United StatesDepartment ofAgriculture

Forest Service

IntermountainResearch Station

General TechnicalReport INT-259

April 1989

Wilderness CampsiteMonitoring Methods:A SourcebookDavid N. Cole

THE AUTHORDAVID N. COLE is Project Leader and research biologistwith the Intermountain Station’s Wilderness ManagementResearch Work Unit at the Forestry Sciences Laboratory onthe University of Montana campus, Missoula. Dr. Cole re-ceived his B.A. degree in geography from the University ofCalifornia, Berkeley, in 1972. He received his Ph.D., also ingeography, from the University of Oregon in 1977. He haswritten many papers on wilderness management, particularlythe ecological effects of recreational use.

PREFACEThe original objective of this report was to provide a hand-

book for managers on how to develop and use a campsite-monitoring system. The state of knowledge appeared to besufficient to make this possible. As the project advanced,however, it became clear that many problems and unan-swered questions remain. Therefore, the objective of theproject turned toward compiling a summary of knowledgeand identifying problems and areas in need of research.Managers able to recognize and modify ideas with merit andwho are careful to avoid the problems identified should finduseful techniques in this report. Managers looking for asystem that can be applied without modification or creativeapplication will find this report frustrating. Researchersshould find the discussion of problems and needed researchof value in identifying important projects.

The review and discussion that follows, then, is an attemptto summarize and evaluate experience with campsite moni-toring systems, to identify situations where more research isneeded, and to provide additional sources of information.

The discussion of existing systems purposefully is critical.My intent is to identify limitations and weaknesses, as wellas to suggest useful approaches. Despite these shortcom-ings, those who have developed existing systems deservemuch credit as pioneers in the field of monitoring. Otherscan learn from what has been accomplished and contributeto the development and use of increasingly effectivemonitoring systems. Finally, the opinions in this report aremine alone and as such are open to questioning, which Iencourage.

RESEARCH SUMMARYThis report summarizes information on techniques that

have been developed for monitoring campsites, particularlythose in wilderness and backcountry. lt is organized as aseries of steps as follows: (1) evaluating system needs andconstraints, (2) deciding on impact parameters and evalu-ation procedures, (3) testing of monitoring techniques,(4) training and documentation, (5) collecting field data,(6) analyzing and displaying data, and (7) applying data tomanagement. For each step, existing techniques are de-scribed and evaluated, problems are discussed, and sourcesof information are listed. Detailed examples are included ina series of appendixes.

A wide variety of monitoring techniques have been devel-oped. They range in format from photographic techniques tofield measurement procedures of varying complexities. Thetechniques have been adapted to many diverse environ-ments and many different types of impact. Experience inanalyzing and using monitoring data is less developed.There is a critical need to develop low-cost monitoring sys-tems with sufficiently high levels of precision. Opportunitiesfor further research are numerous.

The use of trade or firm names in this publication is for reader infor-mation and does not imply endorsement by the U.S. Department ofAgriculture of any product or service.

Intermountain Research Station324 25th Street

Ogden, UT 84401

CONTENTSPage

introduction .......................................................................... 1Step 1. Evaluate System Needs and Constraints.. .............. 1

Decision Making ............................................................... 1Evaluation Criteria ............................................................ 2Photographic Techniques ............................................... .3Condition Class Estimates .............................................. .4Measurements on Permanent Sampling Units ............... .5Measurements and Estimates Without PermanentSampling Units ............................................................... 7

Time Estimates ................................................................ 8Research Needs .............................................................. 8Sources of information ..................................................... 9

Step 2. Decide on impact Parameters andEvaluation Procedures ................................................. 10

Decision Making ............................................................. 10Impact Parameter Descriptions.. .................................... 1 1Research Needs ........................................................... .18Sources of information .................................................. .19

Step 3. Testing of Monitoring Techniques ....................... ..2 0Refinement of Techniques ........................................... ..2 0Calibration Procedures ............................ . ................... ..2 0Estimation of Measurement Error ................................ ..2 0Research Needs .......................................................... ..2 3Sources of information ................................................. ..2 3

Step 4. Documentation and Training ............................... ..2 3Documentation ............................................................. ..2 3Training ........................................................................ ..2 4

Step 5. Field Data Collection Procedures ........................ ..2 4Data Forms ................................................................ ....2 4Electronic Field Data Recorders .................................. ..2 5Research Needs .......................................................... ..2 5Sources of information ................................................. ..2 5

PageStep 6. Data Analysis and Display ................................... ..2 5

Analysis of the Current Situation .................................. ..2 5Analysis of Trends ........................................................ ..3 0Automatic Data Processing ........................................... .30Research Needs .......................................................... ..3 1Sources of information .................................................. .31

Step 7. Management Applications of Monitoring Data.......3 1Establishing Management and Budget Priorities ......... ..3 1Management of Specific Sites.. ..................................... .31Relationship to Standards ............................................ ..3 2Use in Developing Visitor-Use Capacities.. ................... .32Research Needs .......................................................... ..3 3Sources of information ................................................... 33

References ...................................................................... ..3 3Appendixes A-K: Selected Procedures Used To MonitorWilderness Campsites ..................................................... ..3 6

A. Photopoint Photography ........................................ .36B. The Frissell Condition Class System ..................... .37C. The Sequoia-Kings Canyon Campsite

Class System .................................................... ..3 7D. The Eagle Cap Method of Measurements

on Permanent Sampling Units.. .......................... .40E. The Sequoia Method of Measurements

on Permanent Plots.. .......................................... .43F. The Olympic Bare Ground Technique ................... .44G. The Great Smoky Mountains Areal

Measurement Technique.. .................................. .46H. The Bob Marshall Rapid Estimation Procedure.. .. ..4 7I. The Canyonlands Rapid Estimation Procedure.. .. ..5 1J. The Delaware Water Gap Rapid Estimation

Procedure .......................................................... ..5 4K. The Hardware and Software Used To Collect

Data at Yosemite ............................................... ..5 7Hardware ....................................................... ..5 7Software ......................................................... ..5 7

Wilderness Campsite MonitoringMethods: A SourcebookDavid N. Cole

INTRODUCTION

According to the Wilderness Act of 1964, recreationaluse of wilderness is to be managed “so as to preserve its[the wilderness’] natural conditions” and such that “theimprint of man’s work [is] substantially unnoticeable.”Natural conditions have been most severely altered byrecreational use on campsites. A first step that must betaken to control campsite impacts is to document camp-site conditions and how they are changing over time. Thisinformation need has spurred considerable interest in thedevelopment of methods for monitoring campsites inwilderness and other backcountry areas.

The first publication to propose a specific method forsystematically monitoring campsites was ‘the Code-A-Sitesystem (Hendee and others 1976). This was followed bya number of papers suggesting somewhat different ap-proaches to campsite monitoring (Cole 1983a; Frissell1978; Parsons and MacLeod 1980; Schreiner andMoorhead 1979). Since the publication of these reports,there has been considerable campsite monitoring activ-ity-most of it unpublished and difficult to access.

This report discusses the monitoring technologies thathave been developed for use on wilderness campsites andsuggests where improvement is needed. The report isorganized into a series of steps that must be taken in de-veloping a monitoring system. The discussion of each stepbegins with a statement of purpose. For steps that re-quire crucial decisions, a sequence of questions or issuesthat must be addressed is laid out. This is followed by adescription of procedural details. Where there are alter-native courses of action, the strengths and weaknesses ofeach alternative are discussed. Areas of needed researchand development are highlighted, and finally, sources ofadditional information are suggested. Detailed descrip-tions of representative examples of monitoring approachesare included in an appendix.

STEP 1. EVALUATE SYSTEM NEEDSAND CONSTRAINTS

The purpose of this step is to determine what type ofmonitoring system is needed and feasible and to establisha priority for the monitoring effort. The procedure is asfollows:

1. Establish the need for a campsite monitoringsystem.

2. Identify the most serious types of campsite impact.

3. Identify the types of information a monitoringsystem needs to provide.

4. Evaluate funding and work force constraints.

5. Decide among several alternative approaches tomonitoring.

The product is the selection of a monitoring approach, adecision that carries with it certain implications for fund-ing and work force needs.

Decision Making

The first question is, “Do I need a campsite monitoringprogram?” Any wilderness that receives overnight useprobably needs monitoring. Even where campsites arecurrently perceived as satisfactory, conditions may dete-riorate or it may be important to document conditions forthose who feel that impacts are excessive. Monitoringsystems are generally most necessary in places with largenumbers of sites or severe campsite impacts, places whereuse patterns are unpredictable or in a state of flux, andplaces where campsite management programs are chang-ing or have not been evaluated. The overall importance ofmonitoring is underscored by the fact that most wilder-ness areas meet at least some of these criteria. Neverthe-less it is valuable to document, in a written format, howcritical campsite monitoring is to management. Thisdecision will guide later decisions about funding for moni-toring, decisions that will influence the quality of theinformation collected.

A related question is, "How do I plan to use this infor-mation?” Management applications of monitoring dataare discussed in a subsequent section of this report. Ifyou can develop a clear picture of how monitoring datawill be used, however, it will be easier to design an effi-cient system.

The next question is, “Do I need an inventory of allsites?” Systems can be established to monitor eithera sample of sites or all sites in the area. Monitoringa sample of sites can identify the kinds of impacts thatare occurring, as well as how conditions are changing overtime. A carefully stratified sample can also provide infor-mation on how impacts and trends in impact vary withsuch factors as amount of use or location. But it is usu-ally desirable to have information on changes in the num-ber or spatial distribution of sites and information on thecondition of all individual sites in the area. To obtain thisinformation, a census of all sites is necessary.

If a census is not needed, sampling can reduce costsconsiderably. Several sites in each of a variety of environ-ments and use situations (different amounts and types ofuse) could be examined. See such studies as Cole (1982,1983b) and Marion (1984) for examples of this design-really more a research approach than monitoring.

The next question-regardless of the decision betweena census and a sample-is, Wow frequently do campsitesneed to be monitored?” It is unlikely that all campsites willneed to be monitored every year. Once every 5 years seemsto be a reasonable frequency for most situations. This is along enough time for subtle changes to develop into meas-urable changes (at least on some sites), but a short enoughtime to identify impacts before they get out of control. Al-though many sites are unlikely to exhibit measurablechanges, if the interval between observations is longer than5 years, there is little opportunity to halt undesiredchanges. Appropriate monitoring frequencies must bedecided on by each area.

Additional questions must be asked to decide whichmonitoring approach to adopt. One of those questions is,‘What types of impact are of most concern and need to bemonitored?” Although the nature of impacts on wildernesscampsites does not differ greatly between areas, manage-ment objectives will differ, and this should be reflected inthe types of impact that are monitored. From field visits torepresentative campsites and the experience of on-the-ground managers, the most important types of impact tomonitor should be identified. Types of impact that can bemonitored, along with specific procedures for each, aredescribed in step 2.

Another question to address is, “About how many camp-sites are there to monitor?” Although there is relativelylittle variation between most wilderness areas in the needfor monitoring, the types of impact that are occurring, andthe importance of monitoring all sites, there can be pro-nounced differences in the number and accessibility ofsites. Some areas have a small number of sites, eitherbecause camping is allowed only on designated sites orbecause use levels are low and only a few places are suit-able for camping. Other areas have thousands of siteswidely scattered over areas as large as several millionacres. Obviously, monitoring systems can be less costlyin those areas with fewer sites.

A rough guess about number of campsites, along withdecisions about monitoring frequency and types of impactto monitor, are needed to answer the question, “How muchtime will it take to complete an inventory using each of thevarious alternative methods?” The basic monitoring ap-proaches available are described below. These descriptionsconclude with an estimation of the time requirements persite, recognizing that such estimates would vary greatly,particularly with the type and number of impact parame-ters being evaluated. Estimate time requirements for eachof these approaches, without making an initial judgmentabout a preferred approach.

As an example, if an area has about 500 campsites, andthey are to be inventoried every 5 years, 100 sites will haveto be visited each year. A technique that takes two people2 hours per site would require 400 staff-hours in additionto travel time. A technique that takes one person only5 minutes would require only 8 to 9 person-hours in addi-tion to travel time. It should be noted that, for many ofthese techniques, travel time may exceed the time spentmonitoring. This makes differences in the time requiredfor monitoring less critical. It also suggests the value ofcombining monitoring tasks with other tasks, such as pa-trol or cleanup, to make the most use of travel time.

With an estimate of the time it would take to get the jobdone using each of these techniques, the next question is,“How much time can I afford to spend on monitoring?”Funding levels for management of wilderness andbackcountry differ greatly between areas, as does theproportion of those funds allocated to monitoring. Areaswith more resources available and fewer sites have anumber of options available; those areas with fewer re-sources and/or more sites have few options. The moreprecise and informative approaches inevitably take moretime and are more costly. Therefore, a fundamental deci-sion about funding priorities must be made.

Once this decision has been made, the final question is,“Of those approaches that I can afford, which will bestmeet my needs?” Review the pros and cons of each ap-proach described below in the context of the types ofinformation needed and the types of impact of most im-portance. Select a technique that maximizes accuracy,precision, sensitivity, and the amount and quality of infor-mation (criteria that will be discussed below) for thosetypes of impact and information of most importance. Ifinformation needs cannot be met with available funds,more funds should be sought. Many monitoring fundshave been wasted because the information collected isinadequate (often reflecting limited available funds).

E v a l u a t i o n C r i t e r i a

In order to evaluate alternative approaches, evaluationcriteria are needed. All acceptable systems must haveseveral basic features. As was mentioned before, a censusof sites is vastly preferable to a sample of sites. Without acensus, there is no information on number of campsites.It is also necessary for any system to be set up in such away that the same sites can be relocated at a later date.Finally, a system will be of little use if it cannot identifychange in the most important impact parameters.

Much of the difference between acceptable systems is intheir relative accuracy and precision. Accuracy describeshow close an estimate is to a true value; precision de-scribes how close several estimates are to each other,regardless of how close they are to the true value. Usinga dartboard as an analogy, accuracy would be measuredby the proximity of the darts to the bull’seye. Precisionwould be measured by the proximity of the darts to eachother, regardless of how close they were to the bull’s-eye.Accuracy is important because we want to assess thecurrent situation for campsites. We want a system thatwill tell us as accurately as possible, for the most impor-tant parameters, how much impact has occurred. Preci-sion is important because we want to identify trends overtime. If techniques are imprecise we will not be able todistinguish real changes from separate imprecise esti-mates of the same value. To monitor trends, precision ismore important than accuracy.

The quality of the information collected is influenced bythe scale of measurement-whether nominal, ordinal, orinterval (Schuster and Zuuring 1986). Nominal measuresinvolve placing observations in categories that do notimply order. An example is noting whether a campsite islocated in a lodgepole pine forest, a Douglas-fir forest, or

2

a fescue grassland. Ordinal measures place observationsin categories that do imply a relative order, but there isno information about the distance between observationsor categories. An example is noting whether trash on thesite is absent, evident, or abundant. We know that siteswith abundant trash have received more impact thanthose on which trash is evident, but we do not know howdifferent they are. Interval measures do provide informa-tion on the difference between two observations. Forexample, a site with 25 pounds of trash has 10 poundsmore than a site with 15 pounds. Not only do we knowwhich site is more impacted, we also know how muchmore impacted it is. Clearly, the amount of informationgenerated by interval measures exceeds that of ordinalmeasures which, in turn, exceeds that of nominal meas-ures. Another advantage of interval measures, as will bediscussed later, is that they can be combined into syn-thetic summary indexes of impact. Although such indexeshave frequently been constructed from ordinal measures(Cole 1983a; Parsons and MacLeod 1980), this procedureis mathematically inappropriate.

Sensitivity, another important criterion, describes howlarge a change must be for it to be identified confidentlyas a change. Sensitivity is dependent on both precisionand quality of information. High sensitivity requires bothprecise measurements and either interval measures orordinal measures in narrow classes. High sensitivity isdesirable because it permits the identification of subtlechanges.

Another important criterion is amount of information.Obviously, a system that generates information on a num-ber of different types of impact is preferable to one thatcollects just one bit of information, as long as both thequality and importance of the information are similar.Sometimes information is collected on several parameters,but the information is combined in a single index. Unlessinformation on each parameter can be disaggregated,such an approach loses all but a single bit of information.

The final criterion, which unfortunately is often the mostimportant, is cost. Although it is possible to design low-cost systems that meet some of our criteria for a high-quality monitoring system, those that meet all of ourcriteria are the most costly.

As a final important note, the techniques described be-low were developed for a variety of purposes. Some wereintended as monitoring systems; others were not. Thosesystems that do not rate highly in this critical review ofeach as the basis for a monitoring system are not necessar-ily “bad.” They are described here because they have mer-its; unfortunately, all systems also have drawbacks. Theimportant thing is to understand each alternative’s prosand cons and to choose and modify a system that willclosely meet specified needs. That is precisely why thisfirst step of evaluating system needs is so important.

Evaluations of the monitoring systems described beloware summarized in table 1.

P h o t o g r a p h i c T e c h n i q u e s

Some of the earliest attempts at monitoring relied pri-marily on photographic techniques. Magi11 and Twiss(1965), for example, describe how repeated photographsfrom permanent camera points can be used to detectchanges in wildland resources, including campsites. Theattraction of photography is that subjectivity can be re-duced; consequently, precision should be high. The fatalflaw in most systems based entirely on photography is thatthe most basic assumption-that the most important typesof impact will be monitored-is seldom met. Surprisinglyfew types of impact can be accurately evaluated in photo-graphs, and essentially none of them can be assigned aninterval level rating. Moreover, contrary to popular belief,photographs can lie. Photographs taken at different timesof the day, under different lighting conditions, with differ-ent films, cameras, and lenses, or from slightly differentvantage points can give misleading impressions.

Table l-Strengths and weaknesses of alternative systems for monitoring campsites

Evaluation criteria

Monitoring system Accuracy PrecisionScale of

measurement SensitivityAmount of

information cost

Photopoints (A)1

Condition class estimatesFrissell (B)Parsons/MacLeod (C)

Permanent measuresCole (D)Stohlgren/Parsons (E)

Nonpermanent measuresSchreiner/Moorhead (F)Bratton (G)Cole (H)Kitchell/Connor (I)Marion (J)

Low

Mod.Mod. high

HighHigh

Mod. highMod.Mod. highMod. highMod. high

High

High OrdinalMod. high Ordinal

HighHigh

Mod.Mod. lowMod.Mod.Mod. low

---

Intervalinterval

IntervalIntervalOrdinalOrdinalinterval

Low

Mod. lowMod. low

HighHigh

Mod.Mod.Mod.Mod.Mod.

Mod. low

LowLow

HighHigh

Mod. lowMod.HighHighMod. high

Mod. low

LowLow

HighHigh

Mod.Mod. highMod. lowMod. lowMod. low

‘Letters in parentheses refer to the appendix that provides a detailed description of each monitoring system.

3

In conclusion, although the accuracy and precision ofphotographs can be high, this is not always the case andaccuracy and precision are irrelevant if important types ofimpact cannot be monitored. Sensitivity is low in mostcases, as is the amount of information collected. Two re-views of available photographic techniques have con-cluded, consequently, that photographs should enhancebut not replace the field measurements that are the foun-dation of most monitoring programs (Brewer and Berrier1984; Cole 1983a).

Photographs can play several extremely important roles,however. They can be an indispensable means of deter-mining if you are on the correct site when returning toreexamine a site. For relocational purposes, it is helpful toinclude in the photograph unusual landmarks or featureslikely to be around for a long time. Photographs are alsoan important tool for teaching evaluators to make consis-tent judgments when monitoring sites. This will be de-scribed in detail in step Documentation and training.

Photographs can also be a useful way to illustratechanges documented with field measurements. This canincrease the effectiveness of written documents and pres-entations in communicating information on conditions andtrends. Consequently, it is a good idea to take periodicphotographs from permanent photopoints on a sample ofsites. The sample should include as wide a range of situ-ations (impact levels, types of impact, environments, andso forth) as possible. Guidance on establishing permanentphotopoints can be found in Magi11 and Twiss (1965),Brewer and Berrier (1984), and in appendix A.

C o n d i t i o n C l a s s E s t i m a t e s

These systems involve assigning each campsite to acondition class category based on defined levels and/ortypes of impact. The presence, absence, or degree ofchange in certain critical parameters is quickly noted andforms the basis for an impact rating, usually between 1and 5. Such systems can provide relatively accurate andprecise estimates of overall impact. Sensitivity is low tomoderate, depending on the number of categories that aredefined. Sensitivity is higher when more classes are recog-nized, but this reduces precision because it increases thelikelihood of differences of opinion about which class a siteshould be assigned to. The critical limitation of this tech-nique, however, is that only one bit of information is pro-vided and this information is only of an ordinal level. Theonly information that can be gleaned from such systems isthe relative overall impact level on each site and whetherconditions have improved or deteriorated enough, overtime, to assign the site to another class. Information aboutspecific types of impact and trends in specific impacts islacking.

