INFORMATION TRANSFER UPDATE RESEARCH AND DEVELOPMENT OFFICE OF INFORMATION TRANSFER

Selecting Means to Measure Range Condition and Trend

Ronald E. Kirby U.S. Fish and Wildlife Service Office of Information Tran sfer 1025 Pennock Place Fort Collins, CO 80524

December 1, 1987 United States Department of the Interior

FISH AND WILDLIFE SERVICE

OFFICE OF INFORMATION TRANSFER 1025 PENNOCK PLACE, SUITE 212 FORT COLLINS, COLORADO 80524

M2,norandum To: Selected Fish and Wildlife Service Personnel From: Chief, Office of Information Transfer Subject: Information Transfer Update

This Information Transfer Update memorandum is provided to disseminate timely natural resources data important to management of the nation's fish and wildlife Resources. It is provided by Service Research and Development Scientists for your use. Updates are developed in response to specific needs voiced by the management arms of the U.S. Fish and Wildlife Service. The intended audience is Regional staff and field station personnel in each of the Service's major areas of emphasis. U.S. Fish and Wildlife Service personnel may obtain further information by contacting the Office of Information Transfer (FTS 323-5401; commercial 303/493-8401). SELECTING MEANS TO MEASURE RANGE CONDITION AND TREND

Ronald E. Kirby Office of Information Transfer U.S. Fish and Wildlife Service 1025 Pennock Place, Suite 212 Fort Collins, Colorado 80524

December 1, 1987 V

Nothing in this review is meant to suggest policy for the U.S. Fish and Wildlife Service with regard to range management, and where the text could be construed to suggest conflict with established Service policy, the latter shall be taken as having precedence. Regional Directors remain the sole arbiters of the suitability of this discussion to Regional management programs.

V

V SUMMARY This report briefly reviews the concepts of measurement of range condition and trend. Its primary purpose is to introduce the topics and to provide suggestions from the current literature on means to coherently address range management issues. A "cookbook" for range surveys thus is not provided, but sufficient discussion and reference to the literature is provided to permit review of current or initiation of new range condition and trend assessments on National Wildlife Refuges. Suggestions for obtaining additional information are provided, and an Appendix containing five recently published reviews of the topic is attached.

iii CONTENTS Summary ...... iii Introduction...... 1

Acknowledgments • • • . • • • • • • • • • • • • • . . • • . • • . • . • • • • • • • • • • • • • . . • • . • . . • • . • • • • • • • 2

A Basic Reference for Range Inventories and Monitoring ••••••••••••••••••• 3 Why Monitor Range Condition and Trend? ...... 3 Range Condition and Trend Terminology...... 4 Determining Range Condition and Trend: The Process...... 5 Determining Range Condition and Trend: Techniques...... 7 Sources of Additional Assistance...... 8 Stati stfcal Advice ...... 8

Range Assessment Techniques • • • . • • • . • . • . • • . • • • . • • . . • • . • • . • . • . • • • • • • • • 8 Additional Readings and References...... 9

Literature Cited .•..•.••••••••• ~ .•••••..••.•.••••••••.•.....•..•••••••••. 10 Appendix--Pertinent Papers From the Literature ••••..•.•..•.•••••••••••••• 13 Range Inventory Standardization Committee (RISC). 1983. Guidelines and terminology for range inventories and monitoring. Report presented to the Board of Directors, Society for Range Management, Albuquerque, NM. February, 1983. 13 pp. Risser, P.G. 1984. Methods for inventory and monitoring of vegetation, litter, and soil surface condition. Pages 647-690 in National Research Council/National Academy of Sciences CB. Delworth Gardner, Chairman). Developing Strategies for Rangeland Management. Westview Press, Inc. Boulder, CO 80301. Piepper, R.O. 1984. A critique of "Methods for inventory and monitoring of vegetation, litter, and soil surface condition. Pages 691-701 in National Research Council/National Academy of Sciences (B. Delworth--i;-ardner, Chairman). Developing Strategies for Rangeland Management. Westview Press, Inc. Boulder, CO 80301. Bonham, C.O. 1984. Sampling and statistical considerations in range resource inventories. Pages 773-787 in National Research Council/National Academy of Sciences (B. Delworth Gardner, Chairman). Developing Strategies for Rangeland Management. Westview Press, Inc. Boulder, CO 80301. Menke, J.W., and M.F. Miller. 1984. "Sampling and statistical considerations in range resource inventories": comment and discussion. Pages .789-807 in National Research Council/National Academy of Sciences· (B. Del worth Gardner, Chairman). Developing Strategies for Rangeland Management. Westview Press, Inc. Boulder, co 80301.

V INTRODUCTION Staff of several National Wildlife Refuges have requested assistance in selecting means to measure range condition and trend. This is a timely topic for at least four reasons: l. Recent controversy surrounding methods of evaluating range condition and trend has not universally been a part of management revtew in the Fish and Wildlife Service. 2. Nonetheless, Fish and Wi1dlife Service policy (cf Refuge Manual 6RM5) requirtt management of grassland habitats to meet wildlife objectives and to maintain quality of natural and manipulated grais1and communities. 3. Management on many lands, particularly western Refuges, involves use of grazin9 1 contro11P.d burning, haying, 1eeding; . mechanica1 soi1 treatments, terti1itation, brush control, and M)(fou!i w1ed contro1 u agents of change in the rang~ ecosystem. All of these treatments are ideally applied in response to identified deficiencies in the habitat, each of which is determined by range condition and trend surveys. 4. Further, grazing and haying activities on Refuges are permitted on a primary basis only when they "enhance, support, and contribute to established wildlife management objectives"; and on a secondary basis when they "wisely utilize a renewable natural resource and do not conflict with established wildlife management activities" (Refuge Manual 6RM9.l). Establishing criteria for permitting these activities requires knowledge of range condition and trend. Integration of range and wildlife management priorities and selection of proper management tools for the range ecosystem are thus intimately dependent upon ability to characterize the "status" of the management unit (range). Therefore, means to efficiently evaluate condition and trend are an integral part of National Wildlife Refuge System habitat management. The fol lowing review provides introductory guidance for those interested in measures of range status. The purposes of this review are to

1 a) outline reasons for measuring range condition and trend, b) provide a review of selected means of measuring condition and trend, and c) suggest '-,I sources of additional assistance. The reader should bear the following in mind when reviewing this material, however, for this review is purposely narrow in its treatment of these topics. First, although pertinent literature is reviewed and Appendix material provides comprehensive supplementary guidance, this review only provides comment on the first steps in range survey up to and including the collection of data. The interpretation of range measurement data for management and planning purposes is a separate process, and this review cannot substitute for review of original literature or consultation with professional range managers. Second, experience, common sense, general knowledge, and professional judgment all play important roles in the art and science of range management. Most importantly, determining range condition and trend is not a once-a-year process, but rather a portion of ongoing management throughout the year. Finally, following surveys, action must be initiated to develop the range as desired. Range improvements are not a portion of this review; see Vallentine (1980) for a recent presentation of range management practices useful for meeting specific range management objectives.

ACKNOWLEDGMENTS This Information Transfer Update was prepared in response to a series of questions initially raised by Refuge Managers and staff in Region 2. Their clear outline of the problems faced by range managers and their patience in waiting for a review was much appreciated. Constructive comments during initial review of this topic or upon one of several early drafts were obtained from C.D. Bonham, S.S. Berlinger, D.L. Franzen, W. McCoy, J. Neaville, and J.L. Oldemeyer; D.H. Cross provided editorial assistance. J. Neaville is particularly thanked for his assistance in review of practical range mangement techniques. The cooperation of the Society for Range Management in providing at cost to the Service copies of the Range Inventory Standardization Committee (1983) report for distribution to field personnel is acknowledged.

2 A BASIC REFERENCE FOR RANGE INVENTORY AND MONITORING In May, 1978, the Society for Range Management established the Range Inventory Standardization Committee (RISC) to promote uniform terminology and methodology for range inventories and assessments. A report representing a consensus viewpoint (RISC Guidelines and Terminology for Range Inventories and Monitoring. 1983) presented substantial discussion of technical, theoretical, and practical value. This report, purchased from the Society for Range Management, 2760 w. Fifth Ave., Denver, CO 80204, is attached as an Appendix for those wishing full conceptual development of range condition and trend processes and philosophy. It emphasizes concepts and a framework within which compatible range inventories can be conducted.

WHY MONITOR RANGE CONDITION AND TREND? Prescribing kinds and levels of use for rangelands is impossible unless something is known of the physical and biological characteristics of the land. In other words, selection of an optimum management strategy requires the ability to predict the consequences of management action. A means to obtain such predictions is to perform condition and trend analyses for the range in question. Range condition is essentially a quantitative evaluation of range "health." Knowledge of condition alone, however, cannot lead to immediate ·choice of management action. This is because a poor range that is still deteriorating requires different remedial action than a poor range in the process of improving. Thus, the trend, or direction of change in condition, must also be assessed. In range concITtion and trend surveys, successional and community dynamics concepts are used to compare range sites within the range ecosystem with accepted standards and capabilities. These standards and capabilities are, in turn, based upon a site's long-term and present levels of production relative to its potential for production. Note that range condition compares existing to potential conditions and should not be confused with measurement of immediate availability of forage, which usually reflects recent weather conditions. Improved management can halt, then reverse the direction of the rate of change in condition of deteriorating range. It can likewise usually

3 increase the rate of improvement in positively changing range. Throughout, it is important to remember that poor range condition does not necessarily mean that current management is wrong, \,.,.) for the trend may be upward. The reverse is also true.

RANGE CONDITION ANO TREND TERMINOLOGY There are three major philosophical approaches that may be used to evaluate present condition of a range (Schmautz 1984): 1) comparison of present communities with climax or potential natural communities of a site; 2) comparison of present communities with existing successional stages on the site; and 3) evaluation of existing communities with regard to the management objectives of a site. Unfortunately, the almost universal use of qualitative terminology (excellent, good, fair, poor) that equates excellent range condition with the climax stage of succession has led to connotations of value for ecological stages in the plant community. Additionally, these adjectival ratings often have been viewed in terms of ability to produce forage for livestock, presenting a "cow" bias in classification (Smith 1984). In current usage, two separate phenomena are included in the meaning of the term range condition: ecological status and resource value rating. Ecological status is defined as "the present state of vegetation and soil protection of an ecological site in relation to the potential natural community of the site" (RISC 1983:6), Resource value is the value of vegetation or other features of a site for a particular use or benefit (RISC 1983:7). Each actual or potential use has a unique resource value rating which estimates the suitability of the vegetation at a site for a specific use assuming full use is possible. Range condition may therefore be best considered as a generic term encompassing all aspects of current range status. Ecol ogi cal site is the term suggested by the RISC Report for the basic unit of rangeland classification. It is "a kind of land which differs from other kinds of land in its potential natural community and physical site characteristics, and thus also differs in its V

4 ability to produce vegetation and its response to management" (RISC 1983:2). The Potential Natural Conrnunity is defined as •the biotic commun1ty that would become established if all successional sequences were completed without interference by man under the present environmental conditions" (RISC 1983:6)a. Trend actually refers to two concepts. One is trend in ecological status, which may be evaluated by comparison of the vigor and reproductive status of plants representing later versus earlier seral stages on a site (RISC 1983:7). The other is trend in resource value rating, which evaluates change with respect to one (or many) particular uses to which the site may be subject. The inventory data from which trend is determined are generally those from which condition is evaluated, except that data are collected on two or more occasions. Management potential for a site is simply the potential resource value rating for that site.

DETERMINING RANGE CONDITION AND TREND: THE PROCESS What steps should the manager take once a decision has been made to evaluate range condition and trend? First, before beginning a new monitoring program, the system under current use should be critically evaluated. The advantages of continuing with an established system are: major baseline data have already been obtained; historical trends in land use have already been assessed via known techniques; and current procedures are already known and accepted. Unless there are known disadvantages to current practice, i.e., precision and accuracy are unacceptable, data are uninterpretable in relation to Refuge goals and objectives, the detail or the scale of current

aThe Potential Natural Community recognizes the past influence of man, accommodates the existence of naturalized exotic species, and recognizes that, following man's influence, the final stage of succession may not reestablish the "original" vegetation, i.e., as normally defined in concepts of "c 1i max.• See Meeker and Merke 1 ( 1984) for a review of climax theories pertinent to range management concerns.

5 schemes is unacceptable, etc., established systems have much to recommend them. An initial consideration in the process of establishing new range condition/trend surveys (or evaluating present schemes) is to obtain a clear statement of the overall goals of land management and the goals of the range monitoring program. General guides for these objectives include: the purposes for which the Refuge was established; historical practices judged of value; Refuge management and master plans; Annual Work Advice; historical planning documents such as Regional Resource Plans; Regional supervisory direction; and the manager's best judgement of means to meet the Refuge's wildlife/habitat objectives. A monitoring program only makes sense in light of its value to assist in management of the range. The wildlife manager must also remember that when wildlife becomes a major concern, some of the standard ecological status measures become much less important than the resource value ratings that determine potential of the habitat to support species of concern. Determining these resource value ratings is often a site-specific task that will minimally require review of literature, discussion with experts on the ecology of the species in question, and review of station records for information on t,istorical presence and abundance. Determining resource value ratings for some species may require specific management studies or research efforts. It thus may not be possible to initiate a complete range management ~rogram until substantial data on wildlife species and their habitat requirements have been collected. Once objectives are defined, the manager should: 1. Identify the number of ecological sites necessary to fully represent the range of conditions management will potentially affect. 2. Identify Potential Natural Cover {Ve etation) {RISC 1983:6) for each eco 1ogical site •. 3. Obtain baseline data on a) the condition of the Potential Natural Cover and b) the range of condition necessary to provide a guide to identifying trend in each ecological site.

6 4. Select techniques for monitoring range condition that can be accommodated by existing manpower, funding, and availability of equipment but that will still meet management objectives. 5. Select ecological sites for monitoring-­ remember that reasons for selection are intrinsic to the management goals of the 1and unit. 6. Obtain ecological status information through application of appropriate vegetation sampling techniques to determine kind, proportion, and amount of plants; obtain similar soil status data. 7. Statistically analyze the data obtained in 11 6 11 as work proceeds to determine sample sizes needed to evaluate condition. 8. Select plant species to monitor for trend analysis. 9. Pre-sample sites for trend data using frequency analysis for grass and shrubs; density and canopy cover measurements for woody vegetation. Calculate sample sizes necessary to obtain desired confidence intervals. (This step may be accomplished concurrently with step 6.) 10. Continue with trend analysis baseline assessment until appropriate samples are obtained. 11. Repeat steps 6-10 at selected intervals, comparing data between years to evaluate trend and summarizing condition data as it becomes available. 12. Reevaluate priorities and needs and revise management action as necessary. Throughout the above 12 steps, complete and accurate records should be maintained. In this way, a9reed-upon management programs may be continued w1thout interruption even following a complete turnover of station staff. Communication with adjacent stations, review of scientific literature, and discussion of issues with professional range managers will permit incorporation of new assessment methodology and new management methods as their success comes to light.

7 DETERMINING RANGE CONDITION AND TREND: REFERENCES Comparative discussion of specific techniques used to inventory and monitor range habitats is scattered throughout the literature. An excellent recent summary (Risser 1984) and critique (Pieper 1984) are included in the Appendix. Likewise, statistical considerations are very important to range surveys. A recent review with direct application to range management concerns (Bonham 1984) and a critique (Menke and Miller 1984) are therefore also included in the Appendix. Smith (1984), Schmautz (1984), and Pendleton (1984) discuss the use of inventory and monitoring data to develop range managment prescriptions. There are a number of general references that review techniques of value to the range manager; a selected list follows. Joyce et al. (1986) provide a description of a rangeland database consisting of all information from Soil Conservation Service Range Site Descriptions in 20 Great Plains, Southern, and Western states. Stubbendieck et al. (1986) is a recent guide to North American range flora. Avery (1975) and Stoddart et al. (1975) are basic range management texts. Vallentine (1980) is a recent review of range development and improvement methods. Mueller-Dombois and Ellenberg (1974) and Grieg-Smith (1983) are two recent texts on quantitative plant ecology. Southwood (1978) and Green (1979) are readable discussions of general sampling methods and data analysis. Zar (1974) is a good single source of statistical information for the biologist and range manager, and Eason et al. (1982) is a good slngle-source review of basic mathematical and statistical concepts. Avery (1975), Hays et al. (1981); Payne and Copes (1986), Platts et al. (1983, 1987), and Hamilton and Bergerson (1984) provide means to measure wildlife habitat variab~es important to the range manager. Thomas (1979), Hoover and Wills (1984), Cooperrider et al. (1986), and Thomas and Maser (1986) provide thorough discussions of inventorying, monitoring, and managing western wildlands for wildlife of concern to the Fish and Wildlife Service. Workman (1986) provides a thorough discussion of the economic aspects of range management.

SOURCES OF ADDITIONAL ASSISTANCE Statistical Advice Fish and Wildlife Service Research and Development-Centers and Cooperative Research Units have considerable in-house statistical capability. V Several may be able to provide advice regarding

8 sampling design and data analysis on a time-available basis. Before contacting Research and Development for such assistance, first define your objectives for overall management of the land unit and for the range survey. Then, identify available manpower, dollars, and time at your station to conduct range surveys. Following, contact the Office of Information Transfer with your request for assistance. OIT will assist you to locate appropriate contacts to discuss development of your program. Range Assessment Techniques This review has not been intended to provide a techniques manual for field assessment methods, but a number of references have been provided that provide entry to all areas of concern. In addition, easily accessible sources exist for this information within local Bureau of Land Management, County Extension Agent, and Soil Conservation Service offices; among the staff of departments of range science, agriculture, wildlife management, and botany at local universities and colleges; and among Fish and Wildlife Service personnel with range management experience at other field stations. Personnel at each of these are excellent sources of additional, site-specific information. Specific contacts in each of these categories, if unknown, may be obtained by contacting the Office of Information Transfer. Finally, the scientific journals of the Societies most involved in range management (Journal of Range Management, Soil and Water Conservation) provide current reviews, reports of recent research, and evaluations of methods from throughout North America. Additional Readings and References The Office of Information Transfer continues to offer assistance to all Service field personnel interested in obtaining technical information of any sort. Whether involving a more in-depth review of the topics addressed herein, a review of additional techniques, or an investigation of related ecological relationships such as herbivore-range relationships, waterfowl nesting requirements in grasslands, etc., OIT can provide advice regarding knowledgeable Research and Development contacts, non-Service experts, pertinent publications, and the status of current Service and other research. Call or write: Office of Information Transfer U.S. Fish and Wildlife Service 1025 Pennock Place, Suite 212 Fort Collins, CO 80524 Telephone (FTS and Commercial) 303/493-8401

9 LITERATURE CITED

Avery, T.E. 1975, Natural resources measurements. Second edition. McGraw Hill Book Co., New York, NY. 329 pp. Bonham, C.O. 1984. Sampling and statistical considerations in range resource inventories. Pages 773-787 in National Research Council/National Academy of Sciences (B. Delworth Gardner'; Chairman). Developing Strategies for Rangeland Management, Westview Press, Inc. Boulder, CO 80301. [Included in the Appendix of this document] Cooperrider, A.Y., R.J. Boyd, and H.R. Stuart. 1986. Inventorying and monitoring of wildlife habitat. U.S. Dep. Int., Bur. Land Manage. Service Center, Denver, CO. xviii, 858 pp. Eason, G., C.W. Coles, and G. Gettinby. 1982. Mathematics and statistics for the bio-sciences. Reprinted, revised edition. Ellis Harwood Ltd., Chichester, West Sussex, England. 578 pp. Green, R.H. 1979. Sampling design and statistical methods for environmental biologists. John Wiley and Sons. New York, N.Y. is1 pp, Grieg-Smith, P. 1983. Quantitative plant ecology. Third edition. Studies in Ecology Vol. 9. Univ. California Press, Berkeley. 355 pp, Hamilton, K., and E.P. Bergersen. 1984. Methods to estimate aquatic habitat variables. Cooperative Fishery Research Unit, Colorado State Univ., Fort Collins. Miscell. pagination. '.._,/ Hays, R.L., C, Summers, and W. Seitz. 1981. Estimating wildlife habitat variables. U.S. Fish Wildl. Serv., Biol. Serv. Prog. FWS/OBS-81/47. 111 pp. Hoover, R.L •• and D.L. Wills, editors. 1984. Managing forested lands for wildlife. Colorado Division of Wildife in cooperation with U.S.D.A. Forest Service, Rocky Mountain Region, Denver, CO. 459 pp. Joyce, L.A., D.E. Chalk, and A.O. Vigil. 1986. Range forag~ data base for 20 Great Plains, Southern, and Western states. U.S. For. Serv., Gen. Tech. Rep. RM-133. 17 pp. Meeker, D.O., Jr., and D.L. Merkel. 1984. Climax theories and a recommendation for vegetation classification--a viewpoint. J, Range Manage. 37:427-430. Menke, J.W., and M.F. Miller. 1984. "Sampling and statistical considerations in range resource inventories": comment and discussion. Pages 789-807 in National Research Council/National·Academy of Sciences (B. Delworth --­ Gardner, Chairman), Developing Strategies for Rangeland Management. Westview Press, Inc, Boulder, CO 80301. [Included in the Appendix of this document] Mueller-Dombois, D. arid H. Ellenberg. 1974. Aims and methods of vegetation ecology. John Wiley and Sons, New York, NY. 547 pp, '-"

10 National Research Council/National Academy of Sciences cs: Delworth Gardner, Chairman). Developing Strategies for Rangeland Management. Westview Press, Inc. Boulder, CO 80301. 2022 pp. Payne, N.F., and F. Copes, technical editors. 1986. Wildlife and fisheries habitat improvement. U.S.D.A., For. Serv. Wildl. Fish. Admin. Rep. (unnumbered). Miscell. pagination. Pendleton, D. T. 1984. "Use of inventory and monitoring data for range management purposes": a critique. Pages 855-865 in National Research Council/National Academy of Sciences (B. Delworth Gardner, Chairman). Developing Strategies for Rangeland Management. Westview Press, Inc. Boulder, CO 80301. Piepper, R.D. 1984. A critique of "Methods for inventory and monitoring of vegetation, litter, and soil surface conditions. Pages 691-701 in National Research Council/National Academy of Sciences CB. Delworth Gardner, Chairman). Developing Strategies for Rangeland Management. Westview Press, Inc. Boulder, CO 80301. [Included in the Appendix of this document] Platts, W.S., C. Armour, G.D. Booth, M. Bryant, J.L. Bufford, P. Culpin, S. Jensen, G.W. Lienkamper, G.W. Minshall, S.B. Monsen, R.L. Nelson, J.R. Sedell, and J.S. Tuhey. 1987. Methods for evaluating riparian habitats with applications to· management. U.S. For. Serv. Gen. Tech. Rep. INT-221. 177 pp. Platts, W.1., W.F. Megahan, and G.W. Minshall. 1983. Methods for evluating stream, riparian, and biotic conditions. U.S. For. Serv. Gen. Tech. Rep. INT-138. 70 pp. Range Inventory Standardization Committe (RISC). 1983. Guidelines and terminology for range inventories and monitoring. Report presented to the Board of Directors, Society for Range Management, Albuquerque, NM. February, 1983. 13 pp. [Included in the Appendix of this document] Risser, P.G. 1984. Methods for inventory and monitoring of vegetation, litter, and soil surface condition. Pages 647-690 in National Research Council/National Academy of Sciences (B. Delworth Gardner, Chairman]. Developing Strategies for Rangeland Management. Westview Press, Inc. Boulder, CO 80301. [Included in the Appendix of this document] Schmautz, J.E. 1984. A discussion of ''Use of inventory and monitoring data for range management purposes." Pages 843-853 in National Research Council/National Academy of Sciences (B. Delworfh Gardner, Chairman). Developing Strategies for Rangeland Management. Westview Press, Inc. Boulder, CO 80301. Smith, E.L. 1984. Use of inventory and monitoring data for range management purposes. Pages 809-842 in National Research Council/National Academy of Sciences (8. Delworth Gardner, Chairman). Developing Strategies for Rangeland Management, Westview Press, Inc. Boulder, CO 80301. Southwood, T.R.E. 1978. Ecological methods with particular reference to the study of insect populations, Second edition, revised. Chapman and Hall, London. 524 pp.

11 Stoddart, L.A., A.D. Smith, and T.W. Box. 1975. Range management. Third edition. McGraw Hill Book Co., New York, NY. 532pp. Stubbendieck, J., S.L. Hatch, and K. Hirsch. 1986. North American range plants. Third edition. University of Nebraska Press, Lincoln. 465 pp. Thomas, J.W. 1979. Wildlife habitats in managed forests: the Blue Mountains of Oregon and Washington. Agric. Handbook 553. 512 pp. Thomas, J.W., and C. Maser, technical editors. 1986. Wildlife habitats in managed rangelands--the Great Basin of southeastern Oregon. U.S.D.A. For. Serv. and U.S.D.I Bur. Land Manage. (Special edition compilation of 14 General Technical Reports from the Pacific Northwest Forest and Range Experiment Station), Binder and loose-leaf materials, miscell. pagination. U.S. Fish and Wildlife Service. 1982 (and subsequent releases). National Wildlife Refuge System Refuge Manual. Division of Refuge Management, Room 2343 Main Interior, Washington, D.C. 20402. Miscellaneous pagination. Vallentine, J.F. 1980. Range development and improvements. Second edition. -Brigham Young University Press, Provo, UT. 545pp. Workman, J.P. 1986. Range economics. (Biological Resource Management Series). Macmillan Publ. Co.; New York, N.Y. 217 pp. Zar, J.H. 1974. Biostatistical analysis. Prentice-Hall, Inc. Englewood Cliffs, N.J. 620 pp.

12 APPENDIX--PERTINENT PAPERS FROM THE LITERATURE

The following papers are provided as additional reference material for those considering development or modification of range inventory and monitoring programs. The RISC (1983) report was made available at cost by the Society for Range Management. The other papers are from a volume in the public domain entitled Developing Strategies for Rangeland Management: a report prepared by the Corrrn1ttee on Deve1op1n-----Strateg1es for Ran~eland - Management (Nationalll:cademy of SC'lences National Research"""'Counci • B.Delworth Gardner, Chairman 1984), a compendium resulting from six workshops during a two-year period that discussed rangeland management issues.

13 Publlah•d in 1984 in ~h• united States o! America by Weatvie~ Press, Inc., S500 Cencral Av.nu•, Boulder, Colorado 80301; Frederick A. ?raeqer. Publisher

LiDrary of Conqr••• Catalog Card Number: 84•51389 ISBN: 0-86Sll-S43-4

Coapo•ition for this book wa• provided l)y the authors Printed atld bound in the united Staees of America

10 9 e 7 6 5 4 l 2

C C ( Guidelines and Terminology for Range Inventories and Monitoring

Report of the

Range Inventory Standardization Committee Society for Range Management

Presented to the Board of Directors of SRM Albuquerque, New Mexico February,1983 Contents Introduction 2 Purpose of Inventory and Monitoring ...... 2 Definition of Inventory and Monitoring ...... 2 Background Information ...... 2 Classification and Mapping of Land ...... 2 Characterization of Present Resource Values ...... 2 Interpretation of Inventory and Monitoring Data ...... 2 Scope of Report ...... 2

Basis for Classifying Land and Defining Ecological Sites ...... 2 Relationships to Range Sites and Habitat Types to Ecological Sites ...... 2 Ecological Sites in Relation to Potential Natural Plant Communities ...... 3 Determining the Potential Natural Plant Community of an Ecological Site ...... 3 Permanence of Ecological Sites ...... 3 Differentiations Among Ecological Sites ...... 3 Identifying Ecological Sites ...... 4 Ecological Sites Descriptions ...... 4 Classification in Relation to Mapping ...... 5 Mapping of Ecological Sites Complexes ...... 5 Data Storage and Retrieval ...... 5 Collection and Interpretation of Data for Management Planning Resource Assessment ...... 5 Guiding Principles ...... 5 Range Condition ...... 6 Measuring and Interpreting Trend ...... 7 Management Planning ...... 9 Proposed Standard Terminology for Inventory Classification and Analysis of Range Ecosystems ...... 10 Literature Cited ...... 12 Appendix ...... 12

. RISC Members Date of Date of Name Af@iation Service Name Affiliation Service

l. John L. Artz Univ. Nevada, Reno and 1978-1983 9. C.B. Rumburg Cooperative State Research 1978-1983 Western Univ. Public Range- Chairman Service (USDA). land Coordinating Committee, ( 1980) Washington, D.C. Reno. NV 10. E.F. Schlatterer U.S. Forest Service 1980-1983 Washington. D.C. 2. John Baker Bureau of Land Management, 1979-1983 l l. Jack Schmautz U.S. Forest Service 1978-1980 Denver, CO Washington, D.C. 3. Richard S. Driscoll lJ .S. Forest Service 1978 1983 12. E. Lamar Smith Division of Range Resources 1978 1983 Rocky Mt. Forest & Range University of Arizona, Chairman Expt. Station Tucson, AZ (1981-83) Ft. Collins, CO Alternates 4. Richard E. Eckert. USDA Agricultural Research 1978-1983 I. Robert Buttery U.S. Forest Service, Region 1980-1983 Jr. Service, Reno, NV II. Lakewood. CO 5. Minoru Hironaka Dept. of Range Resources, 1980-1983 2. Harland E. Dietz Soil Conservation Service, 1980-1983 University of Idaho, Fort Worth, TX Moscow, ID 3. Jim Hancock Bureau of Land 1980-1981 6. Floyd Kinsinger Bureau of Land Management, 1978-1979 Management, Washington, Denver, CO D.C. 7. James 0. Klemmedson Division of Range Resources 1978-1979 4. Richard Hart USDA Agricultural Research 1980-1983 University of Arizona, Chairman Service, Cheyenne, WY Tucson. AZ (1978-79) 5. Jerry Thomas Bureau of Indian Affairs, 1980-1983 8. George Knoll Bureau of Indian Affairs, 1978-1983 Shiprock, NM Phoenix, AZ 6. Ron Wenker Bureau of Land Manage- 1981-1983 u ment, Washington, D.C. 7 .Gale Wolters U.S. Forest Service 1980-1983 (Research), Washington, D.C. Preface

Increased interest and activity in natural resource inventories !he first action of RISC was to write a "orking paper to define and monitoring by federal and state agencies in the United States the objectives of the Committee (see Appendix). The objectives began in the I 970's. Much of this activity was precipitated by were to develop and recommend adoption of: (I) standard termi­ Congressional legislation and court decisions that required federal nology for inventory, classification and analysis of range ecosys­ natural resource agencies to prepare environmental impact state­ tems, (2) a uniform system for classification and mapping range ments .on resource management activities; to prepare, implement ecosystems, (3) minimum standards and guidelines for data collec­ and periodically revise multi-resource management plans on pub­ tion and (4) a common philosophical base for data interpretation. lic lands; and to assess and monitor the extent and status of natural The committee was to address needs for local management, for resources on both public and private lands as a basis for natural regional and national assessments, and for research. The working resource policy decisions. paper was accepted by the SRM Board of Directors and published Although these inventory, planning and monitoring activities in Rangelands, August 1979. involved a wider spectrum of range resource values and uses than RISC then considered the terminology used in range inventory previous efforts, range managers continued to play a key role and monitoring. A list of proposed standard definitions was pub­ because, for many rangeland attributes, procedures by range man­ lished in the August 1980 issue of Rangelands with a request for agers were better than those developed by other disciplines. comment. Based on comments received and those of selected Moreover, they were the only resource managers with much prior reviewers, revisions were made. RISC then prepared written inventory and monitoring data. However, these data were Jess statements on information needs, rangeland classification, and useful than they might have been, because use of different collection and interpretation of data. These statements, along with approaches and procedures among agencies, or even within the the revised terminology, were submitted in April, 1982 to reviewers same agency, made it difficult to integrate, aggregate and interpret selected by RISC. the SRM Board of Directors, and alt SR M data from different sources. Data collected for specific uses were Sections. Comments of the SR M membership also were solicited not always helpful for multiple use planning. Confusion and con­ (announcement in Rangelands, May 1982). All comments were flict occurred because some terms and concepts used in range considered by RISC in preparing this report of the committee. inventory and monitoring were not perceived the same way by all This report presents concepts and a framework within which range managers, not to mention other resource managers. compatible range inventories can be conducted. It is not a "cook­ In recognition of these problems, the Board of Directors of the book" for range inventory techniques. It is intended to serve as a Society for Range Management (SRM) resolved that SRM should conceptual basis and guideline within which specific inventories take a position of leadership to promote a uniform methodology can be developed. Basic information requirements and measure­ and terminology for rangeland inventories and assessments. In ment procedures are presented which should be included in any May, 1978, the Range Inventory Standardization Committee inventory and around which the entire inventory may be developed. (RISC) was established as a sub-committee of the Research Affairs The report is limited in scope to information requirements and Committee for this purpose. Since February, 1981, RISC has terminology for rangelands in the United States. Inventory or functioned as an independent ad-hoc committee; its chairman is monitoring requirements of all possible rangeland resource users appointed by the President of SRM. or values (i.e. for wildlife, timber, recreation, soils, or other asso­ Membership on RISC consists of 9-10 SRM members selected ciated elements) were not considered in this report. to obtain representation from the U.S. agencies involved in range Some international comment was received and considered in inventory and planning, and from the academic community. preparing this document. However, much additional effort will be Agency heads have appointed the member representatives of: Soil required to reach conformity with international usage in either Conservation Service, U.S. Forest Service (one each from National terminology or procedures. RISC supports initiatives of the SRM Forest Systems and Research), Bureau of Land Management, to promote more uniform international terminology and proce­ Bureau of Indian Affairs, Cooperative State Research Service, and dures. Agricultural Research Service. In addition, there have been 2-3 This report is a consensus viewpoint and represents comprom­ representatives from universities. Alternates were named by some ise. It is the result of much debate among people dedicated to the federal agencies to assure representation at all meetings. The com­ concept of compatible range inventories, but with different expe­ position of RISC was a deliberate strategy based on the notion that riences and perspectives. For some readers, some statement, will its work could only proceed effectively with continual input and seem radical and threatening to long-held basic tenets of range feedback from the agencies involved in rangeland inventories. management. To others, the report may appear timid and conser­ RISC is not an "interagency committee": its members have not vative in relation to recent advances in ecological thought, remote served to represent official agency positions nor have the agencies sensing, computer technology, and multi-resource inventory and been obligated to accept RISC's viewpoints. All of the agencies and planning needs. RISC views the report as a significant, but incre­ universities involved have supported RISC's activities by provid­ mental, step low a rd better communication among range managers ing time and expenses for the participation of members in I 0-12 and between range managers and other resource managers by days of meetings per year since its inception. In addition, the U.S. providing more useful and interpretable range inventory data. If Forest Service, Bureau of Land Management, Soil Conservation the report stimulates thought, research and discussion, and leads to Service and Bureau of Indian Affairs have provided additional further improvement of methods and concepts, we will have funding to support R ISC's activities. accomplished our purpose.

