| [ HOME | Forest Faunal Systems Home | Table of Contents | Gamma Home | The Finder | Glossary ] |
Inputs and an Alternative Research Paradigm
Introduction
Input is a category of a general system. Inputs come from other subsystems or from the same subsystem in a prior state. Inputs include such things as information, material, skills, equipment, ideas or insights, data, energy, and labor. Some efficiency is gained each day when workers decide before going afield what inputs they will need for the particular system with which they will work during the day. It is commonsense to think through what things will be needed; the systems approach is commonsense.
Some systems people prefer to think of all inputs as information; information is used synonymously with inputs. They make the analogy of inputs to electrical systems and follow the analogy of inputs as voltage. They are backed up by an abundant literature on information theory, on the relation of information to noise, and on control systems. Discussions of information by systems people studying cybernetics also justify using the concept of information in developing a related body of theory within environmental management.
 |
| It may be fun to study animals but eventually serious students of fauna, facing limited budgets, need to assign priorities to studies and be able to state likely uses of conclusions reached from such studies. |
One example of the flow of inputs is the quarterly report of the wildlifer. Suppose he or she writes 60 lines, passes them along to a supervisor who passes them along to the main office. The report is input to the main office, but when only 4 lines of the 60 line report are analyzed or used, the noise-to-information ratio (56:60) is very high. Noise is costly and often hides the message present in the information being sent.
Another example: a game checking station worker takes 10 minutes to ask 5 questions and takes 5 measurements from each harvested deer passing a station. If the answers to questions are never analyzed and if 8 of the 10 minutes are spent putting a deer on a scale when an equally useful weight can be obtained from a measurement in inches of the circumference of the body behind the front legs (which requires 30 seconds) (CAP411), then the inputs to the manager for characterizing the deer or hunters or making decisions are small indeed. The cost per useful piece of input is even greater. If a manager had decided beforehand what was to be done with the information (defined objectives for an information-collecting subsystem), and if the statistical processes had been decided, then he or she would know precisely which information was needed. Steps could then be taken to collect just enough of it (reduce noise), accurately, at a low cost per piece of information.
One stumbling block to comprehending inputs has been the seeming similarity of outputs and inputs. The problem lies in defining the context of a system. Often the output of one subsystem may become the input to another. For example, tree seed is input to a mouse. The healthy mouse is the results, the system output. The mouse, however, is input to a fox. Analytically, it is essential to separate the subsystem.
Modules
After the first lunar landing, "module" became a household word. Most people recognize modules as being major parts of something else but largely independent or able to stand alone. They are subsystems, entities from which larger structures are built. Modules fit together; they are designed to interact and to be easily attached. Modularity refers to how well they fit or operate together or can be exchanged.
Input systems are best when they are modular. If on a questionnaire there are three groups of data, a systems person would begin thinking about these groups as data modules and would probably be able to avoid collecting the same information over again on another form since one data module from the first form could be used with the second.
Physical system inputs such as tractors or trailers also are best when they are modular - when parts are interchangeable. Modular design of equipment, of data systems, and of computer programs permits changes to be made when desired, without having to start over again. Building with modules is an efficient and effective practice - whether with buildings or with data. Certain types of forest data (e.g., basal area) can effectively be treated as modules and used by different groups in diverse ways in models of wildlife as well as timber. Modules encourage imaginative combinations of inputs, often resulting in creative new processes, solutions, or outputs.
Computer programs or sets of instructions telling the computer what calculations to make or what to do with data, are often developed with modularity. Modular programs on timber production, water yields, and understory vegetation yields, all developed separately, may be united into forest or watershed analytical tools.
Efficient Data Collection
Trying to optimize data collection is a difficult problem about which there is a large literature. The intent in this section is only to encourage basic considerations in order to allow faunal system managers to understand some of the requirements for recording data, to encourage them in their information collecting and storing, and to expedite the analysis of data so it can be rapidly and accurately converted to information for decision making (Giles 1982).
It is difficult to write passively about data collection and to use such terms as"suggest"or"seems likely." The information age has arrived and computers are a reality. They work well (pay-check errors of less than 1 in 100,000 to the contrary). For faunal system managers to stay up with science and social systems and get ahead, they must use computers to deal with the scientific and managerial processes on the lands. To deny the above or resist its implications is both illogical and deadening. I have some sympathy for King Canute who tried to sweep back the tide from his coast with his broom, but no sympathy for analogous response by system managers.
It seems illogical to me for faunal system managers to complain about deficiencies or limitations like the inability to deal with the complexity of ecosystem management and not rejoice when a solution is presented. It also seems illogical to complain about human paper work, yet persist in it when machines can do the work much faster, more accurately, and at the same or less cost. To complain about not having enough time for field work when a superior office clerk, the computer, exists to do so much of the work does not match up well. Similarly, why should a manager fill in the same data on four forms each year when, after the first entry, a simple code would have this called from an electronic memory? Mismanagement claims are on the horizon when data costing thousands of dollars, often $25 per observation, are not analyzed or used. Consider a $15,000 master of science thesis research project, for example, that yields less than 200 field observations that are given a trivial statistical treatment or, worse, filed, because the researcher"could never get around to it." Problems abound, however, and funds are still expended for data collection to help make decisions - decisions made before the results of data analyses are available.
There are major needs for improved input subsystems for the individual. They are needed at the supervisory level to achieve the smallest possible ratio of data required to information extracted. They are also needed in headquarters to make information available when it is needed. It needs to be retrievable in a useable form. These needs can only be met by every person in a wildlife agency analyzing what information is needed (Giles 1979), what would be used if available, what would not be used or used rarely (and even attach some numerical probability to"rarely"), what is worth working for, what is now duplicated, and what can be done to minimize the cost per piece of used information.
There are information systems, and then there are management information systems. In the former class are trapping records and use analyses, the ecosystem descriptions. The management information system has no acceptable definition. Such a system is a major computer-based aid to decision making, one much more than a giant electronic memory. It is a system which can transform data into useful information for managerial decisions on request.
A management information system, being a system, is fundamentally a function of users' objectives. A wildlife agency without objectives should not attempt to develop a management information system (Buffington 1972). Dill (1962:29) said, "Decision-making is one of the major functions that administrators (or managers or executives) perform. It is accepted by many, in fact, as the central activity in management and as a key subject for attention in management training." Martino and Stein (1969:2) said that a management information system must circulate only essential information and channel information wherever and whenever it is needed. Essential and need are key words, for they imply the existence of objectives. Ackoff (1967) emphasized the need to understand the decision-making process in developing an efficient MIS. Martino and Stein said (1969:3):
There has always been a subjective quality to management. Experience, application of knowledge, business acumen and vision are factors that produce a"good manager." Several factors are somewhat vague; we do not know how to reduce them to a definite set of rules and principles. Probably we never will. But there are many decisions in business management which are directly dependent on facts and logical relationships. It is to this area of management that a study of decision patterns can be applied to produce a more scientific method for operating a business and to upgrade the performance of managers. Viewed in this light, such studies will significantly contribute to the managerial revolution. They will add a"new"function to the manager's job; they will change many decision making patterns; they will be another base to scientific management; and they will have a significant impact on the organization.
A management information system is dependent upon knowledge of decision theory, both descriptive, and normative. Information is needed to make decisions and unless this process is well known (or at least being vigorously studied), then inefficiency is likely. A management information system has all of the system components of objectives, inputs, processes, outputs, feedbacks, and feedforward.Buffington (1972) analyzed information needs of the U.S. Wildlife Refuge System, basing his analysis on decisions to be made. He developed a taxonomy of decisions (Table 6.1) which I believe to be widely useful.
 |
| Charles D. Buffington, PhD |
Table 6.1. Classification and descriptions useful in analyzing decisions made in faunal resource ssystems work (after Buffington 1972).
| Decision Level | Coding and Description |
|---|---|
| I. System Function | All relevant decisions made within the faunal systems are those directed toward achieving system objectives. All of these objectives can be classified as primarily directed toward a major system function (vis-a-vis general systems theory) of input, output- objective, process, feedforward, and feedback. Similarly, a decision can be classified as to how it fits into the systems framework of stated system objectives.
