forest interactions regulating pest populations. Criteria and methods are being developed for implementing and evaluating pest management strategies. A large number of investigators are working together within and across organiza- tional and disciplinary lines in each of the programs to accomplish the goal of providing new or improved control methods and materials. Rapid progress is being made. The success of this approach to problem solving is expected to affect the future and direction of urgently needed research and development efforts on other forest problems. bce Discontinuous Stability in a Sawfly,Life System and Its Relevance to Pest Management Strategies ~ ee + J. M.\ McLeod! Evidence is accumulating that many ecological systems are discontinuously stable, i.e., they possess more than one domain of attraction (Holling 1973). Southwood and Comins (1976) cite evidence for multiple equilibria phenomena in a number of insect systems, including eucalyptus aphids, sugar cane borers, cereal aphids, and a spruce sawfly. Holling et al. (1977) have recently developed a powerful simulation model for Management of the spruce budworm/balsam fir (Choristoneura fumiferana Clem./ Abies balsamea Mill.) ecosystem, at the base of which are multiple equilibria concepts, outlined by Morris (1963) in an exquisitely detailed population study of that insect. The fish- eries literature is full of examples of stocks which have collapsed to extremely low levels following intensive exploitation, and which have remained at low levels for extended time periods, even when fishing pressures were released (Beeton 1969). Noy-Meir (1975), using the graphical techniques of Rosenzweig and MacArthur (1963), discussed the theoret- ical implication of discontinuously stable grazing systems as related to plant charac- teristics, to herbivore characteristics, and, most important, to management practices. He presents evidence that some real pasture and grazing systems are discontinuously stable (Perry 1968; Morley 1966). The consequences of discontinuous sta- bility are relatively sudden and unexpected shifts in numbers or productivity when systems are stressed, i-e., the collapse of a valuable resource, or a major outbreak of a pest. Although such events are usually viewed as catastrophic, the reverse may be true. Using an example from an insect/forest system, I here show how a pest manager may reap unex- pected benefits in managing a system which possesses multiple stability domains. Institute of Resource Ecology, University of British Columbia, 2204 Main Mall, Vancouver, B. C., Canada V6TI1W5. 68 System Desertption The insect is a sawfly, Neodtiprton swainet Midd., a defoliating pest of jack. pine, Pinus bankstana |Lamb. (fig. 1). The system has been partly described (McLeod 1968, 1970, 1972, 1973, 1975a), and a more complete description and a detailed process simulation model are in preparation. This sawfly is a colonial defoliator, a discrete generation, univoltine insect, which feeds from early August through September on the previous year's growth of the host plant. It leaves the current foliage intact, except at high population levels if the supply of “old" foliage is exhausted. In September and early October, larvae drop from the trees and spin cocoons in the soil, where they pass the winter. The following spring, in early June, the insect transforms to the pupa while still in the cocoon, and the adults emerge in early July. The female deposits her total comple- ment of about 65 eggs on a single current shoot of jack pine, and the eggs require 1 month to hatch. A number of natural enemies attack the sawfly. The most important of these are shrews, Sorex ctnereus, which feed on the sawfly in the cocoon stage throughout the fall and winter (McLeod 1966); birds which feed on large larvae in late summer and fall, and on emerging adults in July (McLeod 1974); and a complex of univoltine, synchronous, larval parasitoids which attack young larvae in early August and emerge from the cocoon the follow- ing summer in time to attack sawfly larvae of the succeeding generation (McLeod 1975a). Larvae are strongly affected by weather (fig. 1). Development of larvae, especially in northern latitudes, may be so retarded by cold temperatures that they are unable to complete their feeding in the fall (Tripp 1965). Optimum conditions for sawfly develop- ment and survival are found in a narrow band between latitude 46° and 48° in the southern Figure 1.--Major components of the Swaine jack pine sawfly life system. between the forest and the sawfly and between the sawfly and its natural enemies. There are feedbacks Principal driving variables are fire (for the forest) and weather (for the sawfly). extremity of the host plant’s distribution (McLeod 1970). Jack pine is the only host of Weodiprion swainet. Forests of jack pine are typically pure and even aged, and originate from fires (fig. 2). Cones are retained on the tree and will not open to release seeds unless exposed to temperatures in excess of 52° c. In its early years jack pine is one of the fastest growing of Canadian conifers, but the growth rate declines appreciably after about 35 years, and stagnation follows. If undis- turbed, jack pine forests would usually be slowly succeeded by black spruce; but histori- cally, repeated fires have assured the main- tenance of jack pine in climax on poor, sandy, outwash soils where this species grows best (Rudolph 1958; Rowe 1971). Insect survey literature gives a clear impression of strong discontinuities in space and time in the distribution of numbers of sawflies (McLeod 1970). Usually, this insect is a rare component of the jack pine forest. In certain favoured sites, however, repeated outbreaks have occurred at intervals of about 8 to 10 years, especially as trees approach maturity (fig. 2). During outbreak peaks, 69 severe and general defoliation and some top killing may occur, and in the troughs, some defoliation may be noticeable in the upper crowns of dominant trees only. Rarer still are truly catastrophic outbreaks, where whole forests may be killed over thousands of hectares. In fact, the qualitative behavior of sawfly populations is analogous to the qualitative changes observed