'1

C.3

COYOTE CONTROL: A SIMULATION EVALUATION OF ALTERNATIVE STRATEGIES

Russell L. Gum Louise M. Arthur Richard S. Magleby

U.S. Department of Agriculture Economics, Statistics, and Cooperatives Service

Agricultural Economic Report No. 408 CONTROL: A SIMULATION EVALUATION OF ALTERNATIVE STRATEGIES, by Russell L. Gum, Louise M. Arthur^ and Richard S. Magleby. Natural Resource Economics Division, Economics. Statistics,.and Cooperatives Service, U.S. Department of Agriculture, Agricultural Economic Report No. 408-

ABSTRACT

Current and alternative coyote control strategies in the Western United States are evaluated via a computerized simulation model which predicts the economic and socio-environmental impacts of each strategy. A gradual decrease in lamb losses and an increase in net economic benefits are predicted if the 1974 level of coyote control. $7 million. is increased to $20 million. Socio-environmental benefits did not change significantly under that simulation. Beyond the $20 million level of expenditures, net economic benefits are predicted to decline slightly and socio-environmental benefits decline rapidly. At expenditures below 197^ levels, both economic and socio-environmental benefits decline substantially. Changes in mixes of control methods are discovered which permit both economic and socio-environmental benefits to increase. These alternatives include increased use of the M-44 and aerial gunning and decreased use of traps. Key words: Coyotes, Predator control. Sheep losses. Simulation, Environmental evaluation.

ABOUT THE AUTHORS

Russell L. Gum and Louise M. Arthur are, respectively, agricultural economist and social science research analyst. Natural Resource Economics Division, Economics, Statistics, and Cooperatives Service (ESCS), U.S. Department of Agriculture, stationed at the Department of Agricultural Economics, University of , Tucson. Richard S. Magleby is an ESCS agricultural economist in Washington, D.C.

ACKNOWLEDGMENTS

The authors appreciate the data and consultation from the U.S. Fish and Wildlife Service, U.S. Department of the Interior, and the helpful suggestions from W. E. Martin, University of Arizona, and the many other reviewers of this manuscript.

Washington, D.C. 20250 July 1978 FOREWORD

Ihig report presents results of a simulation evaluation of the economic, environmental, and social benefits and costs of present and alternative policies for control of coyotes in the Western United States, The simulation results depend heavily on hypotheses about relationships among coyote control, coyotes, sheep and larri) losses, and other environmental effects. These hypotheses and the analyses are based on the most reliable data available and on judgments of experts in coyote control. As more data become available, the hypotheses can be revised to describe the control situation more precisely. Meanwhile, this simulation is considered a reasonable approximation of me current situation, with the results providing valuable information for decisionmaking on the control issue. The research was conducted by the Economic Research Service, U.S. Department of Agriculture, in cooperation with State universities, at the direct request of Congress. On January 1. 1978 the Economic Research Service was merged with the Statistical Reporting Service and the Farmer Cooperative Service to form the Economics, Statistics, and Cooperatives Service. Special appropriations by Congress in fiscal years 1974 and 1975 funded the research. Other reports in the series focused on characteristics of the sheep industry, levels of sheep and lamb losses, reasons for the decline of sheep production in the West, and costs and returns of western sheep producers.

CONTENTS

Highlights ii Synopses of Other Reports in this Series ili Introduction 1 The Simulation Model 3 Function of Each Component of the Model 4 Use of the Model's Two Final Outputs 5 Socio-environmental Considerations 6 Economic Considerations 14 Simulation Results 15 Changing Only Expenditure Levels Ih Increasing Expenditures on Each Method in Turn l8 Changing Both Expenditures and Method Mix 18 Conclusions 20 Research Recommendations 22 References 23 Appendix A: Control Methods and 1974 Expenditures 25 Federal Control Efforts 25 Private Control Efforts 25 Total Control Efforts 3I Appendix B: Technical Aspects of the Simulation 33 Socio-environmental Evaluation Methodology 33 Cross Impact Simulation of Control/Coyote/Lamb Loss Relationships . . 37 Impacts of Coyote Control on Other Wildlife 41 Economic Impact Evaluation Methodology 42 Appendix C: User's Guide to the Coyote Control Inforrration System .... 45 HIGHLIGHTS

Coyote control programs in the Western United States involve Federal, State, and private expenditures on various control methods. Throu^ use of a computerized simulation model, predictions are made of the economic and SOCIO-environmental impacts of actual and alternative expenditure levels for these methods. Economic impacts represent those on producers of lambs and sheep, plus those on consumers of lamb and mutton. Socio-environmental impacts include public acceptability of methods (humaneness, selectivity, and cost effectiveness), coyote population changes, and impacts of coyotes on other wildlife. The computerized simulation model facilitates consideration of control alternatives by (1) making explicit the key relationships affecting the impacts of control, (2) permitting changes in the relationships to reflect the best data, judgment, and experience available, (3) rapidly calculating both economic and environmental impacts, and (4) presenting the results so that the impacts and impact tradeoffs of alternatives can be compared easily. However, the model and impacts reported here represent first approximations and should be used as supplements to judgment and experience. Key relationships in the model need validation and refinement as better data and analyses become available. According to the model, increases in expenditures from $7 million to $20 million on the same mix of controls would result in a decrease in coyote population levels, a gradual decrease in lamb losses (from 8 percent down to 4 percent), and a slight increase in net economic benefits. Overall socio-environmental benefits did not change significantly. Beyond the $20 million level of expenditures, net economic benefits would decline slightly and socio-environmental benefits would decline rapidly. At control levels below 1974 levels, both economic and socio-environmental benefits would decline substantially. Alternative mixes of control methods are found for which predicted economic and socio-environmental benefits increased. The mixes of control methods yielding the highest economic benefits included increased use of toxicants—presently limited by legal restrictions—and decreased use of trapping. Predicted socio-environmental benefits are enhanced by decreasing expenditures on trapping and increasing expenditures on the M-44 cyanide ejector and aerial gunning. Research needed to support decisionmaking on predator control should aim for better methods for (1; measuring and monitoring livestock losses on a continuing basis, (¿) predicting losses as a function of control, coyote population, and management practices, and (3) determining impacts on gams species. Additional needs include better data on coyote numbers and coyotes taken by sport hunting and fur trappers.

11 SYNOPSES OF OTHER REPORTS IN THIS SERIES

"Characteristics of Sheep Production in the Western United States" by C. Kerry Gee and Richard S. Magleby Agricultural Economic Report No. 3^5.

About 80 percent of the sheep in the United States are raised in the West, where extensive private and public ranges provide the bulk of the feed. Only about 41 percent of the West's sheep producers have commercial scale operations of 50 head or more, but they own nearly 93 percent of the region's sheep. About one-third of these commercial producers have specialized in sheep, v*iile two-thirds have diversified livestock operations. More than two-thirds operate as sole proprietors, while me rest have formed partnerships and family corporations. Many have substantial equity positions which indicate past profitability. About one-fifth will likely be retiring in the next 10 years, viiich could result in many operations going out of sheep Production. About half of the feed requirement for commercial sheep comes rom private range, while public range supplies one-fifth. Over half of the commercial sheep are grazed under tiie care of herders, usually on open range. Most lambing occurs in late winter and early spring. More commercial producers practice shed lambing than range lambing, but the number of sheep involved is less. The principal marketing problem is the few number of buyers bidding on lambs.

"Sheep and Lamb Losses to Predators and Other Causes in the Western United States" by C. Kerry Gee, Richard Magleby, Warren R. Bailey, Russell L. Gum, and Louise M. Arthur Agricultural Economic Report No. 369 Predators, principally coyotes, are the major cause of lamb and sheep losses in the Western United States, according to 9,000 farmers and ranchers surveyed in 1974. Rates of loss to coyotes varied considerably amjng farmers and ranchers; vhile many had no or minor prédation problems, others reported very high losses. Overall, in the Western United States, losses attributed to coyotes in 197^ numbered 728,000 lambs (more than 8 percent of all lambs born) and 229,000 adult sheep (more than 2 percent of inventory), representing a third of the total lamb deaths to all causes and a fourth of the adult sheep deaths. These losses cost U.S. sheep producers some $27 million in lost returns in 197^, while consumers lost $10 million ." ' ^"^^ because of higher prices for lamb and reduced quantities available.

"Enterprise Budgets for Western Commercial Sheep Businesses, 197^" by C. Kerry Gee ^ ERS-659 Sheep enterprise budgets for 1974 are presented for major producing areas of the 17 Western States. Summaries of production, costs, returns, and operating practices are given for enterprises of various sizes and with different management systems. Most sheep businesses did not have sufficient sales in 1974 to cover all expenses, and about 35 percent are unable to pay cash costs. Businesses in -New Mexico realized the greatest return to invested capital. Small farm flocks in the wheat-corn areas of the Northern Plains States are least profitable.

Ill "Factors in the Decline of Sheep Production in the Western United States" by C. Kerry Gee, Darwin B. Nielsen. Delwin H. Stevens, and Richard S. Magleby Agricultural Economic Report No. 377 Former sheep producers in Colorado, Texas, Utah, and Wyoming were surveyed to determine why they had discontinued sheep production. From 40 to 50 percent were found to have continued in some form of agricultural business, usually involving cattle. The others had retired or taken off-farm employment. Generally, the former sheep producers had smaller scale operations, more equity in the business, higher prédation losses, lower earnings, and were older ttian producers continuing in the sheep business. Factors which they rated of greatest importance in their decisions to discontinue sheep production were high prédation losses, low lamb and wool prices, of good hired labor, and their own age.