Input (1) - The decision provides inputs to the total system. Output-Objective (2) - A decision that identifies a desired output or establishes an objective. Process (3) - The decision is the process by which a stated objective is sought or reached. Feedback (4) - The decision will provide feedback for the system in evaluating, controlling, or improving system performance or actions. Feedforward (5) - The decision will adjust and adapt the present system for the predicted future. |
| II. Objective | The decision has or will affect single or multiple objectives of the system: Single Objective (1) Multiple Objective (2) |
| III. Frequency | The frequency with which a particular decision is made: Frequent (1) - Three or more times/year Infrequent (2) - Less than three times/year Once (3) - The only time this decision has been made |
| IV. Structure | Unstructured (1) - A unique decision, no previous experience, has a creative component, may use standard decision techniques but no identical decision is foreseen or has been made. Structured (2) - More programmed than the unstructured, the operating procedures are more orderly. Highly-Structured (3) - A decision of a repetitive nature, may have been made a number of times previously, solved with standard operating procedures |
| V. Occurrence | Regular Interval (1) - The interval of occurrence of this decision is predictable and at regular intervals. Irregular Interval (2) - The interval is not predictable, or if this is the only (first) time this decision appears, then the interval is irregular. |
| VI. Initial Influence | The domain of initial influence. (This does not include magnitude of influence or consequence of the decision having been made): Physical-Biological (1) Political-Socio-Economic (2) Both of the above (3) |
| VII. Long-term Influence | The domain of long-term influence. (This does not include magnitude of influence or consequence of the decision having been made: Physical-Biological (1) Political-Socio-Economic (2) Both of the above (3) |
| VIII. Magnitude of Impact | Effects over a reasonable number of years; an index of importance: Large (1) - May have a significant effect on the effective management of the system. Small (2) - Negligible effect on the system. |
| IX. Location | Where in the administrative structure of the system the decision is made: Central Office or Above (1) Regional Office (2) Supervisor (3) Local Manager (4) |
The taxonomic classes provide places for decisions, then places where levels of importance of each type of information may be assigned. Inputs can be evaluated on the basis of the number of decisions in which they are used, as well as the importance of the types of decisions.
Often managers will list "soil conditions" as information needed. It is evident that he or she has aggregated pH, top-soil depth, soil moisture, available nitrogen, etc. Information systems need to have the specific items stated that are used by the manager. If all are used, then they should be listed. Likely a half-dozen will be critical and then only two or three used in most decisions.
Some inputs overlap in their usefulness. Some are used in big and little decisions. Some inputs are essential to a decision; others only generally operative. Some are essential at one time; others always needed. Buffington (1972) showed a matrix approach to analyses using decisions ranked in order of the number of inputs required. This provided indices to frequencies of use. High value was implied for inputs with high frequencies of use. But inputs do not have equal cost! Buffington then used his taxonomy as a basis of estimating importance. If some one input item, say number 50, was used in 110 decisions, then, based on frequency, it would be judged very important. But if input item 50 was not used in a single decision classified among those of "Large Magnitude of Effect," it would not be very important and insignificant in most management information systems. Another approach is to study groupings of inputs. Since certain groups conspicuously operate together, lack of one can make the others useless. Aware that all decisions do not have equal priority and that ability to make the most decisions is not an objective, Buffington (1972) developed decision information priority based on the weighted importance of each type of decision. A management information system should be capable of aiding in making the most important decisions.
A management information system should not be developed to supply information for decisions that are made only once (unless they are of the magnitude of a space shot or involve great risk to hundreds of lives) (see Level III Table 6.1) or unstructured (Level IV). It is evident that computer search routines are needed to perform the above analyses. These are processors within the input subsystem. The wildlifer can thus employ these and other analytical tools, such as factor analysis, for improving the input subsystem.
A systems approach attempts to input to a management information system (or isomorphically, any system) only what is necessary for a particular output, to verify the need for such output, to assure costs of an information system do not exceed the benefits from its use, to assure information is available when needed but that time for retrieval (or time, from request to delivery) is realistic, to provide results in a form that can be used, and to assure that the management information system undergoes continuing evaluation and improvement. All else results in waste, inefficiency, or system entropy. Controlling inputs is a difficult problem. Wildlifers are beset by tendencies to throw in "the kitchen sink" when they do not get the results expected from analyses of limited inputs. The computer is loaded with all available data and correlation and regression analyses run in an attempt to find some relationship. This is wasteful. It is often an admission of inability to set objectives, to hypothesize, to predict, and to engage in simultaneous risk-taking that is just as foolish as covering all spaces on a gambling table to assure a win.
All forest faunal systems require inputs of labor, assumptions, commitment, funds, equipment and supplies, lands, climatic factors, people and their interests, and precise scientific observations - all in the right amount, in the right place, at the right time. The input subsystem is complex. It will not be dictated from higher up; it requires questioning, suggesting, demonstrating, and implementing to achieve improved inputs at all levels.
A Research Subsystem
Research is costly!? Try Ignorance.
Phenomenal sums of money are spent on research on the biology of wild animals. Few are spent on learning how to manipulate the structure, dynamics, and relations of wild populations, animal spaces, or resource users. While the results of wild animal research may one day contribute to decisions about management, there is no guarantee. The usual expression associated with research is a fond hope for a re-discovery from a library shelf of a conclusion written years ago.
The faunal system usually has a research-cash-flow problem (Giese 1988). There are insufficient funds to be invested for the long run. There is not enough money to solve the current problems. There are enormous problems, inadequate time, great uncertainty, and changing human interests as well as natural systems. As a consequence, a very difficult and unpopular decision is needed within the forest faunal system, perhaps all of "wildlife management." Easily misunderstood, the decision is to be made, not because it will result in the best condition, but only because of the constraints that now exist and that will not be changed appreciably within several decades. The decision needed is for those who are within the system or closely related to it to do only research directly and demonstrably related to managerial decision making. This in no way suggests other research is not needed; it suggests no value judgment on studies done elsewhere by others. It puts no bounds on the literature and conferences where faunal system managers will look for answers. It only suggests that given the present and foreseen situation of forest fauna, and given the ratio of needs to resources available, faunal system managers must see their research efforts as clearly and precisely directed to producing:
F.F. Darling observed that:
Various factors in U.S. political and economic life and college structure tended to channelize research in wild-life (sic) management, forest ecology and range ecology to rather narrow and macroscopic ends. Funds from the sport, ammunition and forest industries, the bleak fact of a state game commission being the probable career of post-graduate research workers, and the philanthropic ethos of the United States pointed men to the immediate, the useful and the obvious (Darling 1964:2-3).
After the above system reorientation is accomplished, then and only then should so-called basic research be done within the field. Not heretical in intent, this statement recognizes the claim of those who say the above cannot be done without the building blocks provided by basic research. Some can be, and I suggest moving as far along as possible, then when the "wall" is finally hit and other major problems seem under control, then and only then should the more basic studies be undertaken. I learned recently that Gershinowitz (1972:380) had suggested approximately the same ordering, namely:
(i) the use of existing knowledge to produce the desired change; (ii) when existing knowledge is incomplete or insufficient, applied research to find the necessary knowledge; (iii) basic research for the understanding of nature (including man and his works), with particular emphasis on those areas in which lack of fundamental understanding limits the scope of applied research; and (iv) basic research without regard to its possible relation to any area of application.
"Basic" is much discussed. Some suggest it means following any curiosity; study "just because"; climbing some intellectual or problematic mountain"because it is there." I suggest these are worthy and important efforts in a rich, unbounded, and crisis-free environment with surplus expertise, time, or resources. They are known to pay off, eventually, but few people have counted the costs. Others discuss "basic" as fundamental studies, those which engage the elemental properties and processes of systems, seeking to understand them at chemical and physical levels. Faunal system managers must rely upon others for these understandings. The distance between such levels (most of the time) and the time for making the managerial decision is so short that it is usually rare that timely use can be assigned to the performance of the research subsystem. The criterion for performance, as discussed in Chapter 4, is expected system performance increase or reduced risks per dollar invested.