by range managers cited by Noy-Meir (1975), which he attributes to discontinuities in the range’s stability domains. Detailed Historical Data and Perturbations Further evidence for the existence of multiple stability domains in the sawfly’s life system are obtained from precise histor- ical records of population trends in Quebec between 1956 and 1974 (McLeod 1972; 1975a). The data are presented as a number of sawfly eggs per square metre of ground surface and were obtained annually from each of seven 4-hectare permanent plots, as the product of the following: 1. the number of sawfly egg clusters per tree from 75 to 100 randomly selected living whole trees per sample, 2. the number of eggs per cluster obtained from approximately 50 randomly selected egg clusters from (1), 3. the number of living jack pine trees per chain-square (20 x 20 metres) block mea- sured annually in each of 25 blocks per plot. “a ~~ 4 6 idispersal <: SSO ’ ‘ ‘ ‘ ‘ ' ’ , , ’ , , , oO iS BSS = o os oO co) — oO © oO oO ie) £ g oe D fo) (oP) > ras Ss wo oO time(years) Figure 2.--Time bounding of the Swaine jack pine sawfly life system. Jack pine forests are pure and even aged and usually regener- ate after fire. Sawfly populations are usually low but may suddenly erupt and persist at higher levels for years, espe- cially as forests reach maturity. Periods between outbreaks are frequent (about 10 years). The means and standard errors of each sample are shown (table 1), as well as the combined errors (fig. 3). Calculated by propagation of error, the combined error estimate is: See [A (Si. Si¢)] + [B (Sey Sz0)] 2. 7 LG (S54 ° Szp)] where A = number of sawfly egg clusters per tree Soa = standard error of A B = number of sawfly eggs per cluster 70 Plot Sep = standard error of B C = number of living jack pine trees per unit area (404.68 sq- metres) Sac = standard error of C A brief history of the population trends and events in each of the plots follows (fig. 4): Plots 1 and 2 -- Aged 51 and 70 years respec- tively in 1974, these two areas were in outbreak condition in 1964, and in 1965 were treated with the insecticide Phos- phamidon at 4 oz. per acre (0.28 kg/ha). Populations of sawflies dropped to extremely low levels in 1966, the year following the one-shot application of insecticide, and have remained at low levels ever since. 3 -- Seemingly out of phase with the majority of Neodtprion swainet infesta- tions in the mid- to late 1960s (McLeod 1970), the infestation in this 60-year- old stand reached extremely high levels and collapsed in 1965 owing to starvation and a polyhedral virus. Tree mortality was cumulative following collapse, and by 1966, virtually all the trees over 2000 hectares had died. This study was terminated in 1967 following a salvage cut. Plots 4 and 6 -- Sawfly populations were maintained at moderately high levels in these two 5l-year-old stands (in plot 4 from 1962 to 1974 and in plot 6 from 1967 to 1974). At the peak of the infestation in plot 4 in 1968, severe defoliation and some tree mortality (<2 percent) did occur, but the trees rapidly recovered as sawfly populations dropped in subsequent years. In plot 6, although absolute numbers of insects were only moderate throughout the study, the impact on the forest was greater than in plot 4 owing to the small number of trees per unit area (table 1). Severe defoliation persisted from 1967 to 1973 with some top killing but insignificant tree mortality. In 1974, insecticide contamination during an operation against the spruce budworm (McLeod 1975b) caused an abrupt collapse of this sawfly infestation. Plot 5 -- Two sawfly outbreaks occurred in this 5l-year-old jack pine stand between 1951 and 1974. The first collapsed dramatically in 1960 following high mortality through shrew predation and larval parasitism. Although defoliation at the outbreak’s peak was severe, no top killing occurred and the trees recovered quickly. The second outbreak Table 1.--Basic statistics for calculating W. swatnet eggs per square metre Living jack pine Egg clusters Eggs pe trees per unit area, Plot Year per tree cluster (404.68 metres sq.) 01 1964 005.540 + 01.290 -- 163.800 Ol 1965 017.293 + 02.466 053.717 + 01.629 157.300 01 1966 000.013 + 00.013 -- 152.900 01 1967 000.000 + 00.000 -- 132.840 + 06.624 01 1968 000.060 + 00.034 -- 126.000 01 1969 000.085 + 00.051 -- 119.560 + 06.453 01 1970 000.000 + 00.000 -- 113.240 + 06.343 01 1971 000.000 + 00.000 -- 108.240 + 05.370 01 1972 000.900 + 00.000 -- 105.080 + 05.730 01 1973 000.000 + 00.000 -- 102.520 + 05.834 01 1974 000.000 + 00.000 -- 102.400 02 1964 018.580 + 03.369 -- 065.100 02 1965 051.311 + 06.770 054.771 + 01.844 063.200 02 1966 000.053 + 00.032 --— 061.840 + 03.913 02 1967 000.102 + 00.044 -- 054.880 + 03.366 02 1968 000.160 + 00.066 -- 051.900 02 1969 000.640 + 00.136 061.562 + 03.125 049.000 + 03.025 02 1970 000.300 + 00.091 --— 045.320 + 02.885 02 1971 000.400 + 00.110 -- 045.440 + 03.094 02 1972 000.167 + 00.069 -- 043.640 + 02.867 02 1973 000.060 + 00.034 -- 041.280 + 02.791 03 1962 009.533 + 01.841 068.383 + 02.228 119.000 03 1963 018.219 + 02.826 081.180 + 02.151 114.400 + 05.832 03 1964 106.980 + 12.973 051.038 + 01.852 099.840 + 0.576 03 1965 009.240 + 01.754 036.220 + 01.304 071.600 + 04.800 03 1966 000.000 + 00.000 --— 019.840 + 04.760 03 1967 000.000 + 00.000 -- 006.357 + 03.385 04 1962 003.350 + 00.676 -- 117.400 04 1963 001.693 + 00.348 -- 113.640 + 08.805 04 1964 006.253 + 01.053 063.260 + 01.571 2 ost00) 04 1965 0185373) 02.959 066.069 + 01.683 109.600 04 1966 020.274 + 02.655 060.603 + 00.834 108.280 + 08.130 04 1967 013.160 + 01.390 070.320 + 02.278 102.240 + 07.782 04 1968 039.210 + 05.092 071.200 + 01.891 092.520 + 07.243 04 1969 008.733 + 01.055 063.940 + 02.283 094.240 + 07.605 04 1970 004.387 + 00.626 063.060 + 02.163 085.800 + 06.661 04 1971 006.160 + 00.825 061.877 + 02.697 077.600 + 06.220 04 1972 009.540 + 01.290 057.456 + 02.141 072.360 + 05.761 04 1973 008.240 + 01.155 064.480 + 01.667 070.640 + 06.074 04 1974 004.300 + 00.838 069.559 + 03.341 069.400 05 1961 001.449 + 00.371 065.147 + 01.967 100.000 05 1962 004.828 + 00.874 066.060 + 02.451 095.200 05 1963 004.547 + 00.670 063.738 + 01.284 091.800 + 06.133 05 1964 009.905 + 01.658 066.309 + 00.662 089.400 05 1965 034.117 + 03.736 068.053 + 02.106 084 . 