GUIDELINES AND TERMINOLOGY FOR RANGE INVENTORIES AND MONITORING Introduction the present status of resources (mainly vegetation but including some attributes of animals and soils). Characterizing information Purpose of Inventory and Monitoring on vegetation includes plant species present, their relative abun­ Range management planning and decision-making require dance, present resource values for particular uses, rates of produc­ information about the kind, quality, production potential, loca­ tion, current levels of utilization, and changes in these characteris­ tion, and amount of soil, animal and vegetation resources. Selec­ tics over time. Data also are needed on current degree of soil tion of management alternatives requires the ability to predict protection and perhaps on wildlife species and other resources. changes that will result from different management practices. When vegetation is classified by criteria related to resource Information is needed for planning and management at several values, the present value for a given use may be estimated by simply levels, including specific projects, ranches or other management identifying the vegetative class (range condition class) on a given units, and regional or national assessments. land unit. In this case, detailed information on species composi­ tion, vigor, stand structure, yield, utilization, and soil properties Definition of Inventory and Monitoring are needed only for selected monitoring locations. Range inventory and monitoring are processes for obtaining and Interpretation of Inventory and Monitoring Data analyzing information about physical and biological resources. Available information, including field data, is interpreted in Range inventory establishes the status of resources at a given time, terms of ecological status, present or potential resource values, either by complete measurement or statistical inference based on trends in these values, and probable causes of identified trends. sampling. Monitoring measures change in the status of resources Some of these computations or interpretations may be, and should over time. Monitoring may involve complete reinventory. but be, made in the field when data are collected. However, the differ­ more commonly involves repeated measurements on selected ence between data collection and data interpretation should be areas. Decision-making is not a part of inventory and monitoring recognized. Data collection is objective. Interpretation depends on per se; but it involves economic and social evaluations or value value judgements or state of knowledge and can vary over time and judgements based upon inventory and monitoring data. among different interests. For example, estimating utilization is a Background Information relatively straight-forward procedure, but stating that a given level of forage utilization is without harmful effects is a matter of Broadly defined, range inventory and monitoring involve more professional interpretation. than the actual field collection of data. Some information is obtained by observation, or from public and private records. A Scope of Report partial list includes: The following sections of this report outline an approach and the • Legal description and ownership of land including surface minimum needs for range classification, data collection, and data rights, water rights, and boundaries of the management unit. interpretation for general range management purposes. Land mapping procedures are available through the National Coopera­ • Location and condition of cultural features - buildings, roads, tive Soil Survey and thus, will not be discussed here. It is recog­ fences, water developments, power lines, and others. nized that additional types of information or different intensities of • Natural features important to management - natural waters, sampling will be needed for specific cases. natural barriers to livestock, threatened and endangered spe­ Basis for Classifying Land and Defining cies, and others. Ecological Sites • Weather records. RISC recommends the term ecological site for the basic unit of • Records of actual and historical use by livestock and wildlife, rangeland classification. An ecological site is a kind of land which including kind, numbers, periods of use, and productivity. differs from other kinds of land in its potential natural community and physical site characteristics, and thus also differs in its ability • Existing resource information, such as maps of geology, vege­ to produce vegetation and its response to management. tation, soils, range condition or utilization. Relationships of Range Sites and Habitat Types to Ecologi­ • Data from studies and research, such as soil-vegetation rela­ cal Sites tionships or response to management practices, that will help The concepts of range site and habitat type are similar, but not define and characterize ecological sites and vegetation response identical. According to each concept, land units are defined on the characteristics. basis of their inherent productivity and their climax vegetation. Implicitly, both concepts also include the criterion of response to Classification and Mapping of Land disturbance or management by predictable seral vegetation pat­ Classification of the land into relatively homogeneous units terns. Although similar in concept, range sites and habitat types provides a framework for collection, storage and interpretation of often do not correspond exactly in a given area. This lack of data on vegetation, soils, animals and other elements of the range correspondence is due primarily to differences among attitudes of ecosystem. Site identification provides the basis for predicting individuals or agencies responsible for land classification. potential resource values under alternative management strategies. The relative emphasis placed on floristics, structure or produc­ Mapping land units furnishes a relatively permanent base for tivity of vegetation and on edaphic, topographic and climatic display of data on soils, present vegetation, range conditions and features will prod ucc somewhat different classification units, as trends, utilization and associated resources or values. Classifica­ will the allowable amount of variability within-or required differ­ tion and characterization of land are generally based on studies ences among-classified units. Differences between range sites and and research and are usually refined over a long period of time. habitat types are mainly due to differences in approach to or Classification provides a basis for mapping, but the two tasks may purpose of classification and not to any basic philosophical differ­ be carried out concurrently. ence between the proponents of the two systems. However, since one term or the other has sometimes come to be identified with Characterization of Present Resource Values certain agencies, disciplines, land types or regions of the country, Estimation of present resource outputs and prediction of we have chosen to use the term ecological site rather than attempt response to management alternatives requires characteri,ation of to standardize either of the presently used terms.

2 GUIDELINES AND TERMINOLOGY FOR RANGE INVENTORIES AND MONITORING Ecological Sites in Relation to Potential Natural Plant Permanence of Ecological Sites Communities Ecological sites arc subject to many influences that modify or !"he potential natural plant community of an ecological site is even temporarily destroy vegetation but drought, timber plant community that is best adapted to a unique combination of harvest, grazing, fire, and even short periods of tillage. Unless these environmental factors and that is in dynamic equilibrium with the influences are particularly severe, the potential of the ecological environment. Natural disturbances, such as drought, wild fires, site is not permanently affected and the site remains capable of grazing by native fauna, and insects are inherent in the develop­ supporting the potential natural community. ment of any natural plant community. Plant communities pro­ Deterioration of the plant community is often followed by loss tected from these natural influences for long periods do not typify of soil and fertility, loss of ability of the soil to absorb and retain the potential natural vegetation. The termporential natural plant water, increase of stones on the surface, and other forms of site community is used in lieu of the term climax community to reflect deterioration. The cumulative effect of such detrimental influences realistically the conditions existing today. Most present vegetation reduces the opportunities for reestablishing the original cover and and its physical environment have been altered by past use, includ­ productive capacity of the ecological site. Severe site deterioration ing intentional and unintentional species introductions, such that will significantly alter the potential of the site. A different site is the potential natural community will differ from the original pris­ then recognized and described on the basis of its altered potential. tine "climax" plant community. Differentiations Among Ecological Sites The potential natural community of an ecological site is not a precise assemblage of species for which the proportions are the Differences in the kind, proportion, and production of plants are same from place to place or even in the same place from year to in large measure the result of differences in soil, topography, year. Generally, one species or a group of species is most promi­ climate, and other environmental factors. Marked changes in soil nant on a site. Their prominance in the potential natural commun­ texture, depth, and topographic position usually result in pronoun­ ity does not vary from place to place, or from year to year and can ced differences in plant communities. Environmental conditions be used to help classify and characterize sites. Associated species of associated with a specific ecological site can be used to identify the the potential vegetation may fluctuate greatly from place to place site in the absence of the potential natural vegetation. according to local differences in microenvironment or weather Potential natural plant communities change along environmen­ conditions. Consequently, using associated species alone for site tal gradients. Where changes in soils, topography, or moisture classification can be misleading. In all plant communities, variabil­ conditions are abrupt, plant community boundaries are usuallv ity is apparent in production and occurrence of individual species. distinct. Boundaries will be broader and less distinct where plan"t Variability within reasonable limits is the rule rather than the communities change along gradual environmental gradients. Even exception. Boundaries of communities, however, can be recog­ though plant community changes may be gradual, plant communi­ nized by characteristic patterns of species association and com­ ties can be identified, classified and described. The occurrence of munity structure. these communities is predictable and they are associated with concomitant changes in soil, topography, and climate that also can Determining the Potential Natural Plant Community of an be identified, classified and described. Ecological Site Distinguishing between ecological sites along gradual environ­ The following methods are used in determining the potential mental gradients is difficult. Site differentiation may not be readily natural plant community of an ecological site: apparent until the cumulative environmental impact on vegetation is examined over a broad area. Ecological site differences may be • Evaluation of relict vegetation and associated soils on areas reflected in production or in the kinds and proportion of the plant that have been subjected to minimal disturbances. Both pro­ species making up the core of the plant community or both. Of duction and the species composition of the plant community necessity, boundaries between ecological sites along a gradient of should be evaluated. closely related soils (i.e. catena) and a gradually changing climate • Interpolation and extrapolation of plant, soil, and climatic may be somewhat arbitrary. data from existing relict areas along a continuum to other The criteria used to differentiate one ecological site from another points on that continuum for which no suitable relict is are: available. • Significant differences in the kind and proportion of species • Evaluation and comparison of areas currently grazed, logged groups in the plant community. or otherwise disturbed in varying degrees and comparison of such areas with similar areas that are not disturbed. • Significant differences in the production of the plant com­ munity because of differences in soil properties, climate and • Evaluation and interpretation of research data dealing with topographic position. the ecology, management, and soils of plant communities. • Review of early historical accounts and botanical literature of • Significant differences in soil properties, slope and topogra­ the area. phic position reflecting different use potentials and hazards that are not reflected in the community. Characteristics of a plant community obtained from a single location are not likely to be conclusive. Evaluation of plant com­ Any differences in criteria, either singly or in combination, great munities must consider such factors as drought, unusually favora­ enough to indicate a different use potential or to require different ble moisture years, effects of recent fire, excessive rodent concen­ management, are bases for establishing an ecological site. trations or insect damage, plant disease, or excessive soil removal Environmental factors are largely interrelated and the effect of a or deposition by wind or water. Every effort should be made to change in any single environmental factor will vary, depending on examine plant communitit:s throughout the area of occurrence of the influence of other factors. For example, a deep soil is more the ecological site and at different seasons and during different significant on an ecological site that receives additional moisture years. The initial description of a potential natural plant commun­ from repeated overflow than on a sloping upland site that receives ity should be considered as an approximation subject to modifica­ no overflow. An additional two inches of annual rainfall may be tion as additional knowledge is gained. highly important in an arid climate but of minor significance in a

GUIDELINES AND TERMINOLOGY FOR RANGE INVENTORIES AND MONITORING 3 humid climate. Similar variations in degree of significance apply to some soil taxonomic units occur over broad environmental gra­ most factors of the environment. Consequently, in identifying an dients and thus may support more than one distinctive potential ecological site consideration must be given 'o its total environment natural plant community because of changes in an environmental as well as to the individual environmental components. component such as average annual precipitation or temperature. In evaluating the significance of kinds and proportion of species Where this occurs the soil taxonomic unit should be phrased to or species groups that are prominant in the potential natural plant reflect the different potential plant community. community, the relative importance of a species may indicate Only one name should be given to a single ecological site that V whether one or more ecological sites are involved. On deer winter occurs in adjacent states. If this is not feasible, use the local name in range, for example, the potential natural vegetation may consist of the description of the ecological site and indicate the rn,me used in 30% bitterbrush in one area and less than 10% bitterbrush in the adjacent state. another area with otherwise similar species composition and total A mechanism is needed to systematically correlate ecological annual production. If this truly indicates a different site potential, sites within and among agencies and states to insure uniformity. then it would be important to recognize the difference to insure Correlation could be accomplished by establishing an interagency proper management of the winter range. committee of agency representatives to resolve differences. Identifying Ecological Sites Ecological Site Descriptions Naming Ecological Sites A technical description should be prepared for each ecological Ecological site names should be simple, based on readily recog­ site that is identified and named. Descriptions should be brief, but nizable features and descriptive of the site. Such names will help they should clearly present the features that characterize the site. users and managers recognize and remember the significant inher­ They should not be oriented solely around single resource uses ent differences of the ecological sites in their locality. such a livestock grazing or timber harvest. Other resources of the Ecological sites are named using a two-part, abiotic and biotic site may be highly significant in planning, developing, managing, name. The a biotic portion should be brief and should be based on and monitoring resources. Descriptions should include the follow­ such readily recognized permanent physical features as the kinds of ing, as appropriate, along with other pertinent information: soil, climate, topography, or a combination of three features. Some Full name. The full name of the site should be placed on each examples based on these criteria are Sandy Plains, Clay Upland. page of the description. Saline Lowiand, Gravelly Outwash Plains. Pumice Hills, Granitic Physiographic features. Physiographic features should include: Fluvial Lands, Basalt Plains, and Sandy Skeletal Moraine Lands. I. Position of site in the landscape, e.g., ridgetops, swales, The limited number of permanent physiographic features or south-facing slopes. other features make the repeated use of these terms inevitable. 2. Aspect, shape, length and steepness of slopes. Deep sands, for example, occur in areas of widely divergent climate 3. Range in elevation. and support different natural plant communities. Where repetition Special notation should be made concerning susceptibility to occurs. applicable adjectives descriptive of precipitation zone, bio­ surface water flow, depth of water table, and similar characteristics. tic name or other features must be used to differentiate ecological Climatic Features. Pertinent climatic features include: sites. 1. Extremes and seasonal distribution of annual precipitation The biotic name should consist of two (sometimes three) com­ and, where available, probability of receiving selected amounts of mon names of characteristic, diagnostic, or prominent species. monthly precipitation. Where one layer of vegetation exists, two names should be chosen, V 2. Temperature characteristics, (average length, beginning and e.g. Western wheatgrass/ Green needlegrass. Where more than one ending dates of growing season for major native forage species). vegetation layer exists, names should come from both (or three) 3. Other macroclimatic features such as storm intensity, wind layers. For example, Big sagebrush/ Idaho fescue or Ponderosa velocity, and drought cycles. pine/ Bitterbrush/ Idaho fescue. 4. Microclimatic features such as frost pockets and cold air An example of a complete ecological site name might be a drainage that typify site and relate to its potential. Ponderosa pine/ Bitterbrush/ Idaho fescue-Upland stoney loam Potential Natural Vegetation. Several kinds of information are ecological site or a Upland stony loam-Mountain big sagebrush/­ needed to determine the potential natural vegetation for each Bitterbrush/ Idaho fescue ecological site. The order of the biotic ecological site; these include: and abiotic portion of the name is unimportant. However, the I. Structure and appearance of the community. same combination of biotic and a biotic names should not be used 2. List of major plant species and relative proportion of each in to identify different ecological sites, except as described below. the community. Criteria for identifying relative importance of Ecological sites with similar soils and topography may exhibit plant species must be defined. These criteria may include fre­ significant differences in their potential natural plant communities quency, cover, yield or other measures of relative dominance. because of climatic differences. Two or more ecological sites may be recognized and distinguished by the inclusion of the precipita­ 3. List of tree species, the approximate canopy cover of the overstory, and site index for forested sites. tion zone (PZ) as a manifestation of climate in the naming of sites 4. Brief description of common patterns of succession and ret­ when quantitative evaluation indicates that the amount of vegeta­ tion produced is significantly different, e.g., Mountain big sagebrush/ rogression, including casual agents associated with each pattern Idaho fescue-Upland stony loam ecological site, 16-19 PZ and and a list of plant species most likely to invade if cover deteriorates. Mountain big sagebrush/ Idaho fescue-U pland stony loam ecolog­ Total annual herbage and browse production. Show estimated ical site, 20-25 PZ. If the soils have been described adequately, it is total annual herbage and browse production as median air-dry likely that the soils would differentiate the sites in this case. production and the fluctuations to be expected during favorable and unfavorable years. In areas where examples of the potential Correlating Ecological Sites natural plant community are not available, cite the production of Ecological sites are correlated on the basis of species composi­ the highest ecological condition class for which examples are avail­ tion, production of the potential natural plant communities, and able. Production of individual seral stages is desirable and should soils. Sometimes it is necessary to extrapolate composition and be incorporated as information becomes available. plant production data from one soil to describe the plant commun­ ity on a similar soil for which no data are available. It must be Soils. The name of all soils associated with the site should be given. noted that delineation of two distinct soil taxonomic units does not Because soil names are subject to change, list the soil taxonomic automatically require recognition of two ecological sites. Likewise, units associated with the site on a separate sheet that can be easily V updated. Other information needed includes:

4 GUIDELINES AND TERMINOLOGY FOR RANGE INVENTORIES AND MONITORING I. Key properties of soils associated with site, especially those common definitions, common units of measure, and a minimum that significantly affect plant-soil-water relations. standard data set so that meaningful analysis, evaluations, and 2. Amount of ground cover necessary to protect soil from accel­ comparisons can be made. These needs are being addressed by erated erosion. other working groups. Compatibility of storage format and some Site Interpretations. Site interpretations should explain the degree of computer compatibility also are required. potential importance of the site for each of its major uses, including There are a number of data systems in operatfon. The Soil grazing, timber, habitat for wildlife, recreation, esthetics, and Conservation Service has Range Data System (RDS), The Forest watershed quality. Interpretations should note: Service has Range Management Information System (FSRAMIS) 1. Kinds and classes of livestock and seasons of grazing best and a number of regional systems, and a large number of universi­ suited to the site. ties have developed systems of their own. The system that currently 2. Wildlife species that inhabit the area, along with important has most universal application to a wide variety of ownerships and food and habitat plants for each major wildlife species. conditions appears to be the SCS-Range Data System. RISC does 3. Relative stability of the site for watershed species. not have the expertise nor is it in their charge to evaluate data 4. Unique properties important to management of the ecologi­ storage and retrieval systems, but we recognize that a system is cal site. needed. A single system would greatly aid the correlation process 5. Resource value ratings for expected uses for the known eco­ between states and agencies. logical stages as well as the potential natural community. Other i,iformation. Other pertinent interpretative and descrip­ Collection and Interpretation of Data for tive information may be included. Management Planning and Resource Classification in Relation to Mapping Development of an ecological site classification and the actual Assessment mapping of ecological sites are two distinct processes. Classifica­ Guiding Principles tion is the process of ordering and arrangement of land into Data Collection Compared to Data Interpretation groups-ecological sites-based on similarities and dissimilarities Measurement of range attributes and the interpretation of mea­ in data for the parameters measured. The classification and des­ surement data for management and planning purposes are two cription of ecological sites does not involve mapping. However, distinct processes. Selection of attributes for measurement is inventory and mapping are often used to help refine classified sites. determined by characteristics of the resource, the interpretation to Mapping of ecological sites involves the systematic evaluation of be made, and cost or time requirements. Once attributes are the landscape to stratify it into ecological sites based on similarities selected, their measurement should be as objective as possible. The to the ecological sites described in the classification. attribute measured must be adequately defined and the accuracy Mapping of Ecological Site Complexes and precision of measurement reasonably specified. Data collec­ tion should be objective and free of value judgements. Within a specific area of land, ecological sites can be delineated on maps as a single site, with inclusions of small areas of other sites, Measurements are of value to the manager only when they are interpreted in terms meaningful to goals and human values. Inter­ or as a complex of two or more sites that are so interspersed that pretation involves two basic processes:(]) translation of measure­ separate delineation would not be practical or meaningful. ments into outputs or values, and (2) a comparison or prediction of The names of ecological sites are generally shown on the map within each delineation, but if space is limited, the names can be outputs or values obtained by present management relative to those obtained under an alternative management strategy. represented by appropriate symbols explained in a legend. If a delineation represents a single site or a major site that has minor Kinds of Interpretations inclusions of other sites (making up as much as 15% of the delinea­ Many kinds of interpretations may be needed for planning and tion), the name of the major site is used. If a delineation represents decision-making, especially for multiple use management. We do more than one site (i.e. a complex) the name of each major site and not attempt to discuss them all. A few basic interpretations consi­ the approximate proportion of each is indicated. For example, dered necessary in almost any range inventory or monitoring pro­ Bluebunch wheatgrass/ Idaho fescue-Loamy upland, 65%; Blue­ gram will be defined, the types of data necessary will be specified, bunch wheatgrass/ Sandberg bluegrass-Shallow loam, 35%. and the general approach to collecting the data outlined. Establishment of a uniform minimum acreage for site delinea­ One kind of interpretation needed relates to range condition, the tions is usually not practical. Need for mapping detail varies in present status of the vegetation and soil resource. Another impor­ accordance with relative productivity of a site, size of management tant interpretation is estimation or documentation of change unit, map scale, intensity of use patterns and information require­ (trend) in certain vegetation or soil characteristics and reasons for ments. Land that has relatively high productivity is usually observed changes. Other interpretations necessary for economic or mapped in greater detail than that oflow productivity. Land that is environmental evaluation of present or proposed grazing man­ suitable for many alternative uses also may be mapped in more agement include grazing capacity under alternative management detail. systems, need for and location of range improvements, and com­ Intensity and details in mapping ecological sites, therefore, are peting or complementary relationships among different uses. determined locally on the basis of the kinds of land and the needs for planning. Major consideration is given to management needs Detail Required The need for detailed data collection will vary as the scale and for various uses of the land, including but not limited to timber purpose of inventory and monitoring changes. For broad assess­ harvest, livestock grazing, habitat for wildlife, and watershed pro­ ments the number of sample locations or the accuracy and detail of tection. To insure compatibility of mapping units, soil scientists mapping may have more effect on validity of the results than will and vegetation specialists should work closely together to define precision of measurement at a given location. Even at the man­ mapping units that insure that soils and vegetation information is agement level, e.g., ranch or allotment, simple ranking or assigning coordinated. to classes may furnish data of sufficient accuracy for the decisions Data Storage and Retrieval required. It is primarily for monitoring trends where quantitative Because of the extremely large volume of existing information data amenable to calculation of confidence intervals are needed. and the additional data accumulated annually, some means of Importance of Ecological Sites automated data storage and retrieval is necessary. This will require All data collection and interpretation should be based on ecolog-

GUIDELINES AND TERMINOLOGY FOR RANGE INVENTORIES AND MONITORING 5 ical sites. For broad scale assessments, data from different ecologi­ sion after disturbance does not always reestablish the "original" cal sites can be grouped according to similarity in present or vegetation. potential vegetation, soils, geology, present land use, land owner­ Vegetation Rating. The vegetation rating should be based on ship, administrative policy or jurisdiction, or any other useful kind, proportion and amount of plants in the present community criteria. However, data collected without reference to ecological relative to PNC. Specifying an amount implies that an absolute site, even within one mapping unit, cannot be interpreted in terms measure of plant species is required, rather than a species list or the V of site potential because there is no basis for disaggregating back to composition alone. The present community can be compared with the site level. the PNC in absolute terms by several kinds of measurements, such as cover, density, weight or frequency. Use of cover or weight Long-term vs, Short-term Objectives results in a few major species controlling the degree of similarity to Total inventory may be required for long-range management PNC. The importance of common, but small, plants is emphasized planning. Such inventories may be required only infrequently, 5 to to a greater degree when using density or frequency. Most agencies 20 + year intervals. The long-term effects of implementing a man­ currently use weight by species as a basis to estimate vegetation agement plan are ultimately expressed as changes in range condi­ condition, even though weight data arc more time-consuming to tion and trend. Such changes may be documented only by monitor­ obtain and more subject to error than some other attributes. Fre­ ing over a long period of time. Documenting causes of such quency requires less time to obtain objective and repeatable data. changes, including amount and distribution of utilization, weather Some agency personnel and researchers are developing guides to records and actual use may provide useful information for short­ use frequency for rating vegetation status in place of weight. term adjustments in management. To compare present vegetation to PNC both communities must Range Condition he described in terms of the same attributes. If the present com­ General Considerations munity is described in terms of frequency, the PNC also must be Range condition often has been called "the state of range expressed in terms of frequency. Although the degree of similarity health." There are two approaches to range condition assessment. of present vegetation to PNC will vary somewhat depending on the They involve different concepts of range condition and hence attribute chosen for characterization, in most cases the difference require different evaluations: (1) an evaluation of whether the will not be significant enough to alter interpretation. long-term productive potential of sites is being maintained, and (2) The degree of similarity between the present vegetation and an evaluation of the present level of production relative to the PNC can be calculated by a coefficient of community similarity potential production for a given objective or use of the site. (2w/a+h) where a is the sum of species values for measured In the past, procedures developed for interpreting range condi­ parameters of present vegetation, bis the sum of values in the PNC tion have dealt with one or the other of these approaches, or have and w is the sum of the values common to both (Table 1). Although combined elements of both. Using one term-range condition­ other measures are available, this index is widely accepted and for both concepts has led to confusion. Since the term "range used. Vegetation ratings may be expressed as a percentage. If condition" is so much a part of the language and literature of range vegetation rating classes are used. they should be four in number, management, we recommended that it be used in the generic sense corresponding to 0-25%, 26 50%, 51 75% and 76-100% of the covering the broad concept of range condition and that two terms PNC standard. The reasons for using four classes are: ( I) use of with specific meanings (ecological status and resource value rating) more classes implies a precision of measurement that may not be be used for the two separate concepts. achievable, (2) use of fewer classes would make the rating insensi­ tive to change, and (3) use of different numbers or definition of Ecological Status classes among agencies makes comparisons difficult. Definition. Ecological status is use-independent and is defined as the present state of vegetation and soil protection of an ecologi­ Table 1. Example of calculation of ecological status of vegetation using cal site in relation to the potential natural community (PNC) for coefficient of community similarity on herbage production data. the site. Ecological status is evaluated by two independent ratings, one for successional stage of vegetation and one for stability of the soil. The vegetation rating is an expression of the relative degree to Herbage Production (!bi acre) which the kinds, proportions and amounts of plants in a commun­ Present Potential :\'atural Amount in ity resemble that of the PNC. The soil rating is the relative amount Species Vegetation Community Common of protective cover furnished by vegetation and litter compared to I JOO 200 IOO the level of protection provided by the PNC for the site. Both 2 50 100 50 vegetation and soil are rated relative to site potential. This means 3 JOO 50 50 that proper definition and identification of ecological sites is a 4 200 50 50 prerequisite for interpreting ecological status. 5 50 50 50 The kinds, proportions and amounts of plants in the PNC for Total a=500 b = 450 w =300 each ecological site are determined by sampling stands represent­ Condition Score (2w /a+b) =600/950 =.63 = Late Sera! ing the PNC. Potential natural community (PNC) is defined as "the biotic community that would become established if all succes­ Vegetation rating classes should not be given names implying sional sequences were completed without interference by man value judgements, such as good or poor. Ecological vegetation under the present environmental conditions." We prefer the con­ ratings do not have any reference to values produced or to man­ cept of PNC to that of climax (see definition. p. IO) because it agement goals. Vegetation rating classes may be referred to as early recognizes past influences by man, including past use and intro­ seral, midseral, late seral and PNC or early-, mid-, late-succession duced exotic species of animals or plants. Man's influence is and PNC. The use of "P"IC" for the entire upper class (76-100%) excluded from the present onward to eliminate the complexities of seems justified. Experience of ecologists indicates that variability management. The concepts of climax and PNC both refer to a in plant composition within a "homogenous" community is suffi­ relatively stable community resulting from secondary succession ciently high that any similarity index in excess of 75-80% between after disturbance. Although man may or may not have caused the two stands would indicate that the samples could have been drawn disturbance, succession to climax or P"IC occurs without further from the same population. perceptible influence of man's activity. RISC prefers PNC because Soil Rating. Although the procedures described above for rating this term explicitly recognizes that naturalized exotic species may vt:getation status art: similar to approaches widely accepted and persist in the final stage of secondary succession and that succes-

6 GUIDELINES AND TERMINOLOGY FOR RANGE INVENTORIES AND MONITORING used in the past, no such generally acc<;pted procedure exists for could be described and their associated RVR's approximated. rating soil status. Thus, the suggestions for rating soil status should Data Collection for General Assessment be taken more as a hypothesis than an accepted procedure. Classification of present vegetation provides a basis for estimat­ Deterioration of site potential comes mainly from soil erosion. ing succcssional status of the vegetation and resour~.e value ra.tings Some erosion is natural, and it is important to distinguish between when considered in the context of the ecological site on which 1t natural and induced erosion. Even natural erosion can lead to site occurs. Classification of present vegetation should be based on deteriortion, but as managers we are primarily concerned with criteria related to seral stage and major resource values, such as accelerated rates of erosion induced by management. Since natural forage for cattle, fuelwood production, or sagegrouse nesting erosion rates vary from one ecological site to another, rates consi­ cover. In some cases a classification based on dominant species in dered to represent accelerated erosion must be defined on a site by each layer may be adequate. Jn other cases, additional classes site basis. based on density or cover of some species or life forms may be Erosion rates are difficult to measure directly. On a given site, necessary. erosion hazard is related mainly to effective vegetation, litter and The attributes chosen to classify present vegetation depend on other soil surface cover. Minimum amounts of vegetation and litter management needs and uses considered. Auributes m~y include cover should be determined for each ecological site by comparison cover, relative standing crop, density or height dependmg.on the with areas considered to represent natural erosion rates for the site. character of the vegetation and the resource values considered. These comparisons or standards may have to be adjusted for slope. Visual classification of vegetation types and/ or condition classes, The soil rating can be expressed as the ratio of vegetation-litter usually with preliminary mapping on aerial photos, has formed the cover on the test location to vegetation-litter cover on a reference basis for most range inventories in the past 50 years. For genernl area representing natural erosion for the site. assessment purposes, attributes identified through remote sensmg Resource Value Rating with minimal ground checking arc preferred for economic reasons. RISC proposes a third type of interpretation or rating related to Soil status may be estimated in cover classes directly or by use of the general concept of range condition-Resource Value Ratmgs a score sheet similar to those used by several agencies for many (RVR). RVR is the value of vegetation or other features of an vears. If certain vegetation types on a given ecological site consist­ ecological site for a particular use or benefit. RV R's may be estab­ ~ntly provide adequate soil protection while others .do not, then lished for each plant community capable of existing on an ecologi­ classification of present vegetation alone may provide adequate cal site, including exotic or cultivated species. data on soil status for general assessment. On a given ecological site, each use (or potential use) ha.s a Data Collection for Detailed Analysis separate RVR. This interpretation may be based on species, More precise quantitative information on ecological status, growth form, foliage type or other criteria. For example, an R YR resource value ratings, or soil status generally reqllires sampling for forage useful for a particular kind or class of animal and season attributes such as cover, standing crop, density or frequency usir,g of use could be based on proper use factors (PU F's). The RVR point samples, lines, plots or other measuren:'ent techniques. Selec­ could be based on production, cover, density or frequency of plants tion of an appropriate attribute and technique for measurement with different forage values. RVR 's should be derived for various depends on the nature of the vegetation and purpose of the kinds of animals. An RVR for nesting cover useful to a particular inventory. . . species of bird might be based on density or cover of plants. of Quantitative data are not always more accurate or less subJect1ve certain height or size class, perhaps without regard to plant species. than the ranking or general estimation procedures descnbed An RVR related to scenic beauty might be based on abundance of above. Accuracy of quantitative data can be improved by reducing flowering plants, species with fall color, evergreens, diversity of bias in sample selection and by other means. The high degree of growth form, or other attributes. . . . . variability in range vegetation and soils results in low precision The RVR is intended for rating of a spec1f1c locallon without except on intensively sampled monitoring locations. regard to its relationship to other sites or vegetation .types in the area. For instance, the RVR for summer cattle grazing 1s based Measuring and Interpreting Trend only on the vegetation present at that specific location. If RVR is General Considerations expressed in terms of forage acre factor or acres/ AUM, these te.rms Trend is the directional change in kind, proportion and/ or should not indicate carrying capacity, because such cons1dera\Jons amount of plant species, or soil characteristics. Trend may be as slope, distance from water,juxtaposition of other sites ?r vege­ interpreted in both an ecological context and in terms of resource tation types are not included. The RVR is a measure of smtab1hty value. The principal criteria to mterpret trend m ecolog1cal status or usefulness of the vegetation of an ecological site for a specific use should be the vigor and reproductive success of plant species that assuming full use is possible. Off-site considerations do not influ­ are indicative of later seral stages as compared to those of an earher ence it. seral stage for the site. The potential natural community is used as The RV R may be used in an absolute sense, i.e., without refer­ the reference plant community and trend is described as toward or ence to site potential. For instance, the forage acre factor is nothing away from the potential natural community, or not ~pparent. more than a rating of the amount and usefulness of plants for Trend of soil surface conditions is interpreted from evidence of grazing at the present time. It is usually desirable for planning accelerated soil erosion. purposes to know how the present RVR relates to the potential Trend in RVR, when compared to management ojective(s), RVR for that use on the site. For this purpose the present R YR can refers to the change in utility of vegetation at a particular location be expressed asa percentage of the highest RVR a~tually measured for a specific use. The trend of a particular resource value may be on the ecological site. Or, in the absence of areas Judged to repres­ up, for another use the trend may be down, and not apparent for ent site potential, the potential R YR could be estimated based on still another. The direction of trend is based on whether the measurements of a similar ecological site. The potential RVR for a changes in vegetation and soil conditions are desirable or undesir­ use is the management site potential for that use (see definition, p. able for specific management objectives. 11 ). Bccau.se of the dual interpretation of trend, the type of trend Jn planning it is also necessary to calculate trade-offs among us~s must be specified. either ecological status or resource value for a or potential uses, that is, to estimate RV R's for a number of uses m specific use or both. . . response to a management alternative. Fo~ example,. for each Trend can be interpreted at vanous levels dependmg on the ecological site, an array of possible vegerat10n types, mcludrng amount of detail needed or available. At the macroscale, changes seeded stands, obtained by various management alternatives,