These systems are a good choice for areas with littlefunding per site. Only a few minutes are needed to locateeach site on a map and record its condition class. Thisprovides a gross estimate of impact levels and distribution.Such estimates are likely to be acceptably accurate andprecise without spending much time on each site.

Two examples of condition class systems will be pre-sented. The first example is the system proposed byFrissell(1978), based on his experience in the Boundary

Waters Canoe Area Wilderness and in what is now theLee Metcalf Wilderness. This system consists of descrip-tions of five condition states based on extent of vegetationloss, mineral soil exposure, tree root exposure, erosion,and tree mortality. Frissell’s classes are as follows:

1. “Ground vegetation flattened but not permanentlyinjured. Minimal physical change except for possibly asimple rock fireplace.”

2. “Ground vegetation worn away around fireplace orcenter of activity.”

3. “Ground vegetation lost on most of the site, but hu-mus and litter still present in all but a few areas.”

4. “Bare mineral soil obvious. Tree roots exposed onthe surface.”

5. “Soil erosion obvious. Trees reduced in vigor anddead.”

Each campsite is simply assigned to the class that mostaccurately describes the condition of the site.

One problem with this system occurs when sites do notmeet all of the criteria of any single class. For example, itis not uncommon for sites to have extensive tree rootexposure (class 4), but retain litter and humus in all but a few places (class 3). This problem can be handled by as-signing the site a value equal to the midpoint of twoclasses-for example, 3.5 in the example just described.Having done this, however, it is not possible to tellwhether a 3.5 site has root exposure but little mineral soilor abundant soil exposure but no root exposure. In theBoundary Waters Canoe Area Wilderness, Marion (1986)found sites that fit all five condition classes, as well aseach of the four midpoints between classes.

Another alternative would be to reword the definitionsin such a way that if either of several conditions werefound, the site would be assigned to that class. For ex-ample, the definition of class 4 could be changed to “Baremineral soil obvious or tree roots exposed on the surface.”If either of these conditions occurs, the site is assigned toclass 4.

The problems that result from a combination of moder-ately low sensitivity and the provision of only one bit ofinformation are more serious. In a study of campsites inEagle Cap Wilderness, 71 percent of the sites examinedwere condition class 4 sites, despite considerable variabil-ity in site conditions, amount of impact, and amount ofuse (Cole 1982). In a study of a wide range of campsitesin the Boundary Waters Canoe Area Wilderness, Marion(1986) assigned about two-thirds of all sites to classesbetween 2 and 3. Moreover, dramatic changes in condi-tion must occur before they will be reflected in a change incondition class. Over a 5-year period, only one of the 22Eagle Cap sites changed an entire condition class.

A final problem is that, while Frissell’s system workswell in coniferous forests with conspicuous ground covervegetation and thick organic horizons, it does not apply tomany other environments, such as areas above timber-line, grasslands, or deserts. This problem can be dealtwith by developing similar class definitions that areadapted to these other environments. In wilderness areaswith a variety of structurally distinctive environments,however, it may be impossible to develop readily compa-rable rating systems that work well in all environments.

4

This problem-of a system working well in some envi-ronments but not in others-is avoided with a conditionclass system that was devised by Parsons and MacLeod(1980) for use in Sequoia and Kings Canyon NationalParks. They also recognize five condition classes, in thiscase based on eight criteria: density of vegetation, compo-sition of vegetation, total area of the campsite, barren corearea, campsite development, litter and duff, social trails,and tree mutilations. For each of these criteria, the site isassigned a rating from 1 to 5, based on descriptions (seeappendix C for more detail). The condition class is thenthe closest integer, between 1 and 5, to the mean of theseratings. With practice, the evaluator can simply look at asite and assign it to a condition class without goingthrough the process of assigning a rating to each criterionand determining a mean. This results in a system quitesimilar to that of Frissell in which ‘a class one campsitewould usually be no more than a small sleep site andpossibly a fire ring, with little or no vegetative change ortrampling evident,” while “a class five site would be alarge, heavily used barren area” (Parsons and MacLeod1980).

Compared with Frissell’s (1978) system, this approachavoids the problem of sites not fitting into a single class.By including more impact parameters, as well as a rangeof conditions for each parameter, the distribution of sitesacross the range of condition class values is more equi-table. The Parsons and MacLeod (1980) system is proba-bly less precise than the Frissell system because moredecisions (one for each criterion) must be made before aclass rating can be assigned. Moreover, the practice ofassigning a class rating without evaluating each criterion,once considerable experience with the system has beengained, also increases the likelihood of bias and loss ofprecision.

The Parsons and MacLeod (1980) system is a moreaccurate predictor of impact, however. Two studies havecorrelated condition class ratings with impact indexesderived from careful measurements on campsites. Highcorrelations were found in both the Eagle Cap (Cole 1982)and Boundary Waters Canoe Area Wildernesses (Marion1986), indicating that both systems accurately portrayimpact levels on campsites. In the latter wilderness,where both the Frissell system and a modification of theParsons and MacLeod system were compared, the correla-tion coefficient was considerably higher for the Parsonsand MacLeod method.

Such systems clearly achieve some of the goals of aninventory and monitoring system and can be a good choicein places with severe funding constraints-a commonsituation in wilderness and backcountry. They providerelatively accurate and precise information on (1) thenumber and distribution of sites, (2) changes in the num-ber and distribution of sites, (3) relative impact levels indifferent portions of the wilderness, and (4) relative im-pact levels for individual sites. Their value in monitoringchanges in the condition of individual sites is more lim-ited. Moderately low sensitivity and low information con-tent mean that only sizable changes in overall conditioncan be detected and no information is available on whattypes of impact are particularly severe or either deterio-rating or improving.

This means that monitoring information cannot be usedto develop campsite management programs that targetspecific types of impact in particular places. Problemareas can be flagged, but it will be up to managers toguess how they have changed and to decide, on the basisof field examinations, what should be done.

Two other problems are of a more technical nature.One problem is that use of condition class systems locksmanagers into the current set of impact parameters andtheir implicit equal weighting. If managers change theiropinions on the parameters to be used or their relativeimportance, they will raise serious problems. If theyredefine the condition classes, the information generatedwill no longer be comparable to previous ratings. On theother hand, if they continue to use definitions that are nolonger appropriate, the survey may be largely irrelevant.

The second problem is confined to the Parsons andMacLeod (1980) system. It results from performing themathematically improper procedure of calculating themean of several ordinal ratings. Because the intervalsbetween ordinal ratings are unknown-they undoubtedlydiffer both within and between criteria-a mean cannotbe readily interpreted (Schuster and Zuuring 1986).Although a class 5 site would clearly be more highly im-pacted than a class 1 site, certain class 4 sites may not bemore impacted than class 3 sites. Although the high cor-relations between these inappropriate means and intervalscale measurements suggest that a site with a high meanis usually highly impacted, use of this improper procedureinterjects potential for misleading results.

M e a s u r e m e n t s o n P e r m a n e n t

S a m p l i n g U n i t s

Avery different approach is to take detailed measure-ments of a number of impact parameters on permanentlylocated sampling units, usually quadrats, transects, or theentire campsite. Periodic repeat measures of such para-meters as vegetation cover and composition, mineral soilcover, and bulk density or number of damaged trees canprovide highly accurate and precise measures of impact.If designed properly, such data are highly sensitive andthe amount of information is high because many parame-ters can be examined and interval measures can be ob-tained. Such systems rate highest in all evaluation crite-ria with the exception of cost. Measurements of this typeusually require several people spending several hours oneach site, and additional office time is required for datareduction.

Because costs are so high, it is unlikely that measure-ment systems based on permanent sampling units can domore than sample sites. Therefore, these techniques aremore common in research studies than in a true monitor-ing program. Given a relatively small number of sites,measurements of this type could form the basis for amonitoring program that provides a large quantity ofaccurate and precise information. More commonly, de-tailed measurements on a sample of sites might be used tosupplement less precise rapid estimates taken on all sites.

5

Two alternative designs will be described here. Thefirst design was initially used in a study of changes on asample of 26 campsites in the Eagle Cap Wilderness initi-ated in 1979 (Cole 1982, 1986a). It has subsequently beenused, in modified form, on samples of campsites in variousareas, including the Bob Marshall Wilderness (Cole1983b), Grand Canyon National Park (Cole 1986b), andDelaware Water Gap National Recreation Area (Cole andMarion 1988). It has proven to be a useful design fordescribing both current impact levels and changes overtime. A more detailed description of the procedure isincluded in appendix D.

The procedure calls for a large nail to be buried nearthe center of each campsite. For subsequent monitoringpurposes, this nail can be found with a lightweight pinlocator (a type of metal detector). From this point, thedistance to the edge of the obviously disturbed part of thesite is measured in 16 directions. The polygon enclosed bystraight lines connecting transect end points defines thecamp area, within which damage to trees and density oftree reproduction (number of stems/acre) is evaluated.

Four transects between the center point and campsiteboundary are established at right angles to each other.Nails are buried at the end of each transect to facilitatereestablishment of each transect. Approximately fifteen3.3- by 3.3~ft (l- by l-m) quadrats are located along thesetransects. The number of quadrats on each transect isproportional to the relative length of each transect. Thedistance between successive quadrats decreases withdistance from the center point so that there is less ten-dency to oversample the center of the site.

The canopy cover of total ground cover vegetation, ex-posed mineral soil, and each plant species is estimated foreach quadrat. Measures of the depth of organic horizonsand the penetration resistance of the soil are also taken ineach quadrat. Means for each of these parameters arethen calculated for the campsite as a whole. Four soilsamples are taken in the central part of the campsite toobtain measures of soil bulk density, moisture content,and chemical composition. Finally, four measures of wa-ter infiltration rate are taken close to these soil samples.

While these measures provide information about thecondition of the campsite, they do not indicate how muchchange has occurred on the site. For example, a site witha vegetation cover of 50 percent may be perfectly naturalor it could have lost as much as half of its vegetationcover. To obtain estimates of how much change has oc-curred on campsites, similar measures are taken on envi-ronmentally similar undisturbed sites (controls) in thevicinity.

Selection of suitable controls demands great care. Theidea is to select a site that you think the campsite wouldhave looked like before it was camped on. The best con-trols will be similar to the campsite in tree canopy den-sity, rockiness, and slope, and have ground cover speciessimilar to those surviving in protected places on thecampsite. It is also desirable to find a control as close tothe campsite as possible.

Once selected, the control must be referenced to thecampsite so it can be relocated. Then a nail should beburied at the site center. The most efficient size for acontrol will vary with environmental heterogeneity;

controls should be larger in more heterogeneous environ-ments. In the Eagle Cap, circular controls of 1,000 to2,000 ft2 (100 to 200 m2) were used.

In order to characterize amount of impact, campsiteconditions are compared to those on controls. For ex-ample, the difference between vegetation cover on thecampsite and the control provides an estimate of howmuch vegetation has been lost from the campsite. Tocharacterize change over time, one can either examine thedifference between campsite conditions obtained duringsuccessive observation periods or examine the change inthe amount of impact (the campsite/control comparison).For example, on Eagle Cap sites receiving moderate levelsof use, vegetation cover decreased from 6.1 percent in1979 to 5.7 percent in 1984; vegetation loss (the differencebetween cover on campsites and controls, expressed as aproportion of cover on controls) increased from 75 percentin 1979 to 79 percent in 1984 (Cole 1986a).

A different design was employed to follow change onclosed campsites in Sequoia National Park (Stohlgren andParsons 1986). Refer to appendix E for more detail. Oneach campsite, a 32.8- by 32.8-ft (10- by 10-m) study areais identified. Permanent stakes are placed around theperimeter at 3.3-R (1 m) intervals. Buried nails at each ofthe four corners can serve to permit relocation of thesestakes. String is set up between stakes to establish a gridof 10.8.ft2 (l-m2) squares. Each quarter of each square isthen subjectively stratified into core, intermediate, orperiphery. Core areas are the most denuded places, lo-cated close to the center of the site. Intermediate areasexhibit obvious impact, but have more vegetation cover,less litter and duff pulverization, and some pockets ofintact surface sod. Periphery areas, essentially controls,appear to be unimpacted. Of all the quarter squares-each 2.7 ft2 (0.25 m2 -that fall entirely in one of thesestrata, five to 10 are randomly selected. In each, canopycover of each plant species is estimated and bulk density,soil penetration resistance, litter accumulation, soil mois-ture, and soil chemistry are measured.

Both the Eagle Cap (Cole 1982) and Sequoia (Stohlgrenand Parsons 1986) systems will provide quite precisemeasures of change because each employs replicablemeasures on plots that can be readily relocated. Therecan still be measurement error, however, if for example,estimates of vegetation cover tend to be high one year andlow the next year. If campsites are compared with con-trols, this problem should be reduced because the samebias would be applied to controls and, therefore, canceledout when the differences between campsite and controlare calculated.

Levels of accuracy and information provided are likelyto be very different depending on which of these tech-niques is used. The Eagle Cap technique provides onlyone measure of each type of impact, characteristic of theentire campsite. The Sequoia technique recognizes thatconditions within the campsite are not homogeneous andprovides one measure characteristic of the most highlyimpacted parts of the site and another measure for theparts of the site that are intermediate in impact. Depend-ing on information needs, this added information providedby the Sequoia technique can be useful or it can add un-necessary complexity. It provides a more accurate picture

of intrasite variability but does not provide a single sum-mary measure for the entire site.

Several features of the Sequoia technique reduce itsaccuracy, relative to the Eagle Cap technique. First, meas-urements are taken on only a very small proportion of thesite. Measurements on the core and intermediate areasare taken in a total area of no more than 50 ft2 (5 m2) com-pared to 150 ft2 (15 m2) on Eagle Cap sites. Moreover, onmost sites much of the campsite is likely to be outside ofthe 1,076-R2 (100-m 2) area that was sampled. (The mediancamp area in the Eagle Cap was almost twice this size.)Second, using periphery measures as control values can bemisleading. Areas less than 16 feet (5m) from the center ofthe site are likely to either be quite disturbed or, if theyhave not been disturbed, they are likely to be environmen-tally distinct from the campsite proper (for example, undera tree or among rocks).

Both the Eagle Cap and Sequoia techniques provideuseful, precise, and sensitive information on change to thecampsite. As Stohlgren and Parsons (1986) have shown,the added information provided by stratifying the campsiteinto core and intermediate zones may be preferable forexamining recovery of closed sites because intermediateparts of the site recover more rapidly the core parts. Butthe Eagle Cap technique appears more likely to provideaccurate estimates of how much impact has occurred to thesite as a whole because the entire campsite is included inthe sample, the size of the area sampled is larger, andcontrols are likely to be more representative. It appearsthat it should also require less time and therefore shouldbe less costly.

Other variations on these techniques can and have beentried (Echelberger 1971; LaPage 1967; Leonard and others1983; Magi11 1970; Merriam and others 1973). The mostuseful ones will provide accurate measures of how muchimpact has occurred to the campsite and be designed tofacilitate precise replication of measurements on the samesampling units. It is the care that goes into precise repli-cation that takes so much time and makes these tech-niques costly.

M e a s u r e m e n t s a n d E s t i m a t e s W i t h o u t

P e r m a n e n t S a m p l i n g U n i t s

With the preceding techniques much time is spent estab-lishing permanent sampling units on campsites. Severalalternative systems take measurements and estimates ofimpact on campsites without establishing permanent plots.Schreiner and Moorhead (1979) developed such a systemin the early 1970’s for use in Olympic National Park. Theymeasured the distance to the first live plant along eighttransects radiating from the center of the site. The aver-age distance was used as the radius of a circle to calculatebare ground area. They also counted the number of accesstrails radiating from the site to water, the main trail, orother campsites, as well as the number of places that hadbeen trampled heavily by horses, within 100 feet of thesite.

In a study of campsites in Eagle Cap Wilderness, thebare radius proved to be highly correlated with a syntheticindex of change in a number of impact parameters

(Cole 1982). This suggests that it does provide an accu-rate indicator of overall impact. Precision is more of aproblem, however. Centerpoints were not permanentlymarked and slight shifts in the center resulted in sizabledifferences in bare radius and area. With a large meas-urement error, only sizable changes in bare radius or areacan be interpreted as definite changes in impact; conse-quently, sensitivity is only moderate. Only three types ofinformation were collected. Although interval scale meas-ures were taken, the advantage of interval measures overordinal measures is reduced by the sizable measurementerror. Costs are moderately low because these techniquesrequire little time on the campsite. One variation of thissystem required that a scale map be drawn; this requiredconsiderable time, making it costly as well.

Another system based on area1 measures was used onbackcountry campsites in Great Smoky Mountains Na-tional Park (Bratton and others 1978). In this system, thelengths and widths of various types of impact were meas-ured. The types of disturbance measured were bare rock,mud, slope erosion, and bare soil (all of which could becombined in a measure of total bare soil), leaf litter (areasin which litter remained but all vegetation was lost), andtrampled vegetation. In addition, trash, tree damage, andfirewood clearing were quantified by measuring out alongat least two axes to where these disturbances ceased andthen calculating the area involved.

The data generated by this system are only moderatelyaccurate. As Bratton and others (1978) noted, defining allof these disturbances as rectangles overestimates damage.Moreover, the only data provided are area1 measures ofimpact. No data can be generated, for example, on howmany trees have been damaged or how much vegetationhas been lost. In addition, the lack of controls makes itimpossible to estimate how different campsites are fromundisturbed areas.

The precision level of this system is likely to be moder-ately low. No permanent center points or transects wereestablished. In future remeasurements, new centerpoints and axes will be selected. The resultant area1measurements are likely to be very different, even if nochange in impact occurs. Given the sizable measurementerror, differences in area1 measures would have to also besizable before they could be interpreted as definitive evi-dence of a change; therefore, sensitivity is only moderate.Compared to the Schreiner/Moorhead technique, informa-tion is collected on more types of impact, although manyimportant types (such as tree damage, soil compaction,access trails, or change in vegetation composition) are leftout. Again the value of interval scale data is reduced bythe sizable measurement error. Finally, it appears thatthis technique would be quite time consuming, suggestingthat the cost of implementation would be moderatelyhigh.

In evaluating these systems in relation to the perma-nent plot approach, the question that arises is whether ornot the lower cost is worth the loss of precision and sensi-tivity that comes with not establishing permanent sam-pling units. If the cost savings is worthwhile, would it beworthwhile to cut costs even further by relying less ondetailed measurements?

Much of the progress in recent years has come in thesearch for techniques that will provide a suitable compro-mise between costly measurement systems and systemsthat provide little specific information. The nagging prob-lem with all of these techniques is their relatively lowprecision levels. How can one rapidly estimate impactparameters while still keeping measurement error small?

A series of estimation systems have been developed thatdraw heavily on earlier ideas, particularly some of thoseadvanced by Schreiner and Moor-head (1979) and Parsonsand MacLeod (1980). The major differences have beenattempts to (1) collect information on more parametersthan Schreiner and Moorhead did, (2) avoid the probleminherent to the Parsons and MacLeod system of not beingable to disaggregate data on each parameter, and (3)maximize precision levels but use interval measureswhere possible. Progress has been made, but problemsremain. Despite these problems, I believe that once re-fined, these systems will provide the best compromise forthe budgets of most backcountry areas.

A system developed by Cole (1983a) is patterned closelyafter that of Parsons and MacLeod (1980). The most im-portant change is that each parameter is recorded sepa-rately in the field. While this adds a few minutes to thetime it takes to complete the form-requiring an averageof about 10 to 15 minutes for experienced evaluators tomeasure a site-it increases the amount of available in-formation greatly. Other changes include (1) use of moreprecisely defined techniques for evaluating change invegetation density and litter and duff cover, (2) deletionof the vegetation composition parameter, (3) separation ofthe mutilation parameter into a count of both trees thathave trunk damage and trees with exposed roots, and (4)separation of campsite development into both develop-ment and cleanliness parameters. See appendix H for adetailed description of the procedure.

The accuracy of this system was evaluated by Marion(1986) and found to be moderately high, as the Parsonsand MacLeod (1980) system was. Precision is less high,however; it is only moderate. This results from the factthat many more judgments must be made in the field.The fact that categories are broad suggests that, withtraining, judgments should be acceptably precise, butdifferences of opinion will be more common than withcondition class systems or measurements taken onpermanent plots.