500 05 1966 002.053 + 00.291 060.729 + 01.023 082.080 + 05.205 05 1967 014.293 + 02.546 077.140 + 02.350 079.400 — 05 1968 017.693 + 02.555 070.564 + 01.863 076.640 + 04.957 05 1969 016.547 + 01.799 060.700 + 02.503 074.720 + 04.805 05 1970 005.293 + 00.688 063.380 + 02.496 072.120 + 04.673 05 1971 001.080 + 00.192 050.341 + 02.219 070.120 + 04.538 05 1972 001.140 + 00.258 059.340 + 02.203 068.560 + 04.369 05 1973 000.120 + 00.046 049.913 + 01.423 067.240 + 04.346 05 1974 000.380 + 00.106 081.720 + 02.098 066.600 — 06 1967 O22e 7713 025370 063.800 + 02.715 045.760 + 03.353 06 1968 017.270 + 02.034 058.860 + 01.488 040.960 + 03.023 06 1969 021.027 + 02.749 066.354 + 02.026 038.840 + 03.848 06 1970 013.680 + 01.397 064.437 + 01.902 039.280 + 05.709 71 Table 1 (continued) Egg clusteys Plot Year per tree 06 1972 027.080 + 04.841 06 1973 025.760 + 04.198 06 1974 013.760 + 01.907 07 1970 000.060 + 00.034 07 1971 000.000 + 00.000 07 1972 000.000 + 00.000 07 1973 000.000 + 00.000 07 1974 000.000 + 00.000 ; With standard error. No sample of less than 25 egg clusters 3 the series (63.76) is used. Samples with no variance estimates are index curves for jack pine. reached a peak in 1965, but declined in 1966 owing to accidental drift of insec- ticide from an operation in a neighbour- ing stand. The populations rebounded to pre-spray levels in 1967, but events were complicated in the early 1970s by a massive outbreak of a pitch nodule maker, Petrova spp. (McLeod and Tostowaryk 1971) which caused extensive top killing and deformation in the tree crowns. Because Petrova is a sap-feeder, it was able to compete successfully with the sawfly for food. Survival of sawfly larvae declined dramatically in the Petrova outbreak years, and by 1974 sawfly populations were at a very low level. Plot 7 — Monitoring in this 5l-year-old stand was carried out from 1970 to 1974. Only in 1 year of the 5 were sawflies recorded. Casual observations in this stand from 1964 to 1971 would also indicate that sawfly populations were very low during that period. The sharp discontinuities in the distri- bution of sawfly numbers may best be seen by replotting the data of figure 4 a a recruit- ment factor curve, i-e., 10819 /N.. ) vse logyg (Ny ) where N. is the a0 coe sawfly eggs per square metre (fig. 5). The outlying points to the right at (a) trace the population of plot 3 which com- pletely destroyed its food supply (table 1) and collapsed from starvation. Below this is a collection of points with a negative slope crossing the replacement line at about 10816 2. An additional cluster of points Living jack pine Eggs pes trees per unit area cluster (404.68 metres sq.) 058.687 + 02.250 030.240 + 02.696 063.592 + 01.965 029.480 + 03.012 087.420 + 02.350 027.000 — --— 092.480 + 04.177 -- 090.600 — -- 086.000 + 03.498 -- 090.600 + 03.878 -- O8E3500m included. For missing entries, the mean for extrapolations from Plonski's (1960) site z .-¢ Ww = we fo} ° > 2) .- 4 ui n LOG,, SAWFLY EGGS/SQ. METRE Figure 3.--Standard error of mean as function of mean for N. swainet eggs/sq- metre as calculated by propagation of error (see text). Only complete data sets of table l are used. is found two orders of magnitude lower, cross- ing the replacement line just above log N = 0. Between these two clusters are Cheese trajectories resulting from system perturba- tions. At (b) is the dramatic one-generation shift from the high to low density caused by insecticide application (plots 1 and 2). At (c) is the more gradual decline to the lower density resulting from competition with an associated insect (plot 5). Finally, the clockwise circularity in time of the points in each cluster would seem to suggest oscilla- tions resulting from lag effects. 72 presume cause, measure the effects in a +=PLOT 1 A=PLOT3 W=PLOTS =PLOT7 simulation model, and compare the results with heh pee met OSeuri the observed world. These presumed causes are: 1. competition among groups of animals at different trophic levels for a common limiting resource; 2. a "Type III" (Holling 1965) density response of a group of predators which saturates at an extremely low sawfly density: in the sawfly life system, these are insectivorous birds, feeding on larvae and adults (fig. 1); log, (eggs/sq.metre +1) 3. a second group of predators exhibiting a Type III response saturating at high prey Malte’ ed ae densities, just below the level at which Figure 4.--Trends in Neodiprion swatnet egg food begins to be limiting to the sawfly: populations in seven localities in Quebec, these are cinereus shrews (fig. 1); 1956 to 1974. 4. an upper equilibrium set by foliage: this equilibrium is stable on the insect plane but conditionally unstable on the forest plane (Jones 1975, 1977; Holling et al. 1978), if the replacement rate of foliage is not fast enough to prevent widespread killing of trees as the sawfly reaches this level. I know of no statistical averaging techniques which would permit the replication either of the discontinuities, or of the trajectories following perturbation of the observed populations. Instead, I shall (Ne aa/ Ne) RECRUITMENT l 1) LOG,, | L Oo 1 LOG 1 2 9 SAWFLY EGGS/ SQ. METRE (N,] Figure 5.--Data of fig. 4 expressed as recruitment; log (N /N,.) vs. log Nee Arrows show direction in time. t+1 73 5. system "clocks" or time constants caused by lags resulting from the effects of competition: one of these is the re- placement of foliage parts following defoliation at very high densities, but a second more powerful lag is caused by the numerical response of the synchronous univoltine parasitoids (McLeod 1975a); 6- random inputs from a powerful driving variable such as weather. With this simple set of assumptions, most of the major components in the sawfly life system (fig. 1) may be modeled. Not included, however, are forest growth, forest replacement (driven by fire), or insect dispersal. The forest model is not included because in this paper we are concerned only with events after the forests mature- Also, the very short time frame of the outbreak periods (ca. 10 years) would not suggest their dependence upon forest replacement. Similarly, sawflies, unlike spruce budworm (Morris 1963), have only one dispersal stage (adults) and they are rela- tively poor fliers. Dispersal would be important only in the event of forest replace- ment» = SAWFLY CYCLE — —— = FOLIAGE CYCLE FECUNDITY AND LARVAL SURVIVAL PROPORTIONAL TO FOLIAGE RATION LARVAE REMOVE FOLIAGE MK =X-aX1 FALL FOLIAGE LOSS =X-4X2 NEW FOLIAGE ADDED M=X+ax3 Figure 6.