The Type-I objective of the research subsystem may be to make useful inputs to faunal system decisions.
Research, when designed well and operating, is a total system. The output from it becomes input to the larger faunal management system. "To conduct applied research" is too vague to be meaningful. Type-I objectives, are, by definition, vague, so lower level objectives need to be developed.
One way to approach decisions about whether research is to be done is to ask: What would I do with the results from this project? Sometimes the results cannot be seen clearly. Then the questions become: what would I do if the results were generally A? generally B? By the criteria recommended here, if a substantial change (or stabilization) in practice, policy, or use of a technique cannot be easily imagined, the next research project proposal should then be considered.
Much faunal research euphemistically-labeled "administrative studies" is best conceived as information-generating subsystems. Such a point of view focuses the intent of the systems manager; it brings into the open the question of "why find out?" and answers it positively and simply by: "to improve the amount and quality of inputs into faunal resource-oriented decisions." Pressing further: why? A usual answer is: To explain the conditions. Why? I suspect the final answer is to increase predictability and thus, finally, to reduce risks in achieving objectives.
As a result of this point of view, i.e., action decisions, the pathway out of the "basic" research vs. "applied" research maze can become clear. Little distinguishes the two concepts except the time between reporting research and the application of the results. In general, with basic research there will be a longer time between final reporting and application of results than with applied research. Research subsystems can be designed to achieve an objective which classically appears as: to test hypothesis X. The manager needs conclusions. If planned, they become inputs to decisions. A research subsystem might use the best statistical methodology available to test if there is a significant difference between producing food in a forest wildlife clearing by using 600 pounds of 5-10-5 fertilizer or using 800 pounds of 2-12-10. The study of local fertilizer effects must be approached as carefully and with the same scientific objectivity and critical skepticism as a large-scale research job. While the results of administrative studies probably will not have the same widespread application as some findings, the results, nevertheless, are information for making management decisions. If simple fertilizer experiment results were collected systematically over many areas with many associated variables, predictive equations could one day be developed that would largely preclude the need for future managers to repeat, wastefully, most fertilizer experiments.
The critical questions for small information-generating subsystems are:
These are the type of questions that systems-oriented people will ask as they decide among and develop input-generating subsystems.
Scientific research is very difficult to comprehend. Easy to read about, the total research phenomenon is nearly impossible to understand. A 6-year old person probably has little comprehension of the concept of sexuality. It takes maturing, time, physical changes, experiences, and teaching. Similarly, there is a transcendence in science. Even though some have read about it, even done experiments, there is little comprehension of it, even among those said to be Masters of it. Being "raised in the tradition" I have learned to respect and honor the accomplishments of scientists and all those who engage the scientific method. It is largely practical as a 5-step method: (1) idea formation; (2) hypothesis development; (3) tests of the hypothesis including observation, experimentation data collection, and analyses; (4) conclusion; and (5) communication of results. It is largely an inductive process but usually expanded in general discussions to show how deduction has occurred, influenced ideas or methods, and how a second effort to solve a problem was informed by the first efforts, even if they were inconclusive or flawed. There is no doubt about the power and strengths of classical inductive science or the more realistically described classical inductive- deductive paradigm or broadly operating procedural pattern.
Taught to "love" science, the meaning of "philos-" of philosophy, it is painful to suggest an alternative paradigm and thus to reject the classical one. There is an alternative for faunal systems work. I do not suggest it for all areas of science or wherever the scientific method is used, but the alternative is likely to be more relevant to other areas than would be suspected from the narrow confines of faunal system management.
The concept of an alternative paradigm finally matured for me about 25 years after being awarded a Ph.D. My slowness in seeing it was that the alternative was hidden by the brightness of the successes of the classical science paradigm. An experience also helped me to see the alternative. I testified in court, at some risk, opposing a dam, and learned of 20 more dams on the planning boards. I could not win one case or, needless to say, all cases involving some undesirable portion of the 20. The analytical, site-specific work on the horizon was beyond my time, monetary, employer-tolerance, and energetic limits.
I had learned that so-called environmental decisions were not dependent upon scientifically obtained conclusions or observations. "Our biologists have found x but we shall do what we proposed anyway for p, q, and r reasons," is not uncommonly heard in wildlife agency meetings. If knowledge of biological factor x is only one factor and likely to be small or insignificant in a decision, it seems reasonable that faunal system managers need to gain knowledge of p, q, and r, the powerful factors. The faunal system manager often emphasizes fauna in his or her name, thus biology, not acknowledging or even realizing there is a large, complex, multifactor system to be run with payoffs and performance measures far more social than biological. It is irrational for a model builder to proceed when he or she knows that factor z is essential in the model and yet it cannot be observed or entered in any way. It seems just as limited to build a biological construct for a wildlife management agency or operation when it is evident that other factors are as influential as knowledge of owls or otters. There must be knowledge about the system to be managed. However, the only knowledge needed is not ecological. To pretend that it is (or assert that it should be) suggests some failures in observation or perception of reality about faunal systems or human decision systems.
After my experience with the dams, I created a state wildlife data base for 1000 creatures with about 200 factor fields for each. I soon learned that all data were not available, that the costs of getting things known into a computer were very high, and that half of the need-to-know fields (not just nice-to-know) were empty. At an absurdly low estimate of $50,000 per research-person-year and 100% success for each item in one year (also absurd), the costs to complete the data base would be over $2.5 billion! Wildlife management researchers will never see that much money. To act as if they might seems irrational. An alternative is needed. Assuming unlimited funds but reasonably large (twice that at present) and qualified staff, it will take over 2,000 years to complete the work! It seems irrational to continue with the same paradigm.
In my Ph.D. studies of insecticide effects on forest wildlife (Giles 1970) I tried very hard to collect all insect species in a forested area. In trying, I learned that it was nearly impossible. I also learned that in an Ohio woodland in 1965 there were species never even collected! I wondered about ecosystem studies that could not name all of the parts, much less the function or role of the parts. Then I went to China and India within one year (1989) and the magnitude of the work ahead for the forest faunal system manager became almost oppressive. The forest faunal system only really makes sense in a world context. Timber prices are world phenomena; they influence individual islands of trees and the animals and plants within them. To continue as if the classical paradigm can work, as if experiment after project after program can one day meet the real needs of people or fauna seems unlikely. To continue as if it could seems inappropriate, a denial of logical rigor which is a hallmark of the scientific community.
It is difficult to criticize a respected servant, the scientific method and even more difficult to find or create a replacement. Untested and unproven, the inductivist will find no evidence as proof of the suggestion's goodness. The arguments for an alternative must be deductive and pragmatic. The evidence, as it accumulates, may become useful to those who are still science-bound.
The alternative I call the neo-pragmatic paradigm. (Giles et al. 1993.) It is a search for knowledge useful in making decisions in faunal systems management. It is an input system the output of which are answers to realistic, immediately evident questions of resource managers. The processes are those of epistemology, of heuristic convergence of knowledge (Giles, in preparation). The results are always perceived as limited and tentative..., but probably useful if the question was posed correctly, heard well, and has not changed while the search was conducted.
The scientific method has 5 parts, relatively simple, and appropriate for teaching in the grade schools. The neo-pragmatic paradigm, on the otherhand, has at least 22 parts. The number is very inconvenient, but essential. The parts are as follows:
1. Starting at the End - A simulation mentality prevails within the new paradigm. A question posed is assumed to have been answered and that at least one answer is known with great confidence. Then a question must be answered about what will be done with the answer (the input to some decision). If truth has been discovered, so what? If a clear answer is not available, if there is great hesitancy, then there are other questions "standing in line" that can be served better. Given limited time, skill, resources and situations of extremely high risk of species loss, it is essential to exercise unpleasant, stringent control over the knowledge gathering and producing system.