The primary advantage of the system is the largeamount of information that can be collected in a shortperiod of time. Of all systems, this produced the mostinformation per unit of cost. Although the system ini-tially consisted entirely of ordinal estimates, it was even-tually modified to consist of interval estimates that wererecorded and then used to assign each parameter an ordi-nal ranking. This provides even more information, al-though the precision of the interval estimates must bequestioned. Measurement error needs to be minimized(see step 4) and quantified (see step 3).

With all of this information, it is clearly advantageous

weights). These products are then summed to provide asummary rating. Dividing these ratings into, for ex-ample, five classes will produce five condition classes.Changes over time in summary rating can be followed, ascan changes in the individual impact parameters.

As was the case with the Parsons and MacLeod (1980)system, this procedure improperly sums a series of ordi-nal rankings. This makes the summary ratings difficultto interpret. Widely divergent ratings should accuratelyreflect the relative impact of different sites; however,interpretations of even relative differences between siteswith proximate summary ratings are suspect. A site witha rating of 50 may be less impacted than a site with arating of 47, for example. This merely reinforces the ma-jor problem with this type of system-a precision levelthat is lower than desirable, and a paucity of informationon just how low precision is.

Despite these problems, modifications and similar sys-tems have been developed for use in such widely diver-gent situations as canoe campsites in the Boundary Wa-ters Canoe Area Wilderness and the readily accessibleDelaware Water Gap National Recreation Area (see ap-pendix J), sites on whitewater rivers such as the MiddleFork of the Salmon and Colorado Rivers, and backpackersites in the deserts of Canyonlands (see appendix I) andGrand Canyon, as well as sites in mountainous areassimilar to those where such systems were first developed.Although parameters and procedures differ (these will bedescribed more fully, with examples, in step 2), the ap-proach of making rapid estimates of a number of parame-ters remains consistent.

At the Delaware Water Gap, however, problems withlow precision have led managers to base as many methodson counts and measurements as possible. As methods forthe area have developed, managers have felt it necessaryto spend more time and to use more precise techniques(see appendix J).

T i m e E s t i m a t e s

Clearly, the time required to use any of these ap-proaches will be highly variable. Nevertheless, it seemsimportant to provide some general estimates of time re-quired. Photopoint systems might take one person about30 to 60 minutes per site, depending on the number ofphotographs to be taken. The condition class systemstake one person from 3 to 5 minutes per site. The systemsof measurements on permanent plots usually take twopeople between 1 and 3 hours per site, depending on theparameters to be measured. The measurement and esti-mation systems without permanent plots are highly vari-able; the Cole (1983a) system might take one person 10 to15 minutes per site, while a system similar to that usedby Bratton and others (1978) might take several peopleseveral hours per site.

R e s e a r c h N e e d s

to assign a single summary impact rating (essentially acondition class) to each site. This is done by multiplying

The primary research need is to refine rapid estimation

each parameter by a weight assigned to each parametertechniques that provide information on a number of sepa-rate impact parameters but do not take very much time

(more important types of impact are assigned higher per site. The problems with these techniques are low

8

precision levels, uncertain measurement errors, and use ofinappropriate summary impact ratings. Some means ofincreasing precision levels are described in steps 2, 3, and4, steps that involve the development, testing, and docu-mentation of specific procedures. Research needs to gofurther toward identifying the most precise estimationprocedures and suggesting means of maximizing precision.The issue of measurement error and its estimation is dealtwith in step 4. Very little evaluation of error has beenconducted, however; more research is needed on this sub-ject. Several appropriate ways to calculate summary im-pact ratings are described in step 2; more work on thissubject would also be valuable.

S o u r c e s o f I n f o r m a t i o n

Bratton, Susan Power; Hickler, Matthew G.; Graves,James H. 1978. Visitor impact on backcountry camp-sites in the Great Smoky Mountains. EnvironmentalManagement. 2: 431-442. (Describes the system of area1measurements used to measure the condition of camp-sites in Great Smoky Mountains National Park.)

Brewer, Les; Berrier, Debbie. 1984. Photographic tech-niques for monitoring resource change at backcountrysites. Gen. Tech. Rep. NE-86. Broomall, PA: U.S. De-partment of Agriculture, Forest Service, NortheasternForest Experiment Station. 13 p. (Describes and evalu-ates a variety of photographic techniques that can beused to monitor conditions on campsites and trails.)

Cole, David N. 1982. Wilderness campsite impacts: effectof amount of use. Res. Pap. INT-284. Ogden, UT: U.S.Department of Agriculture, Forest Service, Intermoun-tain Forest and Range Experiment Station. 34 p. (De-scribes measurement techniques for monitoring changeon permanent plots. Also correlation between thesemeasures of impact and individual indicators of impact,including Frissell’s condition class.)

Cole, David N. 1983. Monitoring the condition of wilder-ness campsites. Res. Pap. INT-302. Ogden, UT: U.S.Department of Agriculture, Forest Service, Intermoun-tain Forest and Range Experiment Station. 10 p.(Describes the uses and desirable characteristics of amonitoring system. Includes a detailed description ofa system based on rapid estimates of many differentimpact parameters.)

Cole, David N. 1984. An inventory of campsites in theFlathead National Forest portion of the Bob MarshallWilderness. Unpublished report on file at: U.S. Depart-ment of Agriculture, Forest Service, IntermountainResearch Station, Forestry Sciences Laboratory,Missoula, MT. 19 p. (Describes a monitoring system,based on rapid estimates, that was used in the BobMarshall Wilderness. Includes form, instructions, re-sults of the inventory, and suggestions for further work.)

Cole, David N. 1985. Ecological impacts on backcountrycampsites in Grand Canyon National Park. Unpublishedreport on file at: U.S. Department of the Interior,National Park Service, Western Regional office, SanFrancisco, CA. 96 p. (Describes techniques for monitor-ing campsite conditions on permanent plots. Also de-scribes a monitoring system based on rapid estimates ofa number of different parameters.)

Echelberger, Herbert E. 1971. Vegetative changes at Adi-rondack campgrounds: 1964 to 1969. Res. Note NE-142.Upper Darby, PA U.S. Department of Agriculture, For-est Service, Northeastern Forest Experiment Station.8 p. (Describes changes on automobile campgrounds.)

Frissell, Sidney S. 1978. Judging recreation impacts onwilderness campsites. Journal of Forestry. 76: 481-483.(Describes the Frissell condition class monitoringsystem.)

Hart, James B., Jr. 1982. Ecological effects of recreationuse on campsites. In: Countryman, David W.; Sofranko,Denise M., eds. Guiding land use decisions: planningand management for forests and recreation. Baltimore,MD: Johns Hopkins University Press: 150-182. (Dis-cusses impacts on developed campsites and describes aseries of impact measurements.)

Hendee, John C.; Clark, Roger N.; Hogans, Mack L.; Wood,Dan.; Koch, Russell W. 1976. Code-A-Site: a system forinventory of dispersed recreational sites in roaded areas,backcountry, and wilderness. Res. Pap. PNW-209.Portland, OR: U.S. Department of Agriculture, ForestService, Pacific Northwest Forest and Range Experi-ment Station. 33 p. (Describes the Code-A-Site systemfor monitoring campsites.)

Kitchell, Katherine P.; Connor, Jeff. 1984. Canyonlandsand Arches National Parks and Natural BridgesNational Monument draft recreational impact assess-ment and monitoring program. Unpublished paper onfile at: U.S. Department of the Interior, National ParkService, Moab, UT. 80 p. (Describes rapid estimatemonitoring systems for river, four-wheel-drive, andbackpack campsites.)

LaPage, Wilbur F. 1967. Some observations on camp-ground trampling and ground cover response. Res. Pap.NE-68. Upper Darby, PA: U.S. Department of Agricul-ture, Forest Service, Northeastern Forest ExperimentStation. 11 p. (Describes changes over a 3-year periodon a previously unused automobile campground inPennsylvania.)

Leonard, R. E.; McBride, J. M.; Conkling, P. W.;McMahon, J. L. 1983. Ground cover changes resultingfrom camping stress on a remote site. Res. Pap. NE-530.Broomall, PA: U.S. Department of Agriculture, ForestService, Northeastern Forest Experiment Station. 4 p.(Describes changes over a 2-year period on previouslyunused sites on islands off the Maine coast.)

Magill, Arthur W. 1970. Five California campgrounds. . .conditions improve after 5 years’ recreational use. Res.Pap. PSW-62. Berkeley, CA: U.S. Department of Agri-culture, Forest Service, Pacific Southwest Forest andRange Experiment Station. 18 p. (Describes changes inthe condition of automobile campgrounds.)

Magill, Arthur W.; Twiss, R. H. 1965. A guide for recordingesthetic and biologic changes with photographs. Res.Note PSW-77. Berkeley, CA: U.S. Department of Agri-culture, Forest Service, Pacific Southwest Forest andRange Experiment Station. 8 p. (Describes the use ofphotopoints to monitor changes in resources, includingcampsites.)

9

Marion, Jeffrey L. 1986. Campsite assessment systems:application, evaluation, and development. In: Popadic,Joseph S.; [and others], eds. Proceedings, 1984 riverrecreation symposium. Baton Rouge, LA: LouisianaState University, School of Landscape Architecture:561-573. (Describes the results of tests of the accuracyof the Frissell and Cole rating systems and describessome refinements to the system used in the BoundaryWaters Canoe Area Wilderness.)

Marion, Jeffrey L.; Cole, David N. 1987. Recreationalimpact assessment and monitoring systems: past, pres-ent, and future. Paper presented at the National ParkService Science Conference; 1986 July 13-18; FortCollins, CO. (Describes the historical development ofmonitoring systems, as well as recent refinements,innovations, and areas of needed research.)

Merigliano, Linda. 1987. The identification and evalu-ation of indicators to monitor wilderness conditions.Moscow, ID; University of Idaho. 273 p. Thesis. (De-scribes a variety of monitoring techniques for manyresource parameters, including vegetation and soils).

Merriam, L. C., Jr.; Smith, C. K.; Miller, D. E.; [and oth-ers]. 1973. Newly developed campsites in the BoundaryWaters Canoe Area: a study of 5 years’ use. Stn. Bull.511, For. Ser. 14. St. Paul, MN: University ofMinnesota, Agricultural Experiment Station. 27 p.(Describes changes on previously unused campsitesover a 5-year period.)

Moir, William H.; Lukens, William M. 1979. Resourcemonitoring system, Chiricahua National Monument,Arizona. In: Linn, Robert M., ed. Proceedings, firstconference on scientific research in the national parks;1976 November 9-12; New Orleans, LA. Transactionsand Proceedings Series No. 5. Washington, DC: U.S.Department of the Interior, National Park Service:1189-1200. (Describes a system of measurements onpermanent plots.)

Parsons, David J.; MacLeod, Susan A. 1980. Measuringimpacts of wilderness use. Parks. 5(3): 8-12. (Describesthe campsite class rating system used at Sequoia andKings Canyon National Parks.)

Schreiner, Edward S.; Moorhead, Bruce B. 1979. Humanimpact inventory and management in the OlympicNational Park backcountry. In: Ittner, Ruth; Potter,Dale R.; Agee, James K.; Anschell, Susie, eds. Proceed-ings, recreational impact on wildlands; 1978 October27-29; Seattle, WA. R-6-001-1979. Portland, OR: U.S.Department of Agriculture, Forest Service, PacificNorthwest Region: 203-212. (Describes the monitoringsystem at Olympic, based primarily on measures of barearea.)

Schuster, Ervin G.; Zuuring, Hans R. 1986. Quantifyingthe unquantifiable: or, have you stopped abusing meas-urement scales? Journal of Forestry. 84: 25-30. (Dis-cusses the proper uses of different measurement scalesand common abuses, particularly in index construction.)

Stohlgren, Thomas J.; Parsons, David J. 1986. Vegetationand soil recovery in wilderness campsites closed tovisitor use. Environmental Management. 10: 375-380.(Describes measurements taken on permanent plots tomonitor change on closed campsites. Advocates strati-fying sites into core and intermediate parts of the site.)

Sydoriak, Charisse A. 1987. Yosemite’s wilderness trailand campsite impacts monitoring system. Paper pre-sented at the National Park Service Science Conference;1986 July 13-18; Fort Collins, CO. (Describes the camp-site monitoring system used at Yosemite NationalPark-a modified version of the Parsons-MacLeodsystem.)

STEP 2. DECIDE ON IMPACTPARAMETERS AND EVALUATIONPROCEDURES

The next step is to decide what to monitor and how toconduct the monitoring. At this point a monitoring ap-proach should already have been selected and some of themost important types of impact should have been identi-fied. Clearly it is most important to monitor the impactparameters considered to be the most critical.

Decision Making

If you have selected a photopoint system, step 2 can beskipped, although it might be worth reading for hints onwhat to photograph. Details on how to establish a photo-point system are included in appendix A. Some of thediscussion in step documentation-is also relevant tophotopoint systems.

For all of the other approaches, decisions must be madeabout which impacts are most critical and, within theconstraints of the three basic types of systems-conditionclass, measurements with permanent plots, and measure-ments or estimates without permanent plots-how eachtype of impact should be monitored. While it is not neces-sary or even desirable to collect information on all pos-sible types of impact, it would be worthwhile to collectinformation on types of impact that require different man-agement strategies. For example, because the size of thesite and number of access trails might be controlled withsite management techniques that are very different fromthe behavioral controls needed to deal with tree damage,it would be worthwhile to assess each of these differenttypes of impact.

If a condition class system was selected in step 1, it istime to select the impact parameters to form the basis forthe rating. The parameters Frissell(1978) used are wellsuited to forested areas with abundant understory vegeta-tion and thick soil organic horizons; other parameters willhave to be selected in very different environments, suchas those above timberline, in grasslands, or in the desert.Considerable thought and creativity need to go into devel-oping a system of this type. A sequence of progressiveimpact stages must be described, followed by definitionsof each stage in terms of the most important impact para-meters. It is important that categories are mutually ex-clusive so that all sites can be assigned to one and onlyone category.

It is a simpler matter to develop a condition class sys-tem similar to that described by Parsons and MacLeod(1980). Each type of impact is described separately sothat the relationships between different types of impactneed not be understood. Moreover, the large number of

10

parameters examined makes it likely that many differenttypes of environment can be evaluated with this tech-nique. Decisions must be made about which parametersto use, how each parameter will be evaluated, and how asummary rating will be derived. Assessment techniquesfor impact parameters and means of deriving a summaryrating are described below.

If an approach similar to that of Parsons and MacLeod(1980) is selected, however, it is worthwhile to take thelittle additional effort to separately record information oneach parameter. This increases the amount of informa-tion available many times. Although this requires moretime, the added time is usually minor in comparison tothe time spent in transportation. Separately recordingeach parameter should also increase the objectivity andprecision of summary ratings because the procedure ofdeciding on a rating is not shortcut.

This simple change-recording each parameter sepa-rately-results in the type of system described by Cole(1983a, 1984) and used in such places as the BobMarshall, Canyonlands (Kitchell and Connor 1984), andthe Boundary Waters Canoe Area. As with the Parsonsand MacLeod (1980) system, parameters and evaluationprocedures can be selected from the following examples(although other parameters or procedures could certainlybe developed). Generally, with these systems, each indi-vidual estimate does not require much time. Therefore,once on a site, it is usually worth collecting informationon as many important parameters as possible.

If a system based on measurements in permanent plotswas selected in step 1, each measurement can be quitetime consuming, so it is more important to select only themost critical parameters. Much of the requisite time onthe site, however, is involved in establishing and relocat-ing permanent sampling units. Given this investment oftime, it would be unfortunate to not collect information onas many important parameters as is feasible.

Impact Parameter Descriptions

In the following sections, impact parameters that havebeen included in monitoring systems will be described.A final section will discuss the derivation of summary rat-ings and condition class ratings. More detail on some ofthese procedures can be found in the appendixes.

Campsite Area-One of the most obvious measures ofimpact is the area affected by camping. Assuming a simi-lar level of vegetation loss, soil compaction, and so forth,larger campsites can be considered to be more heavilyimpacted. For example, the quantity of vegetation lostfrom a 200-ft2 campsite that has lost 50 percent of itscover is twice that of a 100-ft2 site that has also lost 50percent of its cover. In addition, size of the campsite isone of the most useful parameters for distinguishing be-tween lightly and heavily used sites and it is more likelythan most parameters to change over time (see, for ex-ample, Cole 1982, 1986a; Marion and Merriam 1985).

Although it is an obvious impact parameter, camp areacan be difficult to measure accurately or precisely. Twoproblems contribute to inaccuracy. First, it can be diffi-cult to define the edge of the site. In areas of dense fragile

vegetation, a boundary can be consistently defined by acontrast between untrampled vegetation and trampledvegetation or bare ground. Where vegetation is sparse,boundary definition can be difficult unless obvious distur-bance of organic horizons can be used. In areas of rockoutcrops, sand, and gravel, or in resistant vegetation suchas dry grassland or meadows, it can be virtually impos-sible to define a nonarbitrary campsite boundary.

Even where a boundary can be defined, another prob-lem stems from the difficulty of measuring the area of anirregular figure. If a site is a perfect circle or rectangle,its area can be measured precisely and quickly. Becausethis is never the case, accurate and precise estimates ofthe area of sites that differ greatly from simple geometri-cal shapes is time consuming.

The method that Cole (1982) used to monitor changeson sites in Eagle Cap is relatively precise. A permanentcenter point is used. From this point, 16 radial transectsare extended out to the edge of the site. The distancesfrom center point to boundary are plotted on radial mapsand then the area of the site is determined with a polarplanimeter. Accuracy increases with the number oftransects; precision is increased by using a permanentcenter point. The Schreiner and Moorhead (1979) system,on which the Eagle Cap technique was based, is an ex-ample of a less accurate and precise system. It does notuse a permanent center point, it uses only eight transects,and area was calculated on the basis of a mean radius-an assumption that would only produce an accurate resultif the site was circular.

Cole’s (1982) technique requires about 90 staff-minutesper site. The Schreiner/Moorhead technique might re-quire only about 25 staff-minutes per site because it is notnecessary to find the buried center point and it is notnecessary to map the transect end points or use theplanimeter.

If campsites are relatively regular in shape (withoutmany peninsulas and islands that are not trampled),Cole’s technique can be used to measure area, withoutplotting transects or using the planimeter. The 16 radialtransects define two sides of 16 triangles; a straight linebetween transect end points is the third side. The area ofthe campsite is the sum of the area of the triangles, eachof which can be calculated as 0.5 times the sine of theangle (0.383 for the 22.5-degree angles created by 16transects) times the product of the two transect lengths.Any number of transects could be used, although theappropriate angle would vary. Use of this procedurewould reduce time requisites considerably. Managers atDelaware Water Gap plan to use this technique in aneffort to obtain more precise area1 estimates than theyhave in the past (see appendix J).

More rapid area1 estimates are also possible. If thearea can be considered to approximate either a circle or arectangle, or a combination of simple geometric forms, aradius or lengths and widths can be paced off. Using theappropriate formulas, areas can be calculated quickly.While these areal estimates may not be very precise, theycan be completed in a few minutes by one person. More-over, in places where it is difficult to define a nonarbitraryboundary, these estimates may be as precise as the mostcareful measurements.

11

If rapid estimates of area are used, it is important toinvest considerable time in training and in deriving arough estimate of measurement error. This measurementerror can be used to establish confidence limits around anestimate. If future estimates are not beyond these limits,the interpretation should be that there is a high likelihoodthat no real change in area has occurred. More discussionof measurement error, its importance and use will bepresented in step 3.

Another way to both reduce estimation times andhandle problems with precision is to merely assign eachsite to a class based on a range of areas. For example,in the Bob Marshall Wilderness (Cole 1984), threeclasses were defined: 0-500 ft2 (0-46 m2); 501-2,000 ft2

(47-186 m2); and >2,000 ft2 (>186 m2). The system used inSequoia and Rings Canyon National Parks recognized fiveclasses: 0-20 ft2 0-2 m2); 21-100 ft2 (3-9 m2); 101-500 ft2

10-46 m2); 501-1,000 ft2 (47-93 m2); and >l,000 ft2

(>93 m2). With experience, it is usually possible to assigna size class to many campsites without any pacing ormeasurement. As the number of classes increases, preci-sion decreases-because the likelihood of assigning a siteto the wrong class increases-but sensitivity increases.More classes are preferable, if relatively high precisioncan be maintained through training and calibration ofevaluators. Ideally, classes should be as broad as theconfidence interval around estimates, for whatever tech-nique is used.

A final option, utilized in the Bob Marshall Wilderness(Cole 1984), is to estimate the area of the site and to thenassign each site to a class, on the basis of this estimate.The ordinal class rating is used to compare the size ofdifferent sites or to evaluate change over time. The inter-val estimate is a help in interpreting change, particularlyif estimates are close to the boundary of classes. For ex-ample, a change of one class is less meaningful if theoriginal estimate was close to the boundary of the classassigned during a subsequent rating.

Although campsite area is an important measure ofimpact and has been monitored in most systems in use,its problems are considerable. Where defining nonarbi-trary boundaries is difficult, it might be best to not meas-ure area. Often there is a strong correlation between thedevegetated area of the site and the total area of the site.Where devegetated area can be determined much moreprecisely than total area, it might be monitored instead.Across a variety of environments, however, devegetatedarea cannot be considered a surrogate for camp area. Ingeneral, the proportion of the camp that is devegetatedshould increase as vegetation fragility, amount of use, androughness of the local environment increase.

Other problems with defining campsite area includehow to handle separate sites that have grown together,satellite tent pads, and places offsite where stock havebeen tethered. The problem of intermingled sites is dis-cussed in detail in step 5-field data collection proce-dures. Satellite sites and stock-holding areas either canbe ignored (this reduces the accuracy of area1 estimates ofdamage) or their area can be measured or estimated andthen either reported separately or added to the total camparea. Problems can arise with deciding to which site oneshould assign a satellite site or stock holding area. One

should avoid contaminating a precise estimate of camparea with an imprecise estimate of satellite area. If tech-niques with different precision levels are used, the twoestimates should be kept separate.

Campsite Development-Another common item toaddress is the extent of development; that is, the numberand elaborateness of either agency or user-built facilitiessuch as fire pits, rings, grates, or places; seats; tables;shelters; and so on. The usual technique has been torecord the presence or absence of each, with or withouta count of the number of each type of facility. In addition,development classes have been suggested by Parsons andMacLeod (1980) and by Cole (1983a) (see appendix H).The difference between the two is primarily in the num-ber of classes and that Cole chose to separate cleanlinessaspects (such as amount of trash) from development.

The ability to quantify this parameter is limited.Several systems require a count of firepits. A rapid esti-mate system developed for the desert environment ofCanyonlands National Park (Kitchell and Connor 1984)also records the number of rocks larger than 6 inches indiameter that have been moved either to create flattertent pads or to construct tables or seats. This number isthen used to assign the site to one of four classes for rockdisplacement.

Cleanliness-Cleanliness refers to trash, humanwaste, horse manure, and campfire remnants, particu-larly charcoal. Again, quantification is difficult. In astudy of campsites along the New and Delaware Rivers,the number of 33-gallon trash bags of garbage found oneach site and the number of separate piles of toilet paper,with or without feces, within 164 ft (50 m) of the center ofthe site were counted (Cole and Marion 1988). This didnot effectively distinguish between clean sites and oneswith a small amount of trash. Counts of number of trashitems have been used-for example, in the Canyonlandssystem (Kitchell and Connor 1984)-but this fails to dis-tinguish between large items, such as a tarp, and smallitems, such as a cigarette butt. For human waste,Kitchell and Connor (1984) count pieces of toilet paperand piles of feces and note the presence or absence ofurine odor. Such estimates vary with the eyesight of theevaluator and the time spent searching.

Problems with quantification can be dealt with throughclassification. For backpacker sites, Kitchell and Connor(1984) suggest classes for trash (none, l-3 pieces, 4-6pieces, and >6 pieces), toilet paper (none, l-2 pieces, 3-4pieces, and >4 pieces), and feces (none, 1 pile, 2 piles, or>2 piles). They have different categories for sites used byvisitors in four-wheel-drive vehicles and sites used byboaters. Cole’s (1983a) system deals simultaneously withall aspects of cleanliness. Categories are: (1) no morethan scattered charcoal from one firering; (2) remnants ofmore than one firering or some litter or horse manure;and (3) either some human waste evident or much litteror horse manure.

Cleanliness and extent of development can have a pro-found effect on visitors. For example, Lee (1975) foundthat cleanliness and development were the site conditionsmost critical to the enjoyment of visitors in the Yosemitebackcountry. But they are not a lasting ecological impacton the land; they are easily removed. A strong argument

12

can be built for noting cleanliness and development butseparating them from ecological impact parameters.Counts of facilities, including firepits and the number ofplaces where fires have been built, seem worthwhile. Theproblems with quantification of trash and human wastesuggest, however, that a system of classes (such as none,some, and abundant) provides as much information andas high a level of precision.

In a study of impacts on camping beaches in GlenCanyon National Recreation Area, Carothers and others(1984) developed techniques for quantifying trash andcharcoal accumulation. Trash accumulation was estab-lished by counting, removing, and weighing all humantrash within 16 ft (5 m) of a 49- to 164-ft (15- to 50-m)transect running through the center of each site. Sanddiscoloration was used as a measure of ash incorporationinto the sand. A 50-mL surface sample of sand was col-lected in each of 10 quadrats located along the transectand passed through a sieve (mesh size 150 microns).These samples were then shaken 75 times against a circu-lar disk of coarse Fisher Filter Paper. A discolorationindex was then obtained by matching the color of the filterpaper to colors obtained from a series of beach sands con-taining known charcoal-ash concentrations. The indexranged from 1 (sand with no ash or charcoal) to 16 (sandthat contains 10 percent residue by volume). It might bepossible to design a similar technique for use in environ-ments other than sand beaches.

A unique parameter used at Canyonlands (Kitchell andConnor 1984) is the abundance of ants, flies, and rodentson and around campsites. The idea is that sites withmore of these pests tend to be more highly impacted.Four classes were derived for backpacker sites from "nopests within 50 feet [15 ml of the site” to ‘greater than1 ant colony; ants throughout site; numerous signs of ro-dents, tracks, burrows and nests within 20 feet [6 m] ofsite.”

Damage to Overstory Trees --Most systems attemptto monitor the extent of damage to over-story trees oncampsites. Some of the questions that need to be ad-dressed include how many different types of damage toassess separately, whether or not to consider only treeswithin the campsite boundaries, at what height or diame-ter does a tree become an overstory tree, at what level ofdamage should a tree be considered damaged, how to dealwith felled trees and stumps, and whether to provide aninterval level count of trees or a damage classification.One problem in some situations is distinguishing recrea-tional damage from "natural” damage.

In the Eagle Cap and Bob Marshall Wildernesses, Cole(1982, 1983b) counted all trees greater than 4.5 ft (140cm) tall on the campsite, noting the extent of damage toeach tree. The number and percentage of trees that hadbeen felled or that had trunk scars or exposed roots wascalculated. The number and percentage of trees with anydamage was also noted.

Marion (1984; Marion and Merriam 1985) distinguishedfour levels of damage to standing trees: none (no treedamage other than from obviously natural causes); slight(nails, nail holes, small branches cut off or broken, smallsuperficial trunk scars); moderate (large branches cut off

or broken, trunk scars and mutilations that may be numer-ous but do not total more than 1 ft2 [0.09 m2] of area); orsevere (trunk scars that total more than 1 ft2 [0.09 m2] orcomplete girdling of the tree). Each standing tree withinthe campsite boundaries was counted and classified accord-ing to damage level. In this system, trees with a diameterat breast height greater than 0.8 inch (2 cm) were consid-ered to be trees. A damage index was calculated by multi-plying the number of trees in the "none” category by 1 andthe number of trees in the slight, moderate, and severecategories by 2, 3, and 4, respectively. These products weresummed, and this sum was divided by the total number oftrees. The index, which varies between 1 and 4, can beinterpreted as the average damage to trees on the site. Forexample, an index of 2.5 suggests that the average tree isabout halfway between slight and moderate damage.Felled trees were recorded separately, although theymight, more appropriately, have been placed in a fifthdamage class.

A similar classification of root exposure was as follows:none (no root exposure other than from obviously naturalcauses); slight (the tops of many of the major roots exposedor more severe exposure on only one or two major roots);moderate (the tops and sides of many of the major rootsexposed or very severe exposure on only one or two majorroots); and severe (the tops, sides and undersides of manyof the major roots exposed). An index was calculated as fortree damage.

One problem with this approach is that many of thedamaged trees are found offsite. Visitors usually tie theirhorses offsite, causing root exposure, or fell trees for fire-wood or tent poles offsite rather than onsite. Thus, a countof damaged trees onsite will usually vastly underestimateamount of damage. The problem with counting trees off-site is that different evaluators may go different distancesfrom the site to search for damaged trees. This reducesprecision.

Bratton and others (1978) quantified tree damage interms of the maximum distance from the center of the sitealong at least two axes. This measure is difficult to inter-pret (for example, it says nothing about the proportion ofdamaged trees or the severity of damage) and is not likelyto be very precise.

Other systems have assigned rankings to sites based onthe extent of tree damage. With the Parsons and MacLeod(1980) system, the number of mutilations is counted, re-gardless of their nature or whether they occur on one ormore trees. Highly obtrusive mutilations are distinguishedfrom other mutilations. Categories are: none, l-2; 3-5;6-10, or l-2 highly obtrusive; and >10 or >2 highly obtru-sive mutilations. Apparently, only trees on the campsiteare included, although this and the definition of a highlyobtrusive mutilation is not included in their paper.

In the system developed for the Bob Marshall, Cole(1984) had evaluators count and record the number of treesthat have been scarred or felled, as well as the number thathave exposed roots that are obviously associated with thesite being examined (regardless of whether they are onsiteor not). Then sites are assigned to a class. For damage,the classes are: no more than broken lower branches,1-8 scarred trees or 1-3 badly scarred or felled trees, and>8 scarred trees or >3 badly scarred or felled trees.

13

To be “badly-scarred,” the surface area of scars must ex-ceed 1 ft2 (0.09 m2). The classes for root exposure are:none, 1-6, and >6 trees with exposed roots.

There are problems with any means of evaluating treedamage. Confining the monitoring to the campsite itselfincreases precision, but reduces the accuracy of the esti-mate of impact. Perhaps the best compromise for a low-cost system would be to count trees both onsite and off-site, but only count trees with pronounced damage, suchas those in the moderate and severe classes that Marionand Merriam (1985) define. Recording the numbercounted, as well as an impact class, also seems a usefulcompromise between the low-information classificationoption and the problem, with a count, of having the num-ber appear more precise than it really is. Both can add tothe interpretation.

Tree Reproduction- Loss of tree reproduction hasbeen monitored using measurements on permanent plots,but I know of no examples of rapid estimates. The tech-nique is to count reproduction-defined in the Eagle Cap(Cole 1982) as trees between 6 and 55 inches (15 and 140cm) tall-n the campsite and then calculate a density(such as number of stem&a). Seedlings are then countedon a control-a 538ft2 (50-m2) circle in the Eagle Cap-and density is calculated again. Subtracting campsitedensity from control density provides an estimate of theamount of reproduction per unit area (such as, per hec-tare) that has been lost on the site. Dividing this value bythe control density provides an estimate of the proportionof reproduction that has been lost.

One problem with this technique is how to handle re-production growing in protected places on the campsite.Often all reproduction on the site can be found in pro-tected clumps that are never trampled. In the Eagle Capstudy, reproduction in "untrampled islands” was excludedas being unrepresentative of the trampled portion of thecampsite. Excluding reproduction in untrampled islandsoverestimates impact on the campsite and reduces preci-sion because it requires a judgment about whether or nota clump of reproduction should be included While count-ing all reproduction within the campsite boundaries mayunderestimate impact on the trampled portion of the site,this estimate should be more precise and it does providean accurate representation of impact to the entire site.

Although there appear to be no examples, it should bepossible to evaluate the density of reproduction in classessimilar to those for density of vegetation in the Parsonsand MacLeod (1980) system. These classes could be: sameas surroundings, moderately less dense than surround-ings, and considerably less dense than surroundings.Because tree reproduction is often more patchily distrib-uted than ground cover vegetation, it can be more difficultto get an accurate and precise estimate.

Shrub Damage -If large resistant shrubs are a sig-nificant component of the vegetation, it can be worthwhileassessing damage to shrubs. This might be particularlyimportant in places where trees are lacking and shrubsform the tallest vegetation layer. The only cases I know ofwhere shrub damage has been assessed are in desertenvironments.

On backcountry campsites at Grand Canyon NationalPark, Cole (1985) counted number of shrubs and calcu-lated shrub densities on campsites in twenty-five 10.8-ft2

(1-m2) permanently located quadrats. Shrub density wasdetermined in a circular 538-ft2 (50-m 2) control plot. Onlyshrubs rooted in plots were counted. There was often aproblem deciding consistently how many individualshrubs to count in a shrub clump. The decision rule usedwas to count each clump of stems as one shrub. The mainproblem with this technique is that shrub density may notreflect impact because trampling damage often results ina higher density of smaller plants. In such cases, coverestimates might be a better choice.

At Canyonlands, classes have been delineated for bothdamage to shrubs and for root exposure (Kitchell andConnor 1984). On backpacker sites, damage classes are:none show any damage; 10 percent of shrubs show dam-age (for example, broken limbs, crushed appearance); 10to 30 percent of shrubs show damage or one or two shrubsshow reduced vigor as a result of damage; and >30 per-cent of shrubs show damage, more than two show reducedvigor, or dead and dying shrubs are present. Root expo-sure classes are: no roots exposed, roots exposed on oneshrub; roots exposed on two shrubs; and roots exposed onmore than two shrubs. Including both counts and per-centages in a single definition can cause problems. Forexample, if there are only a few shrubs on the site, one ortwo damaged shrubs might represent damage to a major-ity of shrubs.

Damage to Ground Cover Vegetation-- Severaldifferent aspects of ground cover vegetation damage areregularly monitored. The most common are the area thatis devegetated, reduction in vegetation density or cover,and change in species composition. Each of these can andhas been measured in permanent sampling units andestimated rapidly.

The Eagle Cap technique (Cole 1982), in addition tomeasuring the distance from permanent center point tothe edge of the campsite, measured the distance fromcenter point to the first significant amount of vegetation,along each of the 16 transects. This was a modification ofSchreiner and Moorhead’s (1979) use of eight transectswithout a permanent center point. These end points wereplotted on a radial map and the area of the polygon (de-vegetated area) was determined with a planimeter. Asmentioned before, summing the area of triangles is a morerapid means of calculating area.

It is important to decide on a vegetation cover or den-sity that will be considered a "significant amount” of vege-tation. The definition used in the Eagle Cap was at least15 percent cover in a 1.09-by 3.28-ft (0.33- by l-m) quad-rat oriented perpendicular to and bisected by the tape.This boundary can usually be determined more consis-tently than the edge of the campsite, so estimates of de-vegetated area should usually be more precise than esti-mates of camp area. Only where ground cover is verysparse and patchily distributed is it difficult to determinethe boundary of the devegetated area.

This technique only measures the area of a devegetatedcore close to the center of the site. Sometimes there areseveral devegetated places on a single campsite. These

14

could all be measured in the same manner, although thiswould be quite time consuming. The area of these otherplaces could be estimated and added to the central area,without too much loss of accuracy, if they were muchsmaller than the central area that was measured. Itwould be unfortunate, however, to significantly contami-nate a careful measurement of the central core with roughestimates of other barren places. It is also important todefine how large a barren area must be to be included insuch an estimate. Perhaps only devegetated areas largerthan, say 10 m2, should be included.

The other alternative is to estimate the area of eitherthe central devegetated area or all devegetated areas.As with camp area, radii, lengths, and widths can be esti-mated and then the appropriate area formulas can beused to determine area. In most systems only the size ofthe central area is estimated. This probably results inmore precise estimates, although this provides a lessuseful estimate of vegetation loss on the entire site. Esti-mates can be either to the closest whole unit of measure-ment (for example, meter) or sites can be classified. Forexample, Parsons and MacLeod’s (1980) classes for barrencore area are: absent, 5-50 ft2 (0.5-4.6 m2), 51-200 ft2 (4.7-18.6 m2), 201-500 ft2 (18.7-46 m2), and >500 ft2 (46 m2).Rapid estimate systems for the Bob Marshall (Cole 1983a,1984) and the Grand Canyon (Cole 1985) also have classesfor size of the devegetated center.

Most monitoring systems have used ocular estimates ofvegetation cover rather than more precise measurements(for example, by using tools such as a point-frequencyframe [Chambers and Brown 1983; Mueller-Dombois andEllenberg 1974]) of cover or density. Most systems differprimarily in the size of the sampling unit for which coveris estimated.

With the more precise systems, cover is estimated inpermanent quadrats-in fifteen to twenty 10.8-ft2 (1-m2)quadrats in the system that Cole (1982, 1983b, 1986b) hasused in various places, or in ten to twenty 2.7-ft2 (0.25-m*)quadrats in the system that Stohlgren and Parsons (1986)have used. Usually cover is estimated to the nearest 5 or10 percent. The mean cover from all the quadrats pro-vides an estimate of cover on the site. Such estimatesshould be quite precise; if the sample size is large enoughand the samples are properly distributed, the estimateshould also be quite accurate.

Few systems have attempted statistical determinationof an adequate sample size. Such techniques do exist (see,for example, Chambers and Brown 1983; Mueller-Dombois and Ellenberg 1974). A sample of 5 to 10 percentof the site would probably be adequate in most cases.Because of the pronounced disturbance gradient on camp-sites from center to periphery, a systematic placement ofquadrats provides a more accurate assessment for a givensample size than does a random placement.

The other quantitative alternative is to estimate coveron the entire site. This method is less precise simplybecause it is difficult to visualize cover of the entire siteat one time. This reduction in precision should be re-flected in how the data are displayed. For example, itwould be misleading, although quite feasible, to record

15

cover to the nearest percent if it could not be accuratelyestimated even to the nearest 10 percent. For the BobMarshall system, Cole (1983a, 1984) recommends esti-mating cover in the following classes: O-5 percent, 6-25percent, 26-50 percent, 51-75 percent, and 76-100 percent.This might be improved by dividing the latter categoryinto 76-95 percent and 96-100 percent classes. The mid-points of each category can be used as a single estimate ofcover.

With either of these quantitative estimates of cover, theimpact parameter of concern is not cover itself, but loss ofcover. This can be assessed by estimating cover on anundisturbed control plot and then comparing campsitecover, to cover on a neighboring undisturbed control site.Identical techniques should be employed on both campsiteand control. Vegetation loss can be expressed as eitherthe difference between campsite and control as a propor-tion of the cover on the control. For example, if vegetationcover was 40 percent on the campsite and 80 percent onthe control, loss could be either 40 percent (80 percent- 40 percent)--the amount less on the camp-or 50 per-cent ([80 percent - 40 percent]/80 percent), indicating thathalf of the vegetation has been lost.

In the Bob Marshall rapid estimate system (Cole 1983a,1984), vegetation is estimated in categories, both oncampsites and on a nearby unused comparative area. Therating for vegetation loss is based on the difference innumber of coverage classes between campsite and control.The site is rated 1 (if there is no difference); 2 (if there is adifference of one class); or 3 (if there is a difference of twoor more classes). In the Canyonlands system (Kitchelland Connor 1984), cover loss classes, when compared withcontrol, are: <10 percent reduction, 10-30 percent reduc-tion, 31-60 percent reduction, and >60 percent reduction.

Other modifications of this technique could be devel-oped. The important elements are to make estimates inclasses that reflect the precision of estimates, to makeestimates both on campsites and controls, and then toexpress vegetation loss in terms of a comparison betweenthe two.

Another alternative is to express vegetation loss interms that are not quantitatively defined. The Parsonsand MacLeod (1980) system asks for a rating of vegetationdensity as: same as surroundings, moderately less densethan surroundings, or considerably less dense than sur-roundings. Although not quantitatively defined, thesecategories probably reflect quantitative differences simi-lar to those just noted.

Vegetation composition has also been assessed in simi-lar terms. Parsons and MacLeod (1980) ask for a rating ofcomposition, relative to surrounding vegetation, of: sameas surroundings, moderately dissimilar, or significantlydissimilar.

In Canyonlands (Kitchell and Connor 1984), composi-tional changes are indicated by the proportion of the coverthat consists of exotic and/or disturbance species (Kitchelland Connor 1984). For example, a site with none of thesespecies is given the lowest rating, while a site on which>50 percent of the cover consists of exotic and/or distur-bance species is given the highest rating (most impacted).