--Schema of Weodiprton swatnet simulation model. The Model The schema (fig. 6) follows the life history and mortality schedules of the sawfly outlined in McLeod (1975a). The only dynamic variables are the sawfly, its larval parasitoids, and foliage replacement. Functional responses are of the Michaelis-Menten type: a eNom a“4no™ where NA = number of prey attacked 8 = the saturation level of the response NO = the number of prey available for attack a@= the prey density at which 1/2 satur- ation occurs n = an exponent. When n=l, the response is Type II (Holling 1965). When n>l, the response is sigmoid (Type TT) 6 SAW FLIES LAY EGGS PARASITOIDS ATTACK LARVAE BIRDS ATTACK LARVAE SHREWS ATTACK COCOONS SAWFLY PARASITOID ADULTS ADULTS EMERGE EMERGE BIRDS ATTACK ADULTS The forms of the processes are shown in graphs, at each step: No, number of prey available; P_, percent mortality; Uc, units of foliage consumed; F+S, sawfly fecundity and larval survival; Ra, foliage ration. S = a fixed survival rate, and the circled R = applied to a fixed survival. The circled a log-normally distributed random variable (For details see Appendix.) Competition is modeled with a negative binomial function (Griffiths and Holling 1969), where: NHA = NO: [1-(14NA/NO-K)]* where. NHA the number of prey attacked K = the dispersion coefficient, where K—> o the function becomes the Poisson, eo NA/NO- Weather is simulated by a log-normally distributed random variate acting on young larvae (fig. 6). The maximum number of foliage units available to larvae is four (one for new foliage and three for old). Larvae respond functionally to the available foliage and the response is coupled with a competition equa- tion. Larvae feed first on units 1 to 3 and will not feed on unit 4 (new foliage) unless the ration of old foliage is exhausted. Survival of larvae and fecundity of adults is directly proportional to the foliage ration. The latter is expressed as the total number of units per larva consumed versus the total required (a constant). Details of the model are presented in Appendix I. The expressions listed, and the three initializations (sawfly egg density, parasitoid adult density, and foliage units) are all that are required to replicate the outputs in this report. The derivations of the parameters for the functional and numeri- cal responses and the competition equations will be given in a future report. Valtdatton The usefulness of a process model such as this one is not to make fine-scale year-to- year predictions of changes in insect numbers, but rather to determine what qualitative changes in populations would occur as a result of perturbations. Therefore, the appropriate statistical tests are those which compare the means and variances of the data (table 1), and the model (Appendix I) before and after perturbations (cf. fig. 7). Since changes in numbers are geometric, 10819 trans forma- tions have been used (table 2). Undisturbed Populations (fig. 7A) The data from plots 4 and 6 are used for this validation. The model’s output, for 30 75 generations, shows approximately the same mean density, amplitude, and periodicity as the observations, and this is also supported by the statistical tests which show that the means and variances of the data and the model’s output for undisturbed populations (table 2) are essentially similar. The clockwise circling of the points around the replacement line (log recruit- ment = 0) is set by three events. The func- tional and numerical response of shrews fixes the replacement crossover at about log N, = 2, and the 10-generation cycle is determined by the lag in the response of the larval para- sitoids. Further confirmation of this is seen in plotting parasitoid progeny density and sawfly density in phase (fig. 8). Some damp- ing is also provided by the foliage function peak populations, but by the time outbreaks have reached this level, a crash is almost inevitable in any event because of the lag in the parasitoids’s response. A Natural Perturbation (fig. 7B) Ineplotyos., Enon 19/70 (to l9725. small larval survival averaged 0.2 which is less than one-half the normal rate (Appendix I). This caused a steady decline in numbers, and even when small larval survival returned to high levels in 1972, populations continued to drop and were trapped in the lower domain of attraction. This was simulated in the model by setting the larval survival to 0.15 for three consecutive years, and this provided the same effect as the observations (table 2). Even when larval survival was set back to normal after 3 years, the populations were suf fi- ciently close to the lower stability domain to fall into its zone of attraction. Insecticides Here also, the model’s output is essen- tially similar to observations (table 2). Given a very strong negative perturbation, populations are shifted in only one generation to the lower stability domain and remain locked there for several generations. I conclude that the model faithfully mimics much of the richness of qualitative population behaviours in the real world system including the periodicity and amplitude of oscillations and the shifts between stability domains made under a variety of perturbations. Applteation to Pest Management but may, in fact, sometimes constitute the best management option. Chemical insecticides have come under Clearly, the success of insecticide heavy criticism in recent years because of control against Weodiprion swainet was as much potentially harmful environmental impacts. dependent on the existence of a lower stabil- However, the results of this study would ity domain as on susceptibility to the poison indicate that chemicals may not just be a itself. necessary evil to be used when all else fails, DATA MODEL ae B b T= Pa (0) 2: ——— Se ( = + ~ Zz ‘a Ben <-2 Cc oO = =a =) — O -4 o _ S oO) {e) fo) a a -1 0 1 2 3. A ie} 1 2 3 10G,, sawfly eggs/sq. metre (N+) Figure 7.