IV COYOTE CONTROL: A SIMULATION EVALUATION OF ALTERNATIVE STRATEGIES Russell L..Gum, Louise M. Arthur, and

INTRODUCTION

Coyotes kill large numbers of sheep and lambs each year in the Western United States, Annual losses have been estimated at 8 percent of the lambs born and more than 2 percent of the sheep inventory (5). V In response to these losses. Federal, State, and local governments, as well as individual ranchers and groups of ranchers, carry out coyote control programs aimed at reducing the number of sheep-killing coyotes or coyotes in general. In these programs strategy decisions are made concerning the selection of coyote control methods and the amount spent on each method. The methods most frequently used are trapping with steel leghold traps, aerial gunning with shotguns from either fixed-wing planes or helicopters, and ground shooting (figs. 1 and 2 and App. A). Lesser used metJnods include denning (locating coyote dens and killing the pups), snaring, the M-44 (a device which ejects cyanide poison into the mouth of the coyote), and toxicant 1080 distributed in baits such as cattle carcasses (toxicant use was restricted by Executive Order and Environmental Protection Agency restrictions in 1972). Alternative coyote control strategies (methods and expenditure levels) are evaluated in this report via a computerized simulation model which predicts the economic and socio-environmental impacts of each strategy (fig. 3;- Ihe environmental issue is particularly important due to impacts of control on coyote numbers and other wildlife species. The social issue associated with

Figure 1—Expenditures on each method^ 1974

3 M-44 y//A Snar i ng 3 Toxicants y//////////////Á Ground shoot i na y////////////////^^^ A«riol ounnina y/////A Denning y///////////////^^^^^ Trapp

' 1 1 e Mill ions of do I lar*

Source: Appendix A

1/ Underscored numbers in parentheses refer to an item listed in the ITeferences section. Figur« 2—Coyotes ioken by «och m^ihod^ 1974

M-44 ] SnarIng ^^ Toxicant«

y/////////////M Ground shooting ^^^^^^^ATial gunning Denn i ng

'\r^pp I ng I 8 20 40 Thousands

Source: Appendix A

Figure 3—-Eva I uot I on framework "For coyote control

Control policy alternatives

Predicted socio - \ À environmental impacts \ / ^V^ Comparison of impacts ^ / and Simulation / / tradeoffs Predicted / model / ^ / of real life economic relationships impacts coyote control relates to public interest in humane treatment of animals, cost effectiveness of publicly financed programs, and the protection of wildlife populations. Ihe simulation model is a prototype systems approach to a comprehensive evaluation of the coyote control issue. Relationships which determine the economic and socio-environmental impacts of control measures are explicitly identified and quantified. The quantified relationships are based upon the best data and iudgnent available at the time of the study, 1975-77- A quote fron Boulding's Economics as a Science describes the purpose of this report. ". . . it is a useful first approximation and vAien it comes to evaluating difficult choices it is extremely useful to have a first approximation that we can modify. Without some quidelines, indeed, all evaluation is random selection by wild hunches." (3, p. 129) Major limitations and qualifications of the model are: (1) Several key biological relatifs had to be hypothesized due to the paucity of hard data from well-designed experiments or observations. The hypotheses were developed from available data and interaction with experts on the subject. 2/ The hypotheses are deemed reasonable, but they need verification and/or refinement tnrou^ research and analysis. (2) Ihe following are assumed to be the same for each alternative: (a) impact of fur harvest on coyotes, (b) impact of sport hunting on coyotes, and (c) impact of sheep management alternatives on losses to coyotes. (3) The model does not address: (a) effects on coyote populations or other aspects of the environment outside of the areas in viiich coyote control is practiced and (b) timing and spatial distribution of control. (4) The present model provides reasonable results only for the 15-State western region (States in App. table 1, excluding Oklahoma and Washington). As m^re specific data become available, applications to subregions or States will be feasible. (^) Toxicant 1080 and M-44 cyanide ejectors are assumed to be applied or rigorously controlled by trained professionals so as to minimize danger to human health or domestic animals. (6) Impacts predicted by the model are the medium- to long-run impacts, as opposed to the initial effects of a sizeable change in level or mix of controls. The simulation results presented in the report are tentative because of the model's limitations and should be used as a supplement to judgment and experience. Even so, they represent the most comprehensive description available of the direction and general magnitude of the impacts of alternative coyote control strategies. This description provides decisionmakers with a better perspective of the issue and the economic and socio-environmental tradeoffs involved.

THE SIMULATION MODEL

Coyote control policy alternatives include higher or lower overall expenditures than in 197^ and/or changes in the allocation of expenditures among methods. A range of these alternatives is evaluated in this reoort. The computerized simulation model for this evaluation can t>e used to look at

2/ The experts included biologists, researchers on predator control, and officials involved in management of predator control programs. However, the authors assume responsibility for táie interpretations of the information provided by the experts. any number of other alternatives in addition to those presented and to réévaluate any alternative as new, more reliable information is fed into the system. For additional detail on tine specific computer code, see Appendix C. gtje sirnulation modçl is ^a mathematical approximation of^^the ,real life lological, economic, and social systems which are either affected by coyote control or influence its impacts. Trie general structure of the model is shown in figure 4. Input to the model is the dollar expenditure for each of the various control methods. This information is then fed into various submodels or components of the system, which in turn provide information for other parts of the system. This continues until the various impacts of the control input are estimated and condensed into two final outputs, the socio-environmental index and the net change in economic benefits over actual 1974 control. A brief description of each of these components or submodels is given below, followed by a discussion of important considérâticais made in estimating the economic and socio-environmental impacts.

Function of Each Component of the Model Submodel 1 calculates an index of cost effectiveness for the control input by weigntmg the effectiveness of each method by the expenditures on that method. Submodels 2, 3, and 4 estimate the relative numbers of domestic animals, coyotes otner man tnose that have killed lambs or sheep, and other wild animals that might be accidentally killed by the control measures. Submodel 5 receives information on accidental killings from Submodels 2, 3> and 4 and combines these data into a composite estimate of the relative selectivity of the control input (degree to v*iich it affects only coyotes which kill lambs and sheep). Submodel 6 estimates a humaneness index for the control input by weighting the Bublicly perceived humaneness of each control method by the expenditures on ^at method.

Figure 4—Simulation model structure

1. Cost effectiveness ^^^ f ^"^^-^-..^^ / Accidental killing of ^^■^--^^ 2. Domestic animals 7 Acceptability / 5. Selectivity J 3. Coyotes not causing losses of methods \\ / « 4. Other wildlife ^ // ^ ^É. 16. Socio-environmental // impact / ^ A Input = / T 6. Humaneness ■^^ Í expenditures 10. Coyotes remaining / per method 9. Percent of 11. Coyote population S^ coyotes killed ^ impacts on other wildlife \ ^ ^ 8. Corrective and preventive control N. 12. Index of lamb Economic impacts on= and sheep losses - 14. Producers 17 Net economic 13. Number of lambs 15. Consumers • benefits and sheep lost over 1974 _ _ Submodel 7 takes the cost effectiveness, selectivity, and humaneness indices îransubmodels 1, 5, and 6 and combines them into an index reflecting the general social acceptability of the control input. Submodel 8 divides the original control input into corrective and preventive components because of the differential effect each has on coyotes. Submodel 9 takes the corrective and preventive information and estimates the relative impact of the control input on the coyote population. Submodel 10 then iterates throudi a coyote population model to determine the proportion of the.coyote population remaining after the control input has been consistently applied for several years and the initial shock effects are dissipated. Submodel 11 takes the proportion of coyotes remaining and uses this to determine the probable impacts of the coyote population on other wildlife (such as deer and antelope). Submodel 12 uses the corrective and preventive allocation of submodel 8 to estimate tne probable relative change in lamb and sheep losses compared with

Submodel 13 then appliestoe relative change in losses to actual numbers of lamDS ana sneep lost in 1974 to estimate the new number lost. Submodels 14 and 15 use the number of lambs and sheep lost to determine the quantity or lamDs^ marketed, then to estimate (througji use of a demand equation) the retail and farm prices, and finally to calculate the economic impacts on ,producers (net change in total sales proceeds) and on consumers (net change in consumer surplus). Submodel 16 compiles and condenses the information on social acceptability of metnoas. coyotes remaining, and coyote impacts on other wildlife into a composite index reflecting both social and environmental factors (hereafter called the socio-environmental index). Ihe weights used in combinins; the three items are based on public perception of the relative importance of each. Submodel 17 calculates the net change in economic benefits of the particular control input over 1974 conditions by summing the change in economic impact on producers witii that on consumers and then deducting the change in control costs over 1974 costs. ^ The model then prints out the two composite outputs of submodels 16 and 17 plus other intermediate results, depending upon the degree of detail specified by the operator (see App. C).

Use of the Model's Two Final Outputs The nçdel's.two final outputs, the composite socio-environmental index and the net change in economic benefits, make passible a simple and quick evaluation of alternative control strategies. For example; if additional control expenditures would return less in economic benefits to producers and consumers than IS spent fçr. the increased control, then reducing expenditures is not economically justified.. Further, .^increasing control expenditures is environmentally justified only if socio-environmental quality is also improved. If a change in policy improves both the economic and socio-environmental measures,, that change is considered preferable to current policy, uiiiiculties.arise. when an alternative improves economic conditions but worsens socio-environmental conditions, or vice versa. The model highlights the range of tradeoffs between economic benefits and socio-environmental benefits,, but decisionmakers ^are^.left the responsibility for deciding the relative importance of the two benefits. j ^ u-Lug Socio-environmental Considerations The cost effectiveness, selectivity, humaneness, public acceptability, and the impact on coyote and ofchçr wildlife populations of each control strategy are the important socio-environmental ractors considered by the simulation model.

Cost Effectiveness The cost effectiveness of a control strategy is defined as the expenditures on the strategy divided by the expenditures necessary to provide the same level of control using the m^st cost efficient methods. Thus, a rating of 50 means that the same level of control could be provided by other methods at 50 percent of the cost. Since control can be either corrective (specific control in response to losses) or preventive (control of the general coyote population to prevent future losses), the cost efficiency of each method for each purpose must be considered in calculating the overall efficiency of a strategy (table

For corrective control, both aerial gunning and the M-44 are deemed most effective, with trapping 50 percent as effective and denning 20 percent as effective. For preventive control purposes, toxicant lOoO is deemed the single most effective method, with other methods 12 to 50 percent as effective. Also, the likely use of a method for preventive cr corrective control must be considered. The hypothesized average allocation of each method between the two types of control is shown in table 1. Trapping, for example, is judged to be used an average of 50 percent for corrective purposes and 50 percent for preventive purposes, while toxicant 1080 is deemed entirely preventive. These allocations are based on field experiences of U.S. Fish and Wildlife Service researchers. Ihe control allocation and the effectiveness indices are crucial parameters in the m^del, but are still unrefined. Specific research on these questions is suggested to verify and/or refine these approximations.

Selectivity y Overall, the control metJiod judged to cause the fewest accidental deaths of nontarget animals is ground shooting (table 2). Least selective (causing the most accidental deaths) is toxicant 1080. The individual indices for control impacts on domestic animals, coyotes not causing losses, and otiier wildlife are based on judgnents of Fish and Wildlife Service coyote control researchers. The weights used to combine the three indices into a composite selectivity index for each method reflect the public's relative concern for each class or nontarget animals, obtained in a nationwide survey of public attitudes ( 18).4/ In general, the public is more concerned about impacts on domestic animals and nonooyote wildlife than impacts on those coyotes which are not causing losses.

Humaneness Indices of the relative humaneness of each method (fig. 5) are based on perceptions of the U.S. public, obtained in the nationwide survey of public attitudes (18). The method perceived as most humane is the M-44, which uses a fast acting"^ison (index of 43 out of a possible 100). Perceived as least humane is toxicant 1080, a slow acting poison (index of 13)- No attempt is made to measure biologically the degree of suffering inflicted by each method, if such measurement is even possible. Rather, public perceptions of that suffering are used.