GUIDELINES AND TERMINOLOGY FOR RANGE INVENTORIES AND MONITORING 7 in lifeform of dominant species, gross changes in cover and density technique has many advantages for the determination of trend. of shrubs and trees and marked changes in ground cover can be Frequency data are simple to obtain, objective, statistically relia­ monitored by use of remote imagery and aerial photography. In ble, and relatively inexpensive to collect. As with density and basal such situations, documentation of trend in vegetation and soil area, frequency is rather insensitive to seasonal and yearly varia­ surface conditions can be achieved without detailed measurements tion, a desirable attribute when monitoring long-term vegetational of kind, proportion and amount of major species. Often these trend. Statistically reliable frequency data are less expensive to changes occur over a period of years or decades, unless catastro­ obtain then density or basal area. V phic alterations due to fire or other severe disturbances occur. Less For trend interpretation frequency data generally are inter­ obvious changes in trend are difficult to detect and document preted by analyzing differences in frequency of individual species unless critical on-site measurements of vegetation and soils are over time on a specific ecological site. Key species may be selected made. for dominance, desirability for certain uses or indicator value in the community. Although rooted frequency is not consistently related Apparent vs. Measured Trend to density, basal cover, or distribution pattern, positive statements Apparent trend is the interpretation of direction of change based about trend can be made if change in frequency occurs between two on the evidence that is obtained at a single observation. It should sampling periods. An increase in frequency indicates that new only be done by an experienced observer and should always be individuals have become established. This change in a preferred clearly identified as apparent trend. species can be interpreted as being desirable or showing an upward Measured trend is a quantitative assessment of change based on trend, and a similar change in undesirable species would indicate a repeated measurements over time, of the kind, proportion and/ or down trend. If individual species of interest are recorded by catego­ amount of plant species and soil surface properties. It provides ries. e.g., newly established (exclude current seedlings), mature, quantitative data for interpreting the direction of change, often and senescent, the change in frequency can be interpreted as to how before it is detectable by repeated ocular examination or repeated the age classes within species are undergoing change. photographs over time. Measured trend provides early feedback to Frequency is dependent upon plot size, plot shape and distribu­ indicate if management objectives are being reached. progress is If tion pattern of the individual species. Thus, change in frequency is unsatisfactory, modification in management practices is needed. difficult to interpret unless the same size and shape of quad rat are Measuring Trend used in each sample period compared. To detect statistically signif­ Early detection of trend involves some risks because vegeta­ icant change, the frequency value should fall in a range of20 to 90% tional properties naturally fluctuate widely within and among for sampling sensitivity. years because of climatic variability and other influences. These Frequency data, like all other quantitative measures, cannot be normal fluctuations must be considered when determining trend. used to evaluate ecological status or resource values rating before Sampling error further confounds the problem of early detection standards are established through prior study. Once standards of trend. have been established frequency data can be used for a rapid, Many techniques are available to monitor trend and each has objective and consistent method of trend analysis. Frequency can pros and cons. A review of these techniques leads to the conclusion be used to indicate a real change in vegetation but it cannot be that plant yield, cover and density are not reliable measures of interpreted to indicate a specific amount nor the specific property trend, particularly for herbaceous species. of a species unless additional information is available. In spite of its Because of the complete renewal of above-ground growth annu­ limitations, frequency is the easiest, least costly and most reliable ally, the varied growth forms, and phenologic differences of indi­ kind of quantitative data to collect to detect change in the role of a vidual species, no single measure of herbaceous plants is best for species in a community. determining trend. Each method has a deficiency. Plant produc­ Frequency cannot be efficiently or meaningfully used in all tion and foliage cover data are highly variable, both seasonally and vegetation types. It is more meaningful in perennial grasslands, annually. Basal cover is more stable, but difficult to measure for and for interpreting change in the herbaceous and small shrub many species. Plant density is a difficult parameter to sample component in shrub-grass vegetation. RISC recommends the use adequately because of varied growth forms and at times the diffi­ of frequency for monitoring trend for these vegetation compo­ culty in identifying what constitutes an individual plant. nents. For large woody plants, canopy cover and density should be Because of stand variability, obtaining an adequate sample of the basic measurements to monitor trend. yield, cover or density with sufficient statistical reliability to detect Soil surface condition can easily be obtained along with fre­ trend is extremely difficult. For example, the number of samples quency by fixing points on the sample frame to record hits on bare necessary to obtain precision of± 10% at a probability of 95% is ground, litter, gravel, total basal cover of vegetation and other frequently unreasonable for most land managers (Table 2). characteristics of the soil surface. However, this method will not usually adequately sample basal cover of individual species Table 2. Number of sample units required to estimate vegetation parame­ because of insufficient number of observations. ters with sampling precision ± 10% at probability or 95%. Accuracy, Precision and Probability Statements Parameter Type of Sampling Number Regardless of the type of data collected to evaluate vegetation Measured Unit Required Source change, interpretation should be supported with reliable statistical Cover 100 sq. ft. plots 46-200 Costello and analysis. Vegetation parameters are estimated by measurement Klipple, 1939 from sampling. Accuracy concerns the nearness of the estimated Cover 100 ft. line intercept 44 Hyder and Sneva, value to that of the actual value. Precision refers to repeatability of transects 1960 the sample estimate. High precision suggests a high degree of accuracy, but this is not necessarily the case when dealing with Cover Points 2400-3600 Clark. et al., 1942 vegetation. Density 96 sq. ft. or 38 and 174, Laycock. 1965 Precision and probability statements are functions of sampling 9.6 sq. ft. plots respectively intensity and population variability. High precision in vegetation Yield 1.917 sq. ft. plots 28-193 Scoop and sampling is generally very costly to obtain because of the large Mcilvain, 1963 number of samples required. For trend analysis a compromise Frequency of occurrence is a measure of the spatial distribution between sampling cost and the risk of an incorrect interpretation of of a species, i.e., its distribution in the community. This sampling data suggests that a precision of± 20% of the mean at a probability of 80% should be the minimum acceptable level. Increasing the V

8 GUIDELINES AND TERMINOLOGY FOR RANGE INVENTORIES AND MONITORING probability to 90% would require an increase in sampling effort of annual, measurement. The location chosen should be where collat­ about 50%. However, specification of an adequate level of statisti­ eral information relative to management objectives can be obtained. cal reliability of data will greatly enhance acceptance of related Establishment ofa continuous trend in soil/vegetation characteris­ decisions. tics in relation to weather, utilization, actual use and other varia­ bles will support a more accurate interpretation of data gl!thered Interpreting Trend Data on an infrequent basis elsewhere. Measured or observed changes in kind, porportion and/ or Another strategy is to pay special attention to designed compari­ amount of plant species on a site, in soil characteristics, or in sons among trend locations. For instance, if vegetation cover is animal populations, are interpreted as trend in ecological status or declining on numerous trend locations irrespective of the man­ resource value ratings. Trends in ecological status or R VR 's should agement system, it may be assumed that weather or factors other establish whether present management is resulting in changes than management are responsible. However, if cover of forage toward or away from management objectives. In order to decide if species declines on an ecological site in one management unit but a change in management is needed to reverse undesirable trends or increases or is static on the same ecological site in an adjacent unit, to accelerate desirable ones, it is important to try to establish a change in management is indicated. causes for trends. Several guidelines for collection and interpreta­ tion of trend data follow. Management Planning lmerpreting Trend at One Location. Differences in measure­ General Considerations ments obtained at different dates on the same location because of A major use of inventory and monitoring data is in development, sampling error, personal bias or lack of adequate training should evaluation and revision of management plans. Inventory and mon­ be minimized, The location and size of the sample area must be itoring data furnish only part of the information needed for plan­ adequately specified. The sample area should not involve more ning purposes. The rest must come from economic analyses, public than one ecological site and sampling design should account for input, and legal or policy constraints concerning priority of uses, heterogeneity in plant pattern, topography, and microclimate. consideration of threatened and endangered species, or other value Sampling method should be amenable to statistical analysis and judgments. Planning may be directed at the individual range (graz­ establishment of confidence intervals. Attributes measured must ing allotment) or broader regional/national assessments. be defined in objective terms such that observer bias is minimized. Management planning requires a knowledge of the present sta­ Interpreting Trend in a Management Unit. it is rarely feasible, tus of the resource in terms of resource outputs and the trade-offs nor is it necessary, to obtain a statistically valid sample of an entire among values produced by alternative management strategies. It management unit (pasture, allotment) for trend monitoring pur­ also requires estimation of different resource values produced by poses. Rather, each monitoring location should be carefully management prescriptions designed to alter the present situation selected with specified objectives developed for each location. Data and of costs of achieving or maintaining such conditions. Research, from different sample locations should not be combined until after studies and experience furnish the basis for making analyses and interpretation of each location is made and then only if it is certain predictions. no information will be lost. The overall trend on a management Identification of ecological sites and their associated RVR's unit cannot be determined by averaging trend data from various form the basis for development of alternative management strate­ locations except perhaps under cases of extremely good or poor gies. Inventories of ecological sites and present plant community management. types are necessary for each planning unit. Such inventories gener­ Collateral Data. Collection of collateral data to aid interpreta­ ally involve maps, i.e., complete inventory, especially at the tion of soil or vegetation change is essential. Weather data col­ ranch/ allotment or project level. For generalized resource assess­ lected on or near each monitoring location are highly desirable. ments sampling data are most generally used. It is possible that Storage gages read monthly or seasonally can be used for precipita­ classification and/ or mapping may be replaced or complemented tion. Max-min thermometers at selected locations may help in the future by gradient modelling or direct remote sensing proce­ explain extreme events. Actual use records of livestock and of dures. Such methods are not generally well-developed for range­ wildlife should be maintained. Utilization should be measured on land uses at this time. each monitoring location whenever trend data are collected and at other times when appropriate and feasible. Utilization data should Grazing Management be collected to represent the same location as other vegetation Estimation of present or potential grazing capacity is an impor­ data. A method should be used which provides quantitative esti­ tant interpretation. Grazing capacity is not a site characteristic as mates of either percentage utilization or residue remaining. Exam­ are RVR's but, rather, is a characteristic of the management unit ples are the grazed-class, stubble height, percentage plants or twigs (pasture, allotment or ranch) as a whole, including the pattern of grazed methods, or any number of other techniques suitable in ecological sites with their respective present or potential RVR's as different situations. Caged plots may be used to ensure that some well as the level of investment oftime, money and energy economi­ ungrazed plants are present for making comparative kinds of cally feasible or compatible with management objectives. measures, but generally the number of cages necessary to obtain Utilization surveys can identify key areas, problem areas and/ or precise estimates make their use impractical for direct determina­ opportunities to improve livestock distribution. Utilization mea­ tion of utilization by harvesting. Observations on populations or surements on properly selected key areas can be used to estimate occurrence of rabbits, rodents, insects, fire or other disturbances proper stocking levels under current management, when actual use also should be made. and climatic data are available. These procedures are recom­ Frequency for Collection of Trend Data. In order to establish mended for estimating needed increases or decreases of livestock the reality of trends and the causes for them, it is highly desirable to when no trend data are available and the need for adjustment is measure trend frequently. This is particularly important where fairly obvious. Utilization data also may be used to furnish quan­ management problems exist, but causes are debatable. Annual titative estimates ofrequired increases or decreases in stocking rate trend measurement is ideal. There is often value in measuring when trend data indicate need for adjustments, or to furnish esti­ utilization more than once a year. However, budgets and man­ mates of additional grazing capacity obtainable by alternative power often dictate that trend monitoring can be done only at grazing systems or range improvements. intervals of two, three, five or more years. In this case a monitoring For management planning or for monitoring of adherence to strategy designed to aid in accurate identification of trends and grazing management systems, highly quantitative methods of their causes is especially important. Two ways are suggested to estimating utilization are not necessary nor feasible. Estimation of overcome the problem of infrequent measurement. utilization on major forage species in 3 to 5 classes usually provides One strategy is to select a few locations for frequent, preferably

GUIDELINES AND TERMINOLOGY FOR RANGE INVENTORIES AND MONITORING 9 a reliable measure of the pattern of use in a management unit. This sites. For each site the change in plant community and the R VR is usually sufficient to identify problem areas or opportunities to associated with the new community must be predicted as a basis for improve distribution. estimating future forage production. Forage prcduction estimates There is no feasible wa, to precisely estimate either present or from inventories of present vegetation or projection of future potential grazing capacity based on a one-time inventory alone. production must be adJusted for present or projected distribution The most satisfactory, and ultimately, the only way to establish patterns of livestock. class or kind of livestock, season of use, and proper stocking rates is to monitor trends in ecological status and needs of other forage users. V RVR's under a given stocking level. If measured trends indicate Forage inventories are time consuming and subject to error progress toward management goals, stocking levels under present because of variability in herbage production in time and space. management can be maintained or increased. Trends contrary to Inventory of forage production can be justified only where local management goals indicate that present management is unsatisfac­ experience and knowledge is lacking or for relatively small, uni­ tory. If the intensity, frequency, distribution or duration of live­ form and very intensively managed areas such as irrigated pasture. stock grazing can be established as the cause for undesirable Utilization studies and the monitoring of trend should be used to trends, than a change in one or more of these is indicated. Changes arrive at proper numbers of livestock rather than forage inventories. resulting from other causes, such as wildlife influences, fire or Other Management Considerations absence of it, weather, plant succession, etc. will require changes in The preceding discussion of management planning has focused management designed to correct such trends or to accommodate on planning of livestock grazing. However, data collected and the undesirable trends. interpretations made will serve for planning other types of range Forage inventories can provide baseline data for management use as well. In fact, these principles and approaches should serve planning especially when an area is presently ungrazed, if grazing equally well for any use dependent on the soil-vegetation resource history is available, or if alternative management is proposed. provided the information needs of these uses are built upon the Estimates of potential grazing capacity based on forage inventories classification of ecological sites and present vegetation, selection of can be used for economic evaluation of range improvements or inventory and monitoring techniques, and sampling design con­ vegetation manipulation. design of grazing systems, accommodat­ cepts previously discussed. ing needs of other consumptive users. and for other planning Site classification and vegetation measurements are intended to purposes. be as use-independent as possible. with use-oriented features res­ Forage inventories can be based on maps, samples of present tricted to interpretation. As knowledge and technology change, vegetation types, on resource value ratings known to be associated and as the number of outputs or resource values increase, classifi with these types, or established by production studies designed for cation of sites and vegetation as well as data collection procedures this purpose. Projection of future forage production from present may need to be refined and modified. or alternative management practices must be based on ecological Proposed Standard Terminology for Inventory, Classification and Analysis of Range Ecosystems Allowable Use-the degree of utilization considered desirable and Classification-the assignment of items or concepts into classes attainable on various parts of a ranch or allotment considering the based on similarity of selected attributes. present nature and condition of the resource, management objec­ Climax-the final or stable biotic community in a successional V tives, and level of management. series; it is self-perpetuating and in equilibrium with the physical Apparent Trend-an interpretation of trend based on a single habitat_ (Odum, 1971 ); the assumed end point in secondary observation. Apparent trend is described in the same terms as succession. measured trend except that when no trend is apparent it shall be Community-an assemblage of populations of plants and/ or described as none. animals in a common spatial arrangement. Available Forage-that portion of the forage production that is Comparison Area-an area with a documented history and/ or accessible for use by a specified kind or class of grazing animal. condition that is used as a standard for comparison. Bare Ground-all land surface not covered by vegetation. rock or Critical Area-an area which must be treated with special consid­ litter. (c.f. ground cover) eration because of inherent site factors, size, location, condition, Basal Area-the cross sectional area of the stem or stems of a plant values, or significant potential conflicts among uses. or of all plants in a stand. Herbaceous and small woody plants are Cryptogam-a plant in any of the groups Thallophytes, Bryo­ measured at or near the ground level; larger woody plants are phytes, and Pteridophytes-mosses, lichens and ferns. measured at breast or other designated height. (Syn. basal cover). Density-numbers of individuals or stems per unit area (Density Bedrock-in-place, solid rock exposed at the surface of the earth does not equate to any kind of cover measurement). or overlain by unconsolidated material. Ecological Response Unit-synonymous to ecological site. Biomass-the total amount of living plants and .,nimals above and below ground in an area at a given time. Ecological Site-a kind of land with a specific potential natural community and specific physical site characteristics, differing from Browse-(n) the part of shrubs, woody vines and trees available for other kinds of land in its ability to produce vegetation and to animal consumption. (v) to search for or consume browse. respond to management. Canopy Cover-the percentage of ground covered by a vertical Ecological Status-the present state of vegetation and soil protec- projection of the outermost perimeter of the natural spread of tion of an ecological site in relation to the potential natural com­ foliage of plants. Small openings within the canopy are included. It munity for the site. Vegetation status is the expression of the may exceed !00%. (Syn. crown cover). relative degree to which the kinds, proportions and amounts of Capability Area-Synonymous with ecological response unit. plants in a community resemble that of the potential natural com­ Carrying Capacity-the maximum stocking rate possible without munity. If classes are used, they should be described in ecological inducing damage to vegetation or related resources. It may vary rather than ultilitarian terms. Soil status is a measure of present from year to year on the same area because of fluctuating forage vegetation and litter cover relative to the amount of cover needed ~ production. (Syn. grazing capacity). on the site to prevent accelerated erosion.

10 GUIDELINES AND TERMINOLOGY FOR RANGE INVENTORIES AND MONITORING &osystem-a complete interacting system of organisms (i.e. Plant Associarion-a kind of climax p!::mt community consisting community) considered together with its environment. of stands with essentially the same dominant species in correspond­ Foliar Cover-the percentage of ground covered by the vertical ing layers. projection of the aerial portion of plants. Small openings in the Potential Natural Community-the biotic community that would canopy and intraspecific overlap are excluded. Foliar cover is become established if all successional sequences were completed always less than canopy cover; either may exceed 100%. without interferences by man under the present environmental conditions. Forage-(n) browse and herbage which is available and may pro­ vide food for grazing animals or be harvested for feeding. (v) to Productivity-the rate of production per unit area, usually ex­ search for or consume forage. pressed in terms of weight or energy. Forestland (Forest}--land on which the vegetation is dominated Proper Use-a degree of utilization of current year's growth which, by trees. Lands shall be classified forestland if the trees now present if continued, will achieve management objectives and maintain or will provide 25% or greater canopy cover at maturity. Lands not improve the long-term productivity of the site. Proper use varies presently forestland that were originally or could become forested with time and systems of grazing. (Syn. proper utilization). through natural succession may be classified as potential natural Range Condition-a generic term relating to present status of a forestland. (Schwartz, Thor and Elsner. 1976.) c.f. tree, potential unit of range in terms of specific values or potentials. Specific natural community). values or potentials must be stated. (See ecological status and Grassland-lands on which the vegetation is dominated by grasses, resource value rating.) grasslike plants, and/ or forbs (c.f. dominant). Non-forest land Range Jnventory-(v) the systematic acquisition and analysis of shall be classified as grassland if herbaceous vegetation provides at resource information needed for planning and for management of least 80% of the canopy cover excluding trees. Lands not presently rangeland. (n) the information acquired through range inventory. grassland that were originally or could become grassland through natural succession may be classified as potential natural grassland. Range Sire--synonymous with ecological site when applied to rangeland. Gravel, Cobble, Stones-as defined in Soil Taxonomy (Soil Con­ servation Service 1975): gravel (2 mm-3 inches), cobble (3-10 Range Type- refers to, and only to, the 18 standard range vegeta­ inches), stones (over 10 inches). (Note: For standard range inven­ tion types recognized by the 1937 Task Force (Interagency Range tory procedures it is recommended that gravel smaller than 5 mm Survey Committee, 1937). in diameter be classed as bare ground in cover determinations.) Rangeland (Range}--land which supports vegetation useful for Grazing Management-the manipulation of grazing and browsing grazing on which routine management of that vegetation is animals to accomplish a desired result. through manipulation of grazing rather than cultural practices. Ground Cover-the percentage of material, other than bare Resource Value Rating (RVR}--thevalue ofvegetationpresent on ground, covering the land surface. It may include live and standing an ecological site for a particular use or benefit. R VR's may be dead vegetation, Jitter, cobble, gravel, stones and bedrock. Ground established for each plant community capable of being produced cover plus bare ground would total 100 percent. on an ecological site, including exotic or cultivated species. Habitat Type-the collective area which one plant association Riparian Zone-the banks and adjacent areas of water bodies, occupies or will come to occupy as succession advances. The water courses, seeps and springs whose waters provide soil mois­ habitat type is defined and described on the basis of the vegetation ture sufficiently in excess of that otherwise available locally so as to and its associated environment. provide a more moist habitat than that of contiguous food plains and uplands. (Adapted from Warner, 1979). Herbage-the above-ground material of any herbaceous plant. Savanna-a grassland with scattered trees, whether as individuals Key Area-a relatively small portion of a range selected because of or clumps, often a transitional type between true grassland and its location, use or grazing value as a monitoring point for grazing forest. use. It is assumed that key areas, if properly selected, will reflect the overall acceptability of current grazing management over the Sera/ Community-a successional community. range. Shrub-a plant that has persistent, woody stems and relatively low Key Species-( I) forage species whose use serves as an indicator to growth habit, and that generally produces several basal shoots the degree of use of associated species. (2) those species which must instead of a single bole. It differs from a tree by its low stature-Jess because of their importance, be considered in the management than 5 meters (16 feet) and nonarborescent form. (c.f. tree). program. Shrub/and-lands on which the vegetation is dominated by Leaf Area Index-sum of total leaf area expressed as a percentage shrubs. Nonforested land shall be classified as shrubland if shrubs of ground surface. Leaf area index may exceed 100%. provide more than 20% of the canopy cover excluding trees. Lands no presently shrubland that were originally or could become shrub­ Litter-the uppermost layer of organic debris on the soil surface; land through natural succession may be classified as potential essentially the freshly fallen or slightly decomposed vegetal natural shrubland. material. Standing Crop-the total amount or number of living things or of Management Site Potential-the kinds or levels of productivity or one kind of living thing in an area at a given time. values of a range site that can be achieved under various manage­ ment prescriptions. Stocking Rate-the number of specified kinds and classes of anim­ als grazing (orutilizing) a unit of land for a specific period of time. Pasture/and-grazing lands, planted primarily to introduced or May be expressed as animals per acre, hectare, or section, or the domesticated native forage species, that receive periodic renova­ reciprocal (area of land/ animal). When dual use is practiced (e.g. tion and/ or cultural treatments such as tillage, fertilization, mow­ cattle and sheep), stocking rate is often expressed as animal unit­ ing, weed control, and irrigation. Not in rotation with crops. s/unit of land or the reciprocal. Phytomass-total amount of plants (including dead attached Tree-a woody perennial, usually single-stemmed plant that has a parts) above and below ground in an area at a given time. (c.f. definite crown shape and characteristically reaches a mature height biomass).

GUIDELINES AND TERMINOLOGY FOR RANGE INVENTORIES AND MONITORING 11 of at least 5 meters ( 16 feet). Some plants, such as oaks ( Quercus Usable Forage-that portion of the forage that can be grazed spp.) may grow as either trees or shrubs. (c.f. shrub). without damage to the basic resources; may vary with season of use, species, and associated species. Trend-the direction of change in ecological status or resource value rating observed over time. Trend in ecological status should Vegetation Type-a kind of existing plant community with distin­ be described as toward. or away from the potential natural com­ guishable characteristics described in terms of the present vegeta- munity. or as not apparent. Trend in a resource value rating for a tion that dominates the aspect or physiognomy of the area. ~ specific use should he described as up, down or not apparent. Vigor-relates to the relative robustness of a plant in comparison Trends in RVR 's for several uses on the same site at a given time to other individuals of the same species. It is reflected primarily by may be in different directions, and there is no necessary correlation the size of a plant and its parts in rel_ation to its age and the between trends in R YR 'sand trend in ecological status. environment in which it is growing. Unsuitable Range-range which has no potential value for, or Yield-( I) the quantity of a product in a given space andj or time which should not he used for, a specific use because of permanent (syn. production). (2) the harvested portion of a product. physical or biological restrictions. When unsuitable range is identi­ fied, the identification must specify what use or uses are unsuitable (e.g .• "unsuitable cattle range").

Literature Cited Clarke, S.E., J.A. Campbell, and J.B. Campbell. 1942. An ecological and Odum, E.P. 1971. Fundamentals of Ecology, 3rd Ed. W.B. Saunders Co., gra7ing capacity :-:.tudy of the native grass pastures in southern Alberta, Phil. & London. 547 pp. Saskatchewan, and Manitoba. Canad. Dept. Agric. Tech. Bull. 46, 31 pp. Schwartz, C.C., E.C. Thor, and G.H. Elsner. 1976. Wildlands planning Costello, D.E., and F.E. Klipple. 1939. Sampling intensity in vegetation glossary. USDA-FS-Gen. Tech. Rept. PSW-13, Pac. Southwest For. surveys made by the .;qua re-foot dem,ity method. J. Agron. 31 :800-810. and Range Exp. Sta., Berkeley, CA. Hyder, D.N., and f',A. Sneva. 1960. Bitterlich's plotless method for sam­ Shoop, M.C., and E.H. Mcilvain. 1963. The micro-unit forage inventory pling basal ground CO\ler of bunchgrasses. J. Range Manage. 13:6-9. method. J. Range Manage. 16:172-179. lnteragency Range Survey Committee. 1937. Instructions for range sur­ Soil Conservation Service. 1975. Soil taxonomy. Agric. Handbook No. veys. Western Range Survey Conference, April 24. 1937. (processed) 436, U.S. Dept. Agric., Washington, D.C. Laycock, "'.A. 1965. Adaptation of distance measurements for range Warner, R.E. 1979. California riparian study program. Background sampling. J. Range Manage. 18:205-21 l. information and proposed study design. Planning Branch, California Dept. Fish and Game, Sacramento, CA. 177 pp.

V Appendix Working Paper for the Range Inventory Standardization Committee Convened by the Research Affairs Committee, Society for Range Management

Background (3) lack of uniform terminology and classification systems, and compatible inventory procedures. On July 22, 1977, the Board of Directors. Society for Range Management, acting on a recommendation from the Advisory Acting on the July 22, 1977, resolution of the SRM Board of Council, re,olved that SRM "take a position of leadership to draw Directors, then-President Thad Box invited appropriate organiza­ agencies, universities and land management organizations together tions to send representatives to an exploratory meeting and to promote uniform methodology and terminology for rangeland assigned SRM responsibility to the Research Affairs Committee. inventories and assessments." The Board went further in Februarv This meeting was held in Denver, Colorado. on Mav 31. 1978. and 1978 with a resolution endorsing efforts to coordinate and improve the Range Inventory Standardization Committe~ (RISC) was range inventory systems in the U.S .. supporting national research established. and development programs on identification, classification and This paper has been developed for guidance of the Committee. inventory of natural ecosystcms through coordinated efforts of and to inform the SR M Board of Directors. SR'vl membership and applicable agencies and institutions, with the recommendation others of the purpose of the Committee. that emphasis of such efforts be addressed to local management needs. Purpose of the Committee Considerable information has been collected by a variety of The purpose is to develop and recommend adoption of: inventory procedures, management studies. experience. and re­ ( I) standard terminology for inventory. classification: search. hut application of that information to the solution nf management problems has been hampered by: (2) a uniform system for classification and mapping of range ecosystems; (I) poor accessibility of information (2) incomplete information (J) minimum standards and guidelines for data collections; and 'v

12 GUIDELINES AND TERMINOLOGY FOR RANGE INVENTORIES AND MONITORING (4) a common philosophical base for data interpretations. Mapping Accomplishment of the above will facilitate communications The basic resource map should be based on integrated ecological among range users and managers. It will also facilitate the collec­ units as defined by the classification system. Criteria for mapping tion and availability of valid and useful data (a) for local manage­ at various scales and mapping intensities should be standardized. ment needs (b) for regional and national assessments and programs Uniform standards should be developed for map display of certain and (3) for research. kinds of resource data and interpretations. Issues to be Addressed Data Collection Terminology Data should be collected by compatible procedures in readily Standard definitions of words and phrases are basic for mutual convertible units and in such a manner as to allow reproducibility understanding and communication. Problems with current termi­ in the characterization of the resource, within identified limits of nology include ( 1) use of the same term to describe dissimilar items error, and facilitate reproducible interpretations of range ecosys­ or concepts and (2) use of different terms to describe the same item tems. or concept. RISC proposes to establish a common core of terms and definitions relating to inventory and classification, and to Data Interpretation encourage its usage. Interpretation of basic data will be made for each ecological unit to determine (a) its potential, (b) its present condition, and (c) the Inventory current trend in the condition. Other interpretations can be made The range resource inventory should be conducted to collect as necessary. These interpretations should be made with a common data necessary for local management purposes as well as for conceptual framework. regional and national assessments and program planning. The inventory should include certain information on basic resources Data !lfanagement (e.g. soil, vegetation, animals, water) collected by all who conduct Basic inventory data and interpretations of potential, condition, and trend of ecological units should be expressed in terms that are range inventories, and such additional information as needed. standard and uniform. This is desirable to permit consistent Classification accumulation, storage and retrieval of data for local management A uniform ecologically-based classification system is needed. needs and is imperative for aggregation of information on ecologi­ Such a system would provide a common base upon which to collect cal units for use in resource planning and assessment at the regional range inventory information, to accumulate and extrapolate man­ and national levels. In planning for the use of computers to do the agement experience and research results, and for the assessment of job of data management for range inventories, it is important that the status and needs of ranges. Such a system would facilitate final interpretations of data be reviewed and approved by qualified storage and retrieval of data and have a hierarchical capability for and locally knowledgeable professional people. aggregation and disaggregation of all types of information about range ecosystems.

GUIDELINES AND TERMINOLOGY FOR RANGE INVENTORIES AND MONITORING 13 ) ) )

Developing Strategies for Rangeland Management A Report Prepared by the Committee on Developing Strategies for Rangeland Management National Research Council /

Committee on Developing Strategies National Academy of Sciences for Rangeland Management

B. Delworth Gardner, Chairman. University of California Board on Agriculture and Renewable Resources, John C. Buckhouse. Oregon State University William Burch, Yale University Commission on Natural Resources, George Coggins, University oi Kansas Law School National Research Council Donald D. Dwyer, Utah State University Sally K. Fairfax, University of California Carlton H. Herbel, USDA-Agricultural Research Service Minoru Hironaka, University oi Idaho William S. Huey, New Mexico Department of Natural Resources James 0. Klemmedson, University of Arizona Bobbi S. Low, University of Michigan Jack Ward Thomas, U.S. Forest Service George Van Dyne, Colorado State University (Deceased) John Workman, Utah State University Jeffrey Brotnov, Staff Officer Deborah Faisun, Pro;ect Secretary

Westview Press / Boulder and London 648

Methods for Inventory and Monitoring of Vegetation, Litter, and Soil Surface Condition Paul G. Risser INTRODUCTION

Abstract Rangelands occupy nearly one-half the earth's land surface, and occur in every continent and large island Decisions about· the optimum management of rangelands throughout the world (Williams et al. 1968). In the depend upon satisfactory methods for inventorying the United States grasslands consitute millions of acres rangeland resource and then monitoring the consequences providing a major portion of food required by livest~ck of the imposed utilization activities. ~umerous tech­ of the country. In addition, rangelands contribute wild­ niques have been developed for field evaluat~on of the life habitat, recreational activities, and various other vegetation, litter, and soil surface conditions, and ~esources such as minerals and water. this paper discusses and analyses most of these method­ . Si~ce rangelands are such an important resource, it ologies. Specifically, vegetation methods include plant is particularly necessary that they be managed to provide cover, density, and biomass as well as methods of measur­ a sustained production of all resource values. Managing ing forage utilization. Methods for measuri~g litter and these grasslands in an enlightened manner requires that soil surface conditions are largely descriptive, but some natural resource characteristics be i~ventorictl and moni­ measurements have proven useful. tored. The purpose of this paper is to discuss various Selection of the appropriate methodology depends on methods that can be employed to inventory and monitor a number of rangeland conditions and the ob;ectives of vegetation, litter, and soil surface condition of the investigator as well as the time and money available. rangelands. Strict comparisons of the results from the techniques are A discussion of ~ethodology would be expected to difficult because the various methods have been tested include considerations of rangeland classifications, ade­ under different conditions and constraints. Furthermore, quacy of data for management interpretations, and the the parameters for·the rangelands are rarely known so the statistical veracity of the methods and resulting data. comparisons must be made on the basis of equivalent These subjects are discussed in other papers in this levels of precision rather than accuracy. ~evertheless, volume and will not be addressed in the present paper. the various methods have been compared and recommenda­ A critical analysis of vegetation, litter, and soil tions are made about the most applicable methods for surf~ce condi~ion i~ventory and monitoring methodology inventorying and monitoring rangeland vegetation, litter, requires consideration of the following topics: and soil surface co~ditions. -Types of data available from each method -Types of rangeland where each method is applicable -Biases in the methodology -costs in terms of time and money for required precision and accuracy.

These items are discussed in each of the three major sec­ tions: vegetation, litter, and soil surface condition. Paul G, Risser's present address: Chief, Illinois Natural History survey, 607 East Peabody Drive, Champaign, Illinois 61820

647

( ( ( ) ) )

649 650

,!here necessary, a distinction has been made between see Cooper 1959) can generally be defined as the propor­ ~ethods for inventory and those for monitoring. tion of the ground covered or occupied by vegetation, or the amount of ground area covered by a vertical projec­ METHODS FOR INVENTORY AND MONITORING VEGETATION tion of the vegetation. Measurement of cover has been made at different heights (i.e. canopy or foliage, speci­ It is important to recognize that rangelands are fied heights, crown, or basal). Descriptions of each .~cosystem~, composed of numerous interacting components. method are abundant in the literature (Avery 1959; Greig­ .he b!~vior ~f these systems is controlled by past and Smith 1964; Heady 1975; Parker and Harris 1959; Rich 195~ ?:evailing clunatic conditions as well as characteristics Stoddart, Smit.~. and Box 1975). ·:• the geo~o~ical sUbstrate and their interactions. Under As in other measures of plant biomass, cover can be ~hese conditions, grasslands develop along a variety of measured directly or estimated, and the resulting values ?~thways, many of which can be controlled by various man­ can be used directly or expressed as categories or classes. ';ement techn1ques. A large pro~orition of these manage­ The following paragraphs consider the degree of satisfac­ ;::nt options concern the vegetation component. The plants, tion offered by several methodologies commonly employed -,th aboveground and belowground, represent (l) the com­ to measure plant cover (U.S.D.A., Forest Service 1963). po~ent consum~d b~ livestock, (21 a large part of the ~a~itats of w~ldlife, and (3) the most consistently con­ Ocular or visual estimates. In this method, the ~~icuous p~rtion of the ecosystem. Therefore, management investigator estimates the cover (canopy and/or basal c:t. vegetation plays a key role in the development and cover) in a constructed plot or with some device for ~ainten~nce of grassland ecosystems. To manage the visually locating a sample area. Cover can be estimated :eg~tation according to recognized ~oals and obJ·ect· for the herbage as a unit, for all the individual species, , ... is ne t . -;, 1.ves, ;; ; cessary o inventory and monitor the vegetation. or for selected species or categories. If a plot or '. ·-r ~ous techniques have been developed for doing so ~uadrat is relatively large (e.g. 2 to S feet in one 1Brown 1954; DeVries and DeBoer 1959). dimension), cardboard calibration cards showing percent­ The 1nv~nto:y process is particularly important when ages of the plot can be used to increase the accuracy of goals and obJect1ves are established. That is th the estimates . _1n•,ento y establishes the range of management ~pti~ns If more than one individual estimates the cover .~onitoring, 7 on the other hand, provides a current estim­ (canopy or basal) of a single plot, the estimates are ate of the consequences of natural or managed influences likely to vary significantly unless precautions are made o~ the rangeland ecosystem. Inventory processes must to calibrate and standardize the estimates (51,,ith 19•14). c_~early measure vegetation attributes which communicate It may also be easier to estimate percentage cover by the state of the resource so that the values are known classes rather than absolute coverage. Daubemire's and can be managed: monitoring should provide an indica­ (1958) classes of l through 6 (0-5, 5-25, 25-50, j0-75, tion of the success of the management regimes. 75-95, 95-100 percentage coverage) have been used exten­ Genera~ly, methodologies appropriate for inventory sively in the western United States . !~e ~ppropri~te for ~onitoring and vice-versa. However, Costello and Klipple (1939) computed the number of • ne informat on required is frequently different. For samples needed to estimate plant cover within 10 percent 7Xample, an_inven~ory7 process might take special interest of the mean at the 95 percent confidence level, using ~: the sp~cies which are present and the amount of vege­ lOO-square-foot plots in ten different forage types in - tive biomass, whereas the monitoring process might Colorado and Wyoming. In the shortgrass rangeland 46 to ~~cus on the aJ?Ount of herbage utilization, the condition 106 samples were required and in other grassland types, s~veral indicator species, or the degree of soil the required sample size ranged from 48 to 260 plots. arosion. Although these data led the authors to suggest sample The subject of vegetation measurement for either or demands could be reduced by stratifying sample areas into both purposes is conveniently divided into four cate- homogeneous units, there is no way to assess the presumed 1~r1es of measurements:. (1) areal cover of plants efficiency of this stratification from this data set. In n~~rs ~r density of plants, (JI plant biomas~, and general, this estimation procedure improves as the ~41 utilization of plants and the resulting range condi­ investigators train in the field (Smith 1944), but after ion. In the following sections, each of these four a week of training, 40 percent of the estimates deviated general methodology categories is discussed. by more than 10 percent of the average estimates compiled by a group of eight investigators. .Plant Cover Plant cover (sometimes called "forage density," but 652 6Sl

Chart guadrats. With the use of a pantograph the herbaceous cover, respectively. In grasslands, Parker cover of plants Ln the plot can be actually recorded to and Savage (1944) found that different observer7 recorded scale on a chart. Although the device accurately trans­ small but·consistently different observations with the lates information to the chart, different observers may line intercept. Sampling in_big sagebrush, ~here high not always outline the same canopy, especially of plants levels of precision and confidence were required, the with canopies which are difficult to discern. Recently line intercept was preferable to small quadrats (Hanley field digitizers have been used to measure cover or count 1978). individuals (Mack and Pyke 1979), but the same ambiguity exists in defining the perimeter of the canopy. Pin or point sampling. Cover may be sampled by Ellison (1942) compared pantograph charting with ocu­ employing pins as physical points and using the percent­ lar estimates and point samples and found that estimates age of pins that touch the vegetation as a measure of the of chart areas were generally higher than estimated from cover of that vegetation (Evans and Love 1957; Goodall the other two methods. Furthermore, estimates between 1952; Park 1973; Robinson 1955; Wilson 1963): S7nce the five observers using the chart methods were significantly points are theoretically dimensionless, the pin size. different. The chart pantograph and visual estimate affects the estimates of cover, although the errors in methods had the lowest coefficients of variation. With this bias can be minimized by sharpening the pins (Long the pin method, coefficients of variation were essentially et aL 1972) and calculating correction factors (Warren~ proportional to sample size. Wilson 1959, 1960). Various investigators have used pins inclined at different angles (Tinney, Aamod~ and Ahlgren Variable radius plots. The Bitterlich (1948) or 1937; Warren-Wilson 1960, 1963; Winkworth 1955). Depend­ variable radius plot method has been used to measure basal ing on the average angle of the foliage, and the dis­ ground cover of bunchgrass (Hyder and Sneva 1960) and tance the pins travel thr.ough the vegetation, the shrubs (Cooper 1957, 1963). In the sagebrush grasslands, efficiency of the various angles can be calcul~ted. Hyder and Sneva (1960) concluded that forty-four 100-foot Angled pins are sometimes regarded as more easily read line intercept transects and twenty-eight variable radius (Tinney, Aamod~ and Ahlgren 1937), but the primary reason plots were needed to sample within 10 percent of the mean for angled pins is to minimize the underestimate of with 95 percent confidence. In general, the variable aerial cover caused by grass leaf blades which_are not =adius methods give higher estimates of cover than ocular horizontal. As a result, recommendations of ~in angle estimates, require less time, and the sample-to-sample for measurement of aerial cover range from 90 to 32.5 0 variation is comparatively low (Kinsinger, Eckert, and (Tinney, Aamod~ and Ahlgren 1937; Warren-Wilson 1960, Currie 1960). Most of the variation in estimates apparent­ 1963). _ ly arises because the shurb canopies are not round and In South Australia, Crocker and Tirer (1948) _found because it is difficult to visually determine the extremi­ that 200 point samples were sufficient for measuring_~. ties of the crown, especially at long distances and if dominants, and if 400 points were used, reliable esti-. the total aerial cover approaches 50 percent. The method mates could be made of some of the less important 7pecie~ is only appropriate where canopies are reasonably distinct If the pins are placed in a frame, the number of p7ns and can be seen such as in shrubs or sparse cover of required for a sample depends upon the number of pins per bunchgrasses. frame (and ~he distance between them) and the spatial pattern of the vegetation. In genera~, if pins are Line intercept. The line intercept is a method com­ placed in a frame, more pins are required to sample a_ monly used for determining estimates of plant cover species to a given level of accuracy than if single pins (Hormay 1949). It usually consists of stretching a tape are used (Goodall 1952). However, the improvement varies or line through the vegetation. This tape is marked into with the species and is usually more than compensated f~r intercepts, frequently 6 inches or 1 foot, and cover is by the improved efficiency of placing and reading pins in estimated by the number of intercepts which are partially a frame rather than individually. Kemp and Kemp (1956), or fully covered by either the vegetation as a whole or a among others, have shown the relationship between sample single species. adequacy and the number of pins per frame. As the number -In sampling mountain muhly (Muhlenbergia montana)­ of pins per frame is reduced, the total nlllru?er of Arizona fescue (Festuca arizonica)grass-forb vegetation, required pins is reduced, but the time required per pin Canfield (1941) used eighty-eight SO-foot line transects. is .reduced with more pins per frame. Coefficients of variation were 7.3, 10.6, 12.1 and 6.5 Rocky MOuntain foothill bunchgrass range was s~led percent for mountain muhly, Arizona fescue, forbs and total systematically and at random with 25, 50 and 100 points per line and the results from the point method were

( ( ( ) ) ) 653 654 compared with those from the line intercept technique vegetation can be harvested and the material analyzed by (Fisser and van Dyne 1966), The authors found both the measurement with a photoelectric device which measures point and line methods provide adequate estimates, but the area of irregularly shaped objects. Dot templets or that (a) to sample sod-forming species, random placement lines have been used to measure cover from photographs of points was more efficient, while (b) sampling the near the ground or aerial photographs (Pierce and dominant bunchgrass required fewer points with systematic Eddleman 1970; Wells 1971). location of the points, especially if the distance between the points was greater than the average basal Plant Density diameter of the plants. In the desert sac;ebrush-grass vegetation, however, ground cover was estimated S.67 In this section, the term "density• refers to the times faster with a point frame than with the line number of.individuals per area (Cooper 1959), Although intercept (Brun and Box 1963). The po~nt frame was 4.l~ the definition of an "individual" is relatively simple, times more efficient in time for sampling ground cover in i:e· h~ving one root system, actual recognition in the the sagebrush-shadescale type. field is very difficult. Determining a single individual Because of its convenience, the step-point method may range from merely cumbersome to virtually im?ossible, (Evans and Love 1957) is often used for inventory especially with shrubs and perennial grasses, even with purposes. This is a single p~n.method wh~re the location ma:~ive exc~vat~on. Therefore, it may be convenient to of the pin is guided by a definite notch in the sole of de-ine density in terms of counting units per area (Dix the worker's boot, the pin is pushed through the notch 1961), where counting units are a single stem or shoot, and plant contacts (or the plant nearest the pin i~ a etc. The actual definition of the counting unit will be 180" arcl is recorded. Total ground cover was estimated determ~ned by the objective of the study. from a small square-foot frame, not from the single pins, Field measurement of density is usually accomplished These authors report that in the California grassland the by. two type~ of techniques, quadrats and distance methods, time required to sample an area with the step-point which are discussed below. method was about one-sixth to one-eighth as much as required by the point frame method at a comparable level Quadrat methods. In the simplest sense, quadrats of precision. are areas in which individuals are counted to arrive at A three-step procedure was developed in the u.s .. an estimate of density (i.e. numbers of individuals per Forest Service to analyze ?lant communities or determine area). Since quadrats represent estimates of the total change in vegetation (Parker 1951, 1952; Parker and !)Opulation, quadrat locations and numbers must ~epresent Harris 1959), The three fundamental steps are (Re?pert the range of conditions in the grassland to be sampled. and Francis 1973): T~us, quadrat locations must encompass the variation in microtopography, range condition, and other influencing i) measure vegetation and soil stability on perman­ fact~rs •. The size of the quadrat depends upon the char­ ent transects grouped in clusters of one to acteri~tics of the.population being studied, the purpose three transects, for which the density data will be used the need to ii) summarize data and classify current range condi­ minimize or at least measure the varian~e around the mean. tion and trend in the field, and reasonable symmetry of the distribution curve, edge iii) take two oblique ground photographs for perman­ effects, as well as the practicality and convenience of ent visual record. sampling (G 7eig-Smith 1964). Counting units actually can be counted in the quadrat or the species ~an be assigned 'Ibe technique allows for an inventory, and then if . to one of a limited number of rank or abundance classes repeated at yearly or multiyearly intervals, trends in (U,S.D.A., Forest Service 1963). range condition can be inferred. Measurement of the vegetation is usually obtained by using~ 3/4-inch loop Distance methods. Distance methods are based on the to determine frequency. (Parker 1951; Strickler 1961). r~alization that the smaller the density, the longer the When no plant occurs within the loop, the soil su7face distances between plants or between a point and the factor covering the loop is recorded. Frequency is nearest plant. All these methods are based on an asswnp­ usually measured at 1-ft intervals along the 100-ft t!,On of random distribution of individuals--closest indi­ transects which constitute a cluster. vidual, nearest neighbor, random pairs, and the point-centered quarter method--and are described by for r~asurin cover. Several other Cottam and Curtis (1956). The angle order method is to measure cover. For example, the presumably not affected by non-random distributions 655 656

(Strickler and Stearns 19631, but this method has not Biomass been used extensively in rangelands. '"1e point-centered quarter method has l>een applied ln rangelands, herbage yield or biomass is normally. in several instances to grasslands to estimate density important not necessarily in i eself, but rather because (Becker and Crockett 1973; Dix 1961; Good and Good 1971; this characteristic of the ecosystem is related to a Penfound 1963). Dix (1961) found the method to be as grazing animal product or some ecological condition of accurate as any other grassland sampling method, in the grazing system. Knowing the species composition and addition to being considerably faster. Kowever, the biomass of any rangeland is likely to provide the most method assumes a random distribution of individual!! and fundamental knowledge needed by a manager (U.S.D.A, Risaer and Zedler (1968) showed that in a Wi,sconsin Forest Service 1963). prairie aggregated species were consistently underesti­ As a simple generalization, measurement of biomass mated by the quarter method. In Oklahoma, Becker and is accomplished by (a) direct measurements of plant Crockett (1973) also found that the point-centered weight, (b) measurements of plant characteristics that quarter method underestimated the density, but that the are related to weights in a known relationship, or (cl angle order method provided more accurate estimates. some combination of these methods. The most basic method Although correction factors for degrees of involves simply harvesting the biomass from a sample area, non-random distributions can be developed, the measure of weighing the biomass, and finally extrapolating to the density itself has not proven particularly useful for whole of the sampled rangeland. All indirect methods rangeland inventory and monitoring. This deficiency which have been developed use this harvest method either arises from the disparate size and growth form of as part of the method or as a standard against which to rangeland plants such that one individual (or counting establish the acceptability of the indirect method. The unit) of a species is not easily related to individuals following paragraphs will describe various techniques of another species. Also, if area methods are utilized, t:iat are available for measuring biomass on grasslands. the size of the quadrat must be adjusted to the size and density of the plants. Altering quadrat sizes can be Harvest or clipped clots. The concept of this cumbersome and obviates some other quantitative measures method is straightforward and unambiguous. Since the of grassland structure. objective is to determine amount of biomass on a defined The term •frequency• or "absolute frequency• has area, and since harvesting and weighing material from the been used in rangeland sampling and refers to the nwnber entire area is impractical, the biomass from representa­ of plots in which a species occurs, divided by the total tive sample plots is harvested, usually dried, weighed, number of plots in the sample. Frequency is always and then expressed as some weight measure per unit area. expressed as a percentage. Relative frequency refers to Although the method is conceptually simple, in ~~e occurrences (number of plots of occurrence) of any practice the method is time-coasuming and, therefore, one species divided by the sum of the occurrences of all relatively expensive. As a result, investigators should the species. Unlike density which can be expressed for consider the following items when either contemplating ~~e vegetation as a whole, frequency is always expressed field sampling or evaluating the results obtained by on the basis of individual species. harvest methods. A number of these items pertain to all Measures of frequency are affected by the size of sampling procedures which depend upon use of quadrats. the plot; that is, the larger the plot, the higher the estimate of frequency for a given species. As quadrat a. Since harvest plots are relatively expensive, size increases, the relative frequency of rare species it should be first determined that biomass increases disproportionately to common species. Several measurements are reaiiy necessary or wnetner studies have been directed at determining optim11111 plot other, less expensive measurements are size and shape (Hyder, Bement.and Terwilliger 1966; Rice sufficient (Wiegert 1962). 1967; Squiers and Wistendahl 1976; Van Dyne, Vogel.and b. As clipped plots represent a consequential Fisser 1963; Wight 1967), but the objective is to have expenditure of time per plot, the nWllber of quadrats small enough so the dominants do not achieve an plots that can be clipped will be smaller than al)solute frequency of 100 percent, but large enough to attain the number of plots measured by most other the assumptions of the Poisson distribution, thus allow­ possible sampling techniques. Because of the ing various indices of dominance and pattern (Looman comparatively small number of samples, the 1979; Poole 1974; Stowe and Wade 1979). investigator must be particularly concerned with (i) sample adequacy and (ii) selecting repre­ sentative locations for the sample plots.

( C ( ) ) ) 657 658

c. sample adequacy is best determined statistically Plots that are the same size as the individual on the basis of sample variance, confidence plants of a given species or the same size as a level of the estimate, and the accuracy of the spatial pattern of a species will maximize estimate, A convenient method is to use the plot-to-plot variation. various hoops and other following formula (Avery 1975): devices can be used (e.g. Kennedy 1972). Plac­ ing circular quadrats into very heavy grass canopies can be difficult. Under these condi­ n • ~ tions, rectangular quadrats, in which one side E2 can be removed while the quadrat is moved into where n • the number of required sample plots place, may be more convenient than circular quadrats. s 2 • variance of the samples for some diag­ f, F~r clipping at plot boundaries, a clear guide­ ~ine must be established for objectively decid­ -2 nostic characteristic, e.g. gm ing whether a plant is inside or outside the plot boundary. Experience has shown that in the t • student's t value for a specified absence of such guidelines, field investigators tend to harvest more biomass than really belongs level of probability and degrees of in.th~ quadrat (Van Dyne, VogeL and Fisser 1963). g. Clipping height should be selected in reference freedom. to the experimental question. Relatively large errors can be introduced into the data bv lack E • the half width, usually expressed as of adherence to clipping height, These errors are greatest in short vegetation or uneven some percentage around the mean. microtopography. h. Belowground biomass also can be sampled by the For example, if one is concerned about the total harvest method, but the mechanical difficulties biomass on a rangeland, samples might be taken are even more formidable. These samples may be until the estimate could be characterized as excavated in a prescr1bed volume with a shovel being within: 20 percent of the ruean at the SO or spade. More often, the belowground biomass percent confidence limit. is harvested using a hydraulic corer similar to those used in mapping soils, extracting a soil d, The location of sample plots is crucial (Costello and Klipple 1939) to provide an effi­ core and then separating the underground portion cient, unbiased, sample and to provide for the of the plants from the soil. Some of the methods necessary random assumptions of many numerical used for belowground biomass sampling are and statistical analyses. Clearly this described below. rationale for sample location must be consistent with the purpose of the monitoring or inventory i) Trench tracing~- A trench is dug at study, A stratified random design is usually least 30 inches wide, in a representative most efficient. With this design, the range to area. The roots are measured by digging a be sampled is stratified into relatively vertical_wall_on which the roots are exposed homogeneous units, each of which is then sampled with an ice pick or suitable tool, the roots randomly. This helps assure adequate represen­ are then drawn to scale and measured. tation of the rangeland as a whole, while at the Collection of t.~ese largely qualitative data same time, preserves the random sample assump­ is a laborious process and provides detailed tion for subsequent data analyses. data on only a small area (Albertson 1937). ii) Trench washing method. This method is like e. Even small errors in establishing plot boundaries--especially with small plots--can ~evious one"'except that a water stream introduce significant errors in the resulting is used to soften the soil and assist with biomass data (Van Dyne, VogeL and Fisser 1963). the separation of the fine roots (Unchurch Circular plots have minimal perimeter and and Lovvorn 1951). High pressure streams of therefore should minimize these errors. Long or water have been used along roadcuts or steep rectangular plots tend to homogenize the data, banks for separating and subsequently measur­ thereby reducing the plot-to-plot variance, in~ root systems (Hellmers et aL 1955). 659 660

l. Once the material is harvested, it should be iii) Soil core method. Cores of various sizes separated into the required categories (individ­ have been used to measure root material. ual species or other groups) and dried immediate­ Although the method of pushing the corer ly. However, with a large number of samples this into the soil varies from small manual may not be possible and the material should be techniques to truck-mounted hydraulic stored in cold temperatures until processing. If devices, the method involves removing a the material is not stared under cold or dried core of soil with known dimensions, washing condition, decomposition begins almost inunediate­ the roots free from the soil, and weighing ly, confounding subsequent weight measurements the root mass which is then expressed and making plant identification very difficult. either as a volume or in terms of the soil m. Harvested plant material may be covered with soil surface area. Various washing machines particles which would obviously affect the weights. have been developed to wash away the soil Therefore, it is necessary to account for soil and the washed root mass may be ignited in contamination before weighing. Root material is a muffle furnace so ash weights can be frequently washed, but as much as 20 percent of subtracted from root weight (Bartos and the total root weight may be lost as fine roots Sims 1974; Ruby and Young 1953). in the washing process (Bartos and Sims 1974). This error can be minimized by capturing the i. Whether the material is harvested from floating material which washes free from each aboveground or belowground, there may be diffi­ sample. In some cases, especially with sandy culties encountered in separating the actual soil, it may be possible to blow small root components to be measured. For example, if the samples free of soil particles with an air stream. experimental design requires separation into Both abovegrour.d and belowground material can be i~dividual species or categories, these separa­ expressed as ash-free dry weight, where aliquots tions may be difficult at the young stages of are ashed in a muffle furnace and the ash re­ the life cylcle, or if separations are not made maining is used to correct the weight estimates in the field, from clipped plant parts which of the remainder of the sample. have been stored at cold temoeratures. If it is not necessary to determine exactly the simple Weight-estimate. Because of the enormous time and proportions of live and dead material, pigment expense in using clipped plots, various other techniques extrac~ions may be a useful separation technique have been devised to collect similar data with less ef­ (Lauenroth, Oodd,and Dickinson 1980). Without fort. The weight-estimate method eliminates routine har­ special procedures, it is difficult to separate vesting of plant material and depends upon the ability of roots by species, although rhizomes can the investigator to simply estimate the weight of the frequently be separated. plants (Goebel 1955; Pechanec and Pickford 1937a; Shoop j. Research designs may require separations be made and Mcilvain 1963; U.S.D.A. Forest Service 1963). Because on dynamic categories, such as "live," "current more quadrats can be estimated than clipped in a given year's growth," "current year dead,"or ~previous amount of time, a greater number of estimated samples can ye~r• ~ dead." Again, without special techniques, be taken, thus improving the sample adequacy and increas­ this 1s virtually impossible belowground and ing the likelihood of including representative plot loca­ requires diligent attention and clear guidelines tions. However, in comparing the efficiency of the for separating aboveground material. clipped plots to weight-estimate plots, evaluations must k. N~er~us d7vices have been used for actually include the training and calibrating time, as well as the clipping biomass, ranging from hand techniques time required to collect and prepare samples for eventual­ to various mechanical arrangements. Although ly converting weight estimates to dry matter. electric sheep shearers, either alone or Typically the ratios of time required to clip quad­ connected to a vacuum collecting device, can rats, separating only a few species, compared with the minimize errors caused by uneven clipping, the time required to estimate weights are 5-10 to 1. Correla­ resulting material is often more difficult to tions between predicted and actual weights are high and separate into species. the coefficient of determination varies around 80-95 per­ cent (Francis et al. 1971). In fact, estimates are gener­ ally within 10 percent of the actual weight over a range of

( ( ( ) ) )

661 662 weights, although estimates tend to overestimate small species; inadequate data are usually available for deter­ weights and underestimate higher weights thus artificial­ mining the regression coefficients for rare species. ly reducing the inter-plot variances. Weight estimates In this technique, there are two estimates: y = have been made of all the vegetation in a plot, on the biomass determined by clipping and x1 = biomass deter­1 basis of individual species, and on the basis of subunits mined by weight estimation. Obviously, an economy is of individual species (Hickey 1961; Hutchings and realized if y can be monitored more precisely than x, but Schmautz 1969; Shoop and Mcilvain 1963). xis less expensive to monitor than y. The steps are as A number of the same concerns discussed for clipped follows (Francis et al. 1971): quadrats also apply to weight-estimate plots (e.g. sample adequacy, selection of representative plots, species . i) Monitor bothy and x on a fixed number n identifications, establishing plot boundaries). In addi­ of plots. tion, there are several items of consideration uniquely applicable to the weight-estimate procedure. ii) Derive a mathematical relationship between y and x based on those n plots. This rela­ a. Field investigators must be trained before field tionship can either be a linear regression work begins and must be recalibrated during the or a ratio estimator (regression through the field estimations. These training and calibra­ origin). tion sessions include comparisons of estimated field weights with actual weights. y = F (x) b. Weight estimates may drift during the day or over longer periods of time; therefore, fre­ iii) Monitor x alone on a fixed number n' -n quent calibration checks are always necessary. of plots. c. Green or wet weights can be estimated in the field and the, through comparisons with col­ iv) Determine the mean value of x for all the lected samples, dry weights can be obtained in n' plots sampled. the laboratory. Although it might be more efficient to estimate dry weights in the field, n' iT a 1 !: x. it is more difficult to calibrate on the basis ;;, 1 of dry weights. i=l d. Perhaps the most serious tendency is that experienced field workers frequently estimate v) Estimate the value of y that one would yields close to that which might be expected for expect to correspond ton' (that is, the a range site and condition and, thus, are reluc­ value of y that one would expect to obtain tant to estimate plots much lower or higher than if y had been observed on all n' plots). the expected values. Plots on either side of the mean are estimated with a strong bias toward y' = F (i(°T) the mean, and therefore the observed variance is lower than the real variance. Double sampling is effective if the variance of y' is less than the variance of the estimate that one would Double sampling by clipped plots and weight-estimat~ expect to obtain if y alone was measured for the same The speed advantage ot visual estimates can be combined total cost as that of the double sampling procedure. with the more rigorously obtained estimates from the Conrad and O'Regan (1973) developed a two-stage clipped plots by the use of double sampling (Ahmed and stratified sampling method to estimate herbage yield on Bonham 1980; Wilm, Costello, and Klipple 1944). Acer­ the San Joaquin Experimental Range. In the first stage, tain number of plots are both estimated for herbage many plants were stratified according to some easily weight and then clipped. A ratio or regression is then observed characters, such as cover, by using a point calculated expressing·tha relation of clipped to estim­ frame. The second stage consisted of clipping the vege­ ated weight for these plots. This equation is then used tation in selected plots, sorting the clippings into to adjust weights for plots on which only estimates have species, and weighing the sorted clippings. Equations been obtained. It may be necessary to recalculate the re­ were developed for determining the sample size and the gressions for each sample period and for each species. distribution of the plots between stages one and two. In Therefore, the method is most appropriate for the common this test case, the results showed that two-stage sampling could produce a 15 percent reduction in variance of 664 663

species (Fletcher and Robinson 1956) and is mean weight compared with that from simple random sam­ affected by differences in water content of the pling. soil, herbage. and air (Van Dyne, Glass, and Opstrup As an alternative approach, the total biomass can be 1968). estimated and then the total is allocated to the species comprising the sample. For example, entire plots have In a variety of grasslands the capacitance meter has been harvested, the total biomass weighed , then estimates been shown to be a rapid, efficient component of double made of the contributions of individual species (Wagner sampling processes although woody stems decrease the 1952). These species estimates may be virtually without accuracy in plots containing herbaceous and woody plants control except the total of all species sums to the known (Carpenter, Wallmo, and Morris 1973). Ratios of l to 8 weight of the total biomass from the harvested plo~. A through l to 5 between clipped and capacitance meter variant is the technique referred to as the dry weight quadrats have been shown to provide estimates of high rank method in which the investigator ranks the species precision (Morris, Johnson, and Neal 1976). in each plot on the basis of dry weight contribution to Measuring the standing crop of shrubs has always the total biomass. On some plots the species are ranked, been difficult, partly because shrubs are frequently the biomass is clipped and hand-separated into species, multi-stemmed, thereby defying methods normally applied and these weights are used as multipliers to convert the to trees, and partly because their elevated crowns are ranks of each species on the ranked-only plots into weight frequently more difficult to measure by grassland tech­ estimates by species. niques. Although both the Bitterlich and line transect methods are used, neither uniformly accounts for the Indirect methods of estimating herbage production. three dimensional configuration of the shrub canopy. A number of other measures of vegetation stature are re­ Therefore, dimensional analysis has been used in a number lated to plant weight with reasonable consistency. of vegetation types (Burk and Dick-Peddie 1973; Chew and Chew 1965; Ludwig, Reynolds, and Whitson 1975). The tech­ •Cover. Foliar cover, but not basal cover, frequent­ nique consists of relating biomass to one or more easily ly may correlate well with the weight of the vege­ measured dimensions (e.g. canopy area and volume) of tation (Arny and Schmidt 1942; Cook 1960; Pasto, sample shrubs by regression analysis, then using the Allison, and Washku 1957; Payne 1974; Poissonet et equation to estimate biomass of shrubs after measuring al. 1973; Van Keuren and Ahlgren 1957a, 1957b). the appropriate dimensions and substituting in the ·Height. The height of individual plants or canopy equation. Although it appears that some of the relation­ may be predictably related to weight, but the rela­ ships are constant over wide geographical areas (Ludwig, tionship may change throughout the season (Carpenter Reynolds.and Whitson 1975: Whittal

( ( ( ) ) ) 66S 666

Methods of interpreting biomass measurements. The descriptions of many of these methods. numerical values obtained for biomass estimates can be In the.mixed grassland of North Dakota, Whitman and used for estimating forage utilization (discussed in a Siggeirsson (1954) compared the line intercept with the subsequent section of this paper). In addition, thes~ polnt method using pins included at 45°. With the pins, estimates can be used to estimate net primary production cover was recorded two ways: all hits (aerial and basal) of aboveground and belowground parts of the grassland eco­ and basal hits only, respectively. Total cover was 20.70, system (Kelley, Van Dyne, and Harris 1974; Kucera, Dahl..rnal, 31.91, and 32.28 for line intercepts, all hits, and basal and Koelling 1967; Lomnicki, Bandola, and Jankowska 1968; hits only. Coefficients of variation for total cover Malone 1968; Singh, Lauenrot~ and Steinhorst 1975). All were 13, 24, and 23 percent for line intercepts, all hit~ harvest methods underestimate primary production unl7ss and basal hits, respectively. In general, estimates of there are precautions to exclude or.account for h7rbivory. cover by point sample were about 50 percent higher than The simplest of the primary production estimates is to by line intercept. Percentage composition computed by measure the biomass at the time of peak standing crop or the three methods showed considerable variation. Using at the end of the growing season. However, usually not all hits favored needle-and-thread grass (Stioa comata) all species in a rangeland achieve peak standing crop while basal hits only favored blue grama (Bouteloua ' simultaneously and in addition, material frequently gracilis). Coefficients of variation were lowest for all senesces and becomes detached throughout the year, thus hits, but different species varied greatly with the great­ leading to underestimates. To minimize these errors, est variation in the scarce species. Fifteen line inter­ various schemes have been used such as summing the peak cept transects, and 2,300 points were used to sample values of all the species or categories of species, or . total cover with equal precision, that is, within 5 per­ alternatively, summing the statistically significant posi­ cent of the mean at the 68 percent confidence level. To tive increases in biomass from sequential samples. The . sample various species to this confidence level, 4,300 to calculation method strongly influences the annual net pri­ 29,600 points were required for both basal and aerial mary production estimates (Singh, Lauenroth, and Stein­ hits; 4,600 to 137,500 points were required if only basal horst 1975), sometimes by as much as 30 percent. hits were counted. In an Oklahoma tall-grass prairie, Becker and Comoarison of Methods for Estimating Crockett (1973) compared the following methods: point, Veaetation Cover, Density, and Biomass line transect, angle order, quadrat, point-centered quarter, and the wandering quarter. Sample "adequacy" In the preceding description of each method, _some was not addressed in the study. Rather, sampling was comparative information has been presented to assist in terminated when approximately 400 shoots were encountered clarifying the techniques. However, it is difficult to Although none of the methods provided relative density draw definitive comparisons between the results of values close to the true values, relative densities for various methods. For example, Ellison (1942) showed that the individual species approximated the importances of cover estimates by the pantograph chart method gave cover the individual species, except that the point-centered values much higher than comparable ocular estimates and quar~er method most severely underestimated aggregated point samples. In North Dakota, the data of Whitman and species. The authors concluded that the point and quad­ Siggeirsson (1954) indicated tha~ point samples ~ene7ally rat methods were most applicable if the dominant species gave jO percent higher cover estLmate than the line inter­ were dense clones: however, the point or line transect cept. It might be tempting to conclude, therefo7e, that methods were more adequate if the dominants were sparsely the estimates of cover by the pantograph chart might be cloned. The lack of a sample adequacy measure and as much as twice as great as those estimates from the line absence of time estimates make this study difficult to intercept! The danger of such a syllogism is apparent interpret. when one recognizes that comparing techniques is only Six methods were compared in a homogeneous grassland possible under comparable considerations of sampling in France (Poissonet et al. 1973). adequacy., vegetation differences, etc. Fortunately there have been a limited numbers of studies which have compared •Needle points--very thin points (used as the refer­ the results from various rangeland inventory and moni­ ence method in this paper). toring methods (e.g. Johnson 1957; Mountier and Radcliffe •Double meter points--vertical sightings on the edge 1964 • Winkworth, Perry, and Rosetti 1962) and Pieper of a 2 m ruler. (1978) presents numerous comparisons with his careful ·Bayonet points--points defined by the edge of a blac,e or bayonet (Poissonet et al. 197:_a •. 667 668

·Line transects--species presence was recorded in guidelines for the optimum utilization for each of the . 25 cm segments of 4 lines of 64 cm each, important forage plants, range types, and for each grazing ·Areas with species ranking--species presence and condition (e.g. Reid, Kovner, and Martin 1963). then order of cover magnitude were tabulated in At first glance, the measurement of and subsequent four sizes of plots. application of range utilization would appear to be a •Harvest plots--40 plots, 1 x 0.25 m, were harvested straightforward matter. Certainly a range manager ~ust. and the species sorted, know how much of the forage is being taken at any point in time and also must know how much can be safely consumed Consistent cuvilinear relationships were found be­ with~ut jeopardizing the long-term productivity of the tween specie$ frequences, cover, and biomass by most of grassland. As will be shown, there are severa~ m7thods by the methods. Furthermore, in this homogeneous vegetation, which utilization can be measured, The more difficult a linear relationship existed between cover as measured task is ta determine the most appropriate standards. After by the point methods and biomass estimates. Therefore, understanding the life history and environmental require­ it was possible to predict biomass (a time-consuming ments of a species, it is then necessary to know the measurement) from the rapid point methods (double meter interactions of different intensities, season~ and £re~ or bayonet paints]. quencies of grazing on these plants. These relationships The loop method was compared to the line intercept are made more complex by the additional interactions of and variable radius plot methods in a shrub community, weather, initial range condition, and factors related ~o where the loop method provided higher caver estimates animal usage such as soil compaction, runoff, .and er~sion. than the other methods (Coak and Box 1961; Kinsinger, Since it is very difficult ta deal comprehensively_with Eckert, and CUrrie 1960). However, the loop method is all these interactions, closely defined range utilization really a very small quadrat which is used for estimating measurements and utilization standards ~ave.remain~d e~u­ frequency, a vegetation description which depends an the sive and mast applications have been primarily subJective size of the quadrat. In this case the bias is related to but based on field experience. . plant size (Hutchings and Homgren 1959; Hyder et al. Although there are numerous methods for measuring_ 1965). A positive bias toward higher loop-frequency as utilization, it should be clear that proper rang~ ~ondi­ to cover estimated by other methods has been reported by tion is not determined directly by percentage utilization, several workers (Francis, Driscoll, and Reppert 1972). That is, a number of factors affect plant response and Loop-frequency is not easily related to other vegetation therefore long-term range condition including current measurements, since it simply provides an estimate of growing conditions, type.of habitat, weather and climate, frequency for a given size of small quadrat. As such, palatability and aggressiveness of ~ssociated species, it is subject.to all the relationships between frequency variation of forage preference by different classes of and quadrat size, plant size, and distribution character­ stock, differences in the amount of graz~ng various plants istics (Grieg-Smith 1964; Hutchings and Homgren 1959; can withstand, differences in accessibility of areas to Smith 1944), and consPquently should not be expected to livestock grazing, differences in location and frequency be correlated with estimates of cover as some authors of livestock water, size and shape of pastures, season of have implied (Francis, Driscoll, and Reppert 1972), A grazing and duration of grazing (Schmutz, Holt, and more serious difficulty with the 3-step procedure is the Michael~ 1963). The percentage utilization measurement realization that litter and soil conditions are usually must be applied to specific plants which are indicators of not recorded when a plant counts as a hit, and secondly, grazing preferences (Dyksterhuis 1949). Using these con­ that the data resulting from the three steps cannot be cepts, rangelands have been placed into var~ous range interpreted in a strai;htfoniard , mechanical manner condition classes (Heady 1949; Stoddart, Smith, and Box (Reppert and Francis 19731. 1975}:

Methods of Measuring Forage Utilization l. Unused, No livestock use. 2. Slight. Practically undisturbed. Forage utilization,or percentage utilization, refers 3. Light. Only best plants are grazed. to the amount or percentage of the current growth of herb­ 4. Moderate. Most of the range grazed, little or no age that has been removed by animals. This term can be use of less palatable plants. applied to single plants, groups of plants, or the range­ s. Proper. Range entirely grazed, primary forage land as a whole (Cook and Stoddart 1953). Over the years plants grazed to the correct degree. . . considerable effort has been devoted to the development of 6. Close, Completely grazed, with some repetition. range utilization methods as well as developing utiliza­ Same use of low-value plants. tion standards for proper utilization, that is, standard

( ( ( ) ) 669 670

7. Severe. Hedged appearance and trampling damage. Primary forage plants almost canpletely used and difficulty of cages has been that: (1) the cages them­ low-value plants carry major portion of grazing selves affect grazing conditions (Cowlishaw 1951; Heady land. 1957b) and these differences are not consistent under all 8. Extreme. Range appears stripped of vegetation. grazing situations (Owensby 1969), (2) the expense of Primary forage plants definitely injured and cages mandates fewer samples and therefore the smaller low-value plants closely grazed. sample size, and (3) the regrowth of herbage outside the 9. Destructive. Considerable death loss of primary cage is not constant throughout the season nor easily species. Only remnants of high-value forage measured and reconciled in the comparative measurement. plants remaining. . Weight of plant units before and after grazing. In The following paragraphs briefly outline techniques th1.s method a plant ''unit" is defined as an easily defin­ which have been used to estimate forage utilization able and recognizable portion of individual plant. The {National Academy of Science/NRC 1962; Stoddart, Smith, unit may vary with the species, and may be a single stem and Box 1975). or an entire plant, but it must be large enough so as never to be entirely consumed. Percentage utilization is _ocular estimate of weight removal from plots. An determined by obtaining the initial weight of these units estlll\ation is made on the amount of herbage that is re­ and then knowing how much of these plant units are con­ moved by grazing. In training, the field investigator sumed. As in the use of cages, this method works best clips plots to estimate grazing, then clips the stubble where forage is grazed for short periods and regrowth is and uses the weight of both clippings to calibrate the minimal (Shoop and Mcrlvain 1963). :stimate of herbage removal. Thirty circular quadrats, oetween 25 and 100 square feet in size, usually give an Reduction in height. In this method, the difference adequate estimate on key species in a homogeneous area in average heights of plants on grazed and ungrazed areas (Clark 1945; Reid and Pickford 1941). But some species is measured and used to calculate percentage utilization. ~ay be inadequately sampled because estimates are only Since plant height and biomass or volume are not uniform­ made on these species in the quadrats. ly related, the method, in spite of its simplicity, is not as effective as might be expected (Pechanec and Ocular estimate of_weight removal from plants. This Pickford 1937) . method is based on estimates of weight removal from indi­ vidual plants rather than directly on plots. Plot rat­ Stubble-height class. Stubble heights are placed ings may be obtained from an average for each species and into classes of height, perhaps 10 classes with Class converted to the contribution of each species to the No.l being stubble height of 0-0.S inches and 10 being plots. Though this method is not as rapid as the plot ungrazed plants. Then using SO to 100-foot transects, technique, it has proved to be quite accurate because plants are recorded in stubble height classes {Canfield observations are confined to single plants, and it is 1944) and their lateral extent is measured at ground possible to obtain an adequate sample of every species level. After the data for each species are compiled by of interest. stubble-height classes with the percentage of plants in each class, the data can be converted to percentage util­ Actual weight difference. Grazing is excluded from ization for each species or the data can be summarized p~ots by cages or enclosures, These plots and adjacent to show mean utilization for all species on the range similar ones are clipped and the difference in weight (Collins and Hurtt 1943). represents the percentage forage consumed. This method has been widely used in range research for a number of Height-weight ratio. For each of the-above methods, years (Bently and Talbot 1951; Boyd 1949; Brown 1954; percentage of weight removed has been used as a measure Cowlishaw 1951; Darland and Weaver 1945; Daubenmire 1940; of utilization. The possibility exists that some other Fuelleman and Burlison 1939; Heady 1957a; Linehan 1952; measurement can be related to weight through regression Nevens 1945; Prendergast and Brady 1955; Rieqel,Albertso~ relationships (Lomrnasson and Jensen 1938)-. It is possible and Hopkins 1950; and Weaver and Bruner 1948). The olots to determine the volume or weight distribution in relation have been of various sizes; smaller ones are advantageous to height of a given species (Crafts 1938). This is done for the acquisition of larger sample numbers, but larger for grasses by collecting the plant in a bundl~ clipping ones have less border effect from the cages. Some workers at one-inch intervals.and weighing the clipped materials _have used electric fence to exclude cattle (Prendergast (Campbell 1943; Lornmasson and Jensen 1943), Samples are and Brady 1955) which has worked well. In general, the taken of both grazed and ungrazed at several heights. 671 672

Then the height is converted to weight using the derived Several methods have been developed to estimate the relationships. utilization of browse species (Aldous 1944; Hormay 1943; Numerous charts and tables have been fashioned for Schuster 1965). In general these methods involve esti­ field use once this relationship for a species has been mating amount of browse which is removed from the shrub determined (Campbell and Cassady 1955; Crafts 1938· or trees, or using the length or diameter of twigs to Valentine 1946). For an estimate of utilization f~r the arrive at an estimate of how much plant material had been entire range, each species must be weighted by its removed. relative contribution to the total composition. This method, as the height reduction method, assumes Photographs. Rangelands showing different degrees that for a given species, the relationship between height of utilization can be photographed (Reid, Strickler and and weight is constan~ between years, seasons, and sites. Hall 1980) and then used as standards. It is possible to In fact, the relationship has been shown to be inconsis­ arrive at consistently accurate estimates, i.e. within 5 t9nt (Caird 1945; Clark 1945; Heady 1950). Some of the percent utilization, especially if the photographs are error can be eliminated by developing separate tables for chosen well and include convenient markers such as Oountia different height classes. Reid and Pickford (1941) rec­ spp. Similarly, workable photograph guides have been ommended the ocular-estimate-by-plot procedure over the developed for a number of key species in the southwestern height-weight ratio, but Lommasson and Jensen (1943) ob­ United States (Schmutz 1971). ~ained more consistent results with the height-weight ratio method. The height-weight ratio is an accurate and METHODS FOR INVENTORY AND MONITORING OF LITTER reliable method on perennial grasses, but construction of the relationship is tedious and may need to be developed The litter or mulch layer exerts a nwnber of influ­ for differences in site and weather (National Academy of ences_in the grassland ecosystem. These include providing Science/NRC 1962). ~ habit~t for.numerous organisms; retaining nutrients; inc~easing soil moisture by affecting infiltration, evapo­ Stem-count. Stoddart (1935) showed that with west­ ratio~, and runoff; retarding infiltration by interception ern wheatgrass (Agropyron smithii), percentage utiliza­ of rainfall; acting as an insulator for solar ~a

( ( ( ) ) ) 673. 614

Class 1. No soil loss or erosion, top soil layer in­ discuss either soil classification or soil forming proc­ tact, well-dispersed accumulation of litter esses. Therefore, the discussion concerning soils will from past years's growth plus smaller be limited to soil surface characteristics as these quali­ amounts of older litter. ties relate :nore directly to inventorying and monitoring Class 2. Soil movement slight and difficult to recog­ range resources. nize, small deposits of soil in form of The qualities of particular interest to the evalua­ fans or cones at end of small gullies or tion of range resources include: rills, or as accumulations back of plant crowns or behind litter, litter not well 1. Erodibility--relative suceptibility to erosion dispersed or no accumulation fran past and usually related to soil stability and the year's growth obvious. degree of soil or vegetation disturbance that Class 3. Soil movement or loss more noticeable; top­ would accelerate erosion. soil loss evident, may be some pedestaled 2. Moisture retentivity--the ability of a soil to or hummocked plants; rill marks evident, retain water at specified levels of moisture poorly dispersed litter and bare spots not tension. protected by litter and/or soil. 3. Permeability--the rate at which a soil conducts Class 4. soil movement and loss readily recognizable, or transmits fluids, frequently in relation to topsoil remnants with vertical sides and the entire soil profile, horizons, or the surface exposed plant roots, roots frequently ex­ of the soil (infiltration). posed, litter in relatively small amounts 4. Compaction--the packing together of soil parti­ and washed into erosion protected patches. cles by instantaneous forces exerted at the soil Class s. Advanced erosion;active gullies, steep side­ surface resulting in an increase in soil density walls on active gullies; well-developed through a decrease in pore space. erosion pavement on gravelly soils; s·. Fertility--the quality that enables a soil to litter mostly washed away. provide the proper kinds and amounts of nutrients. 6. Productivity--capability of a soil to produce In some range evaluation schemes, each of the above yields of specified plants under given manage­ classes have been given point values, ranging from 0-20 ment practices. points. Then the soil surface condition is used as another indicator of the status of the range or, if a Erodibility is difficult to measure precisely but two trend is observed, as a method for monitoring the approaches may be utilized. The first is to measure the condition of the rangeland. . . amount of soil lost from a specified area which is marked By far the most common quantitative method of inven­ with graduated pins or a stretched wire. Alternatively, torying litter is simply collecting the material and and especially at the watershed level, it is possible to weighing the biomass (Hopkins 1954), usually after the measure the amount of ~aterial ~roded from the surface material has been dried and corrected for ash (Hulbert and captured in a sediment collecting device (Allis and 196 9) • Kuhlman 1962; Dragonn and Kuhlman 1968; Trieste and Gifford 1980). The latter technique automatically inte­ METHODS FOR INVENTORY AND MONITORING grates over the entire drainage area, but does not allow SOIL SURFACE CONDITIONS the measurement of eroision on a microscale or under dif­ ferent conditions and management strategies within the Soils, and soil characteristics, are frequently dis­ watershed. As an attempt toward uniform terminology the tinguished in a nurnb'er of ways, but the most common ':'re Soil Conservation Service (1976) has constructed defini­ (National Academy of Sciences/NRC 1962): (1) according tions of erosion, including terms of "land damage" (i.e. to differences in one or more properties, such as texture, land losing soil at rates greater than 33.6 mt ha-1 yr-1). color, or reaction; (2) qualities or attributes.a~ mani­ Since erosion oer ~ is relatively difficult to meas­ fest in behavior or performance, such as erod1b1l1ty, ure, several attempts have been made to.relate erosion to moisture retentivity, permeability, fertility, or produc­ other characteristics of the soil and vegetation (e.g. tivity; and,(3) as natural units of the la~dsc':'pe classed Luxmore and Sharma 1980). Ritchie and McHenry (1973 according to distinctive and relevant combinati~ns of a 19781 have used levels of the fallout radionuclide ll7cs n1.n11ber of soil characteristics. The U,S.D.A. Soil Conser­ in a soil to estimate rates of soil loss, while Bondy, vation Service (1975) has constructed a soil classifica­ Lyles, and Hayes (1980) have developed new procedures for ·tion scheme and it is not the intent of this paper to ccmputing wind erosion by time periods based on erosive 675 676

Meeuwig, and Skau 1974), These are devices which measure wind energy distribution, Renner (1936) related four changes in weight of soil in place or the amount of leach­ classes of erosion to slope gradient, soil type, vegeta­ ate moving through the soil (Tromble, Renard, and Thatcher tion density, rodent infestation, and accessibility to 1974). Tubes can be forced into the surface of the soil, li'lestock, from whence he determined some recognizable a reservoir attached to the upper end of the tube, and the range characteristics and land-use practices that contrib­ rate of water movement into the soil can be measured. Al­ ute to erosion. Croft, Woodward, and Anderson (1943) ternatively, rings, 6-12 inches in diamet r, have been =elated erosion to soil organic matter, moisture equiva­ placed on the soil surface and water applie~7 to the reser­ ~ent, and total nitrogen, finding that organic matter in voirs. sometimes larger surrounding rings have been ~sed the surface inch of the soil was strongly, inversely re­ to minimize the lateral migration of water. In experi­ lated to accelerated erosion. Various soil textural mental watersheds or plots, it is possible to apply a ~ategories. have been related to erosion (Johnson and known amount of water and then measure the amount of run­ ~iederhof 1941) and the relationships between percentage off. bare soil and erosion have been investigated in many compaction can be measured directly by comparing bulk areas (Marston 1952; Packer 1951). The general relation­ densities of soils (Bauer 1956) or indirectly by compara­ ship between vegetation, grazing, and erosion is also tive measurements of pore space or rates of infiltration well-known (Branson, Gifford, and Owen 1972; Dunford 195~ or percolation on the basis of several different devices Martin and Rich 1948; Packer 1953). (Gifford, Faust, and Colthrap 1977). . . Altiough the idea of moisture retentivity is clear, Soil fertility is measured by analyzing the ~oils or measurement techniques and definition of numerical 'lalues by tissue analysis of the plants growing on the soil. This or indices are more complicated. Depending upon soil tex­ is a very large, well-documented field of study and, ~;her ture, organic matter and clay structure, soils can retain than mentioning the subJect as a characteristic of soi~s, different amounts of soil water when suction pressure is will not be discussed in this paper. Soil productivity, applied to simulate the extraction power of roots. His­ however, is measured by the productivity of plants grow­ torically water held between 1/3 and 15 atmospheres of ing on the soil. pressure was considered available water. However, we now know that some plants can extract water held at tensions SUMMARY AND CONCLUSION in excess of 15 atmespheres (e.g. Majerus 1975). There­ fore, the conventional measurements of saturation, field Over the past 50 years, an enormous amount of ef~ort capacity, and ?ermanent wilting point do permit compari­ has been expended toward developing tools for evaluating son between soils, but do not adequately characterize the our rangeland resources. The prece~ing pages ~ave de­ moisture retention capability of the soils as compared to scribed numerous methods for measuring vegetation and the dynamics of range plants. Measurement of saturation is utilization, but fewer methods for describing the litter usually made by weighing or by gypsum blocks, and the i;:er­ and soil conditions. manent wilting point is determined by a standard proce­ In general, the distinction between inventory and dure of growing sunflowers, sometimes in conjunction with monitoring has not been drawn at the outset of attempts a grass, and determining the amount of water in the soil to develop methodology. Inventories vary considerably when the sunflowers do not regain turgor in saturated because the scope, depth, and cost of the process is aerial atmosphere. The introduction of a pressure plate usually determined by the resource ~vailable and_the man­ has allowed the convenient description of moisture date or mission for the study. Monitoring activities release curves, comparing the a~ount of water over a usually receive less support and the employed techniques range of atmospheric tension (Acevedo, Hsiao and are those which appear to provide the most useful (recog­ Henderson 1971). nizable and interpretable) results for the relatively Permeability relates to infiltration or the downward small amount of available resources. entry of water into the soil. Two phases are involved. An obvious question involves the need for so m~ny Intake, or infiltration, is the movement of water into different methods for measuring vegetation. The origin of the soil and percolation is the movement through the soil these numerous methods stems in part because the objec­ mass. A detailed discussion of soil hydrology is beyond tives of various studies are not identical, thus the need the scope of this paper (see Branson Gifford, and OWen for different methodologies; and in part, because no one 1972). However, infiltration rates are closely related method has clearly emerged as the most appropriate.over to rangeland management (e.g. Rauzi and Smith 1973) and a a wide range of vegetation types, e~en for conv~ntionally brief description of the methodology is appropriate accepted goals of inventory and monitoring studies, (National Academy of Sciences/NRC 1962). There never has been an attempt to state clearly the . Infiltration, and percolation, can be measured by a goals of monitoring and inventory, establish rigorous variety of devices such as lysimeters (e.g. Blackburn,

( ( ( ) ) ) 677 678

criteria for optimization of results with expenditures, Acevedo, E., T. c. Hsiao, and D. w. Henderson. 1971. and then to comprehensively determine the most appropri­ Immediate and subsequent growth responses of maize ate techniques. Rather various techniques have been leaves to changes in water status. Plant Physiol. applied in one or several locations without a solid base 48:631-636. for subsequent comparisons. Schultz, Gibbens, and Debano Adams, J. E. 1966. Influence of mulches on runoff, ero­ 11961) compared several methods for measuring cover on sion and soil moisture depletion. Soil Sci. Soc. artificial populations and concluded that accuracy was Amer. Proc. 30:110-114. not improved after about 30-40 transects for the line Ahmed, J. and c. D. Bonham. 1980. DUBSAM. Algori:t.lll!!. and intercept, line points,and loop methods, 40 variable plot computer program for optimum allocation in multivari­ ooints, and about 400 points ior point frame samples. ate double sampling for biomass estimation. Range' However, the studies described in this paper, taken in Science Series No. 33. Range Science Department. the field, under a variety of conditions, cannot uniform­ Colorado State University, Fort Collins, Colorado. ly reach these conclusions and, furthermore, the paramet­ Albertson, F. W. 1937. Ecology of mixed prairie in west ers for comparisons are rarely known. central Kansas. Ecol. Monogr. 7:481-547. For measuring caver, both the point method and the Aldous, S. E. 1944. A deer browse survey method. J. line intercept have been used extensively, and although Manunal. 25:130-136. there are frequently small differences in the results, Aldrich, R. C. 1979. Remote sensing of wildland re­ strict comparisons are difficult to make because of dif­ sources. A state-of-the-art review. USDA For. Serv. ferent experimental objectives and procedures (Pieper Gen. Tech. Rep. RM-71. Rocky Mtn. For. and Range l978l. For measuring production, the most common tech­ Expt. Sta., Fort Collins, Colorado. 56 p. niques include estimation, harvest, and indirect meas­ Allis, J. A. and A. R. Kuhlman. 1962. Runoff and sedi­ ures such as cover. Although the results from harvest or ment yields on rangeland watersheds. J. Soil Water clipped plots are conceptually straightforward and there­ Conserv; 17:68-71, :ore appealing, double sampling techniques have success­ Arny, A. C. and A. R. Schmidt. 1942. A study of the in­ fully improved the precision of these data and resulted clined point quadrat method of botanical analysis of in an economy of time. No concensus has ~een reached with pasture mixtures. J. Amer. Soc. Agron. 34:238-247. the ~easurement of utilization. Although exclosures have Avery, G. 1959. Evaluating understory plant cover from o~en used routinely, the difficulty in the interpretation aerial photographs. In Techniques and methods of of results is well known. Evaluations of litter condi• measuring understory vegetation. U.S. For. Serv. tions, and to a certain de~cee those of soil surface South and Southeast For. Expt. Sta. Proc. pp. 82-83. conditions, are largely descriptive according to various Avery, T. E. 1975. Natural resources measurements. categories. McGraw-Hill Book company, N.Y. 339 p. It wo~ld be desirable to design a nationwide study, Bartos, D. L. and P. L. Sims. 1974. Root dynamics of a establishing objectives and the bases on which methods shortgrass ecosystem, J. Range Mangt. 27:33-36. could be compared. Than the techniques could be applied Bauer, L. D. 1956. Soil physics. John Wiley and Sons, to a range of rangeland types and the results compared, N.Y. 489 p. However, the need for such a massive study is not appar­ Becker, D. A, and J, J. Crockett. 1973. Evaluation of ent and will not be until the detailed· needs for manage­ sampling techniques on tall-grass prairie. J, Range ment decisions are specified in precise terms. Until Mangt. 26:61-65. that time, the experienced range manager will be our Bentley, J. R. and M. W. Talbot. 1951. Efficient use of invaluable resource. annual plants on cattle ranges in the California foot­ hills. U.S. Dept. Agric. Circ. 870. Washington, LITERATURE CITED o,c. 52 p. Bitterlich, W. 1948. Die Winkelzahl probe. Allg. Forest Abar, J. D. and J, M. Melillo. 1980. Litter decomposi­ - u. Holzw. Ztg. 59:4-5 ' tion·: measuring relative contributions of organic Blackburn, w. H., R. o. Meeuwig, and c. M. Skau. 1974. l114ttar and nitrogen to forest soils. Can. J, Bot, A mobile infiltrometer for use on ran9eland. J. --- 58:416-421. Range Mangt. 27:322-323. Bondy, E., L, Lyles, and w. A. Hayes. 1980. Computing soil erosion by periods using wind energy distributior.. J. Soil Water Conserv. 35:173-176. The author would like to express his appreciation to Rex Boyd, D. A. 1949. Experiments with leys and permanent o. Pieper, who made many helpful suggestions on an grass. Brit. Grassland Soc. J. 4:1-10. earlier manuscript. 679 680

Branson, F. A., G. F. Gifford, and J. R. Owen. 1972. Conrad, C. E. and w. G. O'Regan, 1973. Two-stage strati­ Rangeland hydrology. Society for Range Management, fied sampling to estimate herbage yield. Pacific Range Sci. Series No. l. Denver, Colorado. 84 p. Southwest. For. and Range Expt. Sta. Res. Note PSW- Brown, D. 1954. Methods of surveying and measuring vege­ 278. 5 p. tation. Commonwealth Agric. Bureau. Commonwealth cook, c. w. 1960. The use of multiple regression and Bureau of Pastures and Field crops. Bull.. 42. 233 p. correlation in botanical investigations. Ecology 41: arun, J.M. and T. W. Box. 1963. A comparison of line 556-560. intercepts and random frames for sampling desert Cook, C. W. and T. w. Box. 1961. A comparison of the shrub vegetation. J. Range Mangt. 16:21-25. loop and point methods of analyzing vegetation. J. Burk, J. H. and W. A. Dick-Peddie. 1973. Comparative Range Mangt. 14:22-27. production of Larrea divaricata Cav. on three geo­ cook, c. w. and L. A. Stoddart. 1953. The quandary of morphic surfaces in southern New Mexico. Ecology 54: utilization and preference. J. Range Mangt. 6:329- 1094-1102. 335. Caird, R. W. 1945. Influence of site and grazing inten­ Cooper, c. F. 1957. The variable plot method of esti­ sity on yields of grass forage in the Texas Panhandla mating shrub density. J. Range Mangt. 10:111-115. J. Forestry 43:45-49. cooper, c. F. 1959. Cover vs. density. J. Range Mangt. Campbell, A.G., D. S. M. Phillips, and E. D. O'Reilly. 12:215. 1962. An electronic instrument for pasture yield cooper, C.F. 1963. An evaluation of variable plot sam­ estimation. Brit. Grassland Soc. J. 17:89-100. pling in shrub and herbaceous cover. Ecology 44:55- ~ampbell, R. S. 1943. Ecology-progress in utilization 569. standards for western ranges. Wash. Acad. Sci. J. Costello, D. E. and F. E. Klipple. 1939. Sampling inten­ 33:161-169. sity in vegetation surveys made by the square-foot Campbell, R. H. and J. T. Cassady. 1955. Forage weight density method. J. Agron. 31:800-810. inventories of southern forest ranges. U.S. Forest Cottam, G. and J. T. Curtis. 1956. The use of distance Service, Southern For. and Range Expt. Sta. Occas. measures in phytosociological sampling. ~cology 37: Paper 139. 451-460. Canfield, R. H. 1941. Application of the line intercep­ cowlishaw, s. J. 1951. The effect of sampling cages on tion method in sampling range vegetation. J. Fores­ the yields of herbage. Brit. Grassland Soc. J. 6: try 39:388-394. 179-182. Canfield, R. H. 1944. A short-cut method for checking crafts, E. c. 1938. Height volwne distribution in range degree of foliage utilization. J. Forestry 42:294- grasses. J. Forestry 36:1182-1185. . 295. crocker, R. L. and N. s. Tirer. 1948. Survey methods in Carpenter, L. H., O. C. Wallmo, and M. J. Morris. 1973. grassland ecology. Brit. Grassland Soc. J. 3:1-26. Effect of woody stems on estimating herbage weights Croft, A. R., L. Woodward, and D. A. Anderson. 1943. with a capacitance meter. J. Range Mangt. 26:151- Measurement of accelerated erosion on range-watershed 152. land. J. Forestry 41:112-:16. Carpenter, S. R. 1980. Estimating net shoot production Currie, P. o., M. J. Morris, and D. L. Neal. 1973. Uses by a hierarchical cohort method of herbaceous plants and capabilities of electronic capacitance instru­ subject to high mortality. Amer. Midl. Nat. 104:163- ments for estimating standing herbage. Part 2. Sown 175. ranges. Brit. Grassland Soc. J. 28:155-160. Carrel, J. E., K. Wieder, v. Leftwich, s. Weems, c. L. Darland, R. w. and J.E. Weaver. 1945. Yield and con­ Kucera, L. Bouchard, and M.-Game. 1979. Strip mine swnption of forage in three pasture types: an eco­ reclamation: production and decomposition of plant logical analysis. Nebr. Univ. Conserv. and Soil litter. p. 670-676. !!!. M. K. Wali, ed. Ecology and survey Bull. 27. Lincoln, Nebraska. Coal Resource Development. Pergamon Press, New York. Daubenmire, R. F. 1940. Exclosure technique in ecology. Chew, R. M. and A. E .. Chew. 1965. The primary productiv­ Ecology 21:514-515. ity of a desert shrub (Larrea tridentata) community. · Daubenmire, R. F. 1958. A canopy-coverage method of ·i!col. Monogr. 35: 355-375. vegetational analysis. Northwest. Sci. 33:43-64. Clark, I. 1945. Variability in growth characteristics of DeVries, D. M. and T. DeBoer. 1959. Methods used in forage plants on summer range in Central Utah. J. botanical grassland research in the Netherlands and Forestry 43:273-283. their application. Herbage Abstr. 24:1-7. Collins, R •. W. and L. c. Hurtt. 1943. A method for mea­ Dix, R. L. 1961. An application of the point centered suring utilization of bluestem wheatgrass on experi­ q1,1arter method for the sampling of grassland vegeta­ mental range pasture. Ecology 24:122-125. tion. J. Range Mangt. 14:63-69.

( C ( ) ) )

681 682

Dragonn, F. J. and A. R. Kuhlman. 1968, Effect of pas­ Hanley, T. A. 1978. A comparison of the line­ ture management practices on runoff. J. Soil Water interception and quadrat estimation methods in deter­ Conserv. 23:55-57. mining shrub canopy coverage. J. Range Mangt. 31: Dunford, E.G. 1954. Surface runoff and erosion from 60-63. pine grasslands of the Colorado front range. J. Heady, H.F. 1949. Methods of determining utili~ation Forestry 52:923-927. of range forage. J. Range Mangt, 2:53-63. · ~yksterhuis, E. J. 1949, Condition and management of Heady, H. F. 1950. Studies on bluebunch wheatgrass in rangeland based quantitative ecology. J. Range Montana and height-weight relationships of certain Mangt. 2:104-115. range grasses. Ecol. Monogr. 20:55-81. Oyksterhuis, E. J. and E. M. Schmutz. 1947, Natural Heady, H.F. 1957a. The measurement and value of plant mulches or "litter• of grasslands: with kinds and height in the study of herbaceous vegetation. amounts on a southern prairie. Ecology 28:163-179. Ecology 38:313-320. ~llison, L, 1942. A comparison of methods of quadrat­ Heady, H.F. 1957b. Effect of cages on yield and compo­ ting short-grass vegetation. J. Agric. Res. 64: sition in the California annual type. J, Range 59S-614. Mangt. 10:175-177. ~vans, R.A. and R. M. Love. 1957. The step point Heady, H. F. 1975. Rangeland management. McGraw-Hill method of sampling--a practical tool in range Book Company, N.Y. 460 p. research. J. Range Mangt. 10:208-212. Heintz, T. W., J. K. Lewis, and S. S. Waller. 1979. ?i3ser, H. G. and G. M. Van Dyne. 1966. Influence of Low-level aerial photography as a management and number and spacing of points on accuracy and research tool for range inventory. J. Range Mangt. precision of basal cover estimates. J. Range Mangt. 32:247-249. 19 :205-211 Heller, R. c. 1975. Evaluation of ERTS-1 data for for­ Fletcher, J. E. and M. E. Robinson. 1956. A capacitance est and rangeland surveys. USDA For. Serv. Res. meter for examining forage weight. J. Range Mangt. Paper DSW-112. Pacific Southwest and Range Expt. 9:96-97. Sta., Berkeley, California. 67 p. ?rancis, R. C., C. V. Baker, G. M. Van Dyne, and J. D. Hellmers, H., J. S, Horton, G. Juhren, and J. O'Keefe. Gustafson. 1971. A study of the weight estimation 1955. Root systems of some chaparral plants in method of botanical analysis. U.S. International southern California. Ecology 36 667-678. Biological Program. Grassland Biome, Technical Hickey, W. C. 1961. Relation of selected measurements Report No. 117. Fort Collins, Colorado. 147 p. to weight of crested wheatgrass plants. J, Range ~rancis, R. C., R. S. Driscoll, and J. N. Reppert. 1972. Mangt. 14:143-146. Loop-frequency as related to plant cover, herbage Hopkins, H. 1954. Effects of mulch upon certain factors production and plant density. USDA Forest Serv. of the grassland environment. J. Range Mangt. 7: Res. Pap. RM-94, Rocky Mt. Forest and Range E:xpt. 25S-258. Sta., Fort Collins, Colorado. 8 p. Hormay, A, L. 1943. A method of estimating grazing use Fuelleman, R. F. and W. L. Burlison. 1939. Pasture of bitterbrush. Calif. For. and Range Expt. Sta. yields and consumption under grazing conditions. Res. Note 35. Amar. Soc. Agron. J. 31:399-412. Hormay, A. L. 1949. Getting better records of vegeta­ Gierisch, R. K. 1967. An adaptation of the grazed plant tion changes with the line interception method. method of thurber fescue. J. Range Mangt. 20:108- J. Range Mangt. 2:67-69. 111. Hulbert, L. c. 1969. Fire and litter effect in undis­ Gifford, G. F., R.H. Faust, and G. B. Coltharp. 1977, turbed bluestem prairie in Kansas. Ecology 50:874- Measuring soil compaction on rangeland. J. Range 877. Mangt. 30:457-460. Hurd, R. M. and N. A. Kissinger, Jr. 1953. Estimating Goebel, C. J. 1955. The weight-estimate method at work utilization of Idaho fescue (Festuca idahoensisl on in southeastern Oregon. J. Range Mangt. 8:212-213. cattle range by percent of plants grazed. U.S. For. Good, R. E. and N. F. Good. 1971·. Vegetation of a Serv., Rocky Mtn. For. and Range Expt. Sta. Pap. No. Minnesota prairie and a comparison of methods. 12. 5 p. Amer. Midl. Nat. 85:228-231. Hutchings, S, S. and R. C. Holmgren. 1959. Interpreta­ Goodall, o: w. 1952. Some considerations in the use of tion of loop-frequency data as a measure of plant point quadrats for the analysis of vegetation. cover. Ecology 40:668-677. Aust. J. Sci. Res. Serv. B. 5:1-41. Hutchings, s. S. and J. E. Schmautz. 1969. A field test Greig-Smith,' P. 1964. Quantitative plant ecology, of the relative-weight-estimate method for determin­ Butterworth Sci. Pub., London. 198 p. ing herbage production. J. Range Mangt. 22:408-411. 683 684

Hyder, D. N. and F. A. Sneva. 1960. Bitterlich's plot­ Linehan, P, A. 1952. Use of cage and mower-stripmethods less method for sampling basal ground cover of bunch for measuring the forage consumed by grazing animals. grass. J. Range Mangt. 13:6-9. Sixth International Grassland Congress Proc. 2:1328- Hycer, D. N., R. E. Bement, E. E. Re111111enga, and c. 1333. Terwilliger. 1965. Frequency sampling of blue grama. Lomnicki, A., E. Bandola, and K. Jankowska. 1968, Modi­ J. Range Mangt. 18:90-94. fication of the Wiegert-Evans method for'estimation Syder, D. N., R. E. Bement, and C. Terwilliger. 1966. of net primary production. Ecology 49:147-149. Vegetation-soils and vegetation-grazing relations Lammasson, T. and c. Jensen. 1938. Grass volume tables from frequency data. J. Range Mangt. 19:11-17. for determining range utilization. Science 87:444. -:~nns, G. G. and B. R. Watkin. 1965. A modified capaci­ Lommasson, T. and C. Jensen. 1943. Determining utiliza­ tance probe technique for estimating pasture yield. tion of range grasses by height-weight tables. J. II. The effect of different pastures, soil types, and Forestry 41:589-593. dew on the calibration. Brit. Grassland Soc. J. 20: Long, G. A., P. S. Poissonet, J. A. Poissonet, P. M. Dage~ 217-226. and M. P. Godron. 1972, Improved needlepoint frame ;onnson, ~- 195?, A comparison of the line interception, for exact line transects. J. Range Mangt. 25:228. vertical point quadrat and loop methods as used in Looman, J. 1979. On pattern in vegetation. Phytocoeno­ measuring basal area of grassland vegetation. Can. logia 6:37-48. J. Pl. Sci. 37:34-42 . Ludwig, J. A., J. F. Reynolds, and P. D. Whitson. 1975. .-,,hnsan, W. M. and C. ff. Niederhof. 1941. Some relation­ Size-biomass relationships of several Chihuahuan ships of plant cover to runoff, erasion and infiltra­ desert shrubs. Alll. Midl. Nat. 94:451-461. tion on granitic soils. J. Forestry 39:854-858. Luxmore, R. J. and M. L. Sharma. 1980. Runoff responses ,-alley, A, F. 1958. A comparison between two methods of to soil heterogeneity: experimental and simulation measuring seasonal growth of two strains of Dactvlis comparisons for two contrasting watersheds. Water glomerata when grown as spaced plants and in swards. Resources Res. 16:675-684. Brit. Grassland Soc. J. lJ:99-105. Mack, R. and D. A. Pyke. 1979. Mapping individual plants ::ally, J. M., G. M. Van Dyne, and w. F. Harris. 1974. with a field-portable digitizer. Ecology 60:459-461. Comparison of three methods of assessing grassland Malone, C, R. 1968. Determination of peak standing crop productivity and biomass dynamics. Am. Midl. Nat. biomass of herbaceous shoots by the harvest method. 92:357-369. Amer. Midl. Nat. 79:429-435. Kennedy, R. K. 1972. The sickledrat: a circular quadrat Majerus, M. E. 1975. Response of root and shoot growth modification useful in grassland studies. J. Range of three grass species to decrease in soil water Mangt. 25:312-313. potential. J. Range Mangt. 28:473-476. Kemp, C. D. and A. W. Kemp. 1956. The analysis of point Marston, R, B. 1952. Ground cover requirements for sum­ quadrat data. Austral. J. Bot. 4:167-174. mer storm runoff control an aspen sites in northern Kinsinger, F. E., R. E. Eckert, and P. o. Currie. 1960. Utah. J. Forestry 50:303-307. A comparison of the line-interception, variable-plot Martin, w. P. and L. R. Rich. 1948_ Preliminary hydro­ and loop methods as used to measure shrub-crown logic res•1lts, 1935-48, 'Base Rock' undisturbed soil cover. J. Range Mangt. 13:17-21. lysimeters in the grassland type, Arizona. Soil Sci. Klipple, G, E. and o. F. Costello. 1960- Vegetation and Soc. Amer. Proc. 13:561-567. cattle responses to different intensities of grazing Maxwell, E. L. 1976. A remote rangeland analysis system. on short-grass ranges on the Central Great Plains. J. Range Mangt, 29:66-73. U.S. Dept. Agric. Tech, Bull. No. 1216. Washington, Morris, M. J., K. L. Johnson, and O. L. Neal. 1976. Sam­ D.C. pling shrub ranges with an electronic capacitance Kucera, C. L., R. c. Dahlman, and R. Koelling. 1967. instrument. J. Range Mangt. 29:78-81. Total net productivity and turnover on an energy Mountier, N. S. and J.E. Radcliffe. 1964- Problems in basis for tall-grass prairie. Ecology 48:536-541. measuring pasture composition in the field. Part 3. Lauenroth, W. K., J. L. Dodd, and c. E. Dickinson. 1980. In evaluation of point analysis, dry weight analysis, Aboveground biomass dynamics of blue grama in a and tiller analysis. New Zealand J. Bot. 2:131-142. shortgrass steppe and evaluation of a method for Mueller-Dombois, D. and H. Ellenberg. 1974. Aims and separating live and dead. J. Range Mangt. 33:210-21~ methods of vegetation ecology. John Wiley and Sons, Laycock, W; ·A. 1965. Adaptation of distance measurements New Yark. 547 p. for range sampling. J. Range Mangt. 18:205-211. National Academy of Sciences, National Research Council. 1962. Basic problems and techniques in range research. A report of a joint committee of the

( ( ( ) ) )

685 686

American Society of Range Management and the Agricul­ Mangt. 23:218-220, tural Board. NAS-NRC Publ. 890, Washington, D.C. !bissonet, P. s., P. M, Daget, J, A. Poissonet, and G. A, 341 p. Long, 1972. Rapid point survey by bayonet blade. Neal, D. L. and J. L, Neal. 1973. Uses and capabilities J, Range Mangt. 25:313. of electronic capacitance instruments for estimating Poissonet, P, S., J, A. Poissonet, M, P. Godron, and G. A. standing herbage. Part I. History and development .• Long. 1973. A comparison of sampling methods in Brit. Grassland Soc. J, 28:81-89. dense herbaceous pasture. J. Range Mangt. 26:65-67. (:Wensby, c. E. 1969. Effects of cages on herbage yield Poole, R. W. 1974. An introduction to quantitative eco­ in the true prairie vegetation. J. Range Mangt. 22: logy. McGraw-Hill Book Company, New York. 131-132, Prendergast, J, J, and J. J, Brady. 1955. Improved mov­ Packer, P, E. 1951, An approach to watershed protection able cage for use in grassland research. Brit. Grass­ criteria. J, Forestry 49:639-644. land Soc. J. 10:189-190. Packer, P. E. 1953. Effects of trampling disturbance on Rauzi, F, and F. Smith. 1973, Infiltration rates: three watershed condition, runoff and erosion. J, Forestry soils with three grazing levels in northeastern 51: 28-ll. Colorado. J, Range Mangt. 26:126-129. Park, G, N. 1973. Point height intercept analysis. New Reid, E. H. and G. D. Pickford. 1941. A comparison of Zealand J. Bot. ll:103-114. the ocular-estimate-by-plot and the stubble-height Parker, K. w. 1951. A method for measuring trend in methods for determining percentage utilization of range condition on national forest ranges. U.S. For. range grasses. J, Forestry 39:935-941. Serv., Washington, D.C. (mimeo). 26 p. Reid, E. H., J. L. Kovner, ands. C. Martin. 1963. A Parker, K. w. 1952. New trends in standards of range use. proposed method of determining cattle numbers in J, Forestry 50:856-859, range experiments. J, Range Mangt. 16:184-187. Parker, K, w. and D. A. Savage. 1944, Reliability of the Reid, E. H., G, s. Strickler, and W. B, Hall, 1980. line interception method of measuring vegetation on Green fescue grassland: 40 years of secondary succes­ the Southern Great Plains. Amer. Soc. Agron. J, 36: sion. U.S. For. Serv. Res. Pap. PNW-274. Pacific 97-110. Northwest For. Expt. Sta., Portland, Oregon. Parker, K. w. and R, w. Harris. 1959. The 3-step method Riegel, o. A., F. w. Albertson, andH. H. Hopkins. 1950. Yields for measuring condition and trend of forest ranges: and utilization of forage on a mixed prairie in West­ a resume of its history, development, and use. In Central Kansas. Kans. Acad. Sci. Trans. 53:455-472. Techniques and methods of measuring understory vege­ Renner, F, G, 1936. Conditions influencing erosion on tation. U.S. For. Serv., South and Southeast. For. the Boise River watershed. U.S. Dept. Agric, Bull. Expt. Sta. Proc. 1959:55-69. 528. Washington, D.C. Pasto, J. K., J, R, Allison, and J, R. Washku. 1957. Reppert, J, N., and R. E. Francis. 1973. Interpretation Ground cover and height of sward as a means of esti­ of trends in range condition from 3-step data. USDA mating pasture.production. Agron. J. 49:407-409. For, Serv. Res. Pap. RM-103. Rocky Mt. For. and Payne, G, F, 1974. Cover-weight relationships, J. Range Range Expt, Sta., Fort Collins, Colorado. 15 p. Mangt. 27:403-404. Reppert, J. N., M, J, Morris, and C, A. Graham. 1962. Pechanec, J, F. 1936. Comments on the stem-count method Estimation of herbage on California annual-type of determining the percentage utilization of ranges. range. J. Range. Mangt. 15:319-323. Ecology 17:329-331. Rice, E. L, 1967. A statistical method for determining Pechanec, J, F. and G, D, Pickford. 1937a, A weight-esti­ quadrat size and adequacy of sampling. Ecology 48: mate method for the determination of range or pasture 1047-1049. production. J. Amer. Soc. Agron. 29:894-904, Rich, R, w. 1959. Aerial photography as a means of mea­ Pechanec, J, F. and G, D, Pickford. 1937b. A comparison suring plant cover and composition. In Techniques of some methods used in determining percentage utili­ and methods of measuring understory vegetation. U.S. zation of range.grasses. J. Agric. Res. 54:753-765. For. Serv. South and Southeast For. Expt. Sta. Proc. Penfound, W. T, 1963. A modif~cation of the point-cen­ pp, 79-81. tered quarter method for grassland analysis, Ecology Risser, P. G. and D, H. Zedler. 1968. An evaluation of 44:175-176. the grassland quarter method. Ecology 49:1006-1009, Pieper, R. D. 1978, Measurement techniques for herbace­ Ritchie, J, C, and J. R, McHenry, 1973. Determination of ous and shrubby vegetation. New Mexico State Univer­ fallout l37cs and naturally occurring gamma ray sity, Las Cruces, New Mexico lmimeo). emitters in sediments. Int. J, Appl, Red, and Isot. Pierce, W, ·R. and L, E. Eddleman. 1970. A field stereo­ 24:575-578. photographic technique for range analysis. J. Range 687 688

Ritchie, J.C. and J. R. McHenry. 1978. Fallout cesiwn- fects of repeated artificial drying. Can. J. Bot. 52: 137 in cultivated and noncultivated north central 2157-2163. United States watersheds. J. Environ. Qual. 7:40-44. Tinney, F •. w., o. S. Aamodt, and K, L. Ahlgren. 1937. Roach, J. E. 1950. Estimating perennial grass utiliza­ Preliminary report of study of methods used in bo­ tion on semi-desert cattle ranges by percentage ot tanical analysis of pasture swards. J. Amer. Soc. ungrazed plants. J. Range Mangt. 3:182-185. Agron. 29:835-840. Robinson, P. 1955. The estimation of ground cover by Tomanek, G. w. 1969. Dynamics of mulch layer in grass­ the pcint quadrat. Ann. Bot. 19:59-66. land ecosystems. pp. 225-240. In R. L. Dix and R. G. Ruby, E. s. and v. A. Young. 1953. The influence of in­ Beidleman, eds. The grassland ecosystem: a prelimi­ tensity and frequency of clippings on the root system nary synthesis. Range Sci. Dept., Range Sci. Ser. of brownseed paspalum. J. Range M.angt. 6:81-92. No. 2. Colorado State University, Fort Collins, Schultz, A. M., R. P. Gibbens, and L. DeBano. 1961. Arti­ Colorado. 437 p. ficial pcpulations for teaching and testing range Trieste, D. J. and G. F. Gifford. 1980. Application of techniques. J. Range Mangt. 14:236-242. the universal soil loss equation to rangelands on a Schmutz, E. M. 1971. Estimation of range use with per-storm basis, J. Range Mangt. 33:66-70. grazed-class photo guides. Univ. Arizona Coop. Ext. Tromble, J.M., K. G. Renard, and A. P. Thatcher. 1974. Serv. and Agric. Sta., Bull. A-73. 16 p. Infiltration for three rangeland soil-vegetation com­ Schmutz, S. M., G. A. Holt, and c. c. Michaels. 1963. plexes. J. Range Mangt. 27:318-321. Grazed class method of estimating forage utilization. Tueller, P. T., G. Lorain, K. Kipping, and C. Wilkie. J. Range Mangt. 16:54-60. 1972. Methods for measuring vegetation change on Schuster, J. L. 1965. Estimating browse fran twig and Nevada rangelands. Nevada Agric. Expt. Sta. T-16. stem measurements. J. Range Mangt. 18:220-222. Univ. Nevada, Reno, Nevada. 55 p. Shoop, M. c. and E. H. Mcilvain. 1963. The micro-unit Tukey, R. B. and E. L. Schoff. 1963. Influence of dif­ forage inventory method. J. Range Mangt. 16:172-179. ferent mulching materials upon the soil environment. Singh, J. s., w. K. Lauenroth, and R. K. Steinhorst. 197~ Amer. Soc. Hort. ScL 82:6B-76. Review and assessment of various techniques for esti­ Upchurch, R. P. and R. L. Lovvorn. 1951. Gross morpho­ mating net aerial primary production in grasslands logical root habits of alfalfa in North Carolina. from harvest data. Bot. Rev. 41:181-232. Agron. J. 43:493-498. Smith, A. D. 1944. A study of the reliability of range U.S. Department of Agriculture, Soil Conservation Service. vegetation estimates. Ecology 25:441-448. 1975. Soil taxonomy. A basic system of soil classi­ Springfield, K. w. 1961. The grazed plant method for fication for making and interpreting soil surveys. judging the utilization of crested wheatgrass. J. Agric. Handbook No. 436. Washington, D.C. Forestry. 59:666-670. . U.S. Department of Agriculture, Soil Conservation Service. Squiers, E. R. and W. A. Wistendahl. 1976. Sample unit 1976. Report of wind erosion conditions on the Great selection for studies of herbaceous old-field vegeta­ Plains. Washington, D,C. tion. Ohio J. Sci. 76:185-188. U.S. Department of Agriculture, Forest Service. 1963. Stephenson, R. E. and C, E. Schuster. 1945. Effect of Range research methods. Proc. of Symp., Denver, May mulches on soil properties. Soil Sci. 59:219-230. 1962. U.S. Dept, Agric. Misc. Publ. No. 940. l72pp. Stoddart, L. A. 1935. Range capacity determination. Valentine, K. A. 1946. Determining the grazing use of Ecology 16:531-533. grasses by scaling. Ecology 44:528-530. Stoddart, L. A., A. D. Smith, and T. w. Box. 1975. Rang, Van Dyne, G. M. 1960. A procedure for rapid collection, management, McGraw-Hill Book Company, New Yark. processing, and analysis of line intercept data. J. 433 p. Range Mangt. 13:247-251. Stowe, L. G. and M, J, Wade. 1979. The detection of Van Dyne, G. M., w. C. Voegel, and H. G. Fisser. 1963. small-scale patterns in vegetation. J, Ecology 67: Influence of small plot size and shape on range 1047-1064. herbage production estimates. Ecology 44:746-759. Strickler, G. S. 1961. A grid method for obtaining loop Van Dyne, C. M, F. N. Glass, and P. A. Opstrup. 1968. readings on small plots. J. Range Mangt. 14:261-263, Development and use of capacitance meters to measure Strickler, G. S. and F. w. Stearns. 1963. The determin­ standing crops of herbaceous vegetation. Oak Ridge ation of plant density. In Range research methods. Natl. Lab. Tech. Mem. 2247. Oak Ridge, Tennessee. U.S. Dept. Agric. Misc. Pu""°iil. No. 940. Van Keuren, A. W. and a. L. Alhlgren. 1957a. A statisti­ Suffling, R. and D. W. Smith. 1974. Litter decomposi­ cal study of several methods used in determining the tion using mesh bags: spillage inaccuracies and ef- botanical composition of a sward. I. A study of established pastures. Agron. J. 49:532-535.

,( ( ( ) ) ) 689 6!>0

"an Keuren, A. w. and H. L. Ahlgren. 1957b. A statisti­ Winkworth, R. E., R. A. Perry, and C. V. Rossetti. 1962. cal study of several methods used in determining the A comparison of methods of estimating plant cover in botanical composition of a sward. II. A study of an arid grassland cormnunity. J. Range Mangt. 15: several forage mixtures. Agron. J. 49:581-585. 194-196. Von Steen, D. a., R. W. Leamer, and A. H. Gerberman. 196~ Whittaker, R. H. 1966. Forest dimensions and production Relationship of film optical density to yield indi­ 1n the Great Smoky Mountains. Ecology 47:103-121. cators. I: 1115-1122. In Proc. 6th Int. Symp. Remote Sensing Environ.,'"Tnst. Sci. and Tech., Ann Arbor, Michigan. 650 p. l'Jagner, a. E. 1952. Weight estimation and other proce­ dures for measuring the botanical composition of pastures. Sixth International Grassland Congress Proc. 2:1315-1321. warren-Wilson, J. 1959. Analysis of the distribution of foliage area in grassland. In Measurement of grass­ land productivity. pp. 51-6I:" Academic Press, New York. w.. rren-Wilson, J. 1960. Inclined point frames. New Phytol. 59: 1-8. ~•rren-Wilson, J. 1963. Estimation of foliage denseness and foliage angle by inclined point quadrats. Aust. J. Bot. 11:95-105. 'ieaver, J. E. and W. E. Bruner. 1948. Prairies and pas­ tures of the dissected loess plains of central Nebraska. Ecol. Monogr. B:507-549. ;;ells, K. F. 1971. Measuring vegetation changes on fixed quadrats by vertical ground stereophotography. J. Range Mangt. 24:233-236. :·,hitman, w. c. and E. I. Siggeirsson. 1954. Comparison of line interception and point contact methods in the analysis of mixed range vegetation. Ecology 35:431- 436. ·,;ieder, R. K. and J. E. Carrel. 1979. Radioactive car­ bon-14 as an indicator of energy flow in litter decomposition studies. Oecologia 44:31-33. ,viegert, R. G. 1962. The selection of an optimum quadrat size for sampling the standing crop of grasses and forbs. Ecology 43:125-129. Wight, J. R. 1967. The sampling unit and its effect on saltbush yield estimates. J. Range Mangt. 20:323- 325. Williams, R. E., B. w. Allred, R. M. Denio, and a. A. Paulsen, Jr. 1968. Conservation, development, and use of the world's rangelands. J. Range Ma.ngt. 21: 355-360. Wilm, H. G., D. F. Costello, and G. E. Klipple. 1944. Estimating forage yield by.the double-sampling method. Amer. Soc. Agron. J. 36:194-203. Wilson, J. W. 1963. Estimation of foliage denseness and foliage angle by inclined point guadrats. Aust. J. Bot. 11:95-105. !oli.nkworth, R. E. 1955. The use of point guadrats for the analysis of heathland, Aust. J. Bot. 3:68-81. 692

a part of the historical record. Tiiese data can often be used as a starting point for monitacing efforts.

This Critique is aimed at setting the framework for methods which might be suited for inventory and tllOni~oring and co evaluate some of these methods for their efficacy in light of imposed con­ straints. Risser (1981) has reviewed most of the available methods A Critique of "Methods for measuring and estimating vegetational parameters which goea for Inventory and Monitoring beyond the scope of this critique. of Vegetation, Litter, SA.'iPLING REQUIROIENTS Range vegetation varies in both time and space. The reliability and Soil Surface Condition" of sampling is directly related to chis variability, Fev options are available to cope with this variability. One of these is to increase Re., D. Pieper sample size. Risser (1981), Avery (197S) and most statistical texts give simple formulas which may be used to calcul~te sample size if some measure of variAbility is known. However. in many cases it is ABSTRACT not feasible to increase sample ·size because of limitations in time and personnel. Sampling range vegetation is very difficult because of the large area involved and the variability of the vegetation. Methods of coping Scratification offers another ~eans of improving sampling with th• variability include using large sample sizes, stracifica­ reliability (Bollham 1981). Stratification is often based on soil tion, adjusting sample units for the vegetation being sampled, and type or plant communi~Y (Driscoll 1981), Allocation of sample units usin& key are.as co represent larger areas. Inventory data should within a stratum may be based on si~e of area. variability of vegeta­ have low observer variability while monitoring data should account tion, magnitude of parameter estimated, or any combination of these for spatial variability. Plant cover can be determined by estimation (Reppert et al. 1962). The sample unit also may be adjustod to or use of seep-point sampling for extensive areas, but many other minimize sam?le variability, For example, Canfield (1941) recommend­ methods are too time consuming. The weight-estimate methods appears ed a SO ft cr3nsect for vegetation ~~th cover between 5 and 15% and a to be the moat useful method of determining herbage weight, and can 100 ft transect for areas with cover between 0.5 co 3l. In this case be used in conjunction with double sampling procedu~es. Indirect and one must realize that th& population being sampled is made up of all dimension analysis may be used fo~ estimating shrub biomass, but possible sample units, and that each population would have indepen­ developing appropriate equations may ba difficult, Although many dent parameters. For example one should not expect the same para­ methods have been developed to astimate utilization, none is without meters from 1 sq. ft. square quadracs as for 25 sq~ ft. rectangular dr3vbacks. Some workers have advocated using residue left after quadrats. Adjus~ing the sample units to fit the vegetation to be grazing as an index rather than percentage utilization. sampled appears eo be an excellent ~ay of improving sampling effici­ ency. Too often a sample unit has been us~d simply because it has OVERVIEW been used in the past without regard to its suitability for ~egeta­ tion being sampled. For example, a 9.6 sq. ft. circular quadrat has Obtaining reliable inventory and monitoring data upon which to often be&n used for sampling herbage weight largely because of the base land management decisions is extremely difficult largely because ease of conversion to pounds per acre (Pechanec ~nd Pickford 1937). of two factors: the large land area involved and the diversity and Daubenmire's (1959) 2 X 5 dm. quadrat has been used to estimate cover variability associated with the vegetation and soils on rangelands. over a wide range of vegetation types but was originally designed for In the past, land management agencies have been criticized for lack sagebruah-grass vegetation, of basic data for decision making. However, this lack is not likely to be overcome quickly. It is important that those in positions to Considering the wide variation in vegetation and soils present aaka decisions and those leveling cricicizms at the agencie• under­ on western rangelands. it is not possible to design specific sampling stand the magnitude of the problem and what neads to be done to alle­ methods which will be suitable under all conditions. The parameters viate ic, While ecological data aeed to be collected over a long to be estimated may be identified and the general sam~ling scheme time span to accumulate an adequate data base, some adjustments can selected. but latitude for some modifications to meet local condi­ probably ba made on the basis of data collected over fairly thort tions should be provided. Ideally, agencies involved in land manage­ time spans. ment should have staff with sufficient background and training to select methods suitable for their conditions, and the ability to Basic data a~• lacking for JDany U.S. ranges. Hovever, these modify methods when necessary. Specialists may be used to develop ranges have been used for a variety of purposes for a considerable period of time and some information on livestock and gaae numbers is

691

( ( ( ) ) ) 694 69~ vegetational types (Evans and Love 1957. Heady 1956). Some workers workshops and fiald training sassions for field personnel, have advocated counting ehe nearest plant to the point for composi­ tion calculations to increase the number of observations. Recent Even with large sample sizes, stratification, and selection of work (Straus·$ and Seal 1981) indicates that there -may be some bias appropriate s11111ple units, it is not likely that rans• sampling w:l.ll associatad with the step point method, especially for large plants. meet conventional 1tatistical critaria on many rangelands. Another approach is to select representative or key areas for $ampl1ng (Smith The main problem. associated with poinc sampling is a slight 1965, Stoddart, Smith and Box 1975). Such an approach can be suc­ positive bias associated with bluntness of the point (Brun and Box caasful only if the sampler has considerable experience and knova the 1963, Johnston 1957, Whitman and Siggeirsson 1954). This bias may be area well, The approach is exemplified by the U.S. Forest Service accentuated with the seep-point method if the observer does not use clusters of permanent 100 ft. loop transects located 111 111<.ey" areas strict standards in recording hits. A triangular device such as the within alloaaents. The assumption is that the key area is represen­ one described by Owensby (1973) should help minimize some of theoe tative of the whole area and that changes measured cha.re are similar biases. The Bureau of Land ~nagement has incoTporated step-point :o those on the rest of the allotmant (Smith 1980). Changes in sampling into their SVIM (Soil-Vegetation Inventory ~ethods) :nanag ..enc are based on changes in these ke.y areas. procedures.

Although many methods can be used for both inventory and HERIIAGE WEIGHT Qonitoring as Risser (1981) has pointed out, requirem~nca might not be quite the same for the two functions. Inventory methods have a :iany methods of determining herbage weight are tedious and time nremium on being applicable to extensive areas at one time. Tha main consuming. It appears that estimation and indirect methods may be objective is to obtain the best set of data possible to represent the the traost suitable for inventory and monitoring purposes. 4rea in question. Monitoring data, on the ocher hand, are often used to detect trand or vegecational changes over time and may be Selecting an appropriate quadrat size and shape is important in restricted to small, representative areas with satisfactory rasults. he~bage weight determination. Tbe perimeter:aTea ratio determines Often the same personnel ~ill not be involved in making the repeated the number of borderline decisions which must be made. Van Dyne measurements. With this pcocasa it is important to have methods with et al. (1963) contended that plots with high P:A ratios yield data low observer va~iability so chat different samplers will obtain con­ with a positive bias because of the tendency on the part of che sistent results~ Such consistency was reported for loop procedures observers to include more in the quadrat than was justified. Circu­ by Sharp (1954). Sampling co measure vegecational cha.age over time lar quadrats have the lowest P:A ratios, and for the same qua.drat should be conducted to minimize spatial variation. Using perman­ shape, large quadracs have lower P:A ratios than small quadrats ently-marked sample units is one method of assuring this. (Pieper 1978). However, one must also be concerned about the variability and precision of his sample. lt has long been known All these procedures - using large samplas, stratification, that transects or elongated quadrats are more likely to intersect adjusting sample units co fit tha vegetation, selecting key areas, more vegetational types and encounter more species than short ones and using appropriate techniques for inventory and monitoring - may (Costing 1956). Wight (1967) and Pino (1962) have demonstrated the help to obt3in reliable information. However, even under the most considerable reduction in variability among data of. quadrats as the favorable circumstances inventory and monitoring data must be length is increased. Wight's daca can be computed On the basis of tempered with good judgment and some subjectivity. equal si:e quadrats while that of Pino has quadrac size slightly con­ founded with shape. In another study, Papanastasia (1977) found that METHODS FOR DETERHING COVER quadrat size was more important than shape for bunchgr~ss vegetation in Greece. These studies were conducted with clipping as the method Many methods for determining cover are too time-consuming to be of determining herbage weighc. used for either inventory or monitoring extensive range areas. Esti­ mation and point sampling techniques appear co be th• principal Vegetatioaal parameters may be estimated to increase sample methods available which could be used for extensive areas. Estima­ sizes and reduce reliability, but it muse be recognized that vari­ tion techniques are those made without benefit of tools or mechanical ances ""'Y be reduced artifically (Shoop and Mcllvain 1963). Appa~­ devices. However, estimation techniques depend !.lPOn the skill of the encly there is a tendency on the part of the estimator to underesti­ observer and could lead co biased results, Smith (1944) reported mate quadrats with large biomass and to overestimace those with considerable variation among days arid observers for the square-foot­ small biomaa.s. density method on several vagetationa.1 types in Utan. Estimates tended to become progressively lover on the same ~reas each time PTecipitatian has often been used as the independent variable to they were sampled. predict biomass or end-of•saason standing crop of vegetation. In some cases annual or seasonal precipitation accounts for a large The point-step method offers advantages in reducing time amount of the variation in annual production. However. the following required for sampling and is fairly reliable for different 695 696

wandering quarter method appears much moLe suitable for extensive tabulation shows the range in coefficients of variacioQ from several inventories.than the angle-order method (Morishita 1957) which was studies: also theoretically suited for nonrandom populations. Independent Although viewed with suspicion by some -workers, frequency Variable a2 Vegetation Author sampling is another method which might have utility in extensive Annual ppt. 0.89 Saltdesert Shrub Hutchings & Stewart 1953 surveys since only a yes-no decision muse be :nade. Quadrat size and May - June ppt. 0.86 Mixed Prairie Smoliak 1956 shape is critical for fr~quency sampling (Hyder et al. 1964) and June - August o. 78 Desert Grassland Cable 1975 determination of optimum quadrat size may requir;-co"nsiderable effort April 0.41 Annual Grassland & Duncan Woodmansee 1975 if the details have not been worked ou~ for each vegetation type. July - August 0.42 Pinyan- juniper Pieper et al. 1971 Examples of differences in quadrat size needed for different vegeta­ tion are a 9 in. X 9 in. quadrat for sagebush grass vegetation in Sneva and Hyder (1962) have summarized precipitation-yield Oregon (Hyder, et al. 1964) and 16 in. x 16 in. quadrat for short relations for several areas in the !ntermouncain Region and developed grass vegetatio;-i~Colorado (Hyder et al. 1964). Optimum. quadrat som.e composite curve!. These data need to ba developed over several size for Bouteloua gracilis was only°"""z ~. X Zin. Hyder~ al. ye~rs co encompass the range in variation of precipitation. Several (1964) recommended that the quadrat size be adjusted so that the .. malyses need to be conducted co find the best "predictor." If data frequency of most species fell between 5 and 95% and that of the most ~re collected over• period of years with nearly average precipita­ abundanc species exceed 75%. Species-araa curves may be used as an tion, then it may not be possible to predict herbage weight during an aid in determining optimum quadrat size (Cain 1938). Hyder~ al. ~typical year. There are limitations to these methods in Che south­ (1975) was able to demonstrate response of many species in short west because of variations in plant density and cover (Parker 1963, grass vegetation to environmental variables using frequency sampling. ~ussell 1973). However, once the equations have been developed, it However, sampling intensity must be high co dececc changes in minor may be possible to predict he~bqge weight from precipitation data. spe;cies.

I>eteruaining production or biomass of browse species is also vet"Y UTILIZATION difficulc. One ~pproach is to measure canopy dimensions and relate chese to ?lane biomass. Titese analyses have been lumped under the Much effort has been expended on ™ethods for determining ;eneral heading of "dimension analysis,'' Relationships have been utilization (Pieper 1978). If one is interested in regulating graz­

developed for browse t.teight and cr0"'1l diameter, crown volume 9 crown ing pressure to the point where sufficienc photosynthetic tissue •rea, and crown height (Bryant and Kothmann 1979 !iedin 1960. Lyon relllains to ~aintain vigor and productive capacity of individual 1968, Ludwig et al. 1975, Rittenhouse and Sneva 1977, Uresk et al. plants, then it seems intituitively obvious chac utilization is one 1977, Nadabo et al. 1980, and Harniss and l!urray 1976). parameter of primary importance. However. if percentage utili~ation methods are to be used most effectively. then ...,e must have proper use Development of these dimensional relationships is tedious 30d standards for the key species (Hedrick 1958). Development of proper time consuming since the planes must be harvested and weighed after use factors is also a difficult cask because of variability associ­ the ~easuremencs have been made. However, after che relAtionships ated with topography, distance from ~ater. vegec1cion type, season have been determined, then only the appropriate shrub dimensions of grazing, and range condition. Nevertheless. having a good esti­ ~ould have co be measured. Several studies have shown that predic­ mate of utilization may be the best method of determining the tions can be improved by transforming dimension measurements co log inunediate impact of grazing on vegetation. If propar use factors and functions (Rittenhouse and Sneva 1977, Bryant and Kothmann 1979). current utilization levels are known, then adjustments in livestock numbers can be made on the basis of utilization estimates. , DENSITY AND FREQUE.~CY Estimation techniques depend upon the skill of the observer and Density (number of individuals per unit area) h.aa not be~n used may not be suitable under situations where different personnel may be extensively for inventoring and monitoring herbaceous vegetation. involved in the estimations. Ocular estimate by individual plant may Density has been used for specific purposes (e.g* evaluation of seed­ be the best way of assuring an adequate sample for each species of ings), but probably has more promise for shrubs 3nd trees. Some interest. apecies which reproduce vegetatively IDAy be difficult co measure for· density because an "individual" is not distinct. Risser (1981) has Clipping before and after grazing is applicable m&inly during Pinpointed some of the problems associated with distance techniques the dormant season when growth is nae a factor. The cage-comparison used with species with concagious distributions. However. the wan­ method is also fraught with the difficulty of unequal grovth of dering quarter method (Catana 1963) was developed for nonrandomly grazed and protected plants (Cook and Stoddart 1953) if the grazing _distribuced populations and is no~ being used for inventory purposes season includes all or part of the growin& season. Besides, these by the Soil Conservation Service on pinyon-juniper vegetation. the clipping methods probably are coo time-consuming to be used on

( ( ( ) ) 698 697 extensive areas. precise results. We have spent litt·.l:e effort to develop techniques for large scale inventory and monitoring purposes. Unfortunately, The stem counc method {Scoddarc 1935) is nothing more than a information needed for best land management decisions also needs to be accurate and precise. Statistical principles apply equally to ooor indeK of utilization (Pechance 1936). Relatin1 th• percentage planes grazed or ungrazed co actual ucilizacion by weight appar­ small research plots and large grazing allotments. The challenge is of there and opportunities abound for the bright, innovative worker to ently works for some species (Roach 1950, Hurd and Kissinger 1953, Springfield 1961). However, for other species, it is not sensitive develop mare celiable inventory and monitoring techniques. ac utilization levels above about 40% {Springfi•ld and Peterson LITERATURE CITED 1964). Ability to detect higher levels of utilization may be ~nhanced for some specie• such as Festuca thurberi by relatin& per­ Avery, t. E. 1975. Natural resources measurements. Second Ed. centage of plants &razed co certain levels or stubble heights to McGraw-Hill Book Co. New York. utilization (Gierisch 1967). Bement, R. E. 1969. A stocking-race guide far beef production on Haight-weight relations have bean developed.for a large number blue-grama range. J. Range Manage. 22:83-86. Bonham. c. o. 1981. Sampling and statiscical consideracions in of species (Keady 1950, Lommasson and Jensen 1938, 1943, Campbell range resources, NAS/NRC Workshop. Tucson, Aciz. ~937, Caird 1945, Clark 1945 and Crafcs 1938). AIJ Rieser {1981) has Brun, J.M. and T. W. Box. 1963. A comparison of line intercepts pointed out, these relations vary considerably from year co year and and random frames for sampling desert shrub vegetation. J. •it• to site for the saae species. Some of these problems can be Range Manage. 16:21-25. ~inimizad by developing separate curves or tables for plants of Bryant, F. C. and M.M. Kothmann. 1979. Variability in predicting different sizes (Valentina 1946). edible browse from crown volume, J. Range Manage. 32:144-147. Cable, o. R. 1975. Influence of precipitation on perennial grass Despite all cha attempts to measure utilization, there remain production in the semidesert southwest. Ecol. 56:981-986. many problems since ve are trying to quantify a vegetal component Cain, S. A. 1938, The species - area curve. Amer, Hidl. Nae, ·,hich is no longer presenc. Hyder (1953, 1954) has been critical of 19:573-581. "?ercentage" ucilizacion methods because of variations in plant Caird, R. w. 1945. Influence of site and grazing intensity on response introduced by years, seasons of grazing, species, soils, yields of grass forage in the Texas Panhandle. J. Forestry ~ange conditions, climates, and intensities of grazing, etc. Hyder 43:45-49. (1954) stated these concerns as follows: Campbell, R. S. 1937, Problems of measuring forage utilization on western ranges. Ecol. 18:528-532. "Percantaga removal daca appear to have the value of permitting Canfield, R.H. 1941. Application of the line interception method accurate ,omputations of stocking rate changes needed· to balance in sampling range vegetation. J. FoT. 39:388-394. cropping with sustained forage supply. But the sourcea of vari­ Catana, A. J., Jr. 1963. The wandering quarteT method of estimating ability prevent a direct plane physiological interpretation of population density. Ecol. 44:349-360. those percentages, and relegate to near impossible the job of Clark, I. 1945. Variability in height of forage grasses in centTal developing usable prover-use factors and the knowledge to Utah. J. For. 42:273-283. justify their dif~erenti3tion in application. Without accurate Cook, C. W. and L.A. Stoddart. 1953. The quandry of utilization proper-use factors applicable to the individual situations and preference. J. Range Manage. 6:329-331. introduced by the sources of variability, the quantitative con­ Crafts, E. C. 1938. Height-volume distribution in range grasses. cept in percentage utili~acion data may not be used with reli­ J. For. 36:1182-1185 ability far correcting stocking: rates and systems of grazing.'' Daubenmire, R. F. 1959. Canopy coverage method of vegetation analysis. Northwest Sci. 33:43-64. Hyder advocated using amount of residue left after grazing Driscoll, R. S. 1981. Stratification of landscape for purpose of rather than the amount removed as the index of grazing intensity. classification, mapping and data display of basic surface Bement (1969) developed stocking rate guides using amount of residue resources of cangelands. NAS/NRC symposium. Tucson, Arizona. left for shore grass vegetation in Colorado. Unfortunately, we do Duncan, D. A. and R. C. Woodmansee. 1975. Forecasting forage yield not have such &uidelines for other range types. Although, ve do have from precipitation in California's annual rangeland. J. Range much backlround in develOpment and use of proper use factors, the Manage. 28:327-329. residue approach should be considered sarioualy by land mauaaament Evans, R. A. and R. M. Love. 1957. The step point method of a1encies. sampling - a praccical tool in range research. J. Range Manage. 10:208-212. Although we have spent considerable time and effort to develop Gierisch, R. K. 1967. An adaptation of the grazed plant method for techniques foT measuring and estimating range prop•rties, many of detanining utilization of thurber fescue. J. Range Mana&e, these method• hava application chiefly in small research areas where 20: 108-lll. detailed maa•uremant can be takan with reaaonably accurate and 699 700