--Comparison of observed recruitment for Neodiprion swatnet (left) with simula- tion model (right). Model runs are for 30 years. Sawfly egg populations are initial- ized at 55/sq. metre; parasitoid adults at 0.03/sq. metre; trees have full complement of foliage at initialization; standard deviation of log-normally distributed random driving variable = 0.17. Undisturbed populations at upper equilibrium (plots 4 and 6) (A)- A shift from upper to lower equilibrium under natural stress (competition for food by an associated insect) (B); in the model, for years 21 to 23 inclusive, small larval survival was set at 0.15 ("normal" = 0.45. Sprayed plots (1 and 2) (C); in the model, "spray" was applied at year 18, and simulated with small larval survival rate eof 0.005. Arrows show direction in time. 76 Table 2.--Statistical comparison of output of the simulation model (Appendix), expressed as 10916 sawfly eggs/sq.m., with the data of table l Differences Model _, if Data, between model N Meesd N xsd and data 1. Undisturbed populations 30 =62.0128 + 0.4067 21 2.0654 + 0.3008 N.S. (fig. 7A) 2. Natural disturbance ZL Balk, 2 OsSs/27 9 2.0318 + 0.4495 N.S. prior to perturbation (fig. 4B) 3. Natural disturbance 9 0.8306 + 0.6133 bee 029033) + 0K6422 N.S. following perturbation (Giake 7B) 4. Insecticide application O25 498 203786 4 2.4083 + 0.2298 N.S. pre-spray (fig. 7C) 5. Insecticide application LI OSG Z9F+N0R2509 ll 0.0608 + 0.3618 N.S. post-spray (fig. 7C) Differences between 1 and 2 N.S N.S 1 and 4 N.S N.S 2 and 4 N.S N.S 1 and 3} kkk kkk 2 and 3 aK tek 1 and 5 wR aK 4 and 5 RK kK : All variances are homogeneous. Figure 9 shows from the model what might have happened had there been no lower equilibrium. Numbers drop to low levels in the years following spraying, but recovery is rapid, and the risk that sawfly populations would overshoot food resources in the release phase is high. Some suggestion of this is also given by the observations in plot 5 following insecticide drift in that area in 1965 (fig. 7B). The populations dropped to low levels in 1966 [cf. the point (a)] but not sufficiently to be attracted to the lower stability domain. The result was a dramatic recovery of populations to pre-spray levels in the year following. However, the existence of the lower stability domain has furnished an unexpected benefit to the manager, since only one application of insecticide is required to maintain pest populations well below economic levels, for several generations following application. The benefits reaped are low environmental impact and high rate of return on the initial investment. Are discontinuous stabilities the rule rather than the exception in ecological sys- 77 tems? If so, it might be useful to reconsider the trade-offs between single-shot heavy applications of insecticides and the temporary ecological disruption they cause, versus lighter application rates and the potential long-term consequences of repeated applica- tions. However, if the distance between stabil- ity domains is great, for example as in the eastern spruce budworm (Holling et al. 1977), then when the insect is in outbreak condi- tions, no reasonable amount of insecticide would suffice to lower the populations to endemic levels. The best that can be done during an outbreak is to try to save foliage until outbreaks subside, and this usually means repeated applications with all the attendant risks. A better control policy may be to apply treatments when the pests are still trapped in a lower stability region, but in "threat" stage, potentially reaping the same benefits as the sawfly example cited in this report. Indeed, this new way of looking at insect ecosystems points to a better way of “ounce of prevention" pest management. Ww a rs Ww = fe] o > 2 WwW Oo °o c a a) ° E wn a 4 - cs ° ° 2 NO. OF SAWFLY LARVAE/ SQ. METRE Figure 8.--Phase plots showing response of parasitoids to changes in numbers of Weodt- prton swatnet large larvae per sq. metre. Data (plots 4, 5, and 6) (A)- Model (oscil- lating at upper equilibrium) (B)-. These parasitoids are responsible for the cycling around the stability domains seen in figure fs z2 Ww = = > in Oo Ww rea ° (0) 1 2 LOG), SAWFLY EGGS/SQ. METRE Figure 9.--Sawfly recruitment in model under same conditions as fig. 7C, except that bird predation functions are removed. With no lower stability domain, populations are depressed only one year (b) following "spray" application (a). A rapid return to the upper equilibrium occurs over a few years, and the system overshoots at (c) before returning to normal. Severe defolia- tion and some tree mortality would result. Arrows show direction. Appendix--Details of Neodiprion swainet simulation model 7. Arrows show direction in time. Initialization: sawfly eggs/sq. metre parasitoid adults/sq- metre foliage standard deviation of log-normally distributed random variate Start Fixed survival to larvae and log-normal random function Foliage function: UPERL is the mean number of foliage units consumed per larva Set up the first three units to be defoliated 78 if 55 (EGG) 0.03 (PARAD) 4 units { YFOL(4) = 1} 0.17 (RAND), (variate initialized at SEED = 333) EGG = 55 LARV = EGG* 0.45+RAND UPERL = 0.0041 ZFOL = YFOL(1)+YFOL(2)+YFOL(3) Appendix (Continued) eo” (UPERL*LARV/ZFOL) Larvae compete for units of foliage COMP = l and defoliation assigned equally to OLDF = ZFOL*COMP each of three units YFOL(1) = YFOL(1)- {OLDF*YFOL(1)/ZFOL) YFOL(2) = ee ea eons) EE YFOL(3) = YFOL(3)- {OLDF*YFOL (3) /ZFOL) Remaining requirements are competed UNITS = OLDF/LARV for on current foliage = PERL-UNITS) *LARV/YFOL(4 COMP = l-e (Wu aS) PPAR) NEWF = YFOL(4)*COMP YFOL(4) = YFOL(4)-NEWF Calculate the foliage ration RATN = (OLDF/LARV+NEWF/LARV) /UPERL available to each larva Drop the first year of foliage YFOL(1) = YFOL (2) (normal fall loss) and add one unit YFOL(2) = YFOL (3) of foliage the following spring YFOL(3) = YFOL (4) YFOL(4) = 1 Disc equation for functional response NATT = PARAD*. 235*20*LARV/ {1+(. 235%. 041 *LARV)} of larval parasitoids Competition function for parasitism PCOMP = {1*(NATT/LARV*1.68)) 1° 8 : : Deo) 2.5 ASS) Birds prey on larvae with Type III NATT = 20*.098(LARV) G9) +(LARV) functional response Bird attacks on larvae cycled through BLCOMP = 1+(NATT/LARV*1.5)7 2°? a competition function Survival to cocoon stage following COC = LARV*BLCOMP*. 