2. Asked Questions - Not all questions are asked. A thoughtful person can pose and list hundreds of questions. The neo-pragmatic alternative only addresses those asked by faunal resource managers. Part 1 and 2 largely constrain the paradigm; there is little room for the idle or detached question, for basic research to attack a problem just because it is there. The realm of concern is decision making and it is highly applied. Some definitions of basic research suggest that it is limited to studies for which no particular use is now seen. The alternative paradigm turns up the clock speed. Must an answer stay unused for 50 years before it is discovered and used and then hailed as a basic research accomplishment? What if an allegedly basic research finding was used (applied) one month after publication? Other arguments aside, basic research is to be done elsewhere (other than within the faunal system). The neo-pragmatic alternative must face the enormous numbers of possible answers, their complexity, and massive limitations if resource decisions are to improve. Admittedly answers to all basic questions would be nice. The luxury or the resources to get them just do not exist.
The "basic research" topic is confounded in the literature by discussions of "fundamental research." The neo-pragmatic paradigm hypothesizes that if all fundamental knowledge (e.g., photosynthesis, nitrogen fixation, lipid metabolism, soil formation rates) were known, so few managers could assemble answers to questions that the results would be useless. The evidence for this is available from professors who try to teach how to make such assemblages. My emphasis is that a program of only fundamental research will not be sufficient. It can and must be a part, however small, of the new paradigm.
3. Meaningful Inputs - All environmental managers I know are awed by the complexity of the system with which they work. They can believe statements to the effect that modern computers are not big enough to handle all of the variables and interactions of ecosystems. Many are quick to scoff at the ideas that systems as complex as a forest can be analyzed, much less rationally manipulated to achieve some set of objectives. Yet the latter is the objective of faunal system management, and managers have been attempting it for years. Some have been doing it very successfully.
That some managers are successful suggests that they know better than others what to look for or else whatever they observe or sense is put together in some special way that allows them to predict, plan, and take advantage of conditions on their areas. There is meaningful information. From all possible stimuli to which managers are subjected, there are real ones and much noise, and some managers can select well from among them.
Systems people recognize that information has a specific gravity, resisting efforts to raise it to top levels in an organization. There are many reasons why there is resistance. Acceptance of this difficulty as a principle seems more constructive for top management and managers than carping about" failure to communicate "or" failure of the staff to do their homework." One major reason (of many complex reasons) for information not flowing up is the failure of information to be meaningful. All levels within an organization have responsibilities to make information more meaningful. Perhaps some new systems, information "fork-lifts," can be developed to overcome the problem.
Meaningful information is a concept easier to enclose than define. Four examples may be helpful:
1. When you describe the fruit production of a forest stand you may say total food (TF) is a function of three main sources. The manager might write: TF = 0.3 X1 + 0.5 X2 + 0.1 X3. He would be saying that these three species account for 90 percent of the food (i.e., 0.3 + 0.5 + 0.1 = 0.9) and even though there are 5 other types of food, it is varied or sparse and no one would gamble on significant production of food from any of the rest. The results are that a stand of 8 species of soft-mast-producing plants is now meaningfully described by a simple equation. Knowledge of these three values would give a very fine managerial capability.
Purists would continue to argue that 90 percent (also viewed as a probability of only 0.10 of being in error) is not good enough; the inputs are not accurate enough. The counter argument is that if there is a probability of a freeze every 3 years (i.e., 0.33 per year) that will destroy 50 percent of the flowers, then the probability that any one spot on the area will have any habitat at all is only 0.84 - so why quibble about the plus-ness or minus-ness of 90 percent? This is an example of noise; any numbers over 90 percent give a false sense of security. The real information is contained in the figure of probable food production (i.e., 0.90 x 0.84 = 0.76: "I have 76 percent confidence in the total food being X amount.") This figure is meaningful in a relative sense in that if a manager were to do fertilizing, burning, or planting, he or she would invest the high costs in an area where the probabilities were highest of getting the expected changes. Management decisions are almost always made in the face of incomplete information (CAP50).
2. A faunal system manager did a step-wise multiple regression analysis of 40 factors that may influence raccoon production, and after all the computer runs were through she discovered that 6 factors accounted for 85 percent of the predictability of the equation. She would probably be tempted to study those variables more intently, to use them to predict or explain raccoon populations, and to call them more meaningful than the others. The information would be in the 6, the data in the 40.
3. In developing wildlife problems for computer solution, I have repeatedly discovered that what initially seemed to be very complex problems often can be reduced to problems solvable on a calculator. The pursuit of meaningful information is one of the most exciting and worthwhile challenges of the systems person. The other side of this same coin is the concept of the megafactor.
Perhaps a manager has done a population analysis - sex ratios, age ratios, crippling and poaching losses, almost the whole process - and then realizes that there is no worthwhile information on a birth rate. It is easy to get into management situations where there is an abundance of information - many worthwhile variables, much good data, ample analyses with well-presented statistics, but just one thing missing - the essence of a truly effective argument which is the turning point of a decision. The systems analyst is especially helpful in identifying such variables, the profound controlling or overriding influence - the megafactor. A megafactor is that manipulable variable which contributes most to explaining the variance in a dependent variable, usually a desired system output, when analyzed by multiple regression techniques. It is that variable over which a manager has conceivable control and which contributes abundantly to high, significant R2 values in an equation having some objective of the manager's system as the dependent variable.
4. One of the major questions testing the systems approach is that of who is to say what are the data to be used. The question is not easily answered, for it requires some commitment to the decisions of others, a commitment fraught with risk. Any solution must appear evasive. It is simply that decisions at all levels on all topics have risks. A decision about which variable to include and which observations to accept, is as risk-filled as a decision about which tractor to buy or when to visit the legislator. The systems approach, if functioning properly, will have informed people making such decisions, and, with proper feedback, these decisions can be constantly improving. Making such decisions as well as making the system work is a dynamic, on-going task. All other approaches require such decisions sooner or later; the systems approach is the least unsatisfactory. [We now return to the list of the 22 parts of the neo-pragmatic paradigm.]
4. Fundamental - Forest faunal system managers or researchers can never (in real time of less than 2 centuries) address faunal questions species-by-species. Consider the insects! Consider only the butterflies! The answers to questions must be developed in groups. These groups may be suggested in the interactive fields of the tetrahedrons in Fig. 6.1. These are: (1) Process - the questions are about the common processes of many systems such as energy budgeting, foraging, dispersal, calcium-phosphorus utilization, and nitrogen budgeting; (2) Life Group - the questions are about the approximately-alike roles of organisms in forests. At what rates does each median individual in a group change energy and matter from one important state to another and how can the manager gain control over that rate? Not species, but behaviorally- or functionally-similar organisms is the group criterion; (3) Abiotic Factors - all aspects of the system are influenced by abiotic factors such as snow depth, temperatures, and
 |
| Fig. 6.1. The needed fundamental research topic groups can be conceived as being interactive and forming a research volume for contributing inputs to decisions related to forest fauna. An example of the algorithmic needs is the universal set shown to deal with spatial analyses. |
5. Taxonomy - Attention to functional groups is needed, although where individual species are already well known, there is no reason to relinquish those gains. "Approximately similar" has to be the phrase used, no matter how distasteful and perhaps risky. The working premise is that it is better to have all life groups in a managed system, no matter how poorly known, than to have a few well known and some omitted altogether. Life groups are emphasized (as discussed elsewhere in this book). There are greater differences between young and adults of some species than between genera. General algorithms developed will account these changes as readily as mortality and natality. The computational process is little different from changing a young life group to an older one (equivalent to death of one; recruitment of another).
Darling (1964:3) said he had heard American wildlife managers "...say openly that they do not know their dicky-birds - they left that sort of thing to the Audubon Society types." This attitude, said Darling, "has been a dangerous trend, utterly at odds with the outlook of Leopold, Shantz and Sears..." He observed, too soon, that it was a changing attitude. A statistician would probably suggest the conclusion in comparing past to present: "no significant difference."
Great losses of interest and support for classical taxonomy have been experienced. This must be reversed. Expert system technology may enable new uses of existing taxonomic knowledge, but it will not suffice to handle the massive needs in simply naming and listing the dominant plants and animals of working forests and similar systems.