When estimates of the presence and/or cover of eachspecies are available, both on campsite and control, it ispossible to calculate various indexes of the difference incomposition between campsites and controls. Cole (1982,1983b) has calculated an index as follows:

Floristic dissimilarity = 0.5 x

where P1 is the relative cover of a given species on thecampsite and P2 is the relative cover of the same specieson the control. Relative cover is the cover of a speciesexpressed as a percentage of the total cover of all species.It is calculated by summing the cover of all species andthen dividing the cover of each species by this sum. Thesum of the relative coverages of all species on a site willequal 100 percent. Good discussions of indexes that arebased on presence or measures of importance other thanrelative cover, or that utilize other formulas, can be foundin Mueller-Dombois and Ellenberg (1974) and Chambersand Brown (1983).

Most of these indexes range from 0 to 100 percent, withhigher numbers indicating a greater compositionalchange. Interpretation is hampered by the fact that thereis always some dissimilarity between two samples of thesame area of vegetation. This inherent dissimilarity var-ies between vegetation types and with the sampling pro-cedure and type of index. By comparing several replicatesamples of control plots, an idea of inherent variabilitycan be obtained. Unless floristic dissimilarity is substan-tially greater than this inherent dissimilarity, change incomposition should be considered negligible.

An important type of impact in arid environments isdisturbance of the fragile cryptogamic soil crusts that areso prevalent in deserts. Cryptogamic soils are formed byalgae, fungi, lichens, and mosses growing in a matrix ofsoil. They often form conspicuous black pedestaled sur-faces that are readily destroyed by trampling. InCanyonlands (Kitchell and Connor 1984), cryptogamic dis-turbance has been assessed with categories ranging from"no disturbance, (crust) still intact in appropriate habitat”to “>60 percent reduction of crust (when compared to ad-jacent undisturbed area).”

Impacts to Soil Organic Horizons- Three commonmeasures of organic horizon disturbance are reduction inorganic horizon cover, reduction in organic horizon depth,and an assessment of the degree to which the litter andduff has been disturbed. Reduction in organic horizoncover is estimated in a manner similar to that of vegeta-tion loss. Estimates of cover can be made either in a setof quadrats or for the entire site. Cover classes are fre-quently used to reflect the precision level of such esti-mates. In most cases it is exposure of mineral soil thathas been estimated. Mineral soil exposure is inverselyrelated to organic horizon cover because it is only exposedafter organic horizons are removed. Campsite cover iscompared to cover on a control to determine increase inmineral soil exposure.

A problem that surfaces when estimating mineral soilexposure is how to deal with the situation where mostorganic matter has been removed but there are still thinpatches of litter remaining. An explicit judgment must bemade about where to draw the line between bare soil and

litter. I have tended to ignore thin, obviously disturbedpatches of litter and treat such areas as exposed mineralsoil.

There are two common problems with measuring areduction in thickness of organic horizons. The first is theproblem of defining the boundary between organic andmineral horizons. Although this boundary is usuallygradational, with training and calibration, consistentdefinitions can be made. This source of error will be mostserious when organic horizons are quite thin (for example,less than 1 to 2 cm).

The more serious problem relates to where to locatesamples. The thickness of organic horizons frequentlyincreases from the center of the site to the edge. Thispattern is superimposed upon a pattern of random vari-ation in thickness related to litter fall and decompositionrates. Enough samples must be taken to adequately ac-count for the random variation and they must be distrib-uted in a consistent manner, from year to year, relative tothe disturbance gradient. If permanent quadrats havebeen established, a good solution is to take one samplefrom each quadrat. These locations will be precisely repli-cated each year and a sample of 15 to 20 measurementsshould be more than adequate to account for randomvariations. Without permanent quadrats or a largesample size, it is doubtful that this parameter wouldchange rapidly enough to be measured with sufficientprecision.

Again, similar measures must be taken on controls.The difference between measures on campsites and con-trols provides an estimate of the reduction in thickness oforganic horizons.

Parsons and MacLeod (1980) suggest categories basedon evidence of disturbance of the organic layers. Thesecategories range from “trampling discernible; someneedles broken, scattered cones,” to “heavily trampled;(litter) clumped and pulverized; cones absent,” and finallyto "litter, cones and duff completely absent.” Kitchell andConnor (1984) use categories based on the proportion ofthe litter cover that appears to be crushed or broken.They also note the distribution of litter, whether it isevenly distributed or confined to the edge of the site orprotected locations. The highest impact category is as-signed to sites in which >60 percent of the litter cover ap-pears crushed or broken and to sites in which >80 percentof the litter occurs around the edge of the site or stableobjects.

Impacts to Mineral Soil- In contrast to many of thepreceding parameters, most of the methods available forevaluating impacts to mineral soil are time consuming.Consequently, these are seldom estimated except in meas-urement systems on permanent plots. The most commonimpacts to assess are soil compaction, water infiltrationrates, moisture content, organic matter content, andchemical composition.

Soil compaction has usually been estimated by measur-ing either bulk density or the resistance of the soil topenetration. Bulk density is the ratio between the dryweight and volume of a soil sample; penetration resis-tance is a measure of the force required to push a pen-etrometer a given distance into the soil. Either measure

16

increases as soil compaction increases. There are a num-ber of alternative methods and tools for each of thesemeasurements (Gifford and others 1977).

Each measure has advantages and disadvantages.Penetration resistance is a simpler measure to take be-cause there is no need to take a soil sample and, if usinga pocket penetrometer, the instrument is light. In addi-tion, penetration resistance is a more sensitive measure ofimpact than bulk density; more subtle increases in com-paction can be detected with a penetrometer.

Penetration resistance measures are quite variable inspace, and they vary with soil moisture. Therefore, thedifference between penetration resistance measures attwo points in time may reflect soil moisture rather thana change in level of compaction. This problem can be cor-rected by taking measures at a standard moisture level,such as field capacity, but this is not practical in mostsituations; it takes too much time. Another option is tocompare the difference between campsite and control attwo points in time, assuming that the influence of differ-ent moisture levels on campsites and controls would can-cel each other out. Unfortunately, it is likely that themagnitude of the effect of moisture levels would differbetween campsites and controls.

Another problem with penetration resistance measuresis the difficulty of obtaining readings in sandy, rocky, orgravelly soils. Finally, with pocket penetrometers, pene-tration resistance levels on campsites frequently exceedthe maximum resistance that can be measuredA.tons/ft 2 (kg/cm 2). Moreover, the instrument is pushedonly 0.25 inch (6 mm) into the soil. This may not providean adequate representation of compaction. For example,it is not uncommon for a highly compacted soil layer to becovered by a centimeter of pulverized soil. The pocketpenetrometer would detect no resistance in the upperlayer, where most measures are taken. One option herewould be to record penetration resistance at variousdepths in the soil.

Bulk density measures are less variable through spaceor time; however, they are also difficult to take in rocky orgravelly soil. In such soils, it can be impossible to use asoil corer without distorting the sample; instead, a tech-nique such as the paraffin clod or irregular hole method(Howard and Singer 1981) must be used. These are quitetime consuming. Even after using one of these techniquesit may be necessary to remove rocks and gravel from thesoil samples to obtain a meaningful measure of bulk den-sity. Other researchers have used instruments such asthe air permeameter, volumeasure, and gamma ray scat-tering device (Gifford and others 1977), although theseare not practical for use in the backcountry.

As was the case with thickness of organic horizons,there is a gradient of compaction, related to amount oftrampling, from the center of the site to its edge. Thisgradient is superimposed on a pattern of random vari-ation on the site. Therefore, it is important to take anadequate number of samples, to take samples along theentire disturbance gradient, and to make certain thatrepeat measures reflect a similar distribution of samples.

The requisite number of samples will vary from place toplace. It appears that five to 10 samples of bulk densityor 10 to 20 measures of penetration resistance are usually

needed. It might be useful to separate these samples intocore and intermediate locations, as Stohlgren and Parsons(1986) did. They took five to 10 measurements of bothbulk density and penetration resistance from core andintermediate parts of the site, as well as from undisturbedcontrols.

One of the most ecologically significant effects of soilcompaction is a reduction in the rate at which waterinfiltrates soil. As with bulk density, there are standardtechniques available to measure infiltration rates. Unfor-tunately, the most reliable techniques require takingmeasurements on soils at a standard moisture level, usu-ally field capacity. Otherwise, rates will be strongly influ-enced by soil moisture levels. This requirement, alongwith the variability of rates, makes it a time-consumingtask to obtain precise and accurate estimates. Samplesize and distribution would have to be similar to that justdescribed for bulk density.

Measures of soil moisture, organic matter, and chemis-try all require obtaining soil samples that must then beanalyzed in a laboratory. Moisture can be determinedrelatively simply by immediately placing soil samples inairtight containers. Gravimetric moisture is the differ-ence between the weight of the soil sample, before andafter being dried, divided by the dry soil weight. Volumet-ric moisture is the difference in weight divided by thevolume of the soil sample. The other methods are morecomplicated and require more specialized equipment.Again, sample size and distribution should be similar tothat suggested for bulk density.

Although these measures of mineral soil characteristicsprovide relatively accurate estimates of current levels ofimpact, a prohibitively large number of time-consumingsamples must be obtained to identify anything more thanthe most sizable changes over time. Further work on thedevelopment of efficient techniques for monitoringchanges in mineral soil impacts would be worthwhile.

At Grand Canyon, soil compaction is being estimated inthe following categories: “minimal evidence of surfacedisturbance”; "much of surface compacted or loosened, butnot cementlike”; or "most of surface cementlike in appear-ance” (Cole 1985). Sites at Canyonlands (Kitchell andConnor 1984) are evaluated in terms of the proportion ofthe site-from 0 to >60 percent of the site-that has ei-ther compacted fine soils or loosened coarse soils. Theseevaluations are useful in distinguishing between highlycompacted and relatively undisturbed sites. The grosscategories are not likely to be sensitive enough to detectsubtle changes, however. It is also not clear how consis-tent evaluators can be in categorizing campsites.

Erosion- Only a few systems have attempted to assesserosion in a systematic manner. It is not even clear howcommon or severe a problem erosion is, since most camp-sites are located on flat ground. It is particularly difficultto arrive at a quantitative assessment of erosion. Leggand Schneider (1977) used depth to the A2 horizon as ameasure of erosion on campsites. They measured depth tothis distinctive layer at four points, in each of six quad-rats with an Oakfield soil tube. Bratton and others (1978)measured the area on the site with evident erosion. Fewother studies have attempted to do more than state

17

whether or not erosion is evident on the site. This is aparameter that might be assessed with the use ofphotopoints.

Offsite Impacts- It is also possible to assess the se-verity of certain impacts that occur off the main part ofthe campsite. The most common of these are access trailsand the impacts associated with firewood collection, con-finement of packstock, and the loading and unloading ofboats.

Access trails (often called social trails) connect thecampsite to the main trail, water sources, and othercampsites and attractions. The usual procedure for accesstrails has been to count them and note whether or notthey are well developed and/or eroded. In most cases thisinformation is the basis for categorizing sites. Categoriesin the Parsons and MacLeod system (1980), for example,range from none, to two trails discernible, to more thantwo well-developed trails. Problems result when trails areonly discernible at certain times of the year and whendefinitions of the difference between a discernible and awell-developed trail are inconsistent.

Bratton and others (1978) estimated the area disturbedby firewood collection by recording the maximum distancedisturbed by firewood collection from the center of the sitealong at least two axes and then multiplying these meas-ures. They also quantified the reduction in woody fuels,using standard woody fuel inventory techniques on fire-wood collection areas and undisturbed controls (Brattonand others 1982). While providing information on whatimpacts have occurred and how large an area is affected,these techniques are not sufficiently precise to identifysubtle trends.

Shorelines of lakes and streams are disturbed wherevisitors travel by boat. Disturbance occurs when boatsare loaded and unloaded, as well as during recreationalactivities along the shore. Marion (1984) has recorded thelength of shoreline that has been disturbed at campsitesin the Boundary Waters Canoe Area Wilderness. Wherethe extent of disturbance can be consistantly defined (forexample, where vegetation is dense and fragile), thislength can simply be recorded to the closest foot, meter,or other unit of length. Where it is more difficult to definethe extent of disturbance, categories are more appropri-ate. Marion’s categories ranged from <15 feet of distur-bance to >25 feet of disturbance.

Areas disturbed by confined packstock have been in-cluded in estimates of the total area of disturbancearound campsites and in counts of tree damage in theBob Marshall Wilderness (Cole 1983b). Schreiner andMoorhead (1979) specifically counted the number of‘horse trample areas within 100 feet” of the campsite.The value of and problems with these techniques havealready been discussed in the sections on campsite areaand damage to overstory trees.

Summary Ratings- Summary ratings and conditionclass ratings provide a means of summarizing all of theseimpact parameters in a single rating. They can be con-structed, as Frissell(l978) did, by noting the presence orabsence of certain conditions. For example, if vegetationis lost on some of the site, but not most of the site, the site

is given a rating of 2. If tree roots are exposed, but treesare not dead or reduced in vigor, the sits is given a ratingof 4. This is an appropriate procedure and, if the mostimportant impact parameters are included, will providean accurate and precise summary rating of impact levels.Review the discussion of Frissell’s condition class instep 1 for problems with the technique and recommenda-tions for how these problems can be alleviated. A condi-tion class summary rating could be assessed, in additionto estimates or measurements of individual parameters,as an indicator of overall condition.

The summary ratings calculated by summing ordinalratings, with or without determining a mean (Cole 1983a,1984, 1985; Kitchell and Connor 1984; Marion 1984;Parsons and MacLeod 1980), use mathematically im-proper procedures. Categorical ratings cannot be com-bined into a readily interpretable single summary rating(Schuster and Zuuring 1986). Although conclusions aboutthe relative impact level of sites with widely divergentratings are probably valid, conclusions about sites withratings close to each other are suspect. Perhaps the prac-tice of grouping these ratings into four or five categoriesovercomes this problem. Because these systems have notbeen adequately tested, however, we do not know.

Instead of calculating a mean from categorical ratings,classes of sites could be defined in terms of the presenceor absence of certain conditions. For example, a class 5site could be any site that received a rating of 5 for morethan two parameters. A class 1 site could be one that wasrated 1 on at least six parameters, regardless of what theother ratings were. Or a class 1 site could be one thatreceived no ratings higher than 2. Obviously there arecountless ways to define overall ratings. Refer to thesection on calibration in step 3 for description of tech-niques that will facilitate definitions that provide an ade-quate differentiation of site variability.

Research Needs

Evaluation procedures are clearly deficient in severalrespects and could be improved by further research. Morework is needed to increase precision of measurement,particularly for the rapid estimation procedures. Wemust learn to report results in units that reflect the preci-sion of the techniques used. Reporting the area of acampsite to the closest square meter, when it can only bereliably reported to the closest 10 m2, is inappropriate andmisleading. Thus more work must be done on establish-ing precision levels (see step 3).

Of the types of impact that have been assessed, tech-niques for measuring impact to the mineral soil are theleast sensitive, most time consuming, and most difficult tointerpret. Further development of these techniques, withsuggestions particularly for rapid estimation techniques,would be useful.

Finally, more work is needed to develop appropriatesummary impact ratings that provide a meaningful indexof how much impact has occurred. Several options havebeen suggested, but with further thought better tech-niques might emerge. In addition, all of these alterna-tives need to be evaluated.

18

Sources of Information

Bratton, Susan P.; Stromberg, Linda L.; Harmon, Mark E.1982. Firewood-gathering impacts in backcountrycampsites in Great Smoky Mountains National Park.Environmental Management. 6: 63-71. (Describes tech-niques that can be used to evaluate and monitor woodyfuels and firewood-gathering impacts.)

Carothers, Steven W.; Johnson, Robert A; Dolan, Robert.1984. Recreational impacts on Colorado River beachesin Glen Canyon, Arizona. Environmental Management.8: 353-358. (Describes techniques for quantifying sanddiscoloration by charcoal and ash and trashaccumulation.)

Chambers, Jeanne C.; Brown, Ray W. 1983. Methods forvegetation sampling and analysis on revegetated minedlands. Gen. Tech. Rep. INT-151. Ogden, UT: U.S. De-partment of Agriculture, Forest Service, Inter-mountainForest and Range Experiment Station. 57 p. (Provides agood discussion of vegetation sampling techniques,including the calculation of indexes related to change invegetation composition.)

Cole, David N. 1982. Wilderness campsite impacts: effectof amount of use. Res. Pap. INT-284. Ogden, UT: U.S.Department of Agriculture, Forest Service, Intermoun-tain Forest and Range Experiment Station. 34 p.(Describes a variety of impact parameters that aremeasured on permanent plots established on forestedcampsites.)

Cole, David N. 1983. Monitoring the condition of wilder-ness campsites. Res. Pap. INT-302. Ogden, UT: U.S.Department of Agriculture, Forest Service, Intermoun-tain Forest and Range Experiment Station. 10 p.(Describes some techniques for rapidly estimating im-pact parameters.)

Cole, David N. 1984. An inventory of campsites in theFlathead National Forest portion of the Bob MarshallWilderness. Unpublished report on file at: U.S. Depart-ment of Agriculture, Forest Service, Inter-mountainResearch Station, Forestry Sciences Laboratory,Missoula, MT. 19 p. (Describes a monitoring system,based on rapid estimates, that was used in the BobMarshall Wilderness. Includes form, instructions, re-sults of the inventory, and suggestions for furtherwork.)

Cole, David N. 1985. Ecological impacts on backcountrycampsites in Grand Canyon National Park. Unpub-lished report on file at: U.S. Department of the Interior,National Park Service, Western Regional Office, SanFrancisco, CA 96 p. (Describes a variety of impact para-meters that are measured on permanent plots estab-lished on desert campsites.)

Gifford, Gerald F.; Faust, Robert H.; Coltharp, George B.1977. Measuring soil compaction on rangeland. Journalof Range Management. 30: 457-460. (Describes andevaluates several methods of measuring soilcompaction.)

Howard, Richard F.; Singer, Michael J. 1981. Measuringforest soil bulk density using irregular hole, paraffinclod, and air permeability. Forest Science. 27: 316-322.(Describes and evaluates some methods of measuringbulk density.)

Kitchell, Katherine P.; Connor, Jeff. 1984. Canyonlandsand Arches National Parks and Natural Bridges Na-tional Monument draft recreational impact assessmentand monitoring program. Unpublished paper on file at:U.S. Department of the Interior, National Park Service,Moab, UT. 80 p. (Describes rapid estimate monitoringsystems for river, four-wheel-drive, and backpackcampsites.)

LaPage, Wilbur F. 1967. Some observations on camp-ground trampling and ground cover response. Res. Pap.NE-68. Upper Darby, PA: U.S. Department of Agricul-ture, Forest Service, Northeastern Forest ExperimentStation. 11 p. (Describes a technique for monitoringchanges in ground cover vegetation, using permanentquadrats.)

Magill, Arthur W. 1970. Five California campgrounds. . .conditions improve after 5 years’ recreational use. Res.Pap. PSW-62. Berkeley, CA: U.S. Department of Agri-culture, Forest Service, Pacific Southwest Forest andRange Experiment Station. 18 p. (Describes some tech-niques for measuring change in a number of impactparameters on permanent plots.)

Marion, Jeffrey Lawrence. 1984. Ecological changes re-sulting from recreational use: a study of backcountrycampsites in the Boundary Waters Canoe Area Wilder-ness, Minnesota. St. Paul, MN: University of Minne-sota. 279 p. Dissertation. (Describes a variety of impactparameters that are carefully measured, but not onpermanent plots.)

Merriam, L. C., Jr.; Smith, C. K.; Miller, D. E.; [and oth-ers]. 1973. Newly developed campsites in the BoundaryWaters Canoe Area: a study of 5 years’ use. Stn. Bull.511, For. Ser. 14. St. Paul, MN: University of Minne-sota, Agricultural Experiment Station. 27 p. (Describestechniques for measuring change, most notably usingmapping techniques.)

Mueller-Dombois, Dieter; Ellenberg, Heinz. 1974. Aimsand methods of vegetation ecology. New York: JohnWiley. 547 p. (A thorough treatment of measuring vege-tational parameters.)

Parsons, David J.; MacLeod, Susan A. 1980. Measuringimpacts of wilderness use. Parks. 5(3): 8-12. (Describescategories used to rapidly estimate a number of impactparameters.)

Parsons, David J.; Stohlgren, Thomas J. 1987. Impacts ofvisitor use on backcountry campsites in Sequoia andKings Canyon National Parks, California Tech. Rep.No. 25. Davis, CA U.S. Department of the Interior,Cooperative National Park Resources Studies Unit,University of California, Institute of Ecology. 79 p.(Describes and evaluates the monitoring system atSequoia and Kings Canyon National Parks.)

Settergren, C. D.; Cole, D. M. 1970. Recreation effects onsoil and vegetation in the Missouri Ozarks. Journal ofForestry. 68: 231-233. (Describes a number of tech-niques for measuring soil impacts that might be used ina monitoring system.)