38*RATN removal by birds, a fixed mortality, and a reduction proportional to foliage ration Shrew functional response (Type III) NATT = 115*300*coc7/557+c0C* Shrew numerical response (Type III) NSHR = (27*c0c7/757*c0c~)+9 Combined shrew response and conversion NATT = (NATT*NSHR) /10° from hectares to sq- metres Survival from shrew predation SHCOMP = 1-(NATT/COC) Survival to adults following shrew ADULT = COC*SHCOMP*PCOMP*. 33 predation, emergence of parasitoids, and a fixed survival Functional response of birds to sawfly NATT = 20*.018* (ADULT) =" 7°/4(.3)7* >>+(apuLT) >" >> adults Competition equation for birds BACOMP = {14 (NATT/ADULT#1.5)}7 1°? Removal by birds, reduction for sex EGG = ADULT*BACOMP*. 66%. 72*RATN*1 20 ratio, proportion of fecundity realized, reduction for foliage ration and multiplied by potential fecundity to yield eggs of following generation 79 Appendix (Continued) Influx of eggs to prevent extinction Emerging parasitoids, and fixed parasitoid survival, yields parasitoid adults of following generation An influx of parasitoid adults to prevention extinction Ltterature Cited Beeton, A.D. 1969. Changes in the environment and fauna of the Great Lakes. Eutrophication: Causes, Consequences, Correctives. Washington, D.C., Nat. Acad. Sci. Griffiths, K.J.- and C.S. Holling. 1969. A competition submodel for parasites and predators. Canad. Entomol. 101: 785-818. Jones, D.D. 1975. The application of catastrophe theory to ecological systems. [m: New Direc tions in the analysis of ecological systems. Part 2. George S. Innis, ed. Simulation Councils Inc., Vol. 5(2). (Also: Simulation 29: 1-15, 1977.) Holling, C.S. 1965. The functional response of predators to prey density and its role in mimicry and population regulations. Mem. Entomol. Soce Can. 45: 1-60. Holling, .c.S. 1973. Resilience and stability of ecological systems. Ann. Rev. Ecol. and System. 4: 1-23. Holling C.S.; G.B. Dantzig; G. Baskerville; D.-D. Jones; W.C. Clark; and R.M. Peterman. 1977. Quantitative evaluation of pest Management options: the spruce budworm case study. In: Proc. Symp- on Concepts and Practice of Integrated Pest Man. in Forestry. (15th Int. Cong., Entomol., Washington 1976) W.E. Waters (ed.), in press. McLeod, J.-M. 1966. The spatial distribution of cocoons of Neodtprton swatnet in a jack pine stand. I. A cartographic analysis of cocoon distribution with reference to predation by small mammals. Canad. Entomol. 98: 430-447. IF(EGG.EQ.0) EGG = EGG+. 00001 PARAD = ADULT*(1-PCOMP) *.041 IF (PARAD.EQ.0) PARAD = PARAD+. 001 1968. Results of an aerial spraying opera- tion against the Swaine jack pine sawfly, Neodtprton swainet Middleton in Quebec using the insecticide Phosphamidon. For. Chron. 44: 14-20. 1970. The epidemiology of the Swaine jack pine sawfly, Weodiprion swainet Midd. For. Chron. 46: 126-133. 1972. The Swaine jack pine sawfly, Neodtprton swainet life system; evaluating the long-term effects of insecticide applications in Quebec. Environ. Entomol. 1: 371-381. 1973. Information retrieval for the Swaine jack pine sawfly life system: a manual of coded sampling forms. Centre de Rech. For. des Laurentides, Quebec. Can. Dept. Environ. Inf. Rep. LAU-X-2: 95 p- 1974. Bird population studies in the Swaine jack pine sawfly life system. Centre de Rech. For. des Laurentides, Quebec. Can. Dept. Environ. Inf. Rep. LAU-X-10. 86 p. 1975a. ing problem. 1-12. Parasitoid evaluation: the monitor- Melsh. Entomol. Series. 17: 1975b- Possible residual effect of fenitro- thion on Swaine jack pine sawfly following aerial applications against spruce bud- worm in Quebec. Ann- Soc. Entomol. Quebec. 20: 82-85. » and W. Tostowaryk 1971. Outbreaks of pitch nodule makers (Petrova spp-) in Quebec. Laur. For. Res. Centre, Quebec. Can. Dept. Environ. Inf. Rep. LAU-X-24: 17 p. Morley, F.H.W. 1966.- Stability and productivity of pas- tures. Proc. N.Z. Soc. Anim. Prod. 26: 8-21. Morris, R.F. 1963. The dynamics of epidemic spruce budworm populations. Mem. Ent. Soc. Can. 31: 116-129. Noy-Meir, I. 1975. Stability of grazing systems: an application of predator-prey graphs. J. Ecol. 63: 459-481. Perry, R-A. 1968. Australia’s arid rangelands. Ann. Arid Zone. 7: 243-249. Plonski, W.L. 1960. Normal yield tables. Ont. Dep. Lands and For. (Silvic. Ser.). Bull. No. 2: 16-19. 81 Rosenzweig, M.-L. and R.-H. MacArthur. 1963. Graphic representation and stability conditions of predator-prey interactions. Am. Nat. 97: 209-223. Rowe, J.-S. 1971. Forest regions of Canada. Can. Dept. Environ. Can. For. Serv. Pub. No. 1300: 72 pre Rudolph, P.O. 1958. Silvical characteristics of jack pine. Lake States For. Exp. Sta. Paper 61: 31 p. Southwood, T.R-E. and H.-N. Comins. 1976. A synoptic population model. J. Anim. Ecol. 45: 949-965. Tripp, H.-A. 1965. The development of Neodtprion swatnet Middleton (Hymenoptera: Diprionidae) in the Province of Quebec. Canad. Entomol. 97: 92-107. Quantitative Evaluation of Pest Management Options: The Spruce Budworm) Case SEN + Coyesic Hodiimene G. B. Dantzi G. Baskerville, Introduction The boreal forests of North America have, for centuries, experienced periodic outbreaks of a defoliating insect called the spruce bud- worm (Choristoneura fumiferana). In any one outbreak cycle, a large proportion of the mature softwood forest in affected areas can die, with major consequences to the economy and employment of regions like New Brunswick, Canada, which are highly dependent on the forest industry. An extensive insecticide spraying programme initiated in New Brunswick in 1952 has succeeded in minimizing tree mortality, but at the price of maintaining incipient outbreak conditions over an area considerably more extensive than in the past. The present management approach is, therefore, particularly sensitive to unexpected shifts in economic, social, and regulatory constraints, and to unanticipated behaviour of the forest ecosystem. Most major environmental problems in the world today are characterized by similar basic ingredients: high variability in space and time, large scale, and a troubled manage- ment history. Because of their enormous complexity, there has been little concerted effort to apply system analysis techniques to the coordinated development of effective Paper presented by R.M. Peterman, Institute of Animal Resource Eco- logy, University of British Columbia Van- couver, B.C., Canada. 