Harniss, R. O. and R. B, Hurray. 1976. Reducing bias in dry leaf sampli~g herbage weight in grasslands of northern Greece. J. estimates of big sagebrush. J. Range Manage. 29:430-432. Range Manage, 30:446-449, Heady, H.F. 1950, Studies on bluebunch wheatgrass in Montana and Park.er, E. E, 1963, Estimating grass ~erbage production on desert heigbc-weight relationships of certain range grasaes. Ecol. plains grassland range. M.S. Thesis. New Mex, State Univ. Las Honogr. 20:S5-8l. Cruces. 37 p .. Heady, H.F. 1956, Evaluation and measurement of the Califon,ia Pechanec. J. F. and G. D, Pickford. 1937. A comparison of some annual type, J, 11.ange Manage. 9:25-27, methods used in determining percentage utilization of range Hedrick, D. W. 1958, Proper utilization - a problem in evaluating gras,es. J. Agr. Res. 54:753-763, th• physiological response of planes to grazing: a review. J. Pieper, R. o .• J. R. Montoya, and V. t. Grace. 1971. Site charac­ Range Manage. ll:34-43. teristics on pinyon-juniper and blue grama ranges in south­ Hurd, R, M. and~. A. Kissinger, Jr. 1953. Estimating utilization central New Mexico. New Mex. State Univ. Agr. Expt, Sta. Bull, of Idaho Fescue (Festuca idahaensis) on cattle range by percent ;73, of plants grazed. Rocky ~t. For. and Range Expc. Sta. Paper~- Pieper. R. D. 1978. Measurement tec~niques for herbaceous and 12: l-5. shrubby vegetation. New ~1ex. Stat.a Univ. Las Cruces. Hutchings, S.S. and C. Stewart. 1953. Increasing forage yields and Pino, R. 1962. Influence of plot shape for sampling desert grass­ shaep production on intarmouncain winter ranges. U.S. Dept. land vegec..ttion. M.S. Thesis. Xew 'Mex. State Univ. Las Agr. Circ. 925, Cruces. 32 p. Hydar, D, N. 1953. Grazing capacity as related to range condition. Reppert, J, N•• M. J, Reed, and Pinhas Zusman. 1962. An allocation J, For. Sl:206. plan for range unit sampling. J. Range Manage. 15:190-193. lfyder, D, )I. 1954. Forage utilization. J. For. 52:603-604. Rittenhouse, L. R. and F. A, Sneva. 1~77. A technique for estimat­ Hyder, D. ~., C. E. Conrad, P. T. Tueller, C. o. Calvin, C. E. ing big Sagebrush production. J. lange Manage. JO: 68-70. Poulton, ilad F. N. Sneva. 1963. Frequency sampling in Risser, P. G. 1981. ~athods for inve~tory and monitoring of vegeta­

sagebrush-bunch grass vegetation. Ecol. 44:740-746, . tion 1 litter, and soil surface co~dicion. NAS/NRC Workshop. ~yder. D, ~ •• R, E, Bement, and C. Terwilliger. 1964. Frequency Tucson, Ariz. sampling of blue grama range. J. Range ~anaga. 18:90-93. Roach,~. E. 1950. Estimating peren~ial grass utilization on semi­ Hyder, D. ~:., R.. E. Bement, E. E. Remmenga, and o. r. Hervey. 1975. desert cattle ranges by ?ercentage of ungrazed plants. J. Ecolo&ical responses of native pl3nts and guidelines for manage­ Range ~anage. J:182-185, ment of shortgrass range. U.S. Dept. Agr. Tech, Bull. 1503, Russell, John w. 1973. Soil-?lant-U.·.restock-Weather relationships Johnston, A. 1957. A comparison of the line intercept, vertical of a range ecosystem: Interacti~:e timeshare regression and point quadrat and loop ~echada of as used in measuring basal simulation model approaches. ~.S. Thesis. New Mexico State area of grassland vegetation. Canad. J, Plt, Sci, 37:34-42. Univ. Las Cruces. 94 p. Lomm.assan, r. and C. Jensen. 1938. Grass volume tables for deter­ Sharp, L. A. 1954, Evaluation of the loop procedure of the 3-step mining ran~e utilization. Sci. 87:444, method in the salt-desert shrub -;::pe of southern Idaho. J. Lomma.sson, T. and C. Jen~en. 194). Determining utilization of range R.ange Manage. 7:83-88. grasses by height-weight tables. J. For. 41:589-593. Shoop, M. C. and E. H. Mcilvain. 1963. The micro-unit foraga Ludwig, J. A,, J, F. Reynolds, and P. D, Whitson. 1975. Size­ inventory method. J. Range Manage. 16:172-179. biomass relationships of several Chihuahuan desert shrubs. Smith, A, D. 1944. A study of the reliability of range vegetation Aaer. l!idl. Nae, 94:451-461. estimates. Ecol. 25:441-448. Lyon, J. L. 1968. Estim.ating twig production of secvicaberry from Smith, A, D, 1965, Determining comi::on use grazing capacities by crown volumes. J, Wildl, Manage. 32:115-119. application of the key species concept. J. Range Manage. lledin, D, E. 1960, Physical site factors influencing annual pro­ 18: 196-201. duction of true mountain mahogany. Cercocarpus montanus. Ecol. Smith, Lamar. 1980. Use of inventor:, and monitoring data for Range 41:454-460. Management purposes. NAS/NR.C Workshop, Tucson, Ariz. Morishita, H. l9S7. A new method for the estimation of density by Smoliak. S. 1956. Influence of climatic conditions on forage pro­ the spacing method applicable to non-randomly distributed popu­ duction of shortgrass rangeland. J. Range !'fa.nage. 9:89-91. lations. Physiol. and Ecol. 7:1J4-l44. Sneva, F. A., and D. N. Hyder. 1962. Estimating herbaae production Nadaba, S~. R. D. Pieper, and R. F. Beck. 1980. Growth patterns and on semiarid ranges in the Interoountain Region. J, Range biomass relations of Xanthocephalum sarothrae (Pursh) Shinners Manage. 15:88-92. on sandy soils in southern New Me.x. J. Range Manage. Springfield, H, w. 1961. The grazed ?l•nt method for judging the 33:394-397. utilization of crested wheatgras,, J, For. 59:666-670. Owensby, C. E. 1973, Modified step-point system for botanical com­ Springfield, H, W., and G. Peterson. 1964. Use of the grazed plant position and basal cover estimates. J. Range :-tanage. method for estimating utilization of some range grasaes in New 26:302-303. Mexico. U.S. For. Service. Rocky Mt. Forest and Range Expt. Papanastasis, V. P. 1977. Optimum size and shape of quadrat for· Sta. Res. Note RH-22,

( ( ( ) ) )

701

Staddart, L.A. 1935. Range capacity determination. Ecal. 16:531-533. Stoddart, L.A., Smith, A. o •• and T. II. Box. 1975. Range Manage­ ment. Third Ed. McGra .... Hill Book Ca .• New York. Strauss, D., and D. L. Neal. 1981. Biases in the step-point method on bunc~gr ... s ranges. J. Range Manage. (Accepted). Uresk, D. II., R. O. Gilbert, and II. H. Rick.ard. 1977. samplin& big sagebrush phytomass. J. Range Manage. 30:311-314. Valentina, K. A. 1946. Determining the grazing use of grasses by scaling. J. For. 44:528-530. Van Dyne, G. Joi., II. G. Vogel, and H. G. Fisser. 1963. Influence of small plot size and shape on range herbage production estimates. Ecol. 44:746-759. Whitman, W. C. and E. I. Siggeirssan. 1954. Comparison of line incerception and point concact methods in the analysis of mixed grass range vegetation. Ecol. 35:431-436. 774

all conceivable hypotheses, which represents a point of view that has a fundamental appeal to the legal profess­ ion. Katz (1975) also pointed out, however, that testi­ mony using data may be treated as "hearsay• evidence in some cases. Scherer (1979) discussed the pro's and con's for the use of statistics in governmental regulations. He Sampling and Statistical pointed out that while statistical information has value for regulatory purposes, it is also costly. A policy of Considerations in "all useful information• at any cost would be as unreal­ istic as regulations made arbitrarily, Thus, a balance Range Resource Inventories must be struck between benefits and costs, for without good statistical information regulations are likely to be Cilar/es D. Bonham ineffective. While the use of statistics in the legal process has not been extensively defined in range ecosystem manageme.n~ ).llSTRACT the role of statistics must be considered in the formula­ tion of decisions pertaining to allocation of rangeland Sampling techniques and statistical procedures are resources. Decisions relating to utilization of range becoming established as bases for adjudication in many resources must be documented statistically if management :ields. Soon such techniques and procedures will aid in decisions are to be defended in an obJective manner. In determining the adequacy of data used by the Bureau of time statistical data will be submitted as legal evidence ~and Management in formulating and implementing livestock in litigation processes involving public land management grazing plans. and livestock grazing permits. Sampling characteristics of the values obtained for To date, the single most effective use of statistics •:s,getation cover and production depend upon the method­ as legal evidence to justify management decisions is to ology used. Accuracy and precision, in particular, are minimize the emotionally subjective treatment of data. ,:fected by the methods employed for data collection. The litigation process by nature invites considerable ?ariations in time and space can be minL'llized by use of testimony from groups which use data to summarize their ~:;,;,ropriate stratification sampling techniques. Sample feelings. The proper use of statistics, including adequacy considerations are shown to affect the estimate probablistic statements, will serve to minimize the sub­ JC the mean as well as that of the variance. Thus, jectivity associated with emotional oresentations. ~robability levels are not always those assumed for Statistical observations will not answer all precision estimates. questions that may be asked about range resource inventory data. For example, vegetation data for cover and pro­ INTRODUCTION duction alone are not sufficient to deal with questions concerning the interrelations of plant species. At best, Statistical procedures are becoming increasingly data currently collected on rangeland vegetation charac­ important in the judicial proces3. As governmental regu­ teristics is a small portion of all rangeland resources. latory agencies require more quantitative data to make The best of statistical considerations will not provide decisions, statistical criteria for making objective answers to questions asked on uncollected data. decisions from data will be established. The Office of The role of statistics in rangeland resource inven­ Surface ~ining has already placed statistical require­ tories is to assist in collecting, analyzing, and inter­ ments on pre- and post-mining vegetation inventory data. preting data in the most efficient manner. Additionally, Statistics are also certain to play a role in bond inferences can be made from data although they represent release criteria for the surface mining industry. only a small pcrtion of the population. No other set of Statistics will eventually play an important role in techniques, other than statistics, allows these benefits the management of grazing lands. to be gained with probability levels. Katz (1975) discussed the presentation of a confi­ dence interval estimate as evidence in a legal proceeding, ACCURACY-PRECISION CONSIDERATIONS claiming that the interval estimate form of data is almost ideally suited for presentation before a judge, jury, and Measurement of rangeland ecosystem components such legal counsel. Specifically, the use of confidence as vegetation production and cover are often complicated interval estimates provides simultaneous evaluation of

773

( ( ( ) ) )

775 776

par~s is ~tra~ght forwar~. Assume that a rod tapering to because of accuracy and precision considerations. Since a fine point is lowered into the vegetation canopy from t, e accuracy concept involves the size of d':'viat~on'!' of 1 above, and any ~ontact of the point with vegetation is the value from the true mean of the population, it is 7ecorded as a hit. The proportion of hits of vegetation irnoerative to obtain a reliable measurement of the true ~s then an accurate estimation of plant cover relative to "'8&n of the population. For example, if vegetation cover h~re gr~w_id only if no human bias enters into hit and no- were the objective, then the method or procedure used to i~ dec~sions. This particular met~od, then, is one obtain cover measurements should result in a trUe esti­ w~ich gives ~o~.pre~ise and accurate estimates of cover, mation of the mean cover value ,for the area. sine':' epeatibility is high. All other techniques for In many cases, methods employed to obtain biological obta~ning7 cover data that involve visual estimates provide characteristics in rangeland resource inventories do not precise but not accurate estimates of cover. Certainly, ,neet an accuracy requirement. For example, the use of then, one would not use only the small deviations from cover classes which estimate cover to the nearest 101 for ~he sarn?le.mean to judge accuracy of the procedure which soecies estimates will have an accuracy of only 10\ in is statistically the definition of accuracy. This con­ each class interval. Yet, most plant species occurring clusion follo~s from the fact that a number of other ~ver a large area in the arid western United States will errors occur in the measurement orocess and their magni­ have an average cover value of less than 10\. Therefore, tudes vary according to observers, as well as technique :,ethodology must allow for measurement to within 11 level m Errors occ~rring in some measuremer.ts because of a· if cover by individual sgecies is important. ethod or technique can sometimes be compensated for More often than not, the concept of precision is the 7as~ly, while ~thers may remain uncorrected. For example one dealt with in methodology. Precision is defined as it .s muc~ easier to change ~~e property of a method, ' the size of deviation from the mean obtained by repeated such as s~ze.of quadrat, than to c~ange the personal anolication of the sampling procedure. Thus, a measure­ characteristics of an observer, which is technique. In ment can be very precise but not accurate. Yet, accuracy the.latter case, an observer using a visual technique to is the desired end product for a given data set that will estimate cover may tend to overestimate the cover of be used for making decisions on use of the rangeland bunchgrasses.while underestimating the cover of s· 1 - resources. stemmed species. ing e An examcle will illustrate the accuracy concept and One way of compensating for such errors is to train the failure to meet the objectives of obtaining accuracy th~ observer to make visual estimates on quadrats for in accuirinq cover data for plant species, Using a which cover values are obtained with a point method. The method presented by Daubenmire (1959), visual cover esti­ u~e ~f 100 points per quadrat yields estimates of cover mates were obtained and reported on the basis of cover within 11 levels.and ~hus trains the observer to visually classes of 0-51 m 1, 6-25% ~ 2, etc. It was suggested estimate.cover with higher accuracy levels. Another that various observers would not differ significantly in alternative would be to sample by a point method in, say, their estimates of cover when the midpoint of each of tha one o~t of 10 plots and use regression techniques (double classes was used in data· analysis, It should be obvious sam~ling) to adJust errors associated with visual that a large range in cover estimates could be obtained estimates. for Class 2, yet all values would be assigned the mid­ Other errors are associated with plot sizes used to point value of 15%, regardless of whether actual values obtain cover measurements. That is, the larger the plot, were 7 or 25%. Thus, precision in the sense of repeat­ the greater the tendency to lose track of cover estimates. ability is attained with respect to a mid90int, while Therefore, smal;er ~lots should be used to obtain esti­ accuracy remains undetermined for cover. mates of cover if visual estimation procedures are used To evaluate accuracy for cover methods, a more Plo~ sh~pe also can influence the accuracy of visual • objective way of determining cover must be employed. The estima~ions. Four-sided plots are generally more accurate Mante Carlo method is useful for illustrating why the thai_i circular plots for cover measurements because esti­ point method of determining cover provides accuracy, mating a ~roportion of a circular area (such as 1%) is l'hile it is empirical, the logic is inductive. Suppose more difficult than estimating a similar proportion of a that an area of irregular shape occurs within a square square or rectangular area. of known area and the former area needs to be determined. Production estimates obtained by clipping procedures Assume that it is possible to drop a point at random into are often.in error because a given "stubble height" · the square. That is, every region of the square has equal left uncl pped. This height varies with observer an~s probability of being hit by a point. Thus, the fraction compensation7 for such errors can be eliminated by clipping of points falling inside that region of irregular shape to ground level. provides a good estimate of its area relative to that of the square. The analogy to ground covered by vegetation 777 778

sample size, which is the number of observations made important to obtain a more objective estimation of varia­ for a particular measurement, often affects the precision tion in rangeland resource data, it "·ould be expe:lient to estimate obtained. Any randomly-observed measurement can use the best statistical methods for obtaining variance be assumed to be an unbiased estimate of a population estimates. Statistics would be useful to defend the value. Therefore, one random observation is an unbiased estimation method in litigation cases if they arise. estimate of the mean, but it may not be the best estimate The most popular statistical procedure in use today of the mean. on the other hand, large numbers of observa­ for analyzing data from rangeland resource inventories tions (i.e., several hundred) do not necessarily provide assumes a normal distribution of the data. The approxi­ a more precise estimate of the mean than, say, 25-30 mation of data variances based on normal distribution obsorvations, if all sampling is conducted on a random statistics in some cases may not yield the most precise basis. estimates of true variances. For example, the estimation Often the mean remains unchanged (relatively speak­ of variation in plant cover using the normal approxima­ ing) while the variance continues to change as the number tion may not be as accurate as, say, the use of binomial of observations are increased. The mean value is the or Poisson distributions. parameter most needed to make management decisions; the:re­ Confidence intervals as well as sample size adequacy fore, in order to compensate for the sample size needed, and other considerations certainly differ depending upon it may be necessary to increase the size of sampling units the statistical distributions assumed for the data. These and to obtain fewer observations, in contrast to using differences have not often been recogni~ed by personnel small units and obtaining more observations. These involved in inventory of rangeland resources. If cover decisions depend entirely upon the measurement needed. values are treated as normally distributed values, then the confidence interval for a mean total cover (x) of MAGNITUDE OF VARIATIONS 24.5% and a standard error of the mean (S-) of 3.44 would be x Variation Measurements CL 711) (3.44) - 24.5 < cover < 24.5 + (l. 711) (3.44) There are a n111t1ber of ways of obtaining an estimate for numerical variation in ecosystem components. The or most common way of expressing variation in plant cover or production is to take the average distance squared from 1.861

( ( ( ) ) ) 780 779

production from quadrat to quadrat is expected to be of the mean will be detected, the normal estimates of symmetrical about the m~an for quadrats. Therefore, it the mean and variance yield the following: can be concluded that normal distribution statistics can be used without concern to obtain approximations of vari~ ations in standing crop biomass. Caution should still be exercised in using the same procedures for obtaining estimates of vegetal cover. A confidence interval should be used fer variance, just as it is used for the mean. Then ar.y value for a where variance can be used with-a known probability of falling ~ • tabular value for confidence level desired: within a class of intervals for the population variances. 92 "variance obtained from a data set: To date, the estimation of variance by most inventory k • precision level of the population mean desired: and gatherers often ignores the confidence interval of the variance. Yet, for making management decisions concern­ X "average or mean of a data set. ing livestock carrying capacities of rangelands, it is important to calculate the upper and lower bounds of a Then, 2 2 variance to understand the magnitude of variation. For l.711) (17.2) 866 n • ..>,.::.;.;c.:.:~....:..:...:.~( ...... "-,- = 144 observations , example, the confidence interval on the variance of the 2 0 [(.10) (25.4)] cover example is 2 2 (n-1) s 2 (n-1) s , The binomial estimates of the medr. variance will < J < .;.:.;_;;.,....;:;... give the following approximate number of observations for x.025 x.975 an adequate sample: or values for a 90% interval are n = 2 2 (24) (17.2) 2 (24) (17.2) , < J 13. 8 < 36.4 where which is ? • proportion of ground covered by vegetation: 2 q • 1.0 - p" ground not covered by vegetation: and 514.5 < a < 195.l z and k were previously defined. The variance for cover then is itself a source of import­ ant variation. If the smaller bound for variance, 195.l, Then, 2 is used to determine sample adequacy, then obviously the (1. 711) (.245) (. 755) 0.541 number of observations required is considerably lower n " 2 O.OOOG •902 observations. [(.10) (.245)] than when any larger estimate of the variance is used. The bottomline for obtaining an adequate sample is to consider variation as a measureable component of any Because of the discrepancy in the number of observations vegetation attribute, instead of only the mean and its needed to obtain sample adequacy, the decision becomes variation. In the sample adequacy equation, assuming one of selecting a quadrat or a point method for obtain­ normality of cover data, the number of observations ing an estimate of total vegetation cover. becomes Herbage production may be distributed as a lognormal 2 distribution: however, a number of studies conducted on (l. 711) (514.5) 1506 n • & biomass of individual species occurring in grasslands has 2 • -r 251 , not substantiated this observation (Bonham unpublished). [ ( .10) ( 24. 5 l] Therefore, assumptions of normality currently used by the Bureau of Land Management to determine statistical vari­ if the variance is truly at the upper bound of the 901 ability for production is probably not in serious error. confidence level for the estimated variance. In any case, even if biomass for individual species should The space-time aspects of range ecosystem component not be a normally distributed variate, the summation of data are a complication in variations that cannot be biomass for individual species to obtain total production easily dealt with. That is, the topographic, edaphic and would be approximately normally distributed. Total 782 781

intervals and thus contribute less to the variation of a climatic conditions that exist at a particular sample site particular measurement. Productivity, then, has less determine the extent of variation that will be encountered opportunity to vary greatly than does the measurement of over a large area of a rangeland inventory. production. This is manifested in practice by· lower The most efficient way of dealing with these sources variation for small amounts of production than for larger of variation is to stratify the vegetation into smaller amounts. That is, small values have small variances subunits such that homogeneity for any attribute is as compared to large values. For management purposes, then, constant as possible within subunits. Even then,a one it appears that productivity would be the better measure­ space-time measurement will not be representative of all ment to determine livestock grazing carrying capacities sources of variations that actually ocdur in vegetation of rangelands. · characteristics over a long period of time. It follows Another important source of variation occurs in that justification cannot be made for a one-time production measurement, since vegetation production is inventory of rangeland resources for any purpose. In based on the amount of living material per unit volume that case, several measurements are needed, especially more often than not in rangeland resource inventories, if long-term management strategies are to be developed this measurement is treated as a two-dimensional sample and implemented. area rather than a sample of a volume. Therefore, a The magnitude of variations encountered in rangeland major source of variation in obtaining production ecosystem components certainly vary with a number of measurements can occur because clipping is often carried factors, including the particular component being measur­ out at ground level or near the surface instead of clip­ ed. Since the space allowed for this paper is limited, ping from the top of the vegetation toward the surface only vegetation biomass, cover and density parts of the and remaining within the vertical projection of the quad­ rangeland resources inventory will be treated from the rat. Plants rooted within the quadrat will receive more standpoint of sampling and statistical considerations. attention in clipping than others with canopy overhanging Because the principles discussed here apply as well to the interior of the quadrat if only ground area is used. many other attributes of the range ecosystem, discussion In general, the practice of clipping individuals rooted of these attributes alone illustrates the points that within the quadrat has been followed. need to be made. Disappearance of the herbage throughout the growing season, whether it be by herbivores, certain plants or Vegetation Production parts (such as leaves, etc.), contributes to the varia­ tion of actual production. Thus, the amount of oroduc­ Because vegetation production (i.e., weight of herb­ tion available at the time of measurement mav not be an age produced) has been so commonly used to determine accurate estimate of the amount of herbage a~tually avail­ carrying capacity of rangelands, it has been an impor­ able to all herbivores, much less livestock. Moreover, tant attributemeasured in rangeland inventories. Pro­ precision probably cannot be controlled under grazing by duction has variationc that depend upon species composi­ herbivores since repeatable observations are not econom­ tion, spatial patterns of precipitation in both amounts ically feasible for the sampling time required. and times over the growing season, length of growing Costs of data collection vary with respect to con­ season, past use history, and method of sampling used to fidence and precision levels. For example, one-time-only obtain the production estimate. estimates of production using a 90\ confidence level and All of these sources of variation, and perhaps other~ a 101 level of precision for the mean indicate that 0.025 are integrated into the single value that is obtained 01,er man days per acre are required for grassland range sites the growinq 3eason. Productivity {i.e., rate of produc­ and 0.043 man days per acre for shrub-grassland ranges in tion l , tho1,.·.,-~ often confused with production, is a southwestern Wyoming and northwestern New Mexico (Bonham distinct at~ribute of vegetation that is not influenced 1980). A confidence level of BO and 201 precision re­ by all of the sources of variation that affect production. quired 0.009 and 0.018 man days for grassland and shrubs, If standing crop productivity is measured at four differ­ respectively, for grassland sampling. These costs are 2.4 ent times throughout the growing season, then variations times more for shrublands studied. due to the relatively shorter interval of time are includ­ For management purposes, the 801 confidence level ed in each of the four measurements. This is in contrast with a 101 precision should be used unless reasons other to the time differentials that occur in the total combined than economics justify a change. These values are based measurement taken as production. ?n obtaini~g information from O.Sxl.0-m quadrats placed The sources of variation that could affect produc­ in vegetation types ranging from 1000-1300 acres in size _tivity measurements also can be minimized by taking for shrublands and 1700-2100 acres for grasslands. Of shorter intervals of time between each measurement. course, travel time varies between observation points Therefore, weather effects are encountered in shorter

( ( ( )

783 784

In ~he strictest sense, veg.etation cover, like since these coints were randomly located. Other factors production, can change on an hourly basis and certainly such as topography, whether or not the range site ('or changes noticeably over a few days• time interval. veoetation type) is found to occur in a contiguous Theretore, variation in plant cover measurements manner or in several separated units, and size of sampling occurs because of the season or time interval in units all affect the costs (Bonham uncublished data). which the measurements are obtained. Phenological Variations in production are often due in large part stages of snecies vary in combinations to contribute to the cotential of the site. This ootential is based to total veget.ation cover. When one species is ini tiat­ on 1011; climate, carent material, slope, as well as ing growth, another may be maturing, and cover of historical perturbations such as fire:etc. Other sources individual soecies is thus affected. Then, cover by of variation such as the seasonal distribution of plant 9lant s9ecies would be comparable only if all measure­ hiomass is encountered in measuring rangeland vegetation ment were obtained at the same 9henological stage each production, That is, whether a 9lant is a cool-season year. Seldom is this ever accomplished for rangeland or a warm-season variety determines the amount of resource inventories~ ~aterial produced during a given time segment of the . Cover estimates are then hiqhly dependent upon the growing season. time of the growing season in which sampling is conduct­ >lor;,hological characteristics of 9lants (e.g., sod­ ed: These estimates may be accurately obtained by a forming versus bunchgrass ty9e) affect amount of plant ~oint method that is objectively located during the time croduction within a qiven ceriod. Therefore, time of interval selected. The point (actually a very samll samoling will influence the outcome of estimated pro­ quadrat) method is, in fact, the only unbiased method duction for an area. Another source of variation in of obtaining total ground covered by vegetation. All oroduction measurement i.s that found in olant death and other methods have biases associated with them. For decomoosition crocesses whicb vary among·· species and example, the ste9-9oint method used by the Bureau of occur continuously throughout the growing season. 'Land Management to obtain cover of vegetation is Therefore, any estimate of herbage production must biased by the observer because line-of-sight depends acknow~dge that these sources of variation do, in fact, on vision 9arallax and interception of the "toe notch" exiJlt and where oossible these variations should be •,ith the plant part. Photographic methods also are minimized for pro9er inter~retation of the data. affected by the parallax problem and become economi­ TO obtain minimum variance estimates of production cally infeasible if corrective procedures are employed when life forms vary within a range site, production and sam9ling adequacy is met. must be obtained during the time when the life form is . To reduce variation caused by methodology and at 9eak standing crop for the growing season. Sampling incre~se both accuracy and precision of cover estimates, estimates are obtained for cool-season grasses and forbs sam9ling must be repeated for time intervals wherein durinq late spring or early s~er and again during changes in dominant species are at a minimum. These early fall if growth is reinitiated. Warm-season species, time intervals will occur when the morphological on the other hand, must be sam9led at their peak time development of the dominant specie(s) of cool-season which differs by species. An octimum trade-off of the and warm-season ?lants have reached their maximum latter would be to clip all warm-season pla,nt growth cover, respectively. Further, these two time intervals when the major or dominant warl'l-season species are at will significantly ex9lain seasonal cover variation in peak production. most rangeland vegetation. Vegetation Cover ST!\ATIFICATION AND VA.qIATION Vegetation cover is a measure of the amount of Stratification methods can be used to isolate sources ground surface that is occupied by living ?lant material. of variation in range resource inventory measurements. This measurement is obtained mostly from subjective One way of accomplishing this is to map the inventory methods, and as a result the magnitude of variations en­ area according to some criteria such as vegetation countered can be very large. Cover estimates are more types, using d':""inant plant species to describe a type. subject to error associated with individual observers More COl!ll!lonly in rangeland resource inventories the than are production measurements. !-lhile there are range site method is used to describe vegetatic~ and its objective ways of obtaining cover measurements, objectiv­ response to soil characteristics. Soil characteristics ity is often sacrificed because of the costs involved. predo~inate the range site mapping as indicated by the Furthermore, vegetation cover is affected greatly by name attached to the site. In either case the seasonality of the measurement. stratification produces homogeneity of veg;tation cover 785 786

by species and/or production of these species within a estimates from square-meter quadrats should not be given range site, The range site approach to stratifi­ recorded.to the nearest 0.1 g if decisions are made in cation often leads to more variation in vegetation terms of kilograms/ha. The cost of obtaining very response than the vegetation mapping approach. This orecise data may be too high to justify in light of difference occurs mainly because range sites are often ~nagement goals. Because measurement to the nearest combined to include a mapping unit which is a complex 0.1 g is impossible to accomplish with present methods of sites representing a number of different exposures of cliooing and estimation orocedures, the nearest lg and slooe characteristics. The complex mapping unit or perhaps the nearest 5 g record is more appropriate usually occurs because of varying soil types between for determination of livestock carrying capacities. small ridges and valleys of a watershed. For example, The magnitude of variations in the data set·is shallow clay and shallow sand range sites of south­ difficult to interpret since vegetation-environmental western ~yoming are often combined into shallow clay­ data are multivariate. Decisions need to be made on shallow sand sites (actually a complex of two range a multivariate data set, not on a univariate variable sites) since either alone is too small to map individu­ such as total production of forage per unit area, ally as a continguous unit. Therefore, variation is though this total is actually made up of individual larger for vegetation characteristics in the complex species which together constitute a multivariate system. mapping unit than within ma~ping units including only If cost and/or variations are to be minimized in sam­ a single range site. pling, the data must be considered in multivariate Another stratification approach makes use of an fashion. This is not to suggest that each individual overlay technique. That is, soils and vegetation are scecies has to be sampled separately in an optimum ~apped and these are combined by overlay procedures way, but rather that sampling methodology sho~ld be onto topographic maps or maps of other environmental ootimal for obtaining measurements on all variables characteristics such as precipitation and elevation. simultaneously. If double sampling procedures are used A combination of such maps will yield several units to obtain weight estimates for individual species, the that can be uniquely defined in terms of vegetation number of plots clipped in ratio to those estimated response characteristics. These individual map visually must be optimized (Ahmed and Bonham 1980). components which are produced from overlays have been referred to as ecological response units. Ecological ~UALITY AND QUANTITY resoonse units were used to describe 70,000 acres of land managed by the Bureau of Land !lanagement in western The quality of the data collected should never be Colorado (Bonham 1975). Variations between such units compromised due to small sam9le sizes. On the other are minimized since major sources of variation are hand, methods which yield large quantities of data esentially homogeneous within the response unit. because they are rapid to employ are not necessarily Any attempt to stratify rangeland resource variation the most reliable in obtaining estimates of Means and should first consider management unit size. If an area variances. From the statistical viewpoint, a sample of less than 1,000 acres is relatively insignificant size of 25-30 could be considered large, while froM the from the management standpoint, then the minimum size ecoloqical vie~oint this size of sample may be small. for any mapging unit should be at that scale or larger. Although 25 samples at random may provide statistical Any unit which is less than 1,000 acres in size should adequacy, they may not cover all species-habitat not be considered separately as an individual unit and relations occurring within the type. should be combined into the type that best describes Statistics should never dictate principles of the unit with respect to vegetation. Furthermore, assessing significance of ecological data (i.e., consideration must be given to the various management biological significance), but they can be helpful practices that may be applied to the vegetation units in determining when population characteristics have before a stratification approach is chosen. Hanagement been adequately described. For example, a systematic practices that aim to improve water quality, reduce sampling technique which includes one observation per erosion of the soil surface, or increase recreational 40 acres of vegetation ty?e would possibly give an values will each have a different approach to stratifi­ excellent description of vegetation characteristics cation of vegetation. on a large range area, but may not be adequately The quantity and quality of range resource component sampled from a statistical viewpoint. In fact, data vary relative to methodologies used, as previously several hundred randomly-placed quadrats may be needed discussed. Absolute precision criteria for the data in to describe most range site vegetation cover and the inventory is set in each case according to the production characteristics with the 80 and 10\ levels management use of the data. For example, production previously mentioned.

( (' ( ) ) ) 787

S!JH"IARY ANO CONCLUSIONS

Rangeland inventories must be based on statistical considerations such as random representation in order to produce data for valid decision making. Large data sets do not necessarily make the estimates more accu­ rate or more precise, and trust in statistical pro­ cedures must be exercised so that probablistic state­ ments can be made about the data set. Data not meet­ ingstatistical criteria produce subjective decisions. In conclusion, the rangeland resource manager needs to make decisions based on statistical criteria. Only statistically sound inventory procedures can provide the security needed to defend decisions. LITERATURE CITED Ahmed, J., and C. O. Bonham. 1980. OUBSAM. Algorithm and computer program for optimum allocations in multivariate double sampling for biomass estimation. Fort Collins, Colorado: Colo. State Univ. Range Sci. Series No. 33. Bonham, C. O. 1975. ~anagement decision making for BLM natural resources. Final Report. BLM Project No. 52500-CT3-l9. Bonham, C. D. 1980. Unpublished data from several vegetation types studied in southwest Wyoming and northwest New Mexico. Daubenmire, ~- 1959, A canopy-coverage method of vegetation analysis. Northwest Sci. 33:43-46. Katz, L. 1975. Presentation of a confidence interval estimate as evidence in a legal proceeding. The Amer. Stat. 29:138-142. Scherer, F. M. 1979. Statistics for governmental regulation. The Amer. Stat. 33:1-5. 790

half to one-fifth as much variation. wich cv~, ranging from 0.10 to 1.48 and an average of about 0.4.

Carrying capacity estimates based on the product of an inventoried forage weight and five adjustment fa~tors "Sampling and Statistical have very high coefficients of variation, probably in exces5 of 2.5. However, grazing capacity and variance Considerations in estimates provide the range manager useful information for better range ~anagement. More emphasis needs to be Range Resource Inventories": placed on variance estimation in range inventori~s. Comment and Discussion INTRODUCTION John W. Menke and Michael F. Miller Historically, statistical methods have played a minor role in range inventory and analysis procedures. >.BS TRACT Designers of sampling plans for past range inventory methods often were a~are of accuracy, precision, and Sampling and statistical considerations play an efficiency criteria that could be applied, but they increasingly important role in range inventories given found Little need to formally present thesa performance greater ~ultiple-use interest group ptessures. PuOlic criteria as part of the justification for, or confidence land management agencies especially need detailed and in, the inventory results. reliable inventories ~f supplies and demands for range­ land resources. Numerous alternative statistics and Recently, and speci:ically because of multiple-use sampling plans exist, yet only the most rneaningful and interesc-group pressures and law suits, laws and acts ~fficient methods can b~ used today because of cost have been passed by the United States Congress stipulat­ litnit.ttions. ing that public land management agencies will conduct detailed inventories of supplies and demands for range­ Or. Bonham;s pap~r vas reviewed and commented upon. land and forestland resuurces. If the Nation"g rangeland T~o voids were found: t) discussion of rangeland resources are going to be partitioned into a greater sampling plans, and 2) estimation of the magnitude of number of portions for a greater numb~r of consumers. variation exp@cted in components of rangeland ecosystems. clearly we need to ~now, more accurately and more These voids are filled with discussion and evaluation of precisely, potential supplies of resources and their nine sampling plans. Lit~rature was partially searched temporal and spatial distribution under alternative to estimate ecosysteffl component variation. sustained-yield managefflent strategies.

The magnitude of variation expected in components Efforts of land management agencies co inventory of the rangeland ecosystem is great. Coefficients of rangelands as required by law have not been adequately variation for important inventoried variables of renge­ funded by Congress. Statistical methods that result in land such as plant species cover, frequency and density, greater inventory efficiency are critically needed. and_ deer 1iet composition are usually large and range Likewise, given the influx of young ~nd inexperienced from O.OJ to J.19 with an average of about t.3~ yield, staff and the added burden of more complex inventories, biomass, or standing crop values ty?ically have one- only elearly documented and efficient methodology resulting in precise inventory data can be used. With new laws and aces concerning our rangeland resources, it appears the demands placed on these inventories will John W. Menke 11 A•sociate Professor of range ecology, increase. Two brief examples will suffice. First. Department of Agronomy and Range Science, University confrontations between ranchers and the U.S. Department of California, Davia, California 9S616. of Interior. Bureau of Land Management (BLM) over permit reduction• in livestock grazing have led rancher, to .Michael r. Hiller i• Statistician, Diviaian af St•tis• hire private consultant range s~ecialists to gather their own inventory data. These data will certainly be tic~. University of California, Davi1 1 C•lifornia 9S6t6. uaed in court proceedings. Hence, the statistical reliability of d~ta presented by the agency and the 789

( ( ( ) ) ) 791 792

reflect the strong empha9iS of current research and developm~nt programs designed to build multiresource rancher will likely affect which data are uaed to decide c~e outcome. If such activity proliferates, minimum sampling procedures. Planning models that use multi­ v&riate scatistics have been developed (Colvell and standard• for rangeland inventory may be eatablished ad Titus 197&) co allocate samples at v•rious scages in ~oc. multistage sampling using remote sensing, buc little or no use has been made of such procedures on the ground at Second, gi•en the clearcut consu~ptive-use nature the range plant community level. of minin1 the western rangelands for coal, oil shale and hard rock ainerals, and the economic value of the tesources extracted, d~cailed inventory procedures In an earlier book prepared by the Joint Committee of the American Society of Range Management and the ire and will be required to regulate the rehabilitation Agricultural Board (1962) many of these same topic, were ~f these disturbed ~angelands, Research and development discussed in great detail. However, that discussion ·">f better methodologies, funded by private industry, may pay dividends to extensive rangeland inventory method focused more on range research than range management. develop~enc. However, such benefits will not solve the 1 targer extensive rangeland inventory problem because the Adding to Dr. Bonham s discussion of quadrat size ~ere size of the latter problem poses great difficulties. and shape, the most obvious effect of size and shape of quad~at is on frequency data. Presence or absence information is clearly quadrat size dependent, but it is ROLE OF STATISTICAL AND QUANTITATIVE METHODS IN RANGE also quadrat shape dep~ndent. For these and other INVENTORY reasons the value. of fr~quency for regional range inventories is reQuced ~ecause data gathered using different quadrat sizes are not comparable, yet the Dr. B~~~~.LU~'iP.Ai.--9d. an. excellent papeT on samplin(I vegetation patterns and species abundance differences aad stacist_i._~a.J considerations in range inventories. For empnasis, we wilt comment on major points made in his among areas inventoried requires that the quadrat size paper and then fill some gaps he left open. be different. Frequ~ncy data are, however, relatively easy to gather and important for range trend documenta­ tion; as long as a trend plot is measured with the same We agree I that from a legal evidence viewpoint I the most effective use of statistics 1n range management in size and sha~e of quadrat at each monitoring period, it the past has been to minimize the emotionally subjective does not ~atter that different quadrat sizes and shapes tr@atment of data. More importantly. he points out, are used for different trend plots. statistics no~ have a larger role in collecting, aaalyz­ ing, as well as interpreting range resource inventory Considerable effort was devoted to the decision on data. the selection ot quadrat sizes and shapes for uae in the BLH Soil-Vegetation Inventory Method (BLH-SVIM 1979). Or. Bonham ~ade an important distinction between Circular plots of various sizes, dependent on vegetation statu3, as chosen by BLM is an optimal choice. aethod and technique effect on measurement error. too 1 often these words are used synonymously in range sampl­ ing. Form.al methods, such as double sampling, are designed co compensate for most of che human-iaposed Rangeland Sampling Plans (technique) error in biomass sampling, but unpredictable errors due co technique still occu~. Training in It is at the stage of designing the sampling plan sampling technique is often inadequate, probably because o~ strategy in the range inventory process that statisti­ of the areat enphasis placed on more rigorously defined cal considerations play a major role. A list of possible aethod aspects of the process. plans serve• to boun~ the options in our concept of alte1"native,:

Dr. Bonham de1cribed 1 rather thoroughly, the source, of variation and 1tep1 to be taken to reduce 1. Simple random sampling saapling error in estiaating vegetation production and 2. Cluster sampling cover. He did not attempt to eatimate the magnitude of 3. Stratified random saapling with allocation variation expected in various rangeland components which proportional to area ve will att~mpt later in this paper. 4. Stratified random sampling will allocation proportional to productivity Dr. Bonham'• di1cu,1ion of sampling plan9 and 5. Stratified random sampling with optimal alloca- allocation of 1ample1 to strata was brief and does not 793 794

tion homogeneous strata. 6. Systematic tampling 7. Multiphase sampling Syste~acic aod multiphase sampling plans are 8. Multistage sampling practiced as part of simple random, cluster, or scrati­ 9. Multiresource sampling fied random sampling plans, Both are used primarily co increase efficiency and precision of sampling. For Space and time do noc permit a detailed discussion example, when the first sample point i~ established for of each of thesa sampling plans, but some discussion of a transect to traverse a site ~rite-up area (SWA) in the each is needed so that recommendations for improvement BLM-SVI~ (1979) procedure, the sampling plan chosen is a in proposed inventory plan, (BLM-SVlM 1979) can be pseudo~systematic one of cluster sampling. Greater ~ade. Sampling plans l, 2, 3~ 5, ~. and 7 were thorough­ efficiency is achieved by setting the predetermined ly discussed by the Joint Committee of the America« course for the sampler and less time is spent in travel Society of Range M~naement and the Agricultural Board and locating plots. Simple random sampling, cluster (1962). sa~pling~ and various stratified random sampling plans eaeh have different variance formulas; and too often the The order of sampling plans l through 9 is useful simple random sa•pling formula is used inappropriately co help the novice ,e~ the interrelationships among and for the other plans. cha evolution of sampling plans. Simple random sampling statistical principles are the basis for all plans Hultiphase sampling includes a well developed set and its random feature assures that each plan meats a of techniques which can and should be used more often to minimum set of required statistical assumptions concern­ improve. sampling efficiency. In ~ultiphase sampling, ing inventory variable distribution, and that all successive measurement procedures of different rela­ individuals in the !tacistical populations ha~e an equal tive costs and difficulty are used together to maximi:e likelihood of being sampled. sa~pling precision for a fixed cost. Double sampling is the most well developed multiphase sampling procedure. Cluster sampling is the simplest ~escricted random It may take three forms, double sampling with strati­ sampling plan where simple random sampling is perfor~ed fication, with ratio estimation, and with regre,sion within a relatively small or compact representative area (~yers and Shelton 1980). Stratifying a rangeland on ~ithin a cluster or subinventory area. Short transeet­ the basis of some measureable variable using aerial based sampling is a form of cluster sampling. photographs~ and then subsampling some of those sample blocks using a ground survey, is a common example of Stratified random sampling is the logical extension multiphase double samplin~. Where a simple ratio ~f cluster sampling and can take a number of forms. relationship is establish~d between a gtound-measured ~nowledge of ~omogeneous subinventory areas, groups, variable and the same aerial-photograph measured vari­ range types or conditions is used to stratify the able, a ratio estimator is e3tablished to correct che sample, reduce the ·variation within straca, and achieve larger and less expensive sample taken from the aeTial greater accuracy and sampling efficiency. If samples photographs, and one is then using double sampling with 3te taken at random without regard to stratum boundaries, ratio esti~ation. Finally, double $ampling with regres­ samplina intensity in each stratum will on the average, sion uses a small sample to establish a relationship ~e proportional to area of a stratum. Under this between characteristics that are difficult or costly and procedure stratified random sampling (SRS) is referred other characteristics that are easier to measure or less to as SRS with proportional allocation. When a range­ exp~nsive. Relationships from the s~all sample are then land is classified into production strata, ~uch greater used co make predictions based upon the.larger, and precision in carrying capacity estimations can be easier-co-measure, or less expensive sample. achieved more efficiencly by allocating a larger number of •••ples to the •ore productive strata (for example, As an example, during a pilot or training phase, see Reppert et al. 1962), chat is, Sas with allocation ocular forage weight estimates and clipped weights are proportional to productivity, Finally, SKS with optimal both made on a series of randomly located plots within a allocation is the sampling plan where, again, dispropor­ stratum. A linear regression is calculated between the tionate allocation of samples allows minimi:ation of estimated and clipped variable weights. Using the time variance and maximization of precision, but for a fixed budget required to ocularly estimate weight, and clip­ cost. Optimal allocation plans have a substantial and-weigh, the sampler can c. lculate and then use the advantage Qver proportional allocation plans when the optimum ratio of plots to e~-~~ate and clip, verses rangeland is rather heterogeneous bQC can be mapped inta those plots to just estimate f~ ,ge weight. Continual

( ( ( ) ) ) 795 796 tuning of the regression coefficients and the optimum !cosvstem Component Articles ratio as the inventory proceeds ensures adequate sampler training and appropriate changes in c?effi~ient~ and Yield, biomass, or standing crop 7 optimum ratios for particular vegetation 11tuac1ons, Plant species cover 3 Plant species frequency 2 finally, at che frontier of range inventory sampl­ Plant species density l ing ties two sa•pling plan~ chat m~y, ~nd hopefully Livestock forage intake 2 vitl revolutionize range 1nvencor1et 1n the future. Deer diat composition 2 Mutcistage and multire•ou~ce sampling! in concep~, ~ake Liccer weight 2 uae of all previously mentioned sampling plans with two Phenophase l additional dimensions, respectively, Multistage sampl­ 11 11 Precipitation l ing, in its simplest sense, is a multidouble sampling plan using remote sensing imagery at fron one t? three levels (scale1) in addition to the ground sampling The criteria for inclusion were twofold: ( l) the information, to proportionally or optimally allocate range ecosystem component variable studied has direct samples at up to four levels and multiple strata at aach relevance to rangeland inventory, and (2) enough statis­ Level. Saaple planning models have been developed tical information was provided to calculate a coefficient (Colwell and Titus 1976) and implemented in a few of variation, the ratio of the standard deviation to the instances such as the BLM California Desert Plan. sample mean. Coefficients of variation are extensively ~ultistage saapling has often been over-promoted, used when one wishes to cofflpare the variation of two or howaver, and as stated ~Y Aldrich (1979: page 15) the mo~e populations independent of the magnitude of their future benefits may be considerably less than expected: means and regardless what the population is.

The mensurational aspects of resources, other than Largest coefficients of variation (note averages in forests, are generally restricted co determining parentheses) were found for plant species cover (1.3), areal extent Gf delineated soils. landform. and frequency (0 .9), densit:, ( 1.4), and deer diet composi­ plant communities. The scale of aerial photographs tion (l.8). Values for these attributes ranged from and topographic relief have a great effect on the 0,03 to 3.19, but averaged about 1.3. Yield, biomass, accuracy of these measurements. or standing crop (0.4) had one-half to one-fifth as much variation as cover, frequency, and density measu~es. Finally, Multiresource sampling is just what the They ranged from 0.10 to 1.48 and averaged about 0.4. word implies, all of the above and for two or more For the few studies compared, there did not seem co be a resources: livestock grazing, wildlife, timber, water, great difference between coefficients of variation for recreacian, ~ineral and archaeological resourc~s. cover (1.3) 3nd ~requen~y (0.9) measures. However, the Ce~tainly this sampling plan is still on the distant frequency data presented generally had lower variances. horizon. Ory matter Livestock forage intake was found co be MAGNITUDE OF VARIATION EXPECTED IN COMPONENTS OF highly variable within and among range types. Clearly more of this variation should be accounted for in range RANGELAND ECOSYSTEMS inventory and forage allocation procedures. Two relatively detailed articles on deer diets showed thac Eatimation of the magnitude of variation expected selective wildife species ar~ much more variable in their in components of rangeland ecosystems is·.a fo~mid~b!e diet selection than sheep or cattle. Synthesis of task, but this variation must be known sLnce 1c l1mtts wildlife use data co estimate proper use factors will the ac~uracy and precision vith which range ~anagers be difficult, but first approximations will serve to can estimate important parameters needed to.regulate t~e promote additional syntheses and critically needed use of rangeland resources. To devalop a first approxi­ mis1ion-oriented research. mation all issues of th~ Journal of Range Management for 1978 1 1979, and the fir$C five issues ~f 1980, ver~ aearched for range ecosystem component variance data, Finally, variability in phenophase and precipita­ tion data, so important to calculate BLM-SV!M (1979) The re•ults are summarized in Table l and include nine adjustment factors, also were found to be highly vari­ eco1yste• components. The nu~ber of articles found for able. Two consolations were found here, however; each is •.• follows: variation in phenology and cumulative precipitation generally appears co be less than that for forage 797 798

Table !. Mean•, coefficients of variation and sample sizes for nine Meadow herbac eoua biomass (g/m2) Reese et al. ratge ecosystem comf)Onencs. Clipping 296 0.87 8 (1980) Subalpine Component and description Mean CV n Source Double samp 1 ing 296 0.51 8 Utah Dry ~eight estimation 296 0,58 8 Yield, bio... s, or standing crop Plant------1pecies cover !lean annual yield (lb/acre) Rauzi (1978) We1tern wheacgra11 73 0.41 5 Shortgrasa Plant cover (%) Uresk (1978) Blue gram, 342 0.10 5 Wyoming Sagebrush conlllunicy Sagebru_sh- luffalogrua 139 0.29 5 Sandberg bluegrass l. 7 0.83 8 bit cerbrush Dryl.md sedgaa 69 0.28 5 Cheat grass 15.3 0.17 8 grass Forbs 18 0.80 5 Fescue 4.0 0.49 8 Washin1,con Total herbage 646 o. 17 5 Turpentine cymopcerua 0.3 1. 89 8 Hoary aster 0.2 2.83 8 Standing phytomass (kg/ha) McGincy et al. Matted cryptantha 1.8 0.31 8 lle•vy continuous 1270 0.40 14 (1979) Russian chi st le 0 .1 2.83 8 4-pasture 2257 0. 55 14 Midgrass Big sagebrush 33 .4 0, 14 8 Exe lo sure 1907 0.43 14 Texas Total phytomass (kg/ha) Bitterbrush conmunity Heavy continuous 2458 0.35 14 Sandberg bluegrass 5. 2 0. 77 4 4-pasture 50!6 0.50 14 Cheat grass 34. 9 0.05 4 !xclosure 4939 o. 53 14 ~acted cryptantha 0.4 o. 50 4 2 4 (g/m) Wing-nut cryptantha 0.2 o. 50 Cheatgrass biomass by season Uresk et al. 4 1971 (1979) Tansy mus tat"d 0.2 1.00 4 March 8 36 0.26 10 Go-back Palouse Buckwheat 2.7 1.00 9.0 0. 33 4 April l 60 0.30 10 prairie Big sagebrush 29.5 0.16 4 19 127 0.19 10 Uashington Bitterbrush 0. 29 4 .May 7 198 0 .19 10 Rabbit brush 3.S 28 125 0.28 10 Merri 11 et al. Nov. 10 48 0.39 10 Canopy cover (%) Bluebunch wheatgrass 6 0,67 20 (1980) 9 1.33 20 Ponderosa pine 1972 Cheat grass l 1.00 20 under story !larch L4 56 0.23 10 Idaho fescue LOO 20 Idaho April 3 76 o. ll 10 Yarrow 2 14 0.43 20 13 132 o. 16 10 Arrowleaf balsamroot 25 126 0.14 10 Sagebrush canopy cover (%) 68 0.20 250 Neuenschwander May 15 112 0.57 10 (1980) Sagebrush-grass S•gebrush biorn.ass L580 1.48 250 Neuenschwandar Idaho (g/plant) (1980) Sagebruah-grasa ------Idaho Plant species frequency (%) Uresk (1978) Herbago yield (kg/ha) 1375 o. 56 54 Pt1111ph rey ( 1980) Frequency of occurrence (saedad species) Pine woodland Sagebrush coamunity Sagebrush- 0.55 8 bitterbrush Oregon Sandberg bluegrass 21.5 94.0 0.05 8 grass 2 Cheat grass Pine gr••• yield (g/m ) 27 0.28 4 Stout et al. Fescue 75 .0 0.23 8 Washington ( 1980) Turpentine cymopterus 0.8 1.41 8 Douglas fir and Hoary aster 3.3 2.83 8 pinegrass Matted cryptantha 1.0 2.83 8 aricish Columbia Russian thistle 4.5 1.32 8 'Big sagebrush 36,0 0.31 8

( ( ( ) ) ) 799 900

litterbruah coaaunity Percent of diet Arnold and Drawe Sandberg bluegrass 36.S o. 38 4 Cactus 42.5 0.59 8 (1979) Cheat gr as• 98.0 0.03 4 Bros,se 36.5 0. 51 8 Coastal ll&tted cryptantha 15.0 0.59 4 Forbs 11.4 1.47 8 grassland T.an•y euscard. 4.0 1.40 4 Grasses 1.4 1.45 8 Te,ras Buckwheat 16.0 O.S9 4 Big sagebrush 5.5 0.47 4 8itterbruah 23.0 0.32 4 Litter weight

Frequency ( %) Merri 11 et al. Litter phytomass (kg/ha) McCinty et al. Slubunch wheatgras• 8 0.38 20 (1980) Heavy continuous 1188 0.50 l4 (1979) Cheat grass 10 0.80 20 Poaderosa pine 4-p.ast ure 27 S8 0.58 L4 Midgrass ldaho fesctJe l 2.00 20 under story Exe lo sure 3031 0.77 14 Tex.as Yarrow 5 0.40 20 ldllho Arrowleaf balsam.root 9 0.44 20 Litter cover (%) Mer-rill et al. ------~------1974 42 0.45 20 (l 980) 1976 41 0.32 20 Pondero1a pine Plant species density understot'y Idaho Sagebrush density 11,907 1.42 250 Neuenachwander (1980) (plants/ha) Phenophass S.D. (d•I• l .!!. Sagebruah-grass ~ !daho Mean phenophase date West and Gasco ~------~-~- Twigs elongating (1978) Shad scale w April 14 7 Desert shrub Livestock fora~e intake Winter fat 29 April LB 7 Utah Flowers opening Range of intake Shad scale II June 20 7 Sheep Utah Desert range 2.2-3.4 Lb/100 lb Cordova et al. Winterfac 7 June 15 7 C•lif. Dry annual range 1.7-2.Z lb { 1978) Fruit disseminating Steers Nevada Sagebrush-grass 8.4-10.4 lb R.eviaw article Shsdscale 1 Sept. 21 7 Nevada Desert shrub S.0-9.0 lb Winterfat 27 August 23 7 Calif. Dry annual range 10.6-13.l lb Summer dormancy Nebraska Sandhill range 1.8-2.5 Lb/100 lb Shad scale ll Sept. 36 7 Oregon Seeded ~heacgras, 11.9-15.9 lb llfinte-r fat 10 Sepe. 31 7 Cows Colorado Meadow 20.9-26.0 lb ------Oregon Seeded wheatgrass 18.1-39.5 lb Preci pit at ion (..,) Mean CV n Cattle intaka (Lb/day) Ro1iere et al. June S-11 Reifers 15.i 0.4J 4 {1980) September 34 0.62 7 Pumphrey (1980) Cows 24.3 0.58 8 Blue gra•a range October 40 0.60 7 Pine woodland August 6-13 Heifers 13.2 0.20 4 Nev Mexico November 48 0.43 7 Oregon Cows 16.l 0.21 8 January 71 0.45 7 .~~~~-~-----~------~------~------~ March 60 0.46. 7 Kay 49 0.32 7 Deer diet composition Sepc.-Occ. 74 0.53 7 lite• per day C«rpenter et al. Sept.-Nov. 122 0.34 7 lluebunch wheatgrass 1049 0.94 30 ( 1979) . April-May 99 0.37 7 Wyoming big sagebrush 311 1.59 30 Mountain brush- April-June 150 0.32 7 Veacern wheacgra1s 78 l.81 30 1rau s~pt .-June 540 0.21 7 Serviceberry 47 2.63 30 Colorado Creen rabbitbruah 47 3.19 JO llue ~aaa 39 2.72 30 Fringed oage 26 2.38 30 801 802 paraaetera. And second, phenophase error, compound Table 2. An example of SVIM data calculations. (From Artz 1980). thea1elve1 •• the season progresses. ,o it is good that phenoloay correction factor• are of lower magnitude as Bluebunch Big the ••••on progresaes. Step vheatgrass Cheatgras1 sagebrush Total I. Production Some additional effort should be directed toward A. Crams per transect 25 15 35 75 chis cype of analysis. A model is generally only•• (convert plot to precise•• it, molt imprecise component, and this type lb/acre) of relative error analysis clearly demonstrates the iapreci1e 1 or evea aissing coaponencs, in existing B. Green weight lb/ acre 250 150 350 750 aodel1 and procedures. C. Weighted average X X X utilization factor l. 5 l. 20 1.00 !STlHATE OF COMPOUNDED ERRORS--SV!M DATA EXAMPLE D. Total production Table 2 illu1trates hov adjustment factor• are u1ed at inventory 375 in sequential mathematical calculations to determine adjusted air-dry forage production and rangeland carry­ E. Phenology adjust- X X X inc capacity in the BLM-SVIM range inventory method. ment factor 1.10 1.00 l.20 W•ighcs of green forage per plot clipped from the wei&ht e1tiaate plots along che 200-poinc transect within a ., . Total green weight site-writeup-area (SWA) are converted to pounds per acre prod. for season 412 180 420 by aultiplying by a constant. Thia value is then 1equentially multiplied by the weighted average utiliza­ G. % air-dry weight X X X tion factor to e1timace total production at the time of .70 .90 .50 inventory; multiplied by the phenology adjustment factor to astimata peak standing crop for the growing season; H. Air-dry production/ multiplied by the air-dry weight correction factor to acre 288 162 210 660 e1timate air-dry seasonal production; multiplied by the average precipitation factor to estimate air-dry produc­ I. Aver.age precipitation X X X tion in a normal precipitation year; multiplied by the factor l. 2 l. 2 1.2 proper use factor to estimate seasonally available 11 (current year . 10 , av. year . 12") forage; and finally the result is divided by animal intake to estimate grazing capacity. The compounded J. Total adjusted air error or uncertainty in the final outcome from such a dry production/acre 346 194 m 1erie1 of calculations is estimated below, II. Crazing Capacity (for~. without any wildife or wild horses) The following analy1i1 is a first approaiaation or the variation (coefficient of variation) in an initial A. Total adju•ted air- 1razing capacity estimate for • hypot~etic.al range unit dry production 300 180 200 680 withir. an allotment where the variation (coefficient available (within reach) of variation and covariation) in inventoried f~rage welghta and adjustment factors are of a magnitude B. Proper u•e factor •i•ilar to those found in the literature. for 1pring .50 .80 0 Proper use factor Le~ Y be a random variable estimating carrying for tu11met' .40 0 0 capacity where: c. UHable forage lb/ acre Spring 150 144 0 294 y • X. Suna.et' !20 0 0 120 1 D. Acres/AUM - Spring (800 lb Req/AUM) 2.7 n ia ttie product operator, and the xi·• are the SU11111er (800 lb Req/AUM) 6.7 randoa variables: inventoried green forage vei&ht,

C C ( ) ) )

803 804

6 6,5,4 6 K

utilization, phenology, precipitation, p~ope~ u5e. and /.J 0.8 o.3 CV(X1 ) 1• 1,1 aniaal intak• factors. Civeft that th• air-dry vaight 2.5 • I correction factor is a•s~med to be a constant ~ich 77: I little or no variance, the estimated grazina capacity is ! I a product of (k • 6) random variables. I •i• I I 2.0 • I Suppa11 1 for i • 1,2, •.• 1 k, Xi has a lognormal

distribution, and the aean1, varianc11 1 and covariances ' can be estimated. Then the coefficient of variation of Y cau be shown to be: 1.5 • ,./ It I· 2 > ,: CV(Y)•{n [ l +CY (Xi ) l [ ll IT I ; I; i•l i 1.0 • I ; c..., I I 2 I where µ , • mean of X., a . • variance of I., c .. • co- 1 1 1 o· 1 1J I 0.5 ~ , variance of X. and X., CV(X.) • ~ coe!fici~nt of varia- I 1 1 1 ~ i I I tion of Xi' and CCV(Xi, Xj) • cij • coefficient of covaria- I 0 I ; l ui u j •0.2 -0.1 0.0 0.1 tion of Xi and X·. The lognormal assumption allows exact calculation of t~e coeffi~ient of variation of products. le i• not our intant that this coefficienc of variation formula be used, generally, without regard for the Figure l. Coefficient of variation of a $VIM-generated nature of the sample distribution. grazing capacity estimate CV(Y) as a function of a common coefficient of covariation among Suppose, now, that CV(Xi) • CV for every i, and all pairs of variates CCV(Xi, Xj)i~j fork of CCV(Xi, Xj) • CCV for every i and j, then: four to six variates, and a common coefficient of variation CV of 0.3, 0.8 or 1.3 for each of six variates.

of Xi's i, also great (Figure L). As pointed out above, Using this equation for CV(Y}, one can estimate the coefficients of variation for biomass, cover. frequency coefficient of variation of a SVtM-generated grazing and density data may average 0.4 to l.4 in rangeland capacity estiaate -hen k • 6 variacea (Xi~,) versus inventories. Within the CCV range -0.1 to 0.05 becveen fever Ck • S or 4) variates, or less adjustment factors all pairs of six Xi's, CV(Y) ranges from O to 2.S for CV when common CV and CCV •re assumed. For CV• 0.8, CV(Y) values of 0.3 to 0.8 for all Xi's. When coefficient• of baaed on six variates is greater than that based on four covariation between pair• of the six Xi ·s are near zero variates if CCV is greater than about -0.05 (Figure 1). (independence), CV(Y), for CV values of 0.3, 0.8 and

With CCV• -0.04. CV(Y) based on six variates is about 1.3. are approximately 0.8 1 4.l and 19.4, respectively. 2.2 while CV(Y) based on fou~ variate• is about 1.9. As CCV approaches -0.2 from -0.05, CV(Y) baoed on six Thi• analysis clearly illustrates that the reli­ variates. is actually_ lover than that baaed on four ability of• SVIM generated grazing capacity estimate is variates. Likewise, as the coefficient of covariation, quite lov when the estimate is dependent on six or even CCV, declines fro• a positive value, co zero, to a fewer variates, Grazing capacity estimates should neaative value the CV(Y) is reduced. thus, when the always be reported with • variance ~•timate, be it a covariance• betwaen Xi's (adjustment factors and confidence intervalt a coefficient of variation, ot some th• inventoried weight) are positive the CV(Y) is other measure of variabilty. Merely reporting the inflatedi vhea covariances are negative, CV(Y) may be variability for each of the six vari•te1 does not give ~educed 1iinificantly. the variability in the grazing capacity estimate itself, and 1ome level of covariance analysis is necess•ry, Sensitivity of CV(Y) to coefficients of variation 805 806

Conceptually the SVIM procedure contain• all the Cordova 1 F . .J. 1 Joe D. Wallace, and Rex D. Pieper. components needed to account for the grazing capacity of 1978~ Forage intake by grazing livestock: A review. J. R.ange Manage. 31: 430-438. a rangeland. Rowever 1 the often semiarid nature of many of our rangeland• make reliable estimation of carrying capacity virtually impossible. Variance in Greig-Smith, P. 1964. Quantitative plant ecology. grazing capacity is nearly as important a parameter to Butterworth1 1 London. 256 p. estimate as mean carrying capacity. Typically range­ lands with lower carrying capacity have higher vari­ Joint Committee of the American Society of Range Manage­ ability in carrying capacity. This fact must be taken ment and the Agricultural Board. 1962. In: into account when establishing accuracy and preci1ion Chapter 9. Sampling methods with special~eference criteria for rangeland inventories; on lower carrying to range management. Basic proble•• and techniques rangelands lesa stringent sampling requirements must be in ranae research. NAS-NRC Publ. 890. accepted. Kershaw, Kenneth A. 1964. Quantitative and dynamic We caution against the viewpoint that ranceland ecology. Amer. Elsevier, New York. 183 p. inventory. and the resultant carryinc capacity and variance escimacions, are useless activities and McGinty, W. Allan, Fred!. Smeins 1 and Leo B. Merrill. measures. On a district or forest, estimates of 1979. Influence of soil, vegetation, and grazing carrying capacity and their variances on nearby allot­ management on infiltration rate and sediment ments aay provide the most valuable relative measure of production of the Edwards Plateau rangeland. J. gra&ing capacity. Coupled with the insight of an Range Manage. 32:33-37. experienced range manager, grazing capacity information can provide for better ranceland management. Ke-rrill, Evelyn H., Henry F. Hayland, and James H. Peek. 1980. Effects of a fall wildfire on herbaceous vegetation on xeric sites in the Selway-Bitterroot LITERATURE CITED Wilderness, Idaho. J. Range Manage. 33:363-367.

Aldrich, Robert C. 1979. Remote sensing of wildland Myers, Wayne L .• and Ronald L. Shelton. 1980. Survey resources: A state-of-the-art review. USDA For. method• for ecosystem management. Wiley­ Serv, Gen. Tech. Rep. RM-71, 56 p. Rocky Mt, For. Inter1cience1 New York. 403 p. and Range Exp. Sen., Fort Collins, Colo. Meuenschwander, L. F. 1980. Broadcast burning of Arnold, Leroy, A., Jr., and D. Lynn Drawe. 1979. sagebrush in the winter. J. Range Manage, 33:233- Seasonal food habits of white-tailed deer in the 236. south Texu plains. J, Range Manage. 32:175-178. Pumphrey, F. V. 1980. Precipitation, temperature, and herbage relationships for a pine woodland aite in Art&, John L. 1 and Lester McKenzie. 1980. Motes for presentation on range surveys. Susanville Rancher northeastern Oregon. J. Range Mana1e, 33:307-310. Meeting, Feb. 21, 1980. 7 p. (unpublished aimeo). Rauzi, Frank. 1978. HiKh rates of nitrogen change BLH-$VIM. 1979. Bureau of Land Management manual 4412. compoaition of shortgrass rangeland in southeastern Physical re1ource 1tudie1. Soil-vegetation inven- Wyoming. J. Range Manage. 31:366-370. tory method. 55 p. Reese, Gary A. 1 Robert L. Bayn, and Neil!, West. 1980. Carpenter, L. H., 0. C. Wallmo, and a. 8. Gill. 1979. !valuation of double-sampling estiaators of subalpine Forage diver1ity and dietary selection by wintering herbage production. J. Range Manage. lJ:300-306. aule deer. J. Range Manage. 32:226-229. Reppert, Jack M. 1 Merton J. Reed, and Pinha1 Zusman. Colwell, R.. N., aa.d S. J. Titu•. 1976. Saa Houston 1962. An allocation plan far range unit sampling. National Forest inventory and development of a J. Range Manage. 15:190-193. aurvey planning •odel. Fin. Rep. for NASA. Cont.ract 9-14452. Space Sci. Lab (55), 17, 75 p. Ro1iere, R. !., Joe D. Yallace. and Raz D. Pieper. Univ. Calif., Space Sci. Lab., Berkeley. 1980. Forage intake in cvo-year old cows and heifers grazing blue grama summer range. J.

( ( ( ) ) ) 807

Range Manage. 33:7\-7).

Stout, Darryl G., A. HcLeon. B. Brooke, and J. Rall. 1980. Influence of ,imulated gra~ing (clipping) on pinegrass growth. J. Range Manage. 33:286-29\.

Dresk, Daniel W. 1978. Diets of the black-tailed hare in steppe vegetation. J, R.ange Manage, 31:439-442.

Uresk, D. W. 1 J. F. Cline, and W. H.. Rickard. 1979, Growth rates of a cheacgrass community and some associated faccors. J. Range Manage. 32:168-170.

West, Neil !. 1 and Juan Gasco. 1978, Phenology of the aerial portions of shadscale and winterfac in Curlew Valley, Utah. J. Range Manage. 31:43-45.