3/ All methods are assumed safe to humans or they would not be used. 4/ A power function is used for the aggregation; see Appendix B. Table 1—Estimated cost effectiveness of various coyote control methods Control : Effectiveness Method allocation 1/ : of each purpose 2/ Corrective rPreventive : Corrective:Preventive

: -Percent Index 1/

Trapping : 50 50 50 12 Denning 50 50 20 12 Aerial gunning ; 25 75 100 25 Ground shooting : 50 50 25 12 Toxicant 1080 : 0 100 0 100 Snaring : 50 50 50 12 M-44 : 25 75 100 50

1/ Hypothesized normal allocation of dollar expenditures for each method between corrective control (specific control to reduce further losses) and preventive control (control of general coyote population to pre- vent future losses).

2j Effectiveness relative to aerial gunning and M-44, the two most effective methods for corrective control, and toxicant 1080, the most effective method for preventive control.

Source: Cost records of U.S. Fish and Wildlife Service and judgments of coyote control specialists. Table 2—Selectivity indices for each coyote control method

Composite Impact on : Impact on c Impact on other (weighted) Method : domestic animals ; not causing wildlife index 2/

Index 1/

Trapping 70 50 25 44 Denning : 100 10 90 60 Aerial gunning : 100 20 100 72 Ground shooting : 100 40 100 82 1080 toxicant : 80 0 50 20 Snaring : 75 50 20 42 M-44 : 60 50 50 53

1/ The higher the index, the less accidental killing of other animals.

2/ Weighting was done by use of a power function (see page 35) with the following velghts given ~ to the individual impacts.

Impact on domestic animals .42 Impact on coyotes not causing losses .20 Impact on other wildlife .38

1.00

Source: Individual impacts are based on judgments of experts. The weights reflect the public's relative concern about each impact, as obtained in the nationwide attitude survey (18). Figur« 5—Human«n«««

M-44 ^^ Snarl no ^^ Toxicant« Í5J5^55Í5ÍJSSÍS^$^^ Ground «hooi i ng ^:Aíí55S^Síí^ A«rI aI gunn i ng iü5!^5S5SS!Ä?S5Ä^ D«nning ^^ Trapping T r —'—r 1 1 1 r e 50 100 Human«n««« i nd«x

Sourc« i Nai i onaI «ampI« «urv«y of publie ai i i iud««

General Acceptability Humaneness, selectivity, and cost effectiveness indices are combined in the model into one index reflecting the general acceptability of the methods (or control inputs) by weighting the three aspects as follows: cost effectiveness (.16), selectivity (.32), and humaneness (.52). ¿/ The weights represent the publicUDllC ^S relative concernncMnni^'nr^ for-P^r» each,laonV^ asoe obtained^K4"-^•; irw=»^ inn^ thei-U^ survey«,,^ ,. ^^^of publici-T_'^ attitudes Clearly, the humaneness and selectivity conçonents predominate win determining acceptability ; the public views cost effectiveness as a relatively unimportant component of acceptability. Ground shootinc is Yifo?^ f^JÎ^Ç^K^^^^P^i'-^? ^° the^public (fig. 67 and toxicant 1080 as the least acceptable .method, probably.because its humaneness is judged to be low and it sometimes kills nontarget animals.

Control Method Impacts on Coyote Numbers Another socio-environmental factor related to coyote control is the impact of Çi?2^i^2L 2? ^9^^ population levels. These impacts are described in terms of the percent of. the coyote populaticn which remains under a given control strategy. A simulatiai of a coyote population is developed to show the effect of predator control programs on ^the .structure and dynamics of a coyote R^P^i^^^J^- Results çT the model^inçiiçate that significant long-term impacts on coyote numbers require continued killing of large percentages of the covote population in an area. These results,are insistent with those from similar models developed by other researchers (4, 8) and with the observed constancy of coyote numbers, despite control effortsT ^ The structure of the coyote populatiai model is illustrated in figure 7 and ofS=n^if,^ ^gor^tail in Appendix B. First, the percentage use of carrying capacity (PPC) determines the percentage of yearlings and older females 5/A power function is used for the aggregation; see Appendix B. Figur« 6—M«ihod acc^piabiIIiy

555í^^íí5!55!^í^5i^^A^^SA!Í M-44 Í5ÍÍÍÍÍÍÍÍÍÍ5ÍÍ^5Í Snaring ^^?^^^?^^^;^^^ Toxieoni« ¿^5a5¿¿¿^¿¿i5j5í;¿Í55^^ G,«^„d .h««tln« Ä55^^^;5?5^;i5:^?5^^ A.rl«l çuonlnç

Í^A;^;55í5¿A?A5¿A;¿^ D-nning ^^¿^^^¿^^^5^!^ Tr.«pplnç

1 1 1 1 1 1 1 1 1 1 0 50 tee Method occ^ptablI iiy Ind^x

Sourc«: R^suli from «i mu I ai I on mod«!

Figur« 7—Coyol« population submod«I

Number of pups per litter Percentage young females wtielping Percentage older females whelping I Number of pups born Percentage use of carrying capacity I New population

10 producing litters and the average size of those litters. (As PCC increases, the percentage of females producing litters and the average litter size tend to decrease.) Based on the age distribution of the population, the number of pups born is calculated and the size of the population determined. Next, the ^Eä^lÄ^dlltiS^Sl^fä ^nlrSf SÜL^r^it iL |{?§'aPfê"FouP!Î"5uirtl^?^ new PCC is calculated using the surviving populaticai. The process is reiterated for 10 years to reflect the results of a continuous control program. Figure 8 illustrates the response of age distribution of the coyote population model to increasing predator control. At a control rate of killing 1 percent çf the fourth-quarter population, 26 percent of the next year's population is in the 0 to,1 year age group. This group comprises an increasing proportion of the population as control rates increase. Figure 9 shows the relationship between the total yearly average population level and the control rate. The population suffers a fairly constant but slow decline with increasing control rates up to SO percent. Thereafter, the population declines at a faster rate because of the increasing proportion of yearlings m the populatioi, which produce fewer litters than older females. Figure 10 shows that the number of pups born peaks at a rate of about 60 percent. Clearly, coyote populations easily compensate for lower level losses to their populations. To illustrate relative population impacts of the different methods, measures of coyote population impacts are calculated for $10 million spent for each method. The choice of $10 million per method is arbitrary and only Illustrates the relative population impacts of the various methods. The results are shown in figure 11. M-Î14 and 1080 have the greatest impact on coyote populations, while denning and ground shooting have the least impact.

Figure 8—Perceniage of yearlings in population

L 0 >

0

Q) 0 C Ü L 0) Û. T I I I 1 1 1 1 r 0 50 100 Percentage coyotes ki I Ied each year

Source: Result from simulation model

11 Figur« 9—Yearly average populati on

200 X c

c 0

D OL 0

100 Percentage ki I led

Sourc«r Resulfc from simulai i on model

Figure 10—Number of pupe born

200

0

L JÏ6

100 Percenioge of coyotee kl I led each year

Source: ReeuI i from elmuloiion model

12 Figur« ti—P^rc^ni coyoi« populaiion remaining

M-44 SnarIng Toxi cants Ground shoot í ng A«rI o I gunn i ng

Trapp i ng T" 1 1 1 r 8 58 100 Psrcsnt

Source : Result from simulation model

Secondary Impacts on Other Wildlife Changes.in coyote population levels would be expected to affect other wildlife populations due to interrelationships among many wildlife species. Although there is little data concerning relationships between coyote populations and all other wildlife populations in general, some data on coyote impacts on individual species of wildlife provide a basis for hypothesizing general ei leCuo \_¿j _Lii> I^) • Coyotes probably have little impact on animals with extremely high reproduction rates, such as rodents, but there is some evidence that they do have an impact on other wildlife populations, such as deer. Thus, if there were no coyote controls, other wildlife would be adversely affected by the '" " " ' populations. Conversely, if coyote r control, the coyoteas ecological ^ , ^ing in adverse impacts on wildlife in general. Thus, some medial level of coyote control was hypothesized as providing a better ecological balance than either extreme (see App. B, p. 41).

General Socio-environmental Weights As noted earlier, the composite socio-environmental index represents a weighted average of the indices for general method acceptability, impacts on coyote numbers, and secondary impacts on other wildlife. The weights used by the model in combining the three indices are: method acceptability (.30), impact on coyote,population (.24), and coyote population impacts on the wildlife (.4b). 6/ These weights reflect the public's relative concern for each of the three aspects of control, as obtained in the nationwide attitude survey (J8,). Ihe public is primarily concerned about other wildlife and least about decreases in the coyote populat ion.

6/ A power function is used to aggregate the indices; see Appendix B.

13 Economic Considerations The economic impacts of control alternatives depend on the initial levels of sheep and lamb losses to coyotes, the change in number of lamb and sheep losses resulting, from a change in control, and the effects of the chance in lanfc losses on the number of lambs nnrketed and market prices.

Initial Levels of Lamb and Sheep Loss The initial levels of lamb and sheep losses in the model are those estimated for 1974 for the 15-State western region {^). Based on reports from 8,910 farmers and ranchers, these estimates are:

Lamo losses Deiore and Cause after docking Sheep losses inousands Percent of lambs Thousands Percent"oT t)orn inventory^ Coyote prédation 728 8.1 299 2.5 Other causes 1,350 15.1 718 7.9 Total 2,078 23.2 ^47 10.4

Control Method Impacts on Lamb Losses The impacts of control method on lamb losses are based on judgments of coyote control experts and on a simulation which used cross impact analysis (7) of the coyote control-coyote populaticai-lamb loss relationships. This approach was taken after statistical analysis of available data yielded no definitive relationships, only general trends (2). Figure 12,displays the relationships used in the model to predict lamb losses. The vertical axis depicts expected losses while the two horizontal axes represent levels of control effort for preventive control and corrective control. The 1974 levels of control are indicated on the two horizontal axes. At zero levels for both preventive and corrective control (top left in fiÇ. 12; lamb losses are predicted to reach a maximum of I8 percent. 7/ If çnly corrective control is used (left of figure), losses decline faster than If only preventive control is used (top or figure). For combinations of preventive and corrective control, losses decline rapidly until I974 levels of ççntrol are reached, after which losses decline more slowly (center of figure).

Procedures Used in Estimating Economic Impacts 8/ A dollar measure of how much better or worse off producers and consumers would be under various coyote control policies is necessary in order to determine whether a policy is economically justified. The dollar measure of impacts on producers used in this report is based on the difference between producers' actual 1974 gross returns and the returns they would have received had an alternative policy existed in 1974. In calculating this change in gross returns, account is taken of both changes in the quantity of lambs marketed and the resulting change in lamb prices. To account for the loss of adult sheep to coyotes, it is assumed that reolacement lambs are kept off the market to maintain a constant herd size. Adjustments in the calculations of number of lambs marketed are also made to account for the fact that coyotes might

2/ In the complete absence of control, lamb losses to coyotes are assumed to reach 18 percent and sheep losses 5.5 percent, on the average. These upper limits, appeared reasonable in light of actual losses found in no-control situations (JQ). 8/ For additional details, see AT^pendix B.