3 Department of Operations Research, Stanford University, Stanford, California. Department of Forest Resources, Univ- ersity of New Brunswick, Box 4400, Fredericton, New Brunswick, Canada.. is @, Glen,” D.,D. yous.” and R. M. Peterman 82 descriptions of, and prescriptions for, such problems. The budworm-forest system seemed to present an admirable focus for a case study with two objectives. The first, of course, was to attempt to develop sets of alternative policies appropriate for the specific problem. But the more significant purpose was to see just how far we could stretch the state of the art capabilities in ecology, modelling, optimization, policy design, and evaluation to apply them to complex ecosystem management problems. Three principal issues in any resource environmental problem challenge existing techniques. The resources that provide the food, fibre, and recreational opportunities for society are integral parts of ecosystems characterized by complex interrelationships of many species among each other and with the land, water, and climate in which they live. The interactions of these systems are highly non-linear and have a significant spatial component. Events in any one point in space, just as at any moment of time, can affect events at other points in space and time. The resulting high order of dimensionality becomes all the more significant as these ecological systems couple with complex social and econo- mic ones. The second prime challenge is that we have only partial knowledge of the variables and relationships governing the systems. A large body of theoretical and experimental analysis and data has led to an identification of the general form and kind of functional relations existing between organisms. But only occasionally is there a rich body of data specific to any one situation. To develop an analysis which implicitly or explicitly pre- sumes sufficient knowledge is, therefore, to guarantee management policies that become more the source of the problem than the source of the solution. In a particularly challenging way, present ecological management situations require concepts and techniques which cope creatively with the uncertainties and unknowns that in fact prevade most of our major social, “4 economic, and environmental problems. The third and final challenge reflects the previous two: How can we design policies that achieve specific social objectives and yet are still “robust"? Policies which, once set in play, produce intelligently linked ecological, social, and economic systems that can absorb the unexpected events and unknowns that will inevitably appear. These "“unexpect- eds" might be the one-in-a-thousand year drought that perversely occurs this year; the appearance or disappearance of key species; the emergence of new economic and regulatory constraints; or the shift of societal objec- tives. We must learn to design in a way which shifts our emphasis away from minimizing the probability of failure, towards minimizing the cost of those failures which will inevit- ably occur. What follows is a summary of certain elements of the spruce budworm case study. We can only hope to provide the reader here with a brief synopsis of the research; a full discussion of the study is contained in a forthcoming book (Yorque et al. 1978). The Descriptive Analysts The descriptive analysis of the budworm/ forest system is aimed to produce a well- Ww © (2) uJ = ax Oo = a 2 oO we (o) > = 2) Zz ea) a) 30-60+* YEARS Figure 1.--The pattern in time. feet of branch area. 83 tested simulation model that could be used as a laboratory world to aid in the design and evaluation of alternative policies. The key requirement of that laboratory world is that it capture the essential qualitative behaviour of the budworm/forest ecosystem in both space and time. Extensive data concerning forest- pest and economic interrelations had been collected over the past 30 years by Environ- ment Canada as one of the earliest interdis-— ciplinary efforts in the field of renewable resource management. There are many missing elements, but this is an inevitability rather than a drawback. If systems analysis is to be applied successfully to the management of ecological systems, it must be able to cope with unknowns. The essential qualitative behaviour in time has been identified through an analysis of tree ring studies (Blais 1968). At least five outbreaks have been detected since 1/7/70, (fig. 1), each lasting 7 to 16 years, with a 34- to 72-year period between them. During the inter-outbreak periods, the budworm is present in barely detectable densities which, when appropriate conditions occur, can increase explosively over four orders of magnitude during a 3- to 4-year period. The distinctive pattern in time is paralleled by one in space. The historical outbreaks typically initiated in one to three or four local areas of Eastern Canada and from those centres spread and contaminated progres-— 7-16 YEARS Representative historical pattern of spruce budworm outbreak. There have been at least five major outbreaks since 1/7/70. Budworm numbers are per 10 square sively larger areas. Collapse of the out- breaks occurred in the original centres of in- festation in conjunction with mortality of the trees, and similarly spread to the area in- fested at later times. The resulting high degree of spatial heterogeneity in the forest age and species composition is closely coupled to the "contamination" features caused by the high dispersal properties of this insect. The essential first step in the dynamic description of this system is a parsimonious bounding of the problem in terms of prime variables, space and time. The process of bounding the problem from the very start of the analysis is a key activity. Everything else in the analysis flows from these deci- sions, and they profoundly influence the final form and relevance of the policies. The key requirement in bounding the problem in space, time, and