6. Natural History Research - Gilbert (1989) reflected on the complexity of populations, agreed with the possibility of modeling them, but then countered that all such models require estimating the values of 12 to 36 parameters. "This degree of complexity seems to be irreducible. The estimation of so many parameters requires a lot of field work. The main difficulty lies in doing the field work, not in making the model." He argued the case against life history parameters including genetics and suggested potentials in studying life history traits. These show "consequences for lifetime fitness." He observed, that:
Population biology is innately (but not hopelessly) complex. Its history is of futile attempts to simplify the unsimplifiable. Such attempts are still very popular. That way you can become an eminent professor, but the animals and plants will have the last laugh. Brave souls who face up to the complexity, usually get swallowed up and left for dead--but only they can advance the subject.
By working to develop a computerized statewide wildlife information system, I discovered that the zoological community knows precious little about the animals that some people seek to manage. One such information system requires about 200 entries of highly select information about each species. Number of eggs or average adult weight seem reasonable requests! For many species the only entries possible have been status, taxonomy (some still in question), and location observed. Even for game animals and fish, data were lacking on such factor as minimum elevation in their range, water chemistry limits, and food habits.
Years ago, natural history studies were eclipsed by controlled laboratory experiments, pressure for large sample sizes and statistical support, and well-funded large-scale efforts. Natural history studies were hurried to a back room by the reductionists, the rise of molecular biology, and university and agency policy demanding fast results and abundant publications demonstrating returns on investments.
We have never learned that research allocation is a response to a"both...and" question, not "either...or." It is no longer trendy but it seems there are increasing calls for more natural history data. People creating computer models has been swept by fear-chills as they realized that "oh, we'll just use the literature to fill in the values" was an empty wish.
Greene (1986:99) said that natural history focuses attention on organisms and can be considered a blend of autecology (the study of the interactions of one species) and ethology (generally, behavior). He argued that it forms the context for both broad and narrow questions and is the prelude to hypothesis formulation. It may certainly be considered on a level in some scientific hierarchy with comparative anatomy.
There have been complaints about the inadequacies of natural history research, but what activities are perfect? Feedback is the life of research as well as biological systems. "Not enough numbers..." is the frequent complaint, for example, when a student only gives food items counted and not amounts of each. "Not enough..."is relatively relative. "As compared to what?"the natural historian may well ask, for "what did that creature eat recently?" and eventually, after more observations, "what are these creatures known to eat?" "How much?" is only one among the next 100 questions that the thoughtful interrogator would pose in a no-win game against the natural historian's observations which are limited by time, behavior, funds, and the destructive nature of some observations.
Greene (1986:101) said "High quality, publicly recorded natural history is data-in-waiting, simultaneously able to provoke theory and confront any number of previously unforeseen predictions." I have observed that there is not enough of a record, even for a well-studied eastern portion of the USA. What of the less-well studied areas of the world?
Elsewhere I have argued for long term studies, but that is to assure all life phenomena are observed...the entire range. The chance natural history observation...a snake getting water from a droplet on its skin ... a capture strategy ... a food element ... a display or antagonism...an egg laying event ... all are needed as the knowledge base is built. Thousands of observers in the short run may see what a few see during long-run studies. The quest must be for knowledge likely to be used, not just that which is specially processed or packaged by the rules of the current game being played.
Natural history research is needed. It needs to be done by competent, educated researchers with special training, perspective, a prepared mind, and a variety of specialized aids, techniques, and methods. While it contains art (as do other sciences having few practitioners who will admit it), it may, when well done, have as many benefits to others as to the observer.
There are new ways by which natural history research may be done and by which it may gain deserved status. One emphasis is that it costs relatively little compared to the resultant benefits. Another is that it appears as a new option to a group of people deadened by the end-of-the-funnel, specialized environment in which they find themselves after a typical USA university education. Besides emphasis, however, there are computerized wildlife information systems throughout the U.S. and in other countries that have been created that provide the potential store place for observations. This is the key element to the future of natural history research for it allows maximum diversity of interest and input, avoids the readership and specialization problems of journals, bypasses entrenched anti-natural-history policy of journals, allows full access (conventional database searches, e.g., all of X, subject to A and B but not C), provides credits for contributions, and provides news (electronic mail) about updates.
The other need is for expeditions to be mounted to areas to gain transportation and other logistic efficiencies, safety, team assistance, and group motivations. These expeditions may vary in extent from one day to weeks. They are akin to the bygone adventures of such trips to "darkest Africa." These expeditions seem to be one means to gain needed support from agencies and foundations reluctant to support the individual scientist.
7. Long-Range Work - Deer live to be about 15 years old but deer older than 6 years are few. To watch a fawn"crop"for 6 years, then another for 6 years, then another is to be engaged in research on one area for 20 years including 6 months for design and 1.5 years for analysis and write up. Intermediate production, of course, is studied but there are only 3 different sequences of years, only 3 "samples"and hardly enough to give a manager who is aware of the variability in the environment much confidence in a prediction or in a model developed for that purpose.
Animal production in one year in which environmental conditions are poor result in high mortality and reduced skeletal sizes. The reduced skeletal sizes reduce natality. The reduced cohort slides down the survival curve. Three years later when conditions are"perfect,"the population is said to be"at capacity." The population is low; even production is low because of abnormal birth passageway size resulting from food shortage 3 years earlier. The conventional wisdom that great environments will result in great production of young is compromised. Only in long-range research are such phenomena observed and built meaningfully into managerial models.
Mast crop size is said to be a function of precipitation in the previous 4 years. Other systems are cumulative. It seems that some natural systems must get a running start to produce.
Toxicants may slowly accumulate in the forests. Only when some threshold is passed are they recognized or even given the attention they deserve from the first day. Long-term studies are needed to achieve certain baseline conditions and the dynamics of systems.
Salamanders in a University of Maryland study (Douglas Gill) were marked. They disappeared, apparently due to mortality. Eight years later they re-appeared! They had a life stage far removed from the study site. Only if a long-term study was in progress would this phenomenon be known.
Most master of science studies in wildlife biology last only 2 years. These comprise a high proportion of all reported studies in the wildlife literature. Contractors who want results"yesterday,"reluctantly agree to 2-year studies. It is incredibly difficult to stabilize funding for Ph.D. studies because they typically have a 3- to 6-year duration.
Ackoff (1962) suggested a taxonomic node in research planning between simultaneous and sequential. See Fig. 6.2. Over-simplifying, the trade-offs are between time and money.
 |
| Fig. 6.2. Two major research strategies may be seen (Ackoff 1962). A. Trades resources and synergism against time and risks; B. Trades time and research resource expenditure risks against learning with feedback resource (fund) economy and availability. |
With enough money and resources it is possible to do a study"all at once", knowing full-well that events will be missed that occur irregularly. The other strategy is to conduct a study over a long period (greater than 5 years) and probably incur less costs. It may allow improved cash flow, learning with feedbacks, and observations not possible in that short run. Nevertheless, students must live with the knowledge that answers gained earlier would have reduced social risks, provided untold benefits, and that observations cannot be well compared between years...anymore than they can be compared well between areas as in the case of simultaneous studies.
LeCren (1984) noted the advantages of long-term research. Nature can carry out "experiments" with ecosystems or parts of them that would be otherwise impracticable (e.g., varying the degree of predation or temperature on fish). If well-designed, they can be efficient in the understanding yielded for the effort put in. This is akin to the problem of integers. An optimum planting of seed would be 53 kg but the seed may only come in 100 kg bags. Allocation must be in the integer number of bags, not kg. Research may be best spent in 3 days of observation per year but the integer may be a graduate-student year.
LeCren (1984:186) listed 5 special problems with long-term research, namely:
While he recommended that one person be involved, this, by definition of long-term, is not possible and so requires a plan for organization and individuals of overlapping tenure to assure continuity. The problems of communicating research intent, plans for analyzing data, and passing along notebooks, records, and computer programs are truly enormous and sufficient of themselves to prevent any long-term research.