14 Figure 12—Corrective and preventive control impacts on iamb oeses

Note = An index of i reflects 1974 control levels.

kill lambs and sheep that would have died from other causes anvwav. Q/ Impacts due to changes m wool sales are not considered, nor is the S^S^-^ ¿^5^ .i?^ ^ rancher abandoning the sheep producing business in order to reî?eation de^ î^'^'^^^t ^^^ "®^' ^^ cattle production or The dollar measure of the impact.of control on consumers is also based on the price of lamb and the quantities available. If there are fewer losses of Sheep and lambs to coyotes, raore^lanfcs will be marketed, causing the price to ;fi}îoK-;i^?'^"f?^T^ K "^^ benefit from both lower prices and increased ^^ii^'^i-'-^^y^^^ ^^^' ^? net economic impacts, then, are simply the sum of producer and consumer impacts minus any increase in coyote control costs or plus any savings m costs compared to actual 1974 expenditures.

SIMULATION RESULTS

ÎSvofÎ^^n^nfi^.™?*!!^ ^^ ^^^^ to evaluate three .alternative strategies for çcmtrçlsP?^?^i^°"î^''-^ as mn^^ÍQ7Í^"Sfo^^"- 1974,. (2) increased ^"^"^l .^^ expenditureexpenditure on on eachthe controlsame^ mixmethod of individually, and (33 simultaneous changes in both expenditure level and ^x ?L '^HHÍrSi^^, ?^ model's prediction or.results are a^proxinltiœs indiSatSí group.Hî^,r^^^1n/^°" I u/ ^"^ general magnitude of impacts for the 15 Western States as a

^L5~?,-Pr°P°'"^^*^ °^ iê"*? 3^^ §^^eep tiiat died in 1974 from causes other prédation is used as the basis for these adjustments. than JO/ Includes States listed in Appendix table 1 except Washington and Oklahoma.

15 Changing Only Expenditure Levels One set of policy options examined is that of retaining the 1974 mix of methods and simply increasing or decreasing overall control expenditures. As control expenditures are increased in increments from the Í7 million level of 1974 up to $20 million, the first predicted impact is a reduction in coyote population numbers in the control areas from 70 to 50 percent of the natural maximum (fig. 13)- With the decrease in coyote numbers, predicted prédation of lambs decreases from the 8 percent level of 1974 to about 4 percent (fig. 14). This increases the number of lambs marketed, providing economic benefits (after payment of control costs) to society (producers and consumers together) of up to $6 million (fig* 15). Beyond approximately $20 million in expenditures for control, the simulation indicated net economic benefits would decrease, for the additional control costs would begin to exceed the additional dollar benefits of control. Thus, expenditures beyond $20 million for this same mix of controls would not be economically justified. Decreases in control expenditures below the $7 million level of 1974 are predicted to result in a rapidly expanding coyote population and substantial increases in lanto losses and decreases in net economic benefits. For example, the simulation indicates that a 50 percent reduction in control expenditures would permit coyote numbers to increase from 70 to 80 percent of maximum, with the increased prédation losses costing society approximately $20 million in foregone economic benefits. At the no-control extreme, the model predicts lamb losses would reach 18 percent on the average, m^re than double the 1974 level, while producers and consumers would be $45 million worse off than they actually were in 1974.

X 9 c

Milt ion do I lar ttxp«ndi tur«

A—Compos i te cocIo-onvIronmontoI Index B—Coyote Impact« on other wildlife index C—Control acceptabi I ity index D—Coyote population Index

Sowrce c Result from slinij lot Ion model

16 FiQur« 14—Pr«d¡ci«d lamb losses

tt 0

M E 0

c Ü L (L

T ' ' 1 *" e 25 se M i I I i on do I Iar «xp«nd i iun

Source: Result from s i mu lot i on modo I

Figure 15—Net economic benefit» over 1974

15

75 M i I I i on do I I or expend i t ure

Source: Result from simulation model

17 The.socio-enyironnBntal index, a composite measure of public perceptions of environmental quality (App. B , can range from 0 to 100, reflecting low to high quality. This index is estimated by tñe madel to be 64 for the 1974 one. This lack of change reflects the public's ^neral lack of empathy for the coyçte as well as the positive impacts of smaller coyote populations on more highly valued prey species such as deer, antelope, and waterfowl. For increases m expenditures beyond the $20 million level, however, the model predicts a substantial decline in the socio-environmental quality index. This results from the predicted sizeable declines in coyote populations at these higher control levels and possible undesirable impacts on the ecosystem. H/

Increasing Expenditures on Each Method In Turn A each 1974 from economic expenditure 'for denning. ^,The' expenditure for M-44 yieldeOS.^ "million"" iri benefits, coirpared with $4 million for aerial gunning, $2.5 million for snaring and trapping, and $2 million for ground shooting. Impacts on the socio-environmental index are minor.

Changing Both Expenditures and Method Mix A third set of possibilities is that of changing both the level of expenditures for control and the mix of control methods. To evaluate these ^Àh^n^^^"^^^I ^y^^^?l^\:}p variations in the levels of trapping, aerial gunning, 1080 toxicant, and M-44 are made using the 1974 levels of all other methods as a base. Each method is varied (in steps of $2 million) from $0 to $8 million expenditures. All possible^ combinations of expenditures and methods are investigated, resulting in 625 combinations. Any, alternative which is lower than any other alternative on both the socio-environmental index and net economic benefits is judged inefficient and not considered further. Of the resulting set of efficient alternatives, presented m table 3^and figure 16, only three included use of trapping. The Bredominance of 1080 and M-44 in the set of efficient solutions is Based on le assumption of proper use of toxicants and M-44 by trained professionals. Unregulated use or these methods will not result in the level of socio-environmental quality or net economic benefits predicted by the model. Efficient alternatives range from a socio-environmental index of 69 with net benefits .of $12.9 million to a socio-environmental index of 44 with $25.9 million in net economic benefits. This implies an average tradeoff of $520,000 of net economic benefits for an average of one socio-environmental index unit. Six of the alternatives have both a higher socio-environmental index.and a higher net economic benefit than that of the 1974 level of control and mix of inatnods. Variables from the model other than the two composite indicators discussed above can also be taken into account in control strategy evaluation. For example,.suppose a policymaker decided the best political strategy would be to do what IS necessary to reduce lamb prédation losses by about one-half (from 8 to about 4 percent), yet at the same time maintain the socio-environmental index as high as possible and the total control expenditure (mostly from tax dollars) as low as possible. Of the efficient control alternatives in table 3, the one which appears closest to meeting these criteria is number 6.

JM/ This IS based on a hypothesis that the coyote does play a role in the ecosystem, and that too few could disrupt the ecosystem. Future research should define this relationship better.

18 Table 3—Efficient coyote control alternatives Socio-envi- : Net Control expenditures : Total 1/ Alter- ronmental : Lamb : economic : Aerial : 1080 : : expendi- native index : loss : benefits : Trapping : gunning : toxicants : M-44 : tures

: Index Percent Million dollars

1 • 69 4.8 12.9 0 4 0 4 10 2 : 69 5.2 13.4 0 0 0 6 8 3 : 68 4.7 13.4 2 0 0 6 10 4 : 68 4.5 14.5 0 2 0 6 10 5 66 4.2 16.0 0 0 0 8 10

6 64 3.6 16.8 0 2 0 8 12 7 63 3.1 17.2 0 4 0 8 14 8 61 3.8 17.9 0 2 2 4 10 9 61 3.5 19.2 0 0 2 6 10 10 59 3.0 19.8 0 2 2 6 12

11 58 2.7 21.0 0 0 2 8 12 12 56 2.3 21.2 0 2 2 8 14 13 54 2.6 21.5 2 0 4 4 12 14 54 2.5 22.2 0 2 4 4 12 15 54 2.2 23.3 0 0 4 6 12

16 52 1.8 23.5 0 2 4 6 14 17 52 1.6 24.4 0 0 4 8 14 18 49 1.5 24.7 2 0 6 4 14 19 49 1.4 25.2 0 2 6 4 14 20 48 1.3 25.9 0 0 6 6 14

21 44 .8 25.9 0 2 8 4 16 11 11 : 64 8.1 0 3.3 1.4 0 0 6.9 3/ 23 : 68 4.9 10.4 0 10 0 0 12 3/ 24 : 67 4.3 11.3 0 12 0 0 14 3/ 25 : 65 3.8 11.9 0 14 0 0 16

3/ 26 : 63 3.3 12.3 0 16 0 0 18 3/ 27 : 61 2.8 12.5 0 18 0 0 20

\J The following expenditures were held constant: denning, $552,000; ground shooting, $1,126,000; and snaring, $281,000.

11 Actual, 1974.

2/ These are the most efficient alternatives which do not include 1080 or M-44.

Source: Result from simulation model.

19 Figur« 16—Tradeoffs for efficient alternallvei

Cf) L 0

0 ■o c 0

1—I—I—I—\—I—I—I—1—I—I—I—I—r 40 50 60 70 Socio—env¡ronmenloI index

Source« Reeuli from simulai i on model

However, if policymakers were willing to trade off the lamb loss level only slightly (for instance, have a 4.2 percent rather than 3.6 percent loss), ttiey could opt for control alternative 5 which would require a lower control expenditure, $10 versus $12 million, while achieving a slightly higher socio-environmental acceptability index, 66 compared with 64. How would the predicted impacts of alternative controls differ if actual lamb losses to coyotes in 1974 were only about 4 percent of the lamb crop or half those estimated in the OSDA survey? To test the sensitivity of the results to such a change in the base loss level, an additional 625 runs of the model were made. Ihe major difference in the efficient set of alternatives is the lower economic benefits predicted. $8 million, as compared to nearly $26 million (tables 3 and 4). In both sets, the alternatives providing the greatest net economic benefits employ the M-44 and 1080 toxicants, while those providing the most socio-environmental benefits utilize neither trapping nor 1080 toxicants.

CONCLUSIONS

Both the marginal analysis of^adding $1 million to each.control method and the comprehensive analysis of 625 control mix alternatives suggest that 1000, M-44, and aerial gunning are (in that order) the best methods From an economic point of view. If chemical methods of control remain legally restricted, then uie best alternative is aerial control. M-44 and aerial methods are also superior to other methods from a socio-environmental point of view. Therefçre, an economically and environmentally aerial and 1 expenditures for risk of M-44 to humans is very low If M-44 is proven unsafe to humans, then

20 Table 4~Economlc tradeoffs for efficient alternatives under the assumption that 1974 sheep and lamb losses were only 50 percent of the USDA estimated value : Soclo- Control expenditures : Total 1/ Alternative : environmental : Net : : Aerial : 1080 : : expendi- number : index : benefits : Trapping : gunning : toxicants : M-44 : tures

: Index MlliioM-S 1 1 -Í « n dollars — _ ^~'"""" ^^

1 : 69 4.4 0 4 0 4 10 2 : 69 4.9 0 2 0 4 8 3 ! 69 5.7 0 0 0 6 8 4 : 66 5.9 0 0 0 8 10 5 62 6.1 0 4 2 2 10 6 : 61 6.3 2 0 2 4 10 7 : 61 6.8 0 2 2 4 10 8 : 54 7.4 2 0 4 4 12 9 : 54 7.7 0 2 4 4 12 10 : 54 8.2 0 0 4 6 12

1/ The following expenditures were held constant: denning, $552,000; ground shooting, $1,126,000; and snaring, $281,000.