variables is to ruthlessly simplify while still retaining the essential properties of behaviour and needs for management. Bounding Time Because of the pattern of outbreaks shown in figure 1, the minimum time horizon required is that which can contain two outbreaks -- that is 150 to 200 years. In order to capture the dynamics of this system, it is essential to have a time resolution of one year with seasonal events implicitly represented. Bounding in Space As in many pest species, the budworm disperses over long distances. The modal distance of dispersal s about 25 miles from one site, but dispersal distances of several hundred miles have also been recorded. It was thought essential to have a minimum total area based on at least twice this modal distance, leading to a minimum modelled region of 14,000 to 15,000 square miles. The area chosen in this study was a 17,000 square mile area containing much of the Province of New Bruns-— wick (fig. 2). But even events in this size of area are profoundly affected by contagion from outside it. It was therefore necessary to add a buffer zone of approximately 75 miles width around the area in order to compensate for edge effects. The behaviour of this system is as highly heterogeneous in space as it is in time, and because of the contagion problem, spatial disaggregation is essential. There is high variation in the spatial distri- 84 bution of the primary tree species, of har- vesting activities and of recreational poten- tial, in part as a consequence of the histor- ical interplay between the forest and the budworm. The 25-mile modal dispersal distance also suggests a minimum resolution of about one-fifth to about one-tenth of that distance. Hence, the overall area is divided into 265 distinct subregions each approximately 6 by 9 miles (fig. 3). Bounding Variables An ecosystem of this extent has many thousands of species and potential variables. Our understanding of the dominant budworm/ forest dynamics is sufficiently detailed, however, that the system’s relevant behaviour can be captured through the interrelations among five species, each of which represents a key role in determining the major dynamics of the forest ecosystem and its resulting diver- sity. These key variables are summarized in figure 4. The principal tree species are birch, spruce, and balsam fir. In the absence of budworm and its associated natural enemies, balsam fir tends to out-compete spruce and birch, and so would tend to produce a monocul- ture of balsam. Budworm, however, shifts that competitive edge since balsam fir is most susceptible to damage, spruce less so, and birch not at all. Thus, there is a dynamic rhythm, with balsam fir having the advantage between outbreaks and spruce and birch during outbreaks. The result is a diverse species mix. As noted earlier, between outbreaks, the budworm is rare but not extinct. Its numbers are then controlled by natural enemies such as insectivorous birds and parasites (see Morris et al. 1963). But a key feature of this control is that there exists an upper thresh- old of budworm numbers which, once exceeded, allows the budworm to “escape” predation and multiply unchecked. There is, in other words, a distinct but limited stability region at low budworm densities (Yorque et al. 1978). In addition to tree species and natural enemies, there is a key stochastic driving variable, weather, which affects survival of the budworm and can flip the system out of the low density stability region if forest condi- tions are appropriate. Outbreaks cannot occur unless the forest has sufficiently recovered from the previous outbreak to provide adequate O 20 40 60 80 KILOMETERS 100 120 140 160 Figure 2.--Study area within the Province of New Brunswick used in the current study. The hatched area includes the primary forested regions of New Brunswick. food. Even with the food conditions met, however, the budworm remains at low densities under control by natural enemies until the weather shifts to successive years with warm dry summers. Such conditions allow larvae to develop so rapidly that densities above the escape threshold are achieved. An outbreak is then inevitable, irrespective of weather. In summary, the decisions on bounding the problem are as follows: Time horizon Time resolution Spatial area Spatial resolution Key variables to capture behaviour 85 150-200 years l year with seasonal causation 17,000 square miles 265, 6 x 9 mile sub- regions ideally, three tree species, budworm, natural enemies and weather. NID J JH JO Th JR SIN IN IN IN [IN IN (= —N WW J jo {ho (oe) © |0O |OD {CO Ke) W Jo Wo |o (eS i CI ES Ja sel el te ae Piel a rE |O nln re on Sa) No io) b bo Nn Wo (o>) WN a |e | Oo n ~S [on (2) + i) (o) Sa) Ble \© | 00 RIS aa © | co wat ALA 0 | a DID S KH ion) NIM TN] he Flwldys |e oe ee NO [Nh | Nh Bo |m [rR 1 jmnfum Figure 3.--This figure shows the numbering and indexing system for the 265 subregions, or "sites" This bounding of the problem immediately determines the number of state variables, which, in turn, affect the decisions about subsequent analytic steps such as optimiza- tion. Even though the previous steps of bounding seem to have led to a highly simpli- fied representation, the number of state variables generated is still enormous. Table 1 summarizes, for this ideal condition, the in the study area. minimum number of state variables necessary to represent the essential behaviour of the system in space and time. In any one subregion, 107 state variables are required, but of course, for the whole 265 subregions, a total of 107 x 265 or 28,355 state variables are required. Thus even this drastic simplification accomp- lished through the bounding exercise leaves an impossible number of state variables, thus demanding further simplification. It would, of course, be quite possible to develop a simulation model with this number of state variables. This would be expensive and time consuming to run and debug, but it would be possible. Our key goal, however, is to provide a useable and well tested model for exploring behaviour and policy alternates. With such a high dimensionality, the model would become nearly as incomprehensible as the HOST TREES age 1 epoeneetS age 75 RES SS ' BUDWORM