LeCren (1984:186) suggested avoiding"monitoring"in discussions of long-term research. He pointed to a need for personal commitment to data and a view of their usefulness.
"For it to be worthwhile and cost-efficient, any long-term project must be carefully conceived and planned by someone with long-term vision and personal commitment." (LeCren 1984:186) He discussed only briefly the problem of funding long-term research, suggesting it is a responsibility of governmental and large institutes. Perhaps. When they eschew their responsibility, then what? I perceive the need for profit-oriented groups.
Forest faunal systems require high quality inputs from research systems. Some of this research must be long-term. Special techniques must be developed to allow recognition of this need, to show uses of data so gained, and to develop alternative strategies to assure stability of such studies. An insecure"long-term study"is simply another high-risk study with almost sure failure in achieving objectives. To secure stability, possible methods may be:
Non-governmental strategies are needed to assure stability of research enterprises. The study that concludes that more research is needed is already a classical joke. That long-term studies are badly needed is widely recognized. Perpetuation of the good project is essential. The last step of the paradigm is to act to secure long-term, sufficiently stable funding and processes that will allow knowledge to be gained. The lesson comes from biology. Survivors must reproduce themselves. Survivors of the rigors of the process, the selective pressures of those insisting upon certainty before any action, and those surviving battering in the public's countervalent value force field must assure their studies are continued. Conventional short-term graduate degree work does not serve well. Neither do agency career-ladder moves or election-driven shifts in programming and budgets. The neo-pragmatist must develop non-governmental companies, institutions, trusts, foundations ... alternatives that require as much skill and knowledge as how to master an array of instruments in the laboratory. "Drawing a conclusion"or"publishing"is not the end of the neo-pragmatic paradigm. Perpetuating needed studies - and perhaps the paradigm itself ... is.
8. Expeditions - Already mentioned under Natural History Research, expeditions are an important strategy to be revived. Groups of faunal system people working with the neo-pragmatic paradigm will make expeditions en mass to computer-selected representative areas to attempt to master quickly the life groups of an area.
Comprehensive long-range studies are rare, usually cut off before mature. Whole areas of the U.S. are almost unknown by anyone, especially ecologists. Evidently other areas of the world are less well known. The positive interactions, synergism, support, efficiencies, and cost savings from such expeditions are to be exploited.
There has to be a place for the sophisticated faunal system reconnaissance such as Darling and Leopold's (1953) Wildlife in Alaska: An Ecological Reconnaissance.
9. Complex Objective - Attention should be given to the complex objective described in Chapter 4. Research is guided by (1) Benefit Units - the resource units in which system performance is measured (e.g., sightings, pelts, harvest, monetary loss, loss reduced by management); (2) Value - the relative importance of benefit units (those of higher value receiving more study); (3) Demand - the minimum number of units perceived to be needed for human satisfaction (including laws saying an agency will prevent any species from becoming extinct); (4) Risk or Expectation (i.e., 1.0 - Risk), and (5) Substitutability - how a floral or faunal event, experience, or benefit unit may substitute for another. The more substitutable, the less the research need (and as always,"other things being equal."); and (5) Costs - the higher the probable costs or risks per unit of change, the more research is needed. Costs are non-linear.
10. Predictive - Classical science is said to be descriptive, explanatory, hypothesis testing, modeling, and predictive. The new paradigm strongly emphasizes predictive work. All management is related to decisions and decisions are made to achieve future conditions. What will be the consequences of a"treatment,"once it is decided upon and implemented? Each act or project, once selected, influences in the future some aspects of the objectives. Chaos theory suggests that things in nature may not be predictable, only explained. Initial conditions can have profound effect on subsequent performance of a system. Chaos theory helps enlighten prediction and reduces the frustrations in the differences seen in the actual and the predicted outcome. Great differences should not be surprising.
11. Anti-Parsimony - A central premise of science is the search for parsimony, brevity, simplicity with explanatory power. Useful in physics (e.g., the classic example is e = mc2) and chemistry, the search misdirects work in faunal systems. Faunal systems are extremely complex, multifactorial, and more economic, esthetic, and energetic than ecologic. They are rarely linear or normal. To simplify in any way is to oversimplify. Parsimony today guarantees error tomorrow. The anti- parsimony concept typically avoids mathematical formulations of the calculus because observed forest relations are not easily described, integration is very difficult, and unimodality rare. Uniting piecewise difference equations in computer models and using numerical methods seem realistic strategies for the future.
12. Geographic Information Systems - Increasingly, computer-based mapping systems are used in faunal systems work. The needs are to conduct analyses such as for the primeness of space for animals (Giles and Koeln 1983), suitability of areas for plants, and zones in which poaching can be seen or heard. Many resource problems are those of optimal space - best trail or road corridors; best site; placement of agents, waterholes, warehouses and offices, and similar questions of distribution. The GIS is a major tool for such decision making. It does not achieve its potential when it is used only to update or produce elegant maps quickly.
13. The Old Expert - Scientists are trained observers, people seeking the perfect stimulus. There are many untrained people who have made equally valid observations, perhaps not in such numbers, but as real. It is important to seek methods as well as the results of collecting observations made by people in the field, from the "old timers." The types of knowledge gained include minimum and maximum limits, distributions, likelihood estimates, occurrence confirmation, and"places at which x was last seen." Concepts of expert systems (CAP54) are also relevant, but more formal than implied here.
The modern faunal system managers are "on shoulders" people, quick to see the heights to which they can reach or see by standing on the shoulders of giants. There are truly giant managers from whom all may learn and advance. There is nearly criminal-waste of the knowledge of managers who retire or move to other areas and take information with them, on paper or in their heads. For faunal system management to become as great as it can become, meaningful information must be collected and stored so it can be retrieved. Present methods, with few exceptions, do not provide this capability. Unfortunately, resources that are available are seldom used. Abundant, high-quality information is essential for making good, low-risk decisions. These risks can be lowered by: (1) a wise, more frequent use of the existing information resource available, and (2) improved definition, storage, and retrieval of pertinent information.
A secondary aspect of this part of the paradigm is the need to protect and collect the knowledge of the about-to-retire and retired professional. Youngsters are usually ready to fill such spaces, to discard the old, to advance. Perhaps. There remains at least a few observations, things known with high confidence, that need to be collected from the experienced, then stored. They need not be repeated or replicated. They will not be published by journal editors today. The new paradigm seeks out, stores in new random-access databases, protects, and positions such knowledge for future use in decision-making.
14. Sampling - Selecting a proper sample size for any kind of research is a difficult operation. It usually should be supervised by a statistician. Key elements of the decision are the variability of the population to be sampled (s2) and the way it is distributed (by strata) in the forest, the level of confidence needed in drawing conclusions from the sample (t2), and the error allowable in expressing the average value (d2). These are related and I believe can be usefully conceived as in Fig. 6.3. They are located in some constraint
 |
| Fig. 6.3. Sample size, n, is dependent upon and influenced by the interactive estimate of the variance, s2; an index to the desired confidence level, t2; and to the proportion of the mean that can be tolerated as error in the final estimate of the number for which the sample is taken (d2). |
The smaller the sample needed, the better, because of costs. If the calculated sample-size-needed is 200 and each sample costs $200 to collect and process ($40,000), and available funds are $10,000, then what? In my view, the study should not be done. "We'll learn something even if we do not collect the proper samples" has been claimed. The alternative view is that the population will be misrepresented, faulty conclusions drawn, and even though cautiously worded, research "results" will be used (even though it is inappropriate to do so) because managers are hungry for "results." Hidden assumptions about the prosperity of research will be overlooked because of graciousness ("we are all limited"), lack of skepticism, lack of sufficient critical skills, insufficient time for critical work, etc.
The research system needs to assure quality performance leading to effective use of results and this includes justifiable sample sizes leading to conclusions at pre-stated levels of confidence and allowable error.