Source: Result from simulation model. only aerial gunning is both an eooncxnically and environmentally reasonable means of control.

RESEARCH RECOMMENDATIONS

Several priority areas for research on prédation and its control became apparent dxring the development of the simulatiai models. Ihese included: {1) better data for analysis of the relationships among coyote control metiiods and expenditures, coyote numbers, sheep and lamb losses, availability oi alternative prey, ranching practices, vegetaticm typ^, and weather (data could be gathered by simultaneous monitoring of the above variables on a continuous basis during the year for at least selected representative .areas): (2) better data en coyotes taken for sport hunting and fur trapping, and analysis of how sport hunting and fur trapping affect coyote numbers in sheep prodacing areas and are affected by flir prices and bounties; and (^) better data on coyote predatiais of ^me species, particularly deer and antelope. These data and analyses would help refine the coefficients used in this model and, in turn, the accuracy of the madel's predictions.

22 REFERENCES

(1) Arthur, Louise M, 1977. "Predicting Scenic Beauty of Forest Ehvironments: Some Empirical Tests," For. Sei. 23(2): 151-160. (2) 1978. Factors Affecting Coyote Prédation of Sheep and Lambs: A ¿)tatisticai Analysis': u.ö. Dept. Agr., h;con. ¿>tat. uoop. ¿serv. AUEKS-47. Nat. Tech. Inf. Serv., Dept. Conm., Springfield, VA. (3) Boulding, Kenneth E. 1970. Economics as a Science. McGraw-Hill Book Company, New York. (4) Connolly, Guy E., and William M. Lon^^urst. 1975. The Effects of Control on Coyote Populations: A Simulation Model. Div. Agr. Sci. Bull.—1872, Univ. Calif., Berkeley. (5) Gee, C. Kerry, et. al. 1977. Sheep and Lamb Losses .to Predators and other Causes in the Western unicea ozaZësl—AEH-3By,—U.S. Uept.—Kff^j—Ecorï:—Res: Sêf^T:— (6) Gier, H. T. 1968. Coyotes in Kansas. Kans. State Univ., Agr. Exp. Sta. Bull. 3^ij Mannattan. (7) Kane, Julius. 1972. "A PriiiBr for a New Cross-Impact Language—With Examples Shown from Transportation Policy." Tecnn. Forecasting and Social Change 4(2). (8) Knowlton, F. F. 1972. "Preliminary Interpretations of Coyote Population Mechanics with Some Management Implications," J. Wildl. Manage. 36(2):369-382. (9) Linhart, Samuel B., and Weldon B. Robinson. 1972. "Some Relative Carnivore Densities in Areas Under Sustained Coyote Control," J. Mammalogy 53(4):88l-84. (10) Linhart, Samuel B., and others. 1968. "Field Evaluation of Antifertility Agent, Stilbesterol, for Inhibiting Coyote Reproduction." Trans. N. Am. Wildl. Nat. Resour. Conf. 33:316-27. (11) McDonald, John H. 1973. Coping with Bivironmental Complexity: A Computer Simulation Metnoaoio"W: waster's tnesis, Dept. hydroi. water hesour., univ. Ariz., Tiicson. (12) Metfessel, Milton. 197^. "A Proposal for Quantitative Reporting of Comparative Judgments," J. Psvch. 24:229-35.

23 (13) Muñoz, John R. 1977. Causes of Sheep Mortality at the Cook Ranch. Florence, Montana WCD-'it). Master's tnesis, Dept. wiidi. biol., uni v. Montana, Missoula. (14) Robinson, Weldon B. 1953. "Population Trends of Predators and Fur Animls in IO8O Station Areas," J. Mammalogy 34(2):220-27. (15) Robinson, Weldon B. 1961. "Population Changes of Carnivores in Some Coyote-Control Areas," J. Mammalogy 42(4):510-55. (16) Stevens, S. S. 1946. "On the Theory of Scales of Measurement," Sei. 103:677-680. (17) Stevens, S. S. 1966. "A Metric for the Social Consensus," Sei. 151:530-541. (18) Stuby, Richard G., Eldwin H. Carpenter, and Ixiuise M. Arthur. 1978. Public Attitudes Toward Predator Control. U.S. Dept. Agr., Econ. ¿stat. uoop. berv., uortncommg;. (19) Technical Committee. 1974. Water Resources Planning, Social Goals, and Indicators: Methodological Development and mpiricai ïesTi FKWG TTI-T; Utah water ties. Lao., Utah otate um v., Logan. (20) U.S. Department of the Interior. 1972. Predator Control in Transition. Final Rpt., Bur. Sport Fisheries Wildl., Div.—Wildl. Serv.

24 APPENDIX A: œNTROL METHODS AND 1974 EXPENDITURES

Federal Control Efforts The extent of Federal predator control in each State, based on 1974 U.S. ^Fish and Wildlife Service (FWS) annual State reports, is reported in Appendix table 1. 12/ Service funds represent appropriated Federal dollars distributed bv FWS. ^Cooperative funds supplement service funds and are comprised of (1; State funds, (2) county funds, (3) Federal funds other than those from FWS, and (4) private organization funds. FWS does not record expenditures for each control method used, but does separate some aerial expenditures (airplane or helicopter rentals) from expenditures for all other methods. Appendix table 2 shows the amount spent on aerial gunning J3/ and all other methods combined, as well as the number of coyotes taken in"Tnese two categories and the calculated expenditures per coyote. Expenditures in each State per coyote killed range from $35 to $100 for aerial gunning and Troui $49 to $384 from all other methods combined. More than 71,000 coyotes were taken during fiscal year 1974 by various FWS programs (App. table 3). Traps accounted for half the coyotes taken; aerial gunning was second. Traps took the highest number of coyotes in 9 of the States, while aerial gunning took xhe highest number in 5 of the 13 States using aerial gunning. Arizona and Washington did not use aerial gunning during fiscal year 1974; there were no Federal control programs in Colorado or Kansas.

Private Control Efforts Estimates of coyotes taken and control costs by private efforts are made by expanding data obtained from 888 ranchers in various western sheep producing areas who responded to a 1974 cost of production survey. 14/ Coyotes taken bv sport hunting or for fur purposes are not included in these estimates. Cost estimates for private control programs indicate that the largest expenditures were for aerial gunning, trapping, and ground shooting (App. tables 4 and ^). Expenditures per coyote were highest for trapping, followed by aerial gunning

12/ These reports do not present predator control expenditures in terms of ïïôllars spent for controlling various kinds of predators, but in terms of dollars spent protecting resource categories (crops, livestock, urban facilities, and others). Ihe livestock category mDst nearly represents the funds used for predator control only. Thus, this figure is used in place of total dollars spent for State animal damage control programs. 13/ FWS gunning expenditures by States do not include associated expenses, such as salaries, travel, and equipment, but these expenses are estimated in a supplemental report (20). Using the costs .of manpower, equipment,^ and supplies for all Stâïïes using aerial gunning, an average hourly cost was estimated to be $29.27 per hour. This cost was added to the iiourly cost for aerial gunning reported in the 1974 State annual reports (App. ^table 2). Expenditures For all other methods are calculated by subtracting the expenditures for aerial gunning from the total expenditures for predator control listed in Appendix table 1. 14/ A total of 888 of 5,734 ranchers responded to a questionnaire on predator control methods, but only a very few responded to the cost and take items for each control method. Data collected give some indication of cost and take for the various control techniques, but should be interpreted with caution due to the limited response rate. Additional details concerning these surveys and copies of the questionnaire may be obtained by writing to Environmental Economic Studies. Economics, Statistics, and Cooperatives Service, Room 420, GHI Building, U.S. Dept. Agr., Washington, D.C. 20250.

25 Appendix table 1—Expenditures for predator control in 17 Western States, 1974

: Service : Cooperative state funds \j : funds \j Total

1,000 dollars

Arizona ; 75 35 110 California 189 638 827 Colorado 0 0 0 Idaho 217 148 366 Kansas : 0 0 0

Montana 217 213 430 Nebraska : 35 57 92 Nevada 181 344 525 New Mexico 248 204 452 North Dakota 38 85 123

Oklahoma 147 188 335 Oregon 170 288 458 South Dakota 79 106 185 Texas : 300 1,268 1,568 Utah 183 138 321

Washington 46 71 117 Wyoming 234 401 635 Total : 2,359 4,180 6,538

\l See discussion, p. 25.

Source: U.S. Fish and Wildlife Service

26 Appendix table 2—Federal expenditures for aerial gunning and all other methods in 17 Western States,1974

Aerial gunning All other methods : Coyotes : Cost per ; j Coyotes : Cost per State : Total cost : taken : coyote : Total cost : taken : coyote

: 1,000 1,000 : dollars Number Dollars dollars Number Dollars

Arizona : 0 0 0 110 997 110 California : 12 251 49 815 7,258 112 Colorado \J : 0 0 0 0 0 0 Idaho : 157 3,341 47 208 2,151 97 Kansas 1/ : 0 0 0 0 0 0

Montana : 169 2,883 59 261 1,678 155 Nebraska : 11 244 46 81 661 122 Nevada 201 2,480 81 324 1,303 249 New Mexico 67 798 84 385 4,667 82 North Dakota : 68 810 84 55 142 384

Oklahoma ; 51 1,194 43 284 3,932 72 Oregon ; 65 1,723 38 393 4,961 79 South Dakota : 41 377 108 144 926 155 Texas : 109 3,101 35 1,459 13,130 111 Utah : 93 1,784 52 228 1,522 150

Washington : 0 0 0 117 2,383 49 Wyoming : 181 2,587 70 454 2,535 179 Total : 1,225 21,573 57 5,317 48,246 110

\J No Federal programs existed.