Stohlgren, Thomas J.; Parsons, David J. 1986. Vegetationand soil recovery in wilderness campsites closed tovisitor use. Environmental Management. 10: 375-380.(Describes a variety of impact parameters that aremeasured on permanent plots.)

19

Vimmerstedt, John P.; Scoles, Frederick G.; Brown,James H.; Schmittgen, Mark C. 1982. Effects of use pat-tern, cover, soil drainage class, and overwinter changeson rain infiltration on campsites. Journal of Environ-mental Quality. 11: 25-28. (Describes techniques formeasuring soil impacts, some of which might be appro-priate to backcountry monitoring.)

STEP 3. TESTING OF MONITORINGTECHNIQUES

At this stage, impact parameters and procedures forevaluating these parameters have been developed. Al-though it is tempting at this point to rush off into a seasonof monitoring, it is more efficient to invest some time andenergy in testing procedures. This will expose problemsthat are likely to arise before they jeopardize consistentand accurate results. It will also identify subtle modifica-tions that will improve the system. Three different typesof testing can be appropriate here. With the exception ofthose systems based entirely on photographs, all systemswill profit from refinement of field techniques and fromestimation of measurement error. If using a system withcategorical ratings, it is also important to calibrate thesystem (see below for definition of calibration).

Refinement of Techniques

This step is difficult to describe because it can be dealtwith in so many different ways. Basically, it involvestrying out proposed techniques on a few sites to see howwell they work and then coming up with ways to deal withany problems that do arise. A variety of sites differing inenvironmental conditions and levels of impact should beexamined by different people. One common product of thisprocess is a set of quantitative definitions for conditions tobe judged in the field. For example, when measuring dis-tance from the center of the plot to the first significantamount of vegetation, it is necessary to define what a “sig-nificant amount of vegetation” is. Other definitions mightinclude what is or is not a tree mutilation or when a muti-lation becomes “highly obtrusive.” These definitionsshould be developed during the refinement stage. Aftertrying them, some procedures may be determined to be tootime consuming or not sufficiently useful or precise. Theycan be dropped or alternative procedures can be developed.As many of these decisions as possible should be madebefore sending out the field crews.

The end product of this step should be careful descrip-tions of all techniques to be used when evaluating eachimpact parameter, as well as precise definitions of anyterms that might be ambiguous. All of the decisions re-garding techniques and definitions need to be carefullydocumented (see step 4).

Calibration Procedures

Many monitoring systems utilize categorical ratings foreach impact parameter. These can be quantitatively de-fined classes, such as for campsite area: 1 = <500 ft2

(<46 m2); 2 = 500-1,000 ft2 (47-186 m2); and 3 = >1,000 ft2

(>186 m2). They can also be nonquantitative. For ex-

20

ample, development classes could be: 1 = no development;2 = a simple rock firering; and 3 = more than one fireringor other developments.

To be most efficient at differentiating between sites onthe basis of amount of impact, approximately equal per-centages of sites should fall into each category. Marion(1986) describes a method for calibrating categories thatare quantitatively defined. First, inventory a sample ofsites (perhaps 50) that represent the full range of use-,environmental-, and impact-related conditions present inthe area for which the system is being designed. This canalso aid in the refinement of techniques (as describedabove).

Second, display these results as a cumulative frequencydistribution and then divide this distribution into classeswith an equal number of sites. For example, if three cate-gories were to be defined and 33 percent of the campsiteswere <600 ft2 in area, the first category should be 0-600 ft2.The upper bound of the second category would be the areaof the site at the 67th percentile. Although these are themost equitable boundaries, it is worth rounding categoriesso that they are easier to remember and work with. Forexample, if the area of the campsite at the 67th percentilewas 1,061 ft2, it would be a good idea to set the boundariesat either 1,000 or 1,100 ft2.

If categories are not quantitatively defined, calibration ismore difficult and, in some situations, may not be possible.If the majority of sites have no development, for example,there is no way to have an equal distribution among cate-gories for the development parameter. With such parame-ters, try several different sets of categories and select theset that provides the most equitable distribution of sitesamong categories.

Estimation of Measurement Error

Error is inherent to all measurement systems. Formonitoring purposes, we need to be able to distinguish,when comparing measures of the same campsite at twodifferent times, between two different measures of thesame condition (measures that differ due to the biases andinterpretations of two different evaluators) and a realchange in conditions. To do this we need to (1) developtechniques with as little inherent measurement error aspossible, (2) do everything possible-through training anddocumentation-to minimize measurement error, and (3)determine the magnitude of measurement error. While wehave made considerable progress on the first two needs,estimates of the measurement error inherent to any moni-toring system have never been provided. This may be themost critical missing link in our ability to accurately moni-tor changes in campsite condition. At this stage, we havelittle ability to evaluate the probability that an observedchange is either a real change or simply a result of meas-urement error.

The details of how to evaluate measurement error haveyet to be developed. In a report prepared in conjunctionwith this project, Steele (1987) suggests some methods forevaluating and displaying measurement error. Becausemonitoring data consists of only one observation per timeperiod, there is no opportunity to evaluate, statistically,variation and error for actual monitoring data. Therefore,

it will be necessary to conduct separate studies designedspecifically to estimate error. If a number of differentpeople independently evaluate conditions on the same site,these evaluations can be statistically manipulated to esti-mate the error associated with different parameters. Theappropriate number of different people and different num-ber of sites has not been determined. At a minimum, thereshould probably be five to 10 people evaluating 10 to 20representative campsites.

In order to decide on appropriate statistical techniques,the distribution of each variable must be determined.Many tests are only appropriate if the distribution of avariable either approximates a normal distribution or canbe transformed so that it approximates a normal distribu-tion. Many of the count and class variables clearly are notnormally distributed. Some may follow a Poisson distribu-tion; for others it is not clear exactly what statistical teststo apply.

For those variables that are normally distributed, therelative sensitivity of different parameters can be evalu-ated by examining their coefficient of variation (the ratioof the sample standard deviation to the sample mean,expressed as a percentage). A large coefficient indicatesthat the variability of independent estimates is high, thesensitivity of the variable is low, and measurement erroris high. For variables with a large coefficient of variation,the difference between two observations must be relativelylarge before it is safe to conclude that a real change hasoccurred. For such variables, significant deterioration islikely to occur before one can be confident that a realchange has occurred.

Quantitative estimates of measurement error for specificparameters can be approached in several ways. Oneoption is to determine the minimum change that, if ob-served, will permit one to conclude, confidently, that achange has actually occurred. This requires the specifica-tion of a confidence level, the risk one is willing to assumeof stating that a change has occurred when it has not. Bysetting this level (type I error), minimum change can bedetermined through use of the two-sample t statistic(Steele 1987).

It is also worth evaluating the likelihood of not detect-ing a change that has actually occurred. This risk is calledtype II error. Power curves can be constructed (Steele1987) that permit type II error to be determined for anycombination of confidence level (type I error) and magni-tude of real change. For example, given that a real changeof 10 percent actually occurs, and one is willing to accept al-in-10 chance of incorrectly concluding that a change hasoccurred when it has not, the likelihood of correctly detect-ing that 10 percent change is the power ([1 - type II error]* 100 percent), as read from power curves.

The utility of the coefficient of variation, the two-samplet statistic, and power curves can be illustrated with ex-amples from a small sample of campsites taken nearMissoula, MT. Five separate sites were independentlyevaluated by nine graduate students from a University ofMontana recreation management class. The parametersused were generally those of Cole’s rapid estimation proce-dure used in the Bob Marshall (see appendix H); a few ad-ditional parameters were also added. Values may besomewhat unrepresentative because the observers

received less training than normal, not all variables metthe assumptions of a normal distribution, and the numberof sites examined was probably insufficient. Neverthe-less, these examples illustrate some of the ways in whichmeasurement error could be examined; they also suggestwhich variables are most sensitive and the magnitude ofmeasurement problems.

Coefficients of variation (table 2) indicate that all of therating variables, except for trash, are more sensitive thanthose that require a numerical estimate. Two of the threeleast sensitive variables were ones not used in the BobMarshall-the trash rating used at Canyonlands (Kitchelland Connor 1984; appendix I) and the measure of baremineral soil area used at the Delaware Water Gap (ap-pendix J). Impact index proved to be quite sensitive; thestandard deviation was typically only 7 percent of themean index. It should be possible to confidently conclude,then, even for relatively small increases in this index,that an increase represents a true increase-rather thansimply random variation between observers. In contrast,observed increases in the number of trees with exposedroots and the area of bare mineral soil must be large be-fore it is possible to conclude, with much confidence, thatthe increase is a true increase.

A better idea of how large changes must be before onecan conclude, with confidence, that a change has occurredcan be derived from table 3. This table shows the mini-mum magnitude of change that can be detected, giventype I error rates of 0.05 and 0.25; these correspond tol-in-20 and l-in-4 chances of incorrectly stating that achange has occurred when there has been no change.Expressing this “minimum detectable change" as a per-centage of mean observations (the values in parenthesesin table 3) provides another measure of relative sensitiv-ity; the order of parameters in table 3 is roughly compa-rable to that in table 2. A value of 50 percent suggeststhat a future observation must be at least 50 percent

Table 2-Coefficient of variation for monitoring parameters, usingdata collected on five sites by nine recreation manage-ment graduate students

Coefficient of variationParameter Median Range

Camp area (rating) 0 0 - 21Vegetation loss (rating) 0 0 - 36Root exposure (rating) 0 0 - 55Impact index 7 4 - 8Development (rating) 12 0 - 20Barren area (rating) 12 0 - 41Tree damage (rating) 19 0 - 21Mineral soil increase (rating) 20 12 - 38Cleanliness (rating) 22 0 - 36Social trails (rating) 27 0 - 38Camp area (ft2) 31 26 - 34Social trails (number) 31 16 - 47

Fire scars (number) 31 0 - 110Barren area (ft2) 31 27 - 106Tree damage (number) 37 24 - 62Trash (rating) 38 19 - 42Root exposure (number) 44 24 - 163Bare mineral soil area (ft2) 53 31 - 102

21

Table 3-The minimum amount of change that, if observed, canconfidently be considered a “real” change, based on datacollected on five sites by nine recreation managementgraduate students

Parameter

Minimum change1

0.05 2 0.252

Camp area (rating) 0.3 (13) 0.1 (4)impact index 3.3 (15) 1.3 (6)Development (rating) 0.5 (21) 0.2 (8)Root exposure (rating) 0.5 (23) 0.2 (9)Vegetation loss (rating) 0.7 (26) 0.3 (11)Tree damage (rating) 0.9 (34) 0.4 (15)Barren area (rating) 0.9 (36) 0.4 (16)Cleanliness (rating) 0.8 (44) 0.3 (17)Social trails (rating) 1.2 (47) 0.5 (19)Mineral soil increase (rating) 0.1 (52) 0.4 (20)Social trails (number) 3.5 (72) 1.4 (29)Camp area (ft2) 1,556 (72) 628 (29)Fire scars (number) 1.6 (78) 0.6 (30)Root exposure (number) 4 (79) 2 (30)Trash (rating) 1.8 (80) 0.7 (31)Barren area (ft2) 520 (87) 210 (35)Tree damage (number) 12 (92) 5 (38)Bare mineral soil area (ftz) 295 (104) 119 (42)

‘Minimum change is the minimum difference between observations, takenat two different times, that would allow the null hypothesis, of no difference, tobe rejected. The two-sample t statistic, with a pooled estimate of the standarddeviation and 40 degrees of freedom, was used. Values in parenthesesexpress this minimum change as a percentage of mean values and providea measure of sensitivity.

Type I error rates of 0.05 and 0.25 are reported, providing confidencelevels of 95 percent and 75 percent, respectively.

Table 4-The likelihood of correctly detecting a real change of anincrease in rating of 1 for the rating parameters, based ondata collected on five sites by nine recreation manage-ment graduate students

Parameter

Chance of detection1

0.052 0.252

Camp areaDevelopmentRoot exposureVegetation lossCleanlinessTree damageBarren areaMineral soil increaseSocial trailsTrash

100 10090 10090 10075 9565 9060 8555 8550 8040 7520 60

‘Chance of detection is 1 minus the type II error rate, expressed as apercentage, as derived from power curves in which both type I error rate anda given magnitude of change are set. For example, a 90 percent chanceimplies (given that a shift in rating of one has actually occurred) that there isa 90 percent chance of correctly rejecting the null hypothesis, that no changehas occurred.

2Type I error rates of 0.05 and 0.25 are reported. providing confidencelevels of 95 percent and 75 percent, respectively. Using tree damage as anexample, if you are willing to accept a one-in-four chance of saying there hasbeen a change when none has occurred (type I error of 0.25) you have agreater than four-in-five chance of correctly identifying a shift of one; if you areonly willing to accept a l-in-20 chance of saying that there has been a changewhen none has occurred, the odds of correctly identifying a shift of one drop 10a greater than 6-in-10 chance.

larger than the estimate of current condition before it issafe to conclude that a change has occurred. Again it isclear that all of the parameter ratings, except for trash,are quite precise. If an increase in rating of 1 is observed,there is very little chance of concluding incorrectly that anincrease has occurred when it has not. Only for socialtrails and mineral soil increase is there a l-in-20 chanceof making this mistake. Impact index is also relativelyprecise; it is highly unlikely that an observed change ofmore than two or three units (on the scale of 9 to 27) doesnot reflect a real change. In contrast, one out of everyfour times that a change in bare mineral soil area as largeas 119 ft2 (11 m2) is reported, there is likely to have beenno real increase.

The minimum detectable change in vegetation cover(for a type I error of 0.05) was 25 percent on campsitesand 17 percent on controls. The corresponding minimumchanges in mineral soil were 26 percent and 5 percent oncampsites and controls. This suggests that the 25 percentcover classes used in the Bob Marshall (appendix H) areabout right, although it might be desirable to divide the76-100 percent class into 76-95 percent and 96-100 per-cent classes to accommodate the greater sensitivity ofmineral soil estimates on controls.

Although knowledge about the minimum detectablechange is critical, it is also enlightening to examine thelikelihood of not detecting a real change in conditions.Tables 4 and 5 show the likelihood of correctly identifyinga real increase in rating of 1 (for the rating parameters)and a real 25 percent increase in deterioration (for theother parameters), respectively. A rating shift of 1 shouldusually be detected without having to accept too muchrisk of making a type I error. In contrast, for none of theparameters that require a count or estimate is there morethan a 50 percent chance of detecting a 25 percent in-crease in deterioration, even accepting a l-in-4 risk ofsaying there has been a change when none has occurred.

In conclusion, we have only begun to investigate thedifficult and complex issue of measurement error. Theresults reported here should be treated as merely

Table 5-The likelihood of correctly detecting a real 25 percentincrease in deterioration, based on data collected on fivesites by nine recreation management graduate students

Chance of detection1

Parameter 0.05 2 0.252

Impact index 85 100Camp area (ftz) 20 50Tree damage (number) 20 50Fire scars (number) 20 50Barren area (ft2) 15 45Social trails (number) 15 45Root exposure (number) 15 45Bare mineral soil area (ft2) 15 45

‘Chance of detection is 1 minus the type II error rate. expressed as apercentage, as derived from power curves in which both type I error rate anda given magnitude of change are set.

2Type I error rates of 0.05 and 0.25 are reported, providing confidencelevels of 95 percent and 75 percent. respectively.

22

suggestive of the precision levels of the various estimationtechniques used. They demonstrate that most of theinterval scale measures are highly imprecise; they aremuch less sensitive than they appear. For a variable suchas camp area (which is typical of these parameters), areamust virtually double before it is safe to conclude that areal change has occurred. Consequently, there is a highprobability of either stating a change has occurred whenit has not or failing to detect even sizable changes. Rat-ings, while they provide less information, are less mis-leading. When a shift in rating occurs, it is likely to bedetected; conversely, when a shift in rating is observed, itis likely to reflect a real change. Finally, the impact index(used to summarize overall impact) appears to also besensitive and relatively precise. One can be quite confi-dent of detecting changes as small as 10 to 15 percent andconfident that changes of 10 to 15 percent reflect actualchanges.

Research Needs

More research is needed on measurement error. Morestudies with larger sample sizes need to be conducted fol-lowing the format of the study reported here. From theseit should be possible to determine appropriate samplesizes for such studies, as well as the approximate distribu-tions for different parameters. For those parameters withdistributions that are not normally distributed, particu-larly class variables, it will be necessary to find tests com-parable to those available for parameters that are nor-mally distributed.

Once appropriate distributions are determined, itshould be possible to more accurately determine the mag-nitude of measurement error for different parameters.Through research it should be possible to identify thoseparameters that are most sensitive. It might also be pos-sible to suggest ways to increase the sensitivity of para-meters with large errors. Ultimately, managers of indi-vidual areas will have to utilize baseline studies and sta-tistical procedures to determine appropriate error termsfor the procedures they adopt. These errors can be usedto decide how large a change must be before it will be con-sidered a real change.

Sources of Information

Marion, Jeffrey L. 1986. Campsite assessment systems:application, evaluation, and development. In: Popadic,Joseph S.; [and others], eds. Proceedings, 1984 riverrecreation symposium; 1984 October 31-November 3;Baton Rouge, LA. Baton Rouge, LA: Louisiana StateUniversity, School of Landscape Architecture: 561-573.(Describes a procedure for calibrating categorical im-pact rating systems.)

Steele, Brian. 1987. Statistical procedures for the analysisof a campsite monitoring program. Unpublished reporton file at: Systems for Environmental Management,Missoula, MT. 56 p. (Suggests statistical procedures forevaluating measurement error. Provides examplesfrom a sample of five campsites that were evaluated in-dependently by nine people.)

STEP 4. DOCUMENTATION ANDTRAINING

Although it is possible at this stage to plunge into theinventory and monitoring fieldwork, it is important toinvest time in training and documentation of methods.If this is not done, consistency and precision will be low;this will reduce the value of the data collected.

The purpose of this step is to minimize the errors asso-ciated with different people taking measurements andmaking judgments. Two sources of error are common tothese techniques. The first results from problems of defi-nition. For example, when measuring campsite area orcounting tree damage, estimates will be highly divergentif evaluators have very different opinions about how todefine the campsite boundary or what constitutes a “dam-aged tree.” Problems of definition can be just as seriouswhen using precise measurements in permanent plots asthey are when making rapid estimates. Precise defini-tions must be worked out in the field, documented in somemanner, and then communicated to evaluators throughtraining. Separate evaluators should be periodicallybrought together to be recalibrated-to make certain thatdefinitions and judgments remain consistent.

The second source of error is in measurement tech-nique. This error is likely to be more substantial whenusing rapid estimation techniques. In measuring camp-site area, for example, it is difficult to measure the area ofan irregular figure. If one evaluator estimates area onthe basis of radial measures of the distance between cen-ter point and boundary, while another pieces together theareas of several simple geometric figures, results arelikely to be very different. Even when counting damagedtrees, estimates will vary depending on whether or notonly onsite trees are counted. Measurement/estimationtechniques need to be agreed on and used in a consistentmanner. The magnitude of the measurement error willvary with the measurement technique used. This pro-vides another reason for using consistent techniques.

Documentation

Once precise evaluation procedures and definitionshave been established, the consistency of their applicationmust be maintained. This can be particularly difficultwhen field crews and even supervisors change from yearto year. An important tool for dealing with the problem ofturnover is the preparation of an impact monitoring sys-tem manual that documents techniques and definitions.This tool will also increase year-to-year consistency inplaces that do not experience turnover. Without such amanual it is doubtful that anyone monitoring sites, say,20 years from now, will be able to use the date being col-lected today.

Several types of information should be included in sucha manual. Much of the manual will consist of step-by-stepdescriptions of how each impact parameter should beevaluated. These should be described in as much detail aspossible, in simple language. If measurement instru-ments are needed, these should be listed, described, andperhaps even photographed. The more detail, the better.

23

Definitions of ambiguous terms, such as what consti-tutes a "damaged" tree or “highly obtrusive” damage, are acritical part of the manual. Definitions should be quantita-tive where possible. Particularly where quantification isnot possible, photographs of the conditions being describedwill contribute to consistent judgments in the field. Forexample, photographs of a variety of trees with “highlyobtrusive” damage, damage that is not “highly obtrusive,”and no damage at all would help greatly where these termsmust be used. Other examples might include the differ-ence between “some” and "much" litter, or the differencebetween a “discernible” and a "well-worn" access or socialtrail.

Often it is important to document things that will or willnot be included in an estimate. In some places, measuresof “barren area” have been confined to the devegetatedarea around the central core of the site; in other places,several devegetated areas on the site have been measuredand summed. When counting access trails for campsiteslocated at a lake, for example, decisions must be madeabout whether or not to count a fisherman’s trail aroundthe lake that happens to run through the site. Logicallysuch trails might be excluded if they would have beenthere regardless of the campsite’s existence. It is alwayshelpful to explain the rationale behind such decisions.

Helpful “pointers” and “rules of thumb” are also useful.Examples include how to decide whether or not a siteshould be considered a campsite and how to ‘split up” anarea of intermingled sites into separate sites for inventorypurposes. Shortcuts and recommendations for how tospeed up estimates are also useful.

The manual should be updated when new suggestionsand improvements are developed. Field workers shouldcarefully document situations that are not clearly ad-dressed in the manual and suggest means of dealing withthese new situations. These can then be discussed with asupervisor who can evaluate the situation and suggestchanges, as well as the need for revision of the manual.Keeping a copy of the manual on a word processor shouldmake the process of manual revision simpler.

Whenever changes in procedure or definition are made,it is important to evaluate how such changes will affectcomparability with data already collected. If comparabilitywill be lost, a decision on whether or not to make a changemust be dependent on a weighing of the advantages of theimproved method and the disadvantages of lost informa-tion. While one should not be afraid to lose the informa-tion contained in previous measurements, this loss shouldnot be accepted unless improvements are substantial. Ifcomparability is lost, this should be stated in the manual,along with any suggestions for how previous measure-ments might be interpreted in relation to new methods.

Training

There are many useful ways to conduct training, butseveral guidelines can be suggested. Evaluators can studythe documentation manual independently, but they shouldbe trained as a group. Definitions and procedures shouldbe discussed and demonstrated in the field. Examples of

the various situations that require different judgmentsshould be viewed and discussed as a group. Then evaluat-ors should each work independently on a series of sites.Results should be compared and discussed. This processmust be repeated until an acceptable level of consistency isreached. If one or several evaluators consistently overesti-mate or underestimate a parameter, they should be in-structed to compensate their judgments in such a way thatthey are calibrated with the group as a whole. Repetitionmust continue until this compensation leads to consistentresults-within an acceptable measurement error.

Periodic reevaluation of the consistency of judgmentsthroughout the field season will also increase precision. Ifevaluators can be reconvened for a few hours every monthor so, internal consistency can be examined and any prob-lems or suggestions for improvement can be discussed.

Where possible, the same supervisors should do thetraining each year. Where this is not possible, the newsupervisor should be trained, in detail, by the previous one.This is critical if consistent calibration of field workers fromyear to year is to be maintained. The manual will help inthis regard, as would the sharing of training responsibili-ties-so it is less likely that all supervisors will leave inany one year.

STEP 5. FIELD DATA COLLECTIONPROCEDURES

To increase efficiency in the field and accuracy in thereporting of data, it is important to develop efficient datacollection procedures.

Data FormsCarefully constructed data forms can make data collec-

tion much simpler. Spaces for recording informationshould be arranged in the order in which data will be col-lected. Otherwise evaluators may have to flip back andforth between pages. Sometimes it is helpful to have onepage on which data is recorded and a separate page withdetails on methods, judgments that must be made, andcategory descriptions. Forms and shorthand codes shouldalways be standardized so that problems in interpretingdata are minimized. If precipitation occurs frequentlyduring the data collection season, it may be necessary toprint forms on waterproof paper and use pens that canwrite on wet paper.

The Code-A-Site system (Hendee and others 1976) usededge-punched cards in the field. This permitted the use ofneedle-sorting methods for sorting and retrieving data,computers were not needed. Needle-sorting proved to becumbersome and the raw data, once retrieved, often had tobe tabulated and summarized manually. Recent increasesin the accessibility of electronic data processing capabilitieshave rendered this approach virtually obsolete.

The process of taking data off a form and entering it intoa computer is time consuming and subject to error. Prob-lems resulting from this translation can be reduced byusing standard preceded forms with data recorded spatiallyon the form in such a way that they correspond to the

24

columns and data fields established in computer files.It is important, however, for the form to be easily inter-pretable in the field. If actions taken to reduce errors indata entry result in more errors in data collection, nothingis gained. Where possible, data manipulation and trans-formation prior to data entry should be avoided. Most ofthis manipulation can be accomplished through computerprogramming.

Electronic Field Data RecordersRecent technology makes it feasible to carry program-

mable, battery-operated, hand-held microcomputers intothe field. Data can be entered directly, eliminating theneed for forms entirely. Prompts, such as "how manydamaged trees are there?” and “are any of them highlyobtrusive?” can be programmed into these devices. Illogi-cal answers can be flagged as probable errors. Althoughthese devices possess data processing capabilities, theirprimary function is field data entry and temporary stor-age. After leaving the field, collected data are downloadedinto nonportable computers.

Currently (1988), such devices cost between $500 and$1,000. Depending on one’s budget and the number ofworkers that need one, this cost may or may not be pro-hibitive. Cost is likely to decline some in the future. Im-portant criteria for a system include durability, weight andcompactness, battery life between charges, screen size andlegibility, data storage capacity, and the type of operatingsystem. This latter criterion is important because certainoperating systems provide greater flexibility in interfacingwith nonportable computers and are more user friendly.

Research NeedsFurther work in the development of portable data re-

corders and software to facilitate data collection would beworthwhile.

Sources of InformationKrumpe, Edwin E. [Personal communication.] Department

of Wildland Recreation Management, University ofIdaho, Moscow ID. (Has developed a portable data re-corder for field collection of monitoring data.)

Sydoriak, Charisse A. [In press]. Yosemite’s wildernesstrail and campsite impacts monitoring system. Paperpresented at the National Park Service Science Confer-ence; 1986 July 13-18; Fort Collins, CO. (Mentions theportable data recorders used in the field in Yosemite.More information is available in appendix K or by di-rectly contacting the Resources Management Division atYosemite National Park.)

STEP 6. DATA ANALYSIS ANDDISPLAY

Once monitoring data has been collected, it must beanalyzed and displayed. The detail and sophisticationused at this step can be highly variable. Data forms canbe examined in a cursory manner; they can be needle-sorted (if Code-A-Site forms are used); or data can beentered into a computer. In most cases data should besummarized in statistics, graphs, and maps. Althoughthe level of analysis is likely to vary with management ob-jectives and analytical capabilities, both current campsiteconditions and trends in condition should be examined.Some suggested analysis procedures, organized underthese two headings, follow. Any of these analyses thatappear unnecessary can simply be ignored. A discussionof the use of computers and software to facilitate analysisis included in a concluding section.

Analysis of the Current SituationA variety of analyses can be conducted to evaluate the

current condition of campsites. Several of the more im-portant types are as follows:

1. It is useful to be able to retrieve data for any individ-ual site of interest, or for all sites in a destination area ormanagement area. For example, table 6 lists the size ofall campsites on Minisink Island on the Delaware River.The ability to retrieve data for individual sites easilyfacilitates planning for management of both individualsites and larger areas, such as Minisink Island. Wherethere are a large number of sites, however, analysis atthis level of detail is cumbersome.

2. A simple type of analysis, at a more general level, isto calculate summary descriptive statistics for all camp-sites in the entire wilderness, in individual managementunits, or in destination areas within the wilderness.

Table 6-Camp area (m2) for each of the 10 campsites on MinisinkIsland at Delaware Water Gap National Recreation Area

Campsite number Campsite location Camp area

20 Minisink23 Minisink24 Minisink26 Minisink28 Minisink29 Minisink30 Minisink31 Minisink32 Minisink33 Minisink

m 2

7984

22118

507241

72392

72108

25

Separate summary statistics can be calculated for eachindividual impact parameter and a summary rating, ifone was used. An example would be the median andrange for the number of damaged trees on campsites inthe entire area or around a certain lake of concern.Medians are often more appropriate measures of centraltendency than means because they are not skewed byextreme values.

Summary statistics can be used to assess impact levels,both in the entire area and in portions of the area. As anexample, refer to some output from data collected oncampsites in the Delaware Water Gap National Recrea-tion Area. Summary statistics for two river segments arecompared in table 7. Both an idea of impact levels anddifferences in impact between segments are revealed.The only pronounced difference between segments is thatcampsites along the stretch from Bushkill to Smithfieldtend to be larger. Campsites on this segment have lostmore vegetation, but the extent of shoreline damage isl e s s .

The number of campsites in the wilderness provides animportant indication of impact. The number and percent-age of all sites for each management area or destinationarea can be displayed. See, for example, table 8. Theriver segment with the most campsites is Bushkill toSmithfield. Illegal sites, however, are much more com-mon on the Milford to Dingmans and Dingmans toBushkill segments. This suggests that more legally desig-nated sites might be needed between Milford andBushkill.

Further insights into impact levels can be gained bydividing the range of impact into categories and display-ing the number and percentage of sites in various catego-ries. In the example in table 9, almost half of the camp-sites between Milford and Dingmans are in the smallestsize class- <1,076 ft2 (cl00 m2). Successively smallerproportions are found in the larger size classes. Camp-sites between Bushkill and Smithfield were also skewedtoward the smaller size classes, but not as dramatically.Only 40 percent of sites were in the smallest class andmore than 10 percent were in the largest class.

Table 10 shows the number of campsites in each of fivecondition classes for different destination areas inSequoia and Kings Canyon National Parks. Differencesin numbers of campsites are readily apparent. Convert-ing the data into percentages and displaying them in his-tograms makes differences in the relative frequency ofcondition classes more apparent (fig. 1). The Dusy Basinarea has a very large number of campsites, but few of thesites are severely impacted. At Bubbs Creek, there arefewer sites but a large proportion of the sites are severelyimpacted. This type of analysis is most useful for compar-ing levels of impact in different areas.

At Sequoia and Kings Canyon National Parks, overallimpact ratings have been calculated for entire destinationareas (Parsons and Stohlgren 1987). Although individualsites are given ratings between 1 and 5, it is clear thatimpacts on class 5 sites are more than five fold those onclass 1 sites. The ratings 1 through 5 were replaced byweights based on campsite area. For example, a site with

Table 7--Campsite conditions on the Milford to Dingmans (M-D) and Bushkill to Smith-field (B-S) river segments at Delaware Water Gap Recreation Area

River segment

M-D B-SMedian Range Median Range

Camp area (m2)Bare mineral soil (mz)Damaged trees (number)Tree stumps (number)Shoreline disturbance (m)Trees with exposed roots (number)Firerings (number)Vegetation loss (percent)

202 18 - 77588 0 - 483

3 0 - 171 0 - 5

12 0 - 571 0 - 61 0 - 6

35 -25 - 75

286 10 - 3,07195 0 - 1,042

4 0 - 162 0 - 357 0 - 361 0 - 81 0 - 4

40 -25 - 75

Table 8-The number of legal and illegal campsites on different river segments at Delaware Water Gap NationalRecreation Area

River segment

Legal sites

Number Percent

Illegal sites

Number Percent

Total sites

Number Percent

Above Milford 5 2.8Milford to Dingmans 20 11.2Dingmans to Bushkill 29 16.2Bushkill to Smithfield 58 32.4Below Smithfield 4 2.2

Totals 116 64.8

9 5.0 14 7.816 8.9 36 20.120 11.2 49 27.4

7 3.9 65 36.311 6.1 15 8.463 35.2 179 100.0

26

Table 9-Frequency distribution, by campsite area category, forcampsites on the Milford to Dingmans (M-D) and Bushkillto Smithfield (B-S) river segments at Delaware WaterGap National Recreation Area

Campsitearea (m2)

River segment

M-D B-SNumber Percent Number Percent

0- 100 17 47 26 40101 - 300 10 28 22 34301 - 600 5 14 7 11601 - 900 4 11 3 5

>900 0 0 7 11

Total 36 100 65 100

Table 10--Number of campsites, by condition class, in destinationareas in Sequoia and Kings Canyon National Parks(Parsons and Stohlgren 1987)

Condition class TotalDestination area 1 2 3 4 5 sites

Goddard CanyonMcClure Mea&wlonian BasinCartridge CreekRae LakesHamilton LakesHockett MeadowsDusy BasinBubbs Creek

- - - - -

37853559

1163

10216830

- Number of sites - - - - --33 33 19 1587 51 23 1922 13 0 435 4 0 0

106 63 31 53 3 14 2

87 58 20 2394 28 2 024 31 20 13

137265

7498

32325

290292118

an area rating of 5 is, on average, 150 times larger than asite with an area rating of 1. Assuming, then, that totalarea is the most appropriate indicator of total impact, andthat the total area rating will probably be the same as theentire campsite class rating, weights for class 1 through 5sites are 1,6,30,75, and 150. To determine the total im-pact of each destination area, the number of campsites ineach class is multiplied by these weights; then these prod-ucts are summed to get the total weighted value. Figure 2shows the campsites at Lake Reflection, their campsiteclasses, and the calculation of the total weighted value forthe lake.

The total weighted value and the weighted value/siteallow comparisons between different destination areas.The total value provides a perspective on aggregate im-pacts in an area. Destination areas vary greatly in size,however, so areas with a larger total may not necessarilyhave more impact per unit area. The total value/site prov-ides a perspective on how impacted the average site is.

The problem with this procedure is in the selection ofweights. Although basing weights on total camp area isprobably as defensible as any other single criterion, assign-ing interval values, after the fact, to ordinal rankings isinevitably suspect. Is an area with one class 5 site really

( a ) ( b )

1 2 3 4 5 1 2 3 4 5

C O N D I T I O N C L A S S

Figure 1-Percentage of total sites in each conditionclass, for the (a) Dusy Basin and (b) Bubbs Creekdestination areas in Kings Canyon National Park.

CampsiteClass

No.sites

WeightingFactor

W eightedValue

1 6 X 1 = 62 6 X 6 = 363 3 Y 30 = 904 5 X 75 = 3755 3 X 150 = 450

T o t a l W e i g h t e d V a l u e = 957

Figure 2-Distribution of campsites and conditionclasses at Lake Reflection in Kings Canyon NationalPark. Calculation of a weighted value for the destina-tion area is illustrated.

27

Table 11 -Rank ordering and percentiles, accord-ing to size, for campsites on Minisink

Table 12-The 20 destination areas, in Sequoia and Kings CanyonNational Parks, with the most campsites within 25 ft

Island (7.6 m) of water

Site number Size Percentile Area number Area name Number of sites

m2

120 507 85116 392 76119 241 62114 180 58121 152 49115 108 43110 84 38109 79 33117 72 31117 72 31112 50 25111 45 20113 18 7

comparable to an area with five class 3 sites? Theseweighted values contain no inherent “truth”; they are theproduct of mathematically inappropriate procedures and alarge number of subjective judgments. This should neverbe forgotten, despite the seductive apparent objectivity ofthe numbers produced. But if field examinations of anumber of areas for which total values have been calcu-lated suggest that the numbers generated do make sense(that areas with comparable total values appear to havecomparable impact levels), these values can be a valuablemanagement tool.

9301 Mosquito Lakes 486501 Vidette Meadow 374602 JMT - S. Fork Kings 379303 Eagle Lake 366404 Kearsage 1 and 2 365701 Woods Lake 324202 Middle Dusy Basin 276004 Gardiner Pass Lakes 256503 JMT - Bubbs 243305 Colby Meadow 245202 Volcanic Lakes 234203 11393 Lakes 228902 Columbine Lake 228909 Upper Rattlesnake Creek 223303 Evolution Meadow 225403 Lower Granite Lakes 226002 Gardiner Basin 229202 Monarch Lakes 214502 Palisade Basin 218804 Big Five Lakes 21

3. For a perspective on impact levels for individual sitesit can be useful to rank-order sites according to their im-pact level. Table 11, for example, rank-orders campsiteson Minisink Island according to their size. This is an easymeans of distinguishing between more heavily and lightlyimpacted sites. Sites can be rank-ordered within destina-tion areas, larger management units, or the entirewilderness.

conditions and standards. It would be particularly helpfulto be able to “flag” sites or larger areas (depending on howstandards are written-for individual sites or larger ar-eas) that either exceed standards or are close to stan-dards. This might amount to simply a list of sites or ar-eas where either of these conditions applies. Where stan-dards are exceeded, increased management is immedi-ately necessary. Where conditions are close to standards,management should be stepped up as soon as possible.

Assigning each site a percentile rating can help furtherto establish relative impact levels for each site. Under thecolumn labeled “percentile” in table 11, values can rangefrom 1 to 100 percent. A site in the first percentile is inthe smallest 1 percent of sites in the area; sites in the100th percentile are among the largest 1 percent of sites.A value of 70 percent indicates that 70 percent of sites aresmaller. To evaluate impact levels for different destina-tion or management areas, the number and proportion ofsites in categories based on percentiles (for example 20 to40 percent and 40 to 60 percent) can be calculated. OnMinisink Island, for example, only 30 percent of the sitesexceed the 50th percentile (for the entire Delaware Rivercorridor) for size; some other destinations along the riverhave a much higher proportion of large sites.

A similar approach can be taken for flagging any othermanagement situation of concern. For example, table 12lists the management areas in Sequoia and Kings CanyonNational Parks with the most campsites within 25 ft (8 m)of water. This could be obtained by flagging managementareas with more than 20 sites within 25 ft (8 m). Or itcould be obtained by rank ordering management areas ac-cording to this variable. Because sites within 25 ft (8 m)of water are targeted for rehabilitation, such a list estab-lishes priorities for such projects.

4. If standards have been established stating maximumlevels of impact to be tolerated on campsites, it is impor-tant to be able to assess the relationship between current

5. Most of these data will need to be mapped at somepoint to better understand spatial relationships. Some ofthe maps that would be useful include maps of all sites invarious classes, such as all class 5 sites, maps of all sitesthat exceed some level, such as a size of 300 m2, and allsites that exceed standards. Figure 3 shows a map ofcampsites and impact levels in a portion of the BobMarshall Wilderness. Very different management ap-proaches will be needed at George Lake (characterized bya large number of lightly impacted sites), Koessler Lake(with only one site, which is severely impacted), andUpper Holland Lake (characterized by many highlyimpacted sites).

28

I M P A C T I N D E X

L l G H T ( 2 0 - 2 9 )

M O D E R A T E ( 3 0 - 4 0 )

H E A V Y ( 4 1 - 5 0 )

S E V E R E ( 5 1 - 6 0 )

1N

Figure 3-Campsite location of amount and impact in a portion of the Bob Marshall Wilderness.appendix H for a discussion of the impact index.

Refer to

29

The ability to do mapping automatically, using a com-puter, makes this process much simpler than if data needto be mapped by hand. The data from the inventory ofcampsites in the Bob Marshall, for example, are beingintegrated into a Geographic Information System beingdeveloped for the area. This will greatly facilitate displayand analysis of the data base.

Analysis of Trends

Experience with analyzing data on trends in campsiteconditions is more limited. As with analysis of the cur-rent situation, it can be useful to be able to recover dataon individual sites, to provide summary statistics, torank-order sites, to flag certain situations, and to havemapping capabilities. The major difference is that theanalysis involves comparison of observations taken atmore than one time.

1. The ability to easily recover all data for any site, ateach observation period, is a useful way to evaluatechange on each site. Amount of change can be expressedas the difference between two observations. It can also beexpressed as this difference as a percentage of the earlierobservation. For example, if the area of the site increasedfrom 400 to 500 m2, it increased 100 m2, which is 25 per-cent. As the number of sites increases, analyzing changeson individual sites becomes increasingly impractical.

2. Summary statistics provide a perspective onamounts of change for the entire area or potions of areas,such as management or destination areas. Medians andranges are useful statistics for displaying typical changesand variability in response. Table 13 shows changes over5 years on 16 campsites in the Eagle Cap Wilderness(Cole 1986b). Medians at the two observation periods areprovided, as are medians for the difference between theseobservations and this difference expressed as percentchange. Finally, the number of sites that increased ordecreased is displayed in order to evaluate how consistentchanges were, and the statistical significance of thechange is assessed.

Such statistics, in addition to indicating what changeshave occurred, can be used to assess differences inamount of change between management areas within awilderness. Areas with more pronounced changes or agreater proportion of sites experiencing change should beassigned a high priority for management attention.

3. Sites can be rank-ordered according to how muchchange has occurred since the last observation period. Aswith the analysis of the current situation, each site’s per-centile can be determined to gain a perspective on howchange compares to what has occurred on other sites.Those sites in the higher percentiles and those areas witha large number of sites in the higher percentiles are thesites and areas with the greatest need for more intensivemanagement.

4. There are a number of situations that might usefullybe flagged. Those sites that have deteriorated or im-proved most might be identified, as might the manage-ment areas that have deteriorated or improved most.Other situations that might be flagged include those thatstill exceed standards, those that have violated standardsover the observation period, those that are approachingstandards, and those that have improved in relation tostandards.

5. Any of these sets of flagged sites could be mapped.New sites that have developed and sites that are nolonger there could also be mapped. Finally, it can beuseful to classify sites according to level of deteriorationor improvement and then map sites in each of theseclasses.

Automatic Data ProcessingAccess to computer hardware and software makes it a

relatively simple matter to perform a myriad of usefulanalyses. Manfredo and Hester (1983) have developed asoftware package, written for Apple computers, that ana-lyzes and graphically presents impact monitoring infor-mation. The analyses mentioned above, for campsites at

Table 13-Median change in size and tree damage, over a 5-year period, on 16 campsites in the EagleCap Wilderness1

StatisticCamparea

Devegetatedcore area

Damaged Trees withtrees exposed roots

Felledtrees

m 2 - - - - - - - - - - - - - N u m b e r - - - - - - - - - - - - -

Median1979 198 86 9.0 3.5 4.01984 233 104 7.5 3.5 5.0

Difference 22 5 0 0 1.0Change (%) 11 10 0 0 35

Number of sitesIncrease 14 10 3 4 8Decrease 1 5 6 3 4

Significance <O.OOl 0.03 0.17 0.26 0.08

‘Difference is the median difference between 1979 and 1984. Change is difference as a percentage of 1979 values.Positive values indicate an increase between 1979 and 1984. Significance was tested with the Wilcoxon matched-pairs,signed-ranks test.

30

the Delaware Water Gap, were produced using softwaredeveloped by Chuck Robbins and Jeff Marion. They usedthe dBASE III data manager, along with some additionalprogramming, to do most of the analyses just mentioned.

Data base managers make data entry simple, and manyof them have built-in capabilities to derive basic summarystatistics. With some additional programming it shouldbe a simple matter to develop user-friendly, menu-drivensoftware that can easily perform all needed analyses andthen generate maps and graphs to display results.

Research NeedsBecause efforts to monitor campsites are still in their

infancy, we have little experience with analysis of moni-toring data. Experience with trend data is particularlylimited. The preceding discussion presented some ideasand, where possible, some examples. We need to evaluatemeans of analyzing and displaying these data. Oncethese methods are fairly well established, it should bepossible to develop software packages that will simplifythe analysis procedures.

Sources of InformationCole, David N. 1986. Ecological changes on campsites in

the Eagle Cap Wilderness, 1979 to 1984. Res. Pap.INT-368. Ogden, UT: U.S. Department of Agriculture,Forest Service, Intermountain Research Station. 15 p.(Displays results of changes over a 5-year period,)

Manfredo, Michael J.; Hester, Arlene. 1983. A microcom-puter based campsite data system. In: Bell, J. F.;Atterbury, T., eds. Proceedings, renewable resourceinventories for monitoring changes in trends, an inter-national conference. Soc. Amer. For. No. 83-14.Corvallis, OR: Oregon State University, College ofForestry: 731-734. (Discusses the use of data manage-ment systems and presents a software program, forApple computers, that analyzes and displays campsitemonitoring data.)

Marion, Jeffrey L. [Personal communication.] ResearchScientist, U.S. Department of the Interior, NationalPark Service, Mid-Atlantic Region, Star Route 38,Milford, PA 18337. (Has developed dBASE III basedsoftware for analyzing campsite monitoring data.)

Parsons, David J.; Stohlgren, Thomas J. 1987. Impacts ofvisitor use on backcountry campsites in Sequoia andRings Canyon National Parks, California. Tech. Rep.No. 25. Davis, CA: U.S. Department of the Interior,Cooperative National Park Resources Studies Unit,University of California, Institute of Ecology. 79 p.(Describes results of an inventory of campsites,including an evaluation of procedures. Includes amethod for assessing the overall impact for entiredestination areas.)

STEP 7. MANAGEMENTAPPLICATIONS OF MONITORINGDATA

Many of the early attempts to monitor campsites werenot very successful because plans for using the data gen-erated were unclear. This is the reason for all the empha-sis in step 1 on deciding on your needs and which types ofimpact are most critical. It is also important, from thestart, to have a plan for using the data. Otherwise timemay be spent collecting information that is never usedand items of importance may be overlooked. Four impor-tant uses for monitoring data can be described.

Establishing Management and BudgetPriorities

Perhaps the most immediate use of the data is to estab-lish priorities for management projects. The analysis ofthe current situation will identify places where campsiteimpacts are particularly severe and followup measureswill identify places where conditions are deterioratinggreatly. These places should receive a high priority formanagement attention.

Exactly what situation is most undesirable and deserv-ing of management attention will vary from area to area.At Sequoia and Rings Canyon National Parks, where thepolicy is to obliterate campsites located within 25 ft(7.6 m) of water, those places with the most sites close towater receive a high priority for management attention.Where severely impacted sites are not tolerated, placeswith many class 5 sites (or some other measure of severeimpact) would be a high priority. Managers who are par-ticularly concerned with the proliferation of impactsmight assign highest priority to places with the greatestincrease in number of sites. Regardless of managementobjectives, if the appropriate types of measures are taken,it should be a simple matter to identify places that are inparticular need of management attention and, therefore,should receive a high priority in the budgeting process.

Monitoring can also facilitate the budgeting process bymore objectively describing the nature and extent of im-pact problems. Specific problems in specific places can beidentified, making it a simpler matter to determine thelevel of funding necessary to deal with these problems.

Management of Specific Sites

This type of analysis can also be extended to the man-agement of individual sites. It is possible to identify thosesites that are currently most heavily impacted, as well asthose sites that are deteriorating most. Moreover, as longas data on individual impact parameters have been re-corded separately, it will be clear which types of impactare most severe or are deteriorating most. This is impor-tant because very different management actions areneeded for different types of impact. Damage to trees, forexample, is best dealt with through education of campers;total area of the campsite might be dealt with throughlimits on party size or through site management intended

31

to make site expansion more difficult; loss or disturbanceof organic horizons is inevitable with use, although itmight be less severe if campsites were used less fre-quently. Management responses must be tailored to theparticular types of impact that are occurring.

Relationship to StandardsRecently there has been an attempt to make manage-

ment more objective, explicit, and consistent by usingspecific statements of objectives to drive management.This approach, described most completely in a processtermed “limits of acceptable change” (LAC) (Stankey andothers 1985), involves the definition of standards. Stan-dards are precise, usually quantitative, statements ofmaximum levels of impact that will be tolerated (for ex-ample, campsites will be no larger than 1,000 ft2).

Standards are statements of conditions that, at a mini-mum, will be provided. Existing conditions can be com-pared with standards to determine where problems exist.A problem, by definition, is a situation where standardsare violated. Where problems exist, increased manage-ment is required. Conversely, where problems do notexist, management actions that constrain legitimate rec-reational uses should not be required. Once standardsare agreed on, situations where management actions areand are not needed can be agreed on. Often the specificmanagement actions needed are also obvious because thenature and location of problems are quite specific.

Inventory and monitoring are a critical part of thisprocess and most areas will have standards related tocampsite impacts. Through a process such as LAC, camp-site monitoring is formally integrated into the planningprocess. This was successfully done in the Bob MarshallWilderness complex, which will provide the examples inthe following discussion.

During early meetings in the planning process, camp-site impacts were identified as one of the foremost con-cerns in the area. Consequently, development of a camp-site inventory and monitoring procedure was a high prior-ity. Because the area to be inventoried was 1.5 millionacres (0.6 million ha) and there was only a handful ofpeople available to work part time on the inventory, aprocedure involving rapid estimates of a number of sitecharacteristics and impact parameters was developed(Cole 1984).

The first issue was to select the types of impact (indica-tors in the LAC terminology) to write standards for.These were decided on after field trips to identify problemsituations, evaluation of public concerns, and analysis ofdetailed measurements on a sample of 35 campsites (Cole1983b). Frequent problems in need of management wereplaces with excessive numbers of sites, places with largenumbers of highly impacted sites, and individual siteswith excessive amounts of barren soil. The specific indica-tors selected to address these problems were (1) the num-ber of campsites per 640-acre (259-ha) section, (2) thenumber of moderately and highly impacted campsites per640-acre (259-ha) section, and (3) the area of barren coreon any campsite.

Moderately and highly impacted sites were sites with asummary rating of 30-50 and more than 50, respectively.These summary ratings were derived by rating nine para-meters, multiplying these ratings by weights (reflectingthe relative importance of each parameter), and summingthese products. Refer to appendix H for more detail.

The Bob Marshall Wilderness complex was subdividedinto four opportunity classes. Different standards werewritten for each of these opportunity classes. For ex-ample, the standards for maximum number of campsitesper 640-acre (259-ha) section are one in opportunity class1, and 2, 3, and six in classes 2, 3, and 4, respectively.Other standards for opportunity class 1 are no moderatelyor highly impacted campsites per section and no morethan 100 ft2 (9 m2) of barren core on any campsite. Analo-gous standards for opportunity class 4 are no more thanthree moderately impacted sites and no more than onehighly impacted site in any section, and no more than2,000 ft2 (186 m2) of barren core on any campsite.

Whether current conditions violate standards or not caneasily be determined from the inventory data. Barrencore can simply be read off the form or flagged on a com-puter data base. The number of sites per section requiresmapping and then counting of numbers of sites. Thenumber of moderately and highly impacted sites can beassessed in a similar manner, although it is necessary tofirst calculate summary impact ratings.

Defining standards provides a specific focus for camp-site monitoring, as well as for the entire planning andmanagement process. As increasing numbers of areasadopt this framework, the use of monitoring data inrelation to standards is likely to become increasinglyimportant.

Use in Developing Visitor-UseCapacities

Data from campsite inventories have been used to es-tablish visitor use limits at Sequoia and Kings CanyonNational Parks. The procedure is discussed in detail byParsons (1986) and Parsons and Stohlgren (1987). Thefollowing discussion is excerpted from those papers.

Sequoia and Rings Canyon National Parks have beendivided into 52 backcountry travel zones. Each zone hasa daily use capacity, determined largely on the basis ofcampsite inventory data. Zone capacities are controlledby daily trailhead quotas, established through theQUOTA computer model developed at Yosemite NationalPark (van Wagtendonk and Coho .1986).

The decision to base zone capacities on campsites re-flected a major management objective of maintaininghistorical use patterns in the Parks. This assures thattraditional low-use (and generally low-impact) areas willremain, while recognizing the futility of trying to reduceuse levels in traditional heavy use areas, particularlygiven the long periods required for recovery from impact.

The first step in the process of setting capacities was tocount the number of class 3,4, and 5 campsites in eachdestination area (termed management area). The numberof these sites that were unacceptable (either because theywere within 25 ft [7.6 m] of water, within 100 ft [30 m] of

32

another class 3,4, or 5 site, or unacceptable for some otherreason) was determined and subtracted from the originaltotal. This final number was the maximum number ofsites that could be used at one time. Class 1 and 2 siteswere not included, although it was recognized that theywould continue to be used occasionally.

This estimate of the number of acceptable sites wasevaluated by a team of scientists and managers familiarwith the area. In many cases it was agreed that moreacceptable sites were available than could ever be occupiedat one time without exceeding either the peak recorded useor the group’s opinion of what use level was appropriate.In these cases, the maximum number of sites was reducedto a level that seemed appropriate, without causing unac-ceptable crowding or increases in use.

The maximum number of sites was summed for eachdestination area in each zone to obtain a zone total. Thesenumbers were compared with available information on thenumber of parties using specific zones during peak useperiods. If these numbers exceeded the reported peak use,they were reduced accordingly. The final number of ac-ceptable sites that could be occupied at one time wasmultiplied by the average party size to obtain the maxi-mum daily number of persons allowed in each zone.

Parsons and Stohlgren (1987) stress that the rationalebehind this approach stems from a goal of maintaining ex-isting use and impact patterns. Should objectives stresseither more stringent preservation or provision of morerecreational opportunities, underlying assumptions wouldhave to be shifted and the procedure would have to bemodified. Nevertheless, capacities could still be derivedlargely from campsite inventory data

Research NeedsResearch could suggest additional ways that monitoring

data might be applied to management. Further work onwhich types of impacts (indicators) are most useful forwriting standards would be helpful, as would evaluation ofthe success of programs that do utilize monitoring data intheir management programs.

Sources of InformationParsons, David J. 1986. Campsite impact data as a basis

for determining wilderness use capacities. In: Lucas,Robert C., compiler. Proceedings-national wildernessresearch conference: current research; 1985 July 23-26;Fort Collins, CO. Gen. Tech. Rep. INT-212. Ogden, UT:U.S. Department of Agriculture, Forest Service, Inter-mountain Research Station: 449-455. (Describes howcampsite inventory data were used to derive use limitsfor Sequoia and Kings Canyon National Parks.)

Parsons, David J.; Stohlgren, Thomas J. 1987. Impacts ofvisitor use on backcountry campsites in Sequoia andKings Canyon National Parks, California. Tech. Rep. 25.Davis, CA: U.S. Department of the Interior, CooperativeNational Park Resources Studies Unit, University ofCalifornia, Institute of Ecology. 79 p. (Includes a sectionon the calculation of use capacities for zones, on thebasis of campsite inventory data.)

Stankey, George H.; Cole, David N.; Lucas, Robert C.;Petersen, Margaret E.; Frissell, Sidney S. 1985. Thelimits of acceptable change (LAC) system for wildernessplanning. Gen. Tech. Rep. INT-176. Ogden, UT: U.S.Department of Agriculture, Forest Service, Intermoun-tain Forest and Range Experiment Station. 37 p. (Dis-cusses standards, their purpose, and how they can bedeveloped.)

REFERENCESBratton, Susan P.; Stromberg, Linda L.; Harmon, Mark E.

1982. Firewood-gathering impacts in backcountrycampsites in Great Smoky Mountains National Park.Environmental Management. 6: 63-71.

Bratton, Susan Power; Hickler, Matthew G.; Graves,James H. 1978. Visitor impact on backcountry camp-sites in the Great Smoky Mountains. EnvironmentalManagement. 2: 431-442.

Brewer, Les; Berrier, Debbie. 1984. Photographic tech-niques for monitoring resource change at backcountrysites. Gen. Tech. Rep. NE-86. Broomall, PA: U.S.Department of Agriculture, Forest Service, Northeast-ern Forest Experiment Station. 13 p.

Carothers, Steven W.; Johnson, Robert A.; Dolan, Robert.1984. Recreational impacts on Colorado River beachesin Glen Canyon, Arizona. Environmental Management.8: 353-358.

Chambers, Jeanne C.; Brown, Ray W. 1983. Methods forvegetation sampling and analysis on revegetated minedlands. Gen. Tech. Rep. INT-151. Ogden, UT: U.S.Department of Agriculture, Forest Service, Intermoun-tain Forest and Range Experiment Station. 57 p.

Cole, David N. 1982. Wilderness campsite impacts: effectof amount of use. Res. Pap. INT-284. Ogden, UT: U.S.Department of Agriculture, Forest Service, Intermoun-tain Forest and Range Experiment Station. 34 p.

Cole, David N. 1983a. Monitoring the condition of wilder-ness campsites. Res. Pap. INT-302. Ogden, UT: U.S.Department of Agriculture, Forest Service, Intermoun-tain Forest and Range Experiment Station. 10 p.

Cole, David N. 1983b. Campsite conditions in the BobMarshall Wilderness, Montana. Res. Pap. INT-312.Ogden, UT: U.S. Department of Agriculture, ForestService, Inter-mountain Forest and Range ExperimentStation. 18 p.

Cole, David N. 1984. An inventory of campsites in theFlathead National Forest portion of the Bob MarshallWilderness. Unpublished report on file at: U.S. Depart-ment of Agriculture, Forest Service, Intermountain Re-search Station, Forestry Sciences Laboratory, Missoula,MT. 19 p.

Cole, David N. 1985. Ecological impacts on backcountrycampsites in Grand Canyon National Park. Unpub-lished report on file at: U.S. Department of the Interior,National Park Service, Western Regional Office, SanFrancisco, CA 96 p.

Cole, David N. 1986a. Ecological changes on campsites inthe Eagle Cap Wilderness, 1979 to 1984. Res. Pap. INT-368. Ogden, UT: U.S. Department of Agriculture, ForestService, Intermountain Research Station. 15 p.

33

Cole, David N. 1986b. Recreational impacts on backcoun-try campsites in Grand Canyon National Park, Arizona,USA Environmental Management. 10: 651-659.

Cole, David N.; Marion, Jeffrey L. 1988. Recreation im-pacts in some riparian forests of the eastern UnitedStates. Environmental Management. 12: 99-107.

Echelberger, Herbert E. 1971. Vegetative changes atAdirondack campgrounds: 1964 to 1969. Res. NoteNE-142. Upper Darby, PA: U.S. Department of Agricul-ture, Forest Service, Northeastern Forest ExperimentStation. 8 p.

Frissell, Sidney S. 1978. Judging recreation impacts onwilderness campsites. Journal of Forestry. 76: 481-483.

Gifford, Gerald F.; Faust, Robert H.; Coltharp, George B.1977. Measuring soil compaction on rangeland. Journalof Range Management. 30: 457-460.

Hendee, John C.; Clark, Roger N.; Hogans, Mack L.;Wood, Dan; Koch, Russell W. 1976. Code-A-Site: a sys-tem for inventory of dispersed recreational sites inroaded areas, backcountry, and wilderness. Res. Pap.PNW-209. Portland, OR: U.S. Department of Agricul-ture, Forest Service, Pacific Northwest Forest andRange Experiment Station. 33 p.

Howard, Richard F.; Singer, Michael J. 1981. Measuringforest soil bulk density using irregular hole, paraffinclod, and air permeability. Forest Science. 27: 316-322.

Kitchell, Katherine P.; Connor, Jeff. 1984. Canyonlandsand Arches National Parks and Natural Bridges Na-tional Monument draft recreational impact assessmentand monitoring program. Unpublished paper on file at:U.S. Department of the Interior, National Park Service,Moab, UT. 80 p.

LaPage, Wilbur F. 1967. Some observations on camp-ground trampling and ground cover response. Res. Pap.NE-68. Upper Darby, PA: U.S. Department of Agricul-ture, Forest Service, Northeastern Forest ExperimentStation. 11 p.

Lee, Robert G. 1975. The management of human compo-nents in the Yosemite National Park ecosystem.Yosemite, CA: Yosemite Institute. 134 p.

Legg, Michael H.; Schneider, Gary. 1977. Soil deteriora-tion on campsites: northern forest types. Soil Science ofAmerica Journal. 41: 437-441.

Leonard, R. E.; McBride, J. M.; Conkling, P. W.;McMahon, J. L. 1983. Ground cover changes resultingfrom camping stress on a remote site. Res. Pap. NE-530.Broomall, PA: U.S. Department of Agriculture, ForestService, Northeastern Forest Experiment Station. 4 p.

Magill, Arthur W. 1970. Five California campgrounds . . .conditions improve after 5 years’ recreational use. Res.Pap. PSW-62. Berkeley, CA: U.S. Department of Agri-culture, Forest Service, Pacific Southwest Forest andRange Experiment Station. 18 p.

Magill, Arthur W.; Twiss, R. H. 1965. A guide for record-ing esthetic and biologic changes with photographs.Res. Note PSW-77. Berkeley, CA: U.S. Department ofAgriculture, Forest Service, Pacific Southwest Forestand Range Experiment Station. 8 p.

Manfredo, Michael J.; Hester, Arlene. 1983. A microcom-puter based campsite data system. In: Bell, J. F.;Atterbury, T., eds. Proceedings, renewable resourceinventories for monitoring changes in trends, aninternational conference. Soc. Amer. For. No. 83-14.Corvallis, OR: Oregon State University, College ofForestry: 731-734.

Marion, Jeffrey L. 1986. Campsite assessment systems:application, evaluation, and development. In: Popadic,Joseph S.; [and others], eds. Proceedings, 1984 riverrecreation symposium; 1984 October 31-November 3;Baton Rouge, LA Baton Rouge, LA: Louisiana StateUniversity, School of Landscape Architecture: 561-573.

Marion, Jeffrey L.; Merriam, L. C. 1985. Recreational im-pacts on well-established campsites in the BoundaryWaters Canoe Area Wilderness. Stn. Bull. AD-SB-2502.St. Paul, MN: University of Minnesota, AgriculturalExperiment Station. 16 p.

Marion, Jeffrey Lawrence. 1984. Ecological changes re-sulting from recreational use: a study of backcountrycampsites in the Boundary Waters Canoe AreaWilderness, Minnesota. St. Paul, MN: University ofMinnesota. 279 p. Dissertation.

Merriam, L