When an alternative sampling strategy must be devised it might include:
No matter what, do not collect samples when a priori it is known that the number will be inadequate for a conclusion. There are a million, million alternative research projects for which an appropriate sample size can be taken. Research funds are too limited to allow them to be applied to projects for which it can be asserted before the first sample is taken that no conclusion can be reached. Not more research, but alternatives are needed to that in which the design was flawed, samples inadequate, the total gains made in faunal system knowledge-base were zero, and project costs were a loss to the resource system.
15. Alpha Level - The 95% level of confidence in decision making is unjustified in most faunal system work. Often only 51% is needed. I suggest an alpha level in most forest faunal system work of 0.20, not 0.05. Other things equal, the sampling costs will be about 5 times less with such a strategy. Thousands of decisions (say 10,000) are made each year in environmental work. At the 95% level, 1 out of 20 is probably wrong - about 500! Wrongness is a function of non-existent objectives, economic change, ecological dynamics, and catastrophe. Nature has a way of overcoming human ignorance and misuse. Extreme confidence and low risk, while very desirable, just is not and never will be possible. There is recovery power, resilience in most ecosystems.
16. Robustness - Managers easily fool themselves. Ego leads them to a destructive spiral of logic that holds: the more precise the better. The Alpha Level concept of the neo-pragmatic paradigm is relevant throughout the input system. Robustness is a similar viable concept. It embodies complex philosophical, theological, and psychological interaction involving values and how they are stated, risk-taking behavior, and a concept of time on a continuum between the immediacy of crisis and the futurity of geological existence. The daily situational responses of the faunal system manager are consistent with the idea of robustness. Robustness is a concept that can improve inputs to the faunal resource system. It may reduce the frustrations of the neophyte wed to precision and accuracy, give pause to old timers, and increase the rate with which everyone gains effectiveness in faunal system management.
There are several positions from which robustness can be viewed. A total view cannot be gained from one position.
Position I
The first view is that from elementary arithmetic. In any manipulation of numbers (adding, multiplying, etc.) the appropriate number of significant figures is that of the lowest number in the group. For example, to multiply 3.0036 x 3 x 1.00127 and to show anything other than 9 as the answer is incorrect. The reasons are self-evident. It is inappropriate to express more precisely than the least precise component of a set of numbers. It is irrational to display more confidence in a calculation than in the confidence attached to the least accurate or least precise number involved in the calculation.
Consider a small system like a deer herd or a pond. It seems inappropriate to measure deer weights to the hundredths of a pound when the desired output is specified in terms of plus-or-minus 3000 pounds. When analyses are decided upon robustly, when researchers go on "fishing trips" for good variables, it would be peculiar to strive for excessive accuracy or precision.
The only major limitation to this idea is the rule: any percentage of a very big number is still a very big number. To accept robustness is not to advocate across-the-board disregard for precision. The percentage points, decimals, and sizes of numbers used in certain statistical analyses are very important. (One does not watch the growth of many sums of squares before they appreciate the influence of a few digits of precision.) Even in the face of this limitation I contend that most natural resource decisions are made between pairs of alternatives. There is preliminary screening of multiple alternatives, of course (but this too is often done in pairs). The decision maker usually looks for ratios, guides, or key figures as indices to the best alternative. If, in making such comparisons, the relative influential factors contributing to the predicted consequences of each alternative are held to the same degree of precision, the results will be useful. Usefulness is a primary aspect of the concept of robustness within neo-pragmatism.
Position II
The concept of multiple regression is helpful in examining the robustness concept as it was for"Meaningful Inputs (2)." When some desired outcome or benefit is sought from a system, then it may be possible to list the major factors that explain how it can be obtained. Since manipulation is done to achieve future goals, then the list has predictive power.
An objective Q, may be seen as a function of factors A through n:
Q = f(A, B, C, D . . .n).
Using some gross statistical concept of stepwise multiple regression, it is possible to imagine a future value of Q being predicted by:
Q = a + bD + cC + dA + eB.
If Q can be estimated well by making an input of only D, but C contributes a little more, but A and B contribute less than 5 percent, it may be very reasonable only to use D.
To strive for a very precise system to predict achievement of a not-very-clear objective, Q, seems unwise. The system must balance out; equal precision or sloppiness on both sides of the equality seems reasonable. Confidence in the goal statement of about 80% seems best matched with factor D in the equation above. The equation Q = a + bD is robust and appropriate. Further involvement may lead to false confidence, expensive data collection and analysis, and the predictor being "fooled" by him or herself as often as by a fickle public or by nature.
Whether some additional factor which would add 0.5 percent predictability is desirable or not is not important here. The important concept is that desired outcome or benefit is usually a sloppy idea. Even for highly quantifiable systems like the stock markets, the benefits desired are a sloppy, multidimensional concept of returns, taxes, stability of earnings, pride of ownership, risks, portfolio mix, and a host of others. In public forest faunal system management, the concept of desired benefits is even more obscure, has greater dimensions, and is usually specified by commissions or by individuals strongly influenced by quasi-political forces. It is easy to recognize that the"greatest good"is a poorly conceived goal of such decision makers (no matter how vigorously argued). Even more evident is that the concept of"the most people"for whom the benefits are sought is strongly biased by power groups, sampling procedures and lack of good records from which to draw samples, the absence of efficient feedback mechanisms, and the struggles to achieve goals for the mean of the population rather than its median (or, better, some sophisticated utility function for the entire population). "Over the longest time"is a third dimension of the public resource goal that is typically so poorly handled that little need be said. Counterintuitive results of ecological and managerial simulations are the least costly evidence of the failure of decision makers to deal adequately with the time dimension in managing large systems. Ill-treatment of the time dimension clearly demonstrates the probable low quality of most statements of objectives and inputs needed to achieve them.
Position III
Garbage in, garbage out (GIGO) is a cute, over-worked phrase used by computer detractors, more applicable to those doing data processing than those doing modeling. It is a relevant warning, however. In this chapter on inputs, it is essential that the reader see the image of the entire system. Inputs are featured, not separated from a system. GIGO is a separatist's slogan. It denies that processors can screen out erroneous data; that "outlier"data points can be included in one computer run and deleted in another; that a statistic is synthetic and is not complete truth; that a mean and median and answers at upper and lower confidence limit can be presented for decision makers; and that pure inputs (no garbage at all) can be processed improperly. A robust approach to data and models is one that recognizes that the role of processing is as great or greater than data gathering and presentation and has the power to improve inputs or their interpretation.
Position IV
Experienced resource and land managers know all too well the stochastic nature of the systems with which they deal. It is conceivable that a very precise statement of objectives associated with well-researched factors which achieve such objectives could produce failure, and often will. Failure can be produced in the best-known systems by climatic catastrophe, political shifts, and such subtleties as a modest change in bank interest rates. To overdo the precise quantification of systems subject to such profound changes is irrational. The normal rhythmic nature of such systems generates a dynamism interactive with the above to produce in some managers frustration to the point of apathy.
Position V
Systems to be manipulated are subject to three major information problems. Often there is inadequate information (although the accuracy and precision of existing information is high). There are often erroneous data and occasionally even falsified data. The modern need is to synthesize an overabundance of data for many factors from many sources into a useable form in the time available for it to be used.
The argument that too many data are needed is often raised against the systems approach. Much may be needed but in balance. It eventually dawns on almost everyone that techniques for gathering environmental information, e.g., animal units produced, even for three or four wild forest land species, are insufficient to give worthwhile data for comparison against criteria in order to make firm decisions. Even elaborate studies of the multiple interactions and vagaries of weather and soils provide only a few reliable discriminating results. Such studies require large resources and long periods of time and usually must be rejected on the grounds of insufficient funds, time, or of value per unit cost and practicality.
With decisions coming at an increasing rate, the pressure is real. Oversimplified, the purpose of this section is to explain the principle: the confidence, accuracy, precision, risks, stochastic nature of factors, and low quality of statements of objectives makes a robust approach to faunal system inputs - virtually to the entire resource management activity - appropriate and other approaches inappropriate.
Position VI
Also a part of the systems approach and a dimension of the concept of robustness is a heuristic attitude. The attitude results in far more liberal transgeneration of data than in the past, using the computer to develop ratios (magic numbers), linearizations, and groupings of data to fit a management hypothesis. It involves a freer, iterative, discovery mode than the straight-jacketed approach of the classical statistician. The pendulum has swing from the first non-uses of statistics in natural resource management, through acceptance, into super-stat, and now increasingly into the freedom of heuristic data management for crisis decision-making.
In any natural resource decision there exists the half-hearted assumption that the land owner or the faunal system manager will do what the decision maker indicates will be best to do. To advocate relatively sophisticated decision-making processes requires verification on the basis of this assumption. The instant that the resource manager makes such an assumption, he or she will likely recall the"ornery"owner or employee prone to act contrary to recommendations ... just because. The action can be explained on the basis of poorly quantified values, or more explicitly, on the basis of people having very dynamic value and risk-taking systems. Right and safe yesterday may be wrong or risky tomorrow. The concept of robustness recognizes that there is a temporal dimension in a hidden assumption. The resource manager (or the public) must make appropriate trade-offs between preciseness of their recommendations and their estimate of whether a land owner or agency employee will not only perform all recommended acts but also perform them over the entire planning period. The odds are small; the corollary is thus: why demand extreme precision? While the question has a negative connotation, it can be put more positively by saying that managerial variability is fully as great as the variability in the natural system. To operate on the assumption of optimum management (or even stable mediocre management) is neither practical nor realistic. Robustness is a concept of managerial realism.
The word sloppy has been used, perhaps too sloppily. Adoption of robustness is surely not an advocacy of nihilism in environmental affairs. It is a plea for a rational approach to environmental data collection and management, and for an adoption of an attitude of tentative conceptual reckoning. It is a concern for a proper perspective on quantifying the environment. My concern flows from an awareness that there is an ultra-conservatism among conservationists. Their risk-taking behavior can be classified as low on a conservative-liberal scale. The concept of robustness can provide some release that is rational, and may enable some action in time to provide human-kind environmental hope. When the house is afire, there is no time to calculate the hydraulic psi on the hose or the Btu's released from the house. There may be time to get the hose. The interesting may be replaced by the necessary.
The necessary within the forest faunal system is not easy to define, but is easier when the known extras are laid aside. The dynamics of a tentative approach to the environment are not satisfying for those not prone to take risks. Surety and risk-freeness are not a part of the life of the sensitive forest faunal system manager. Their lifestyles typically have feedback operation that enable and encourage them to go ahead and start, given a particular state of knowledge and competence. Once started, there is inherent within it the assurance that the evolving new system can be made to work.
17. Equifinality - Closely related to the Alpha Level part of the paradigm is a concept of systems theory, namely equifinality. It is the recognition of multiple, very different pathways to the same end state (e.g., to a climax forest). Ignoring equifinality leads to extremely high variance among observations, thus unnecessarily large and costly statistical samples being taken. Statistical control is not the issue, but a new form of stratification, a new awareness of the meaningless of the mean. Every organism, every plot of land is truly unique and statistical representations suppress this uniqueness, hide differences, lose information while gaining some. Computers now provide power to deal with such a uniqueness where statistical generalization was once required. The pathways in a potential tree diagram reveal realistic event sequences, any one with a possibility. Possibility means little; probability is sought. Within forests the search is almost impossible because every pathway in a network diagram shows every stand (almost every tree) as unique and that pathway takes 30 to 300 years to complete. "Replications,"so much a part of classical research, are impossible. The probability of a pathway has little meaning at the human scale and temporal dimension of a forest.
18. Clinical - The neo-pragmatic paradigm might be summarized quickly for those unwilling to listen to 20 or more dimensions. It is clinical. It is akin to the action of the good doctor who deals with unique patients, takes knowledge as available, diagnoses tentatively, prescribes, requires consultation in a few days, and keeps abundant records. The research is cumulative, sequential, uncontrolled, probabilistic, and often risky. The risks are reduced by the associated feedback and adaptive processes.
19. Automated Prescriptive Systems - General computer systems can be developed that produce "plans,"reports, and prescriptions for action based on dynamically changing text, data bases, and algorithms. These will become less and less hard-copy oriented. An example is Guidance, under development.
20. Retrospection - Cleaning the files may be one way of expressing a major part of the new paradigm. There is more work to be done retrospectively than can be done by the current research body in forest faunal systems work. The available now-published literature is almost beyond use. It needs synthesis. Abundant raw data needs processing (e.g., federal agent diaries), and old studies need to be re-examined. Often studies have been so restricted that useful conclusions for larger systems have been missed.
The dye sulfanilamide was synthesized in 1908 by I.G. Farben's lab. In 1932, a few years later Gerhard Domagk, Farben's director of therapeutic research, tested a derivative of this substance, prontosil, on bacterial infections in mice and rabbits. It cured bacterial disease in those animals but it had no antibacteriological properties in vitro. In the year I was born, 1933, the dye was first used on humans to treat a staph infection then called"blood poisoning." If the investigators had restricted their work to in vitro tests, the antibacterial properties would not have been seen and the antibiotics revolution in medicine been delayed.
21. Permanence - New efforts need to be expended to increase the permanence and utility of knowledge about forest systems gained at such high costs and often great risk and hardship. New institutional arrangements, data storage systems, hypertext, expert interview schemes, and electronic storage (photo, sound, data, and text) need to become part of the effort.
22. The Pragmatic Exception - Not competition but ecological niche theory suggests the need for specialization. Faunal resource managers cannot master all of the sciences of their realm. They need great help from specialists. They suboptimize their capability, time, and money in attempting to master a topic peripheral to their central forest faunal interest. There may be exceptions to this rule on the grounds of pragmatism.
There are problems that block advancement in faunal system management, no one to solve them, no funds, no early work. They must be solved in order to achieve substantial gains at reasonable costs. Not the perfect arrangement, nevertheless when these problems arise and are identified, they may need to be solved by non-experts within the field. They should be avoided and in-house solution resisted.
Applications
Gershinowitz (1972) emphasized the difference between researcher and manager and their worlds. The difference is substantial and to ignore or discount it leads to low success in applying research findings. He said that where research is successfully applied, formal organizational relationships and methodologies have been set up to reduce the barriers. He emphasized that where agencies work with university facilities, special methodologies are needed. The continuous and intimate interactions of research worker and results-user he described for industry are rare in the university. He noted that faculty who were successful at innovation in practice were often acting outside the university environment as consultants or in agency staff roles. The institutional setting, i.e., the lack of an appropriate meeting ground, is a major barrier to adopting research, not the capabilities of individuals.
In the faunal research system, every problem imaginable exists that can prevent the adoption of research. Agencies separate research and management; management is excluded from research planning; agencies rarely send out"requests for proposals"but respond to random submissions. Adoption of findings is not rewarded; stability is desired. Research agency is pitted against management agency for funds. There are no rewards for team work. In most cases, managerial groups should be buying research results to get specific improvements and capabilities. There is no middle person or staff to help translate research results into practical terms or into potentials (i.e., there is no"marketing"function).
L.E. Hicks said to a U.S. Biological Survey Research Staff meeting in Missouri in 1937 (mimeo report) that research is not complete "...until it has actually been published and made available...in various necessary versions"and ... "distributed to the people that each should reach." Contrary to sentiment within the Wildlife Society about multiple publications, Hicks said, and I agree, that "there are many research reports that should be reworked four or five times into different versions of length or technical complexity."
Though it escapes many graduates, one reason for requiring a master of science or other advanced degree is to develop an experimental or research attitude. This includes a mix of ideas like: activities in the field can be conceived as tests; nothing is final; there are opportunities for change and adoption and these should be based on observations matched with objectives. Transferring research results is not a message-sending activity: "take it or leave it." It is, when effective, a dynamic involvement, with feedback (Giles 1981).
Questions and Work
| Quick Access to the Contents of LastingForests.com |
|---|
This Web site is maintained by R. H.
Giles, Jr.
Last revision May 17, 2001.