Source: U.S. Fish and Wildlife Service Appendix table 3—Coyotes taken under Federal programs in 17 Western States, by method, 1974

: Aerial : Ground State : Total ; Trapping : gunning : shooting : Denning : Snaring : Dogs

: Number Percent

Arizona : 997 86 0 11 3 0 0 California : 7,509 80 3 8 7 0 0 Colorado 1/ : 0 0 0 0 0 0 0 Idaho : 5,492 34 61 4 1 0 0 Kansas \l : 0 0 0 0 0 0 0

Montana : 4,561 24 63 6 6 0 0 Nebraska ! 905 40 27 17 15 0 0 Nevada 5,066 35 49 0 4 0 2 00 New Mexico ! 5,465 75 14 2 3 5 0 North Dakota '• 952 3 85 0 8 0 0

Oklahoma '' 5,126 45 23 16 13 2 0 Oregon • 6,684 55 26 9 8 1 0 South Dakota • 1,303 31 29 13 24 2 0 Texas '• 16,233 58 19 5 4 14 0 Utah : 3,306 15 54 13 17 0 0

Washington • 2,383 83 0 8 10 0 0 Wyoming • 5,122 18 50 11 19 0 0 Total • 71,104 50 30 8 8 4 0

\J No Federal programs existed.

Source: U.S. Fish and Wildlife Service. Appendix table 4—Estimated private control expenditures, by method, 1974

: : Aerial : Aerial : Ground : Toxicants Region 1/ Trapping : Denning : (planes) ; (helicopter)¡shooting : and M-44 Other Total

1,000 dollars

Mountain 74.6 19.6 50.1 213.4 34.0 28.4 2.3 422.4 Plains 1.2 1.2 0.0 0.0 4.3 0.0 7.5 14.2 Northern Plains 23.6 2.0 9.6 6.9 26.8 3.7 5.8 78.4 Texas/New Mexico 87.0 0,0 2.5 0.2 31.4 17.7 5.7 144.5 Pacific Coast 8.4 0.0 0.0 0.0 0.0 2.2 0.1 10.7 California/Arizona 5.1 0.3 0.7 0.0 21.8 0.0 7.7 35.6 Great Basin 32.1 0.0 13.7 34.8 5.3 0.0 3.0 88.9 Total 232.0 23.1 76.6 255.3 123.6 52.0 32.1 794.7

Dolla rs

Costs per coyote 39.0 16.0 12.0 29.0 11.0 10.0 17.0 19.0

Source: Estimates based on data from a sample of 888 sheep producers. Appendix table 5—Estimated number of coyotes taken by private efforts, by method, 1974

Aerial ; Aerial ; Ground ; Toxicants ; Region ; Trapping : Denning (planes) :(helicopter);shooting ; and M-44 ; Other : Total

Number

Mountain 2,534 559 4,315 7,044 1,834 4,631 452 21,369 U) Plains 243 24 0 0 911 131 101 1,410 o Northern Plains 471 810 1,007 695 2,081 91 356 5,510 Texas/New Mexico : 1,350 0 100 356 3,339 484 242 5,871 Pacific Coast 339 0 0 0 35 134 28 536 California/Arizona : 342 10 143 36 2,794 0 702 4,027 Great Basin : 711 40 802 778 371 0 0 2,702 Total ; 5,990 1,443 6,367 8,909 11,365 5,471 1,881 41,426

Source; Estimates based on data from a sample of 888 sheep producers. from helicopters. 1¿/ Expenditures per coyote were relatively low for private gunning fV»om planes, toxicants, and ground shooting.

Total Control Efforts In 1974, more money—Federal and private—was spent on trapping than on any other method, followed by aerial gunning and ground shooting Upp..^..^^.. table 6). Little use was made of 1080 toxicants and M-44 due to severe restrictionsrestric. or^. bans on their use since 1972. Control expenditures and coyotes taken were highest in Texas, California, Wyoming, Montana, and Nevada.

J¿/ Private expenditures, in contrast to Federal expenditures, include no overhead expenses.

31 Appendix table 6—Estimated costs of coyote control and number of coyotes killed by control method and State, 1974 1/ : : Aerial : Ground : 108C . Trapping : Denning ; gunning : shooting ; toxicants : Snaring : M-44 : Total State : Cost : Kill Î nnsf ' If-ill ^ii± i^.i±± : Cost : Kill : Cost : Kill : 1,000 1,00C 1,000 1,000 1,000 1,000 1,000 1,000 : dollars No. dollars No. dollars No. dollars No. dollars No. dollars No. dollars No. dollars No. Arizona : 87 968 3 32 0 0 27 973 0 0 0 0 2 218 119 2,191 California : 698 6,454 55 528 19 251 136 816 12 67 2 22 44 498 987 8,636 Colorado 2/ 15 470 4 99 0 0 7 46 6 806 0 0 2 93 34 1 514 Idaho 182 2,281 9 134 209 3,341 45 654 4 510 2 20 1 50 452 6,990 Kansas Ij 1 69 7 0 0 2 258 0 37 0 0 3 29 6 400 Montana 168 1,849 41 794 228 2,883 96 1,712 7 1 ,007 2 18 3 271 545 8 534 Nebraska 38 424 14 180 14 244 35 412 - 28 1 8 1 34 102 1 330 Nevada 196 2,055 23 237 222 2,480 100 892 0 0 0 0 1 0 542 S (\f\L New Mexico 361 4,743 14 90 68 789 35 1,528 7 458 22 276 2 119 509 8,003 North Dakota : 10 92 25 206 68 810 21 269 0 33 0 0 2 27 126 1,437 Oregon : 272 4,195 39 586 85 1,723 86 943 3 381 6 92 1 45 492 7 965 South Dakota : 60 557 42 535 46 377 54 839 1 40 4 31 2 107 209 2,486 Texas : 1,000 10,228 63 606 111 3,101 196 2,906 11 301 238 2 ,280 3 149 622 19,571 Utah : 77 1,068 70 672 151 1,784 109 960 6 770 0 0 1 75 414 5,329 Wyoming : 152 1,585 150 1,269 234 2,587 178 1,372 6 1 ,039 3 24 2 166 725 8,042 Total : 3,317 37,038 552 5,975 1,454 20.370 1.127 14,580 63 5 .477 281 2 ,771 70 1 ,881 6 ,883 88,092

1/ Includes Federal and private control efforts. Does not include expenditures and kill data for sport and fur hunting and trapping. Allocation of Federal expenditures among other (non-aerial) methods was done in proportion to coyotes taken.

1] No Federal programs. Information for State is based only on ranchers' reports and may therefore be incomplete.

Source: Fish and Wildlife Service 1974 annual report and expanded survey responses from 911 sheep producers. APPENDIX B: TECHNICAL ASPECTS OF THE SIMULATION

This section describes in detail the conceptual and empirical basis for estimating the socio-environmental and economic impacts of coyote control, and some additional technical aspects of particular submodels-

Socio-environmental Evaluation Methodology The socio-environmental concerns treated in the simulation model are organized as illustrated at the top of Appendix figure 1. Control method acceptability is composed of the humaneness, cost effectiveness, and selectivity of the methods. Selectivity (prevention of accidental.deaths of other animalsj.is. in turn, composed of selectivity regarding domestic animals,anim^j.^, other wildlife, and those coyotes not presently killing sheep. This hierarchy of socio-environmental impacts is only part of a larger hierarchy^ consisting of these impacts, measureable aspects of the environment related to these impacts (social indicators), and decision variables related to feasible alternative policies (App. fig. 1). The perceptible impacts in the hierarchy, such as humaneness, are aggregated until one general measure, termed socio-environmental quality, is obtained. To measure this abstract concept in quantitative terms, first the lowest leyel impacts are quantified by linking them to easily quantifiable social indicators. The lowest level impacts can then be aggregated into a general measure of socio-environmental impact.

Appendix figure 1—Hierarchy of socio-environmental impacts^ social indi calore^ and decision variables

Coyote population effects on other wildlife populations

Decision variables 'for policy alternatives'

33 For example, a decision tx) lift the ban on interstate shipment and Federal use P/h..'^9î^^^^^ ^^^. control purposes would result in increased use of 1080 and M-44 (decision variables) and, in^turn, in a larger number of coyotes killed for a given level of expenditures (social indicator). The public would perceive this as a change in control cost effectiveness (related ^SS^^r^"^^^^™?'^*^^.^"^?^^*^ category). If the importance of the change in cost effectiveness IS weighted relative to the importance of accompanying chances in perceived humaneness and selectivity of control methods, an overall measure or acceptability of the control method mix can be achieved. The steps involved in this quantitative evaluation of socio-environmental impacts , of coyote control include: (1) developing a list of socio-^nvironmentai.impacts relating to coyote control, (2; measuring public preferences for various impacts, (3; identifying social indicators (measurable aspects of the physical or social environment expected to be affected by control plans; and measuring their effects on the impacts listed in step one, and (4) measuring the effects of alternative control policies on social indicators.

Developing Lists of Socio-environmental Impacts Detailed, guidelines for developing these impact categories are provided by the Technical. Committee for Water Resources Planning (J1). The hierarchy of socio-environrnental impacts for coyote control were presented in Appendix iigure 1. In general, disaggregation of abstract impacts into more concrete and perceptible impacts should be reasonably exhaustive, although a lower bound should be established at a point at which impact components are still meaningful to.and perceptible by the public. Components of one primary impact should be as independent and mutually exclusive as possible.

Measuring Public Preferences After a set of socio-environmental impacts capable of being perceived and evaluated, by the public is defined, these impacts can be evaluated quantitatively. This step requires intensive public involvement, for the relative values of various impacts depend on their relative importance to society. The most valid and reliable source of a society's evaluation can be provided only by a representative sample of the relevant societal group(s). The social group of interest in the present study included inhabitants or the 40 contiguous States. A stratified random sample of households was selected from, all States and the District of Columbia, a universe of approximately 54 million households. Within each household, respondents were selected on the basis of age (18 or older) and sex. The response rate was 78 percent, resulting in 2,041 respondents. However, all respondents were not asked the questions used m the evaluation or socio-environmental impacts of control. Respondents who had not heard of the coyote control issue or did not think it important were not asked specific questions concerning control. They did, however, provide information on their attitudes on more general wildlife issues related to the control issue (18). Approximately 3^ percent of the respondents did not answer the questions specifically addressing the social and socio-environmental goals related to coyote control.16/ Those questions directly related to socio-environmental impacts of coyote control were formulated to provide ratio measures of preference so that items within each level of the hierarchy (App. fig. 1) could be compared directly and aggregated. The measurement technique selected was Metfessel's General Allocation Test (12), which requires that 100 points be distributed among the impacts at each level of the hierarchy. For example, the tradeoffs among

J6/ Additional details and copies of the questionnaire may be obtained by writing to Environmental Economic Studies, Economics, Statistics, and Cooperatives Service, Room 420, GHI Building, U.S. Dept. Agr., Washington, D.C. 20250.

34 components of control method acceptability—selectiyitY, humaneness, and cost effectiveness—were measured by asking respondents first to rank the three components for importance, then to distribute a^total of 100 points among them to reflect the relative importance of each. Four hypothetical respondents could distribute 100 points among the components of metiiod acceptability m the following manner:

"CómponenT Hvpotneticai responden^

Points Cost effectiveness 0 5"D 60 Specificity 0 50 20 Humneness^ 100 0 20 37

The resulting point allocations provide cardinally scaled estimates of preferences For these impact categories. Point allocations for , all socio-environmental impacts by all respondents to the salient items are listed in Appendix table 7.

Identifying and Measuring Social Indicators Although 67 percent of the respondents indicated their preferences for the impacts of coyote control, even these respondents may not have been familiar enough with specific control alternatives to judge the impacts .of a particular policy. In addition, it would not be feasible to ask the public to evaluate a nearly infinite array of alternative policies in order to anticipate all the information ttiat might be desired by decisionmakers. Thus, the public was asked only to provide preferences for the general impact categories. They were asked, for instance, to rate their preferences for the cost.effectiveness of coyote control efforts relative to the humaneness and selectivity, but not to rate preferences for a certain level of spending allocated among, a specified set of control methods. These technical measures—social indicators— were related to public evaluations of impacts by mathematical equations. In some instances the public may be able to judge the.relationship of a social indicator to a socio-environmental impact category; in other cases, experts may be able to derive the information from a synthesis of public judgments and technical measures (J). Whatever the case, descriptions of relationships between indicators and subgoals are crucial to the use of social indicators as links between policies and goals.

Synthesis Public preference weights (allocated points) assigned to socioTenvironmental impacts and information concerning the effect of policy alternatives on social indicators can be synthesized to predict public responses, to specific coyote control alternatives. The synthesis is accomplished by means or a mathematical preference function: P. * nx. i Pi = preference for an alternative; Xi = measure of impact of an alternative on a social indicator and its related impact category: and ei = the proportion of points allocated to a subgoal. Jl/ Pj can then be used as the measure of impact (Xi) for more abstract impact categories. For instance, when the preference value for method selectivity or a given alternative is determined from the public values assigned to selectivity regarding domestic

17/ Constant elasticity over the range of variable Xi is assumed (_16, Jl) ; that is, the sum of the exponents is 1.00.

35 Appendix table 7—Relative importance of socio-environmental goals as viewed by the national sample 1/

Goals : Mean points allocated

: Points

Primary goals: Coyote population impacts on : 46 other wildlife Method acceptability 30 Extent of control : 24 Total : 100

Acceptability subgoals: : Cost effectiveness : 16 Specificity : 32 Humaneness : 52 Total : 100

Specificity subgoals: : Impact on domestic animals : 42 Impact on coyotes not causing losses : 20 Impact on other wildlife : 38 Total : 100

1/ See App. fig. 1 for goal structure.

Source: National survey; see text footnote 16,

36 aniriBls, other wild animals, and coyotes not causing losses, this preference value can in turn be used in conjunction with the humaneness and cost effectiveness variables to determine that alternative's method mix acceptability, the next highest impact (App. fig. 1).

Cross Impact Simulation of Control/Covote/Lamb Loss Relationships This section contains more details on hew these variables were simulated in the submodels 8-10, 12, and 13-

Submodel 8: Standardization of Control First, expenditures per method are transformed into standard measures of corrective and preventive control. The transformation i"equji;^s ^J^ determination of the proportional^use. and suitability of a method for corrective and preventive control. This is expressed in dollar terms: for preventive control, the dollars reflect cost for a given level of control if only aerial gunning were used, and for corrective control if only steel leghold traps were used, f'or example, a $1,000 expenditure for the M.44 is considered by the model to provide control equivalent to that resulting from a $500 expenditure on traps for corrective control plus a $1,500 expenditure on aerial gunning for preventive control.

Submodel 9: Yearly Impact on Coyotes Next, corrective and preventive dollar equivalents are transformed ^into the percentages of coyote populations killed each year by coyote control. Ine transformation requires two steps: (1) Corrective and preventive dollars are transformed into indices of corrective and preventive control efforts. The transformation is based on the assumption that the 1974 levels of corrective and preventive control were each 40 percent of maximum control, maximum control being that level of control at which sheep losses would be zero, or at which expenditures on control could not be justified on economic or environmental grounds. The ratio oi corrective dollars for an alternative to corrective expenditures in 1974 was adjusted by a nonlinear function to reflect decreasing.effectiveness of increased expenditures for control. The index of preventive control was determined in a similar manner. (2) Next, the two indices of preventive and corrective control are.transformed into a percent of the coyote population killed. The assumed relationship, is: Percentage killed = (.20 * index of corrective) + (.50 * index of preventive). The specific coefficients are determined partially by loçic ^(that is, corrective control kills fewer coyotes than preventive control; and partially from trial and error runs of the entire model. This submodel is calibrated so the entire model coincides with observed real world events. Howeyer, no statistical tests were made to check the validity, as sufficient data in the proper form for calibration of the model do not exist.

Submodel 10: Coyote Population Impacts Submodel 10 relates the impacts of control level (in.coyotes killed) on the coyote population level. The resulting population leyel is expressed in percentage of the coyotes remaining. This calculation requires simulating the coyote population's response to various control death rates. The following variables are used to describe the structure and dynamics of a coyote population: percentage use of carrying capacity^ (PCC), percentage of yearling females producing a litter (YF^,^ percentage of older females producing a litter (OF), average litter size^ (ALS)., deaths, due to natural causes TNDR), deaths due to predator control, age distribution, and number of surviving coyotes.

37 The ipdel begins,with a^ population level of 1,000 coyotes. This coyote population utilizes .100 percent.of the carrying capacity of the habitat and has.reached equilibrium; ^that is, m the absence or predator control activities, the number of births per year equals the number of deaths per year. The.percentage use of .carrying capacity (PCC) is used to determine the variation of three variables: percentage of yearlings producing a litter (YF), percentage of older females producing a litter {(F), and average litter size (ALS). PCC for each time period was calculated as follows: FCC = total number of surviving coyotes Unitiai population ol 1,000; The percentages of yearlings producing a litter was modeled by the function: YF = 77.2 - 0.722 * FCC, (10.0 < FCC < 100.0) The percentage of older females producing a litter was modeled by: OF = 94.4 - 0.44 * FCC, (10.0 < FCC < 100.0) Thus, YF and OF vary inversely with.FCC and rangrange from 5 to 75 percent and 50 to 90_ îrcent, respectively (App. figs. 2 and 3). Over a 17 year period. GierV. ^ I^J.^.^—¿,- ,observed ^'^ that 10•- to-- ^<55 r'-:^,"percent of-: the yearlings,and 60 to 83 percent of the older females were capable of breeding. Knowlton Co) found 48 to 81 percent of all females capable of breeding and calculated a range of 32 to 91 percent based on data presented by Linhart, et al. (10). Knowlton also reports, that, on the average, 92 percent of those ^capable of breeding (ovulating) also implanted embryos. Because these observed ranges may not represent the absolute extremes, they were slightly expanded for the model. Due to the lack of more specific research on coyote population' mechanics, the simplest forms (linear) for these relationships are used.

Appendix figure 2—Percent yearling females whelping

100 Percent of maximum population

Source Expert Judgment

38 Appendix figure 3—Percent older females whelping

188

o £

0 c Ü L

188 Percent of moxlmum population

Sour ce: Exper t J udomen t

Average litter size (ALS), has been observed by Gier (6) to range from 4.5 to ^•7 PyP5 P^"" litter and to vary inversely with density. The function used to model ALS was: ALS = 8.666 - 0.666 * PCC, (10.0 < PCC < 100.0) Thus, ALS ranges from 2.0 to 8.0 (App. fig. 4). death rates, based on natural mortality only, and aœ distributionu ^ ^ ^S^^ V?^ ^^^"^^^ ^^^ uncontrolled coyote populations from data reported, by Knowlton (8). The model was run and death rates adjusted until the initial population of 1,000 remained stable. To reflect replacive mortality, the death rates were reduced as the PCC became lower. Appendix figures 5 and 6 show the natural death rates and the age distribution of the stable population, respectively. The total coyote population was calculated at the end of each quarter. First, PCC was used to determine the percentage of yearling and older females producing a litter and the^average litter size. Using the agç distribution data and the variables described above, the number of pups born in the first quarter was calculated. The population was then decreased on a quarterly basis by implementing the natural and predator control death rates. Predator control death rates were percentages^of the coyote population killed during each quarter of the year. At the end of the fourth quarter, a new PCC was calculated using the surviving population and the process was reiterated for a specified number of years.

Submodel 12: Control Impacts on Relative Losses This submodel relates the level of control in terms of corrective and

39 Appendix figuro 4—Averago litter size

9 N

L 9

Sour ce : Exper t J udgmen t

Appendix figure 5—Natural death rates

crD L Û. 9 O O c 0 L Q.

Sour ce : Exper t J udginen t

40 Appendix figure 6—Age distribution of stable population 25

0 0 c 0 L

Ag<

Source: Result from simulation model

preventive ccxsts to sheep and lamb losses. The basic method used is cross-impact analysis (J, JJ.). The simulation shows diminishing returns to both preventive and corrective control: thus a large increase in expenditures would be necessary to bring losses io near zero levels. See submodels 8-10 and 13 for details.

Submodel 13: Estimation of Lamb and Sheep Losses The index of lamb and sheep losses from submodel 12 is transformed into an estimate of losses by assuming the 197^ levels of lamb and sheep losses correspond to a loss index of .4. Ihe ratio of the estimated index to .4 is multiplied by the 1974 sheep and lamb losses to yield the loss estimates.

Impacts of Coyote Control on Other Wildlife The general relationship between control level, coyote populations, and other wildlife populations was quantified for submodel 11 in the following manner:

Degree oi controT uoyote popuiatiofT impact on wildlife" Percent of maximum Index All out 0 .50 Extensive 25 .25 Medial 50 0 Some .25 None 100 .50

41 Economic Impact Evaluation Methodology This sectiai describes .in. greater^ detail the portion of the system that calculates the net economic impacts of control alternatives, submodels 14, 15, and 17. Sheep and lamb losses are transformed into economic impacts on both sheep producers and consumers. The estimated total net economic impact is the sum or benefits to producers and consumers minus the costs of control.

Submodel 14: Economic Impact on Producers ??Sr,^^^^^^i^ impact of control on producers can be measured in terms of the difference between IQ74 returns and expected returns for that year under an alternative control policy. One approach to determining this value is to use partial budgeting, based on. cost and returns data for individual ranchers, yniortunately, this approach ignores the impact on lamb prices of an increase in the quantity of lamb produced. Returns to the industry should be based on the actual 197^ price and quantity of lamb and on the price and quantity that would result if no lambs were lost to coyotes. To calculate the portion of industry gross returns lost or gained as a result of sheep and lamb losses to coyotes, the price and quantity of lamb marketed under, the alternative .control policy must be estimated. First, the demand function (schedule of prices at various quantities) for lamb was derived.

Demand Analysis 18/ The data for the demand analysis consists of quarterly observations for 1958 through 1974 for the following variables: p = seasonal average retail price of lamb; ql = seasonal average per capita consumption of lamb and mutton; qb = seasonal average per capita consumption of beef; qp = seasonal average per capita consumption of pork; i = per capita personal disposable income; and t = time. The regression equation selected was: p = 65.5 -20.8 * ql -1.29 * qb -.71 * qp + 0.12 * i -.50 * t * ql (3.1) (2.4) (3.2) ^"^ (29.3) (4.7) with a coefficient of determination of .99 and an F ratio (for the test that all coefficients equal zero) of 1170. F-tests for all variables included in the model are significant at the 95 percent confidence level. To develop an estimate of the 1974 demand for lamb, the average 1974 values for qb, qp, i, and t for 1974 are substituted into the estimated demand equatioi. The resulting estimate of the demand is: p = 169. - 53.8 ^ ql at the average 1974 quantity, yielding a price flexibility of .17. Therefore, a 1 percent change in the quantity of lamb produced only results in a .17,percent change in the price of lamb. The estimated demand function for 1974 IS Illustrated in Appendix figure 7. The difference between retail and farm prices is $1.05 per pound, resulting in an average farm price of $0.37 per pound. The 1974 production of lamb of 942 million pounds results in a gross return to the industry of $350 million. These data can he used as a base by which to measure increases or decreases in gross returns to producers for any control alternative. If, for example, the control alternative to be evaluated were to result in no losses to coyotes in 1974 (clearly not a feasible solution, but one which

18/ The demand analysis was provided by Kerry Gee and Muhammed Usman, Dept. oT Economics, Colorado State University.

42 Appendix figure 7—Consumer demand for lomb^ 1974

2

0» Ü La p74

• P* B

q74 q* 1 —I 1 1 r T T r 0.0 0.5 I .0 Pound« per capita p^r quarter

Source: Regrese i on ana I ye Ie

illustrates the lower bound of the analysis), the production of lamb in the West would increase from 615 million pounds to 709 million pounds and the price for lamb would fall $0.01 per pound to $0.36 per pound (assuming a constant farm to retail price margin). This would result in an increase in gross revenue of $29 million, from $204 million to $233 million. It should be noted that the lower price offsets 3 percent of the advantage to producers of lower losses and increased quantity produced. This change in gross revenue represents the maximum amount that industry p ofits could increase if coyote losses fall to zero. The actual value v juld bo somewhat less, due to the small but positive increase in costs as a reí jlt of growing and selling an increased quantity of lamb. The distribution of increased industry profits is important to individual Producers. Producers with zero losses to ooyotes in 1974 will not benefit rom the lowering of all other producers' losses to zero. In fact, their situations will worsen due to the $0.01 per pound decrease in the pnce of lamb. For producers with average and above average losses, increases in production would more than offset the reduction in general price level. Submodel 15: Economic Impact on Consumers The economic impact on consumers (diange in consumer surplus) is related to both changes in the price of lamb and in the quantity marketed. For example. Appendix figure 7 reveals that if the price of lamb falls from p74 to D* and the quantity marketed increases from q74 to q*, the area p74, p*, B, A reflects the savings to consumers ft'om lower prices and the area A, B, C reflects the tienefits to consumers due to the additional quantity being marketed (q* - q74). If a control alternative results in no losses to coyotes, consumers benefit from lower prices and increased quantity of lamb available, for a savings of $10 million.

43 Submodel 17: Net Economic Impact The calculation of net economic impacts is based on the changes from the 1974 conditions in producer plus consumer surplus minus any increased control expenditures for the alternative cr plus any savings due to decreased control expenditure. The net economic impacts can be considered as the relevant measure of the economic impacts to society in general due to changes in coyote control strategies.

44 APPENDIX C: USER'S GUIDE TO THE COYOTE CONTRCL INFORMATION SYSTEM

This appendix describes a computerized information system based on the simulation model reported in this publication. The system provides decisionmakers instant and efficient access to information on the impacts of any control policy alternative. The Coyote Control Information System (CCIS) package involves a central omputer, remote terminals, a data base, and analytical models (App. fig. §I). A decisionmaker desiring information from the system selects an alternative coyote control policy (expenditures for any combination of methods) and a system operator types the information into a computer terminal. The information is transmitted over a phone line to the conputer, which uses the data base and analytical models to generate information on the predicted impacts of the selected policy. This information is then transmitted back to the terminal and displayed, usually within seconds of the request. The system is controlled by a simple vocabulary of terms related to the type of information the decisionmaker wishes to input or retrieve. The possibilities include: Command Result DECI - Asks for user input of expenditures by method. LOSS - Outputs predicted loss level for selected alternative. COYS - Outputs predicted coyote population index for selected alternative.

Appendix figure 8—Coyote Control Information System CCCIS)

Time-sharing Data base computer

Analytical models

Graphic display terminal Information I system operator Decision maker

45 ENVR - Outputs measures of socio-environmental impacts of selected alternative. ECON - Outputs measures of economic impacts of alternative. TRADE - Provides graphic summary of results of all alternatives considered during a session. A summary table ^of predicted impacts of an alternative—socio-environmental index, acceptability of methods, coyote population level, percentage lamb and sheep losses, and producer and consumer surpluses—is automatically provided. To use the system, an cperator inputs DECI to the terminal and the computer responds with a request for expenditures for each control method (App. fig. 9;. Feasible methods of controlling or.reducing sheep and lamb losses due to coyote prédation in 1974 included in the system are trapping, denning, snaring, aerial gunning, ground shooting, 1080 toxicant, and M-44. A decisionmaker can either alter the mix of control methods or evaluate a policy that retains the 1974 mix, but changes the general level of control expenditures. After the user decides how much to spend on each type of control, an operator inputs expenditures and the computer calculates, stores, and displays.a summary or predicted impacts of the decision (App. fig. 10). If the decxsionmaker desires more information than is presented in the summary reportj he,then tells the operator which predictions he desires to review: economic impacts (ECON), loss levels (LOSS), coyote population levels (COYS), or socio-environmental impacts (ENVR). The appropriate command is typed into the computer terminal and the results are displayed (App. figs. 11, 12, and ' 5) •

Appendix figure 9—CCIS decision input

PLEASE INPUT YOUR DECISION FOR

ALTERNATIVE PREDATOR CONTROL LEVELS

THE 1974 LEVELS IN THOUSANDS OF DOLLARS SPENT AND IN NUMBER OF COYOTES KILLED ARE $1000 TAKE $1000

Irappi ng 3349. 37038 NEW VALUE» 1000 Dttnn i ng 552 5975 NEW VALUE« 500 Amr1 a 1 gunn i ng 1453. 29370 NEW VALUE- 3000 Ground shooting 1 126. 14580 NEW VALUE= 1000 Tox i canls 62. 5477 NEW VALUE» 0 Snar1ng 281 . 2771 NEW VALUE= 200 M-44 69. 1881 NEW VALUE= 0

A6 Appendix figure 10—CCIS summary report

Economic impacts $1^000 change Producer impact -4992. Consumer impact -1581. Net sconomic impact —5122.

Env i ronmentaI i mpacts i ndex porcontago chango Environmental index 66. 4. Metl^od acceptob i I i ty 45. 13. Coyote population index 121 . -! .

Other impacts number p«rcttniag« chang« Number of Iambs Iost 848492. 17. Coyotes killed 87927. -21 .

Appendix figure 11—CCIS economic report CHANGE IN CONSUMER AND PRODUCER SURPLUS FROM 1974 LEVELS IMPACT ON PRODUCERS -5. IMPACT ON CONSUMERS -2. PRICE INDEX 1Ö0.29

THE COST OF PREDATOR CONTROL ALTERNATIVE SELECTED 6 1974 LEVEL 7

NET BENEFIT OF CHANGE -5

47 Appendix figure 12—CCIS loss report

THE 1974 LOSSES AND THE ESTIMATED LOSSES FOR THE ALTERNATIVE ARE

1974 LOSSES ESTIMATED LOSSES NUMBER PERCENT NUMBER PERCENT

LAMBS 728200. 8. 14 848492 9.48

SHEEP 230400, 2.54 268460, 2.96

Appendix figure 13—CCIS socio-envIronmentaI report

Soc i o—env i ronmentaI qua I i ty

°" ' ~-. Coyol© population impacts on other wildlife 83 0Metl-)od acceptabi I i ty

Coyotes remaining 67 Cost effectiveness

Spec i f i cIty 71 Humanen Dornest i c an i ma I ®(\ 96) / Coyotesri not 1 causing losses1 25) Otl^er wild! If« 89

48

« U. S. GOVERNMENT PRINTING OFFICE : 1978 261-496/111 The decisionnaker can repeat this process until he has viewed the results of the entire set of alternatives he wishes to consider. At this point, the system's tradeoff option can be used to compare a wide range of impacts of alternatives. For example, to view the relative merits of the alternatives selected with respect to botii socio-environmental and economic impacts, the operator would type in the command for the tradeoff option, TRADE, followed by the appropriate number codes for the predicted impacts. \^/ The results are displayed in graphic form with the numerical labels on each point representing the order in which the alternatives had been selected (App. fig. 14;. The numerical labels of alternatives are also included on the summary tables. From the information presented in the graph, the decisionmaker can eliminate less efficient options. For example, in Appendix figure 14 point 6 has both Içwer economic and socio-environmental desirability than point 3 and should be eliminated from further consideration, unless there are strong institutional and/or political reasons for further consideration of this point. At this point, the decisionmaker can review more detailed tradeoffs (humaneness versus cost effectiveness, for example), specify new alternatives to be evaluated, alter model assumptions and/or calculations, or end the session.

Appendix figure 14—Tradeoff display

(0 L 0

0

c 0

-25 T 1 1 1 1 1 1 1 1 1 1 1 1 1 \ 1 1 T 50 55 60 65 70 Socio-environmental index

19,,/ ^^f^Sí^í^t—.^^...^w*^. ion ^categoriesv,^yv.e,>^.._a.v.o forx^i tradeoffuiQupv^ii options;yuvxywo. (1) environmental quali ^I) coyote population impacts on other wildlife, (3) method aoœDtabilit covotes remaining, (5) cost effectiveness, (6) selectivity, (T) human....^^,S) (ö; selectivity regarding domestic animals, (9) selectivity regarding coyotes not causing losses, (10) selectivity regard;ng other wildlife, til) nunter of lambs lost, (12) number of sheep lost, (13) percentage of lambs lost, (14) percentage of sheep lost, (15) value of lambs lost. (16) value of sheep lost' (17) producer surplus, (l8) consumer surplus, (1$) price index, (20) cost oí" control, (21) coyotes taken, (22) coyotes taken correctively, (¿3) coyotes taken preventively, (24) percentage of coyotes killed, 25) coyote population index, (26) number of coyotes, iZjl number of pups born, (28) number of control kills, (29) empty, and (30) net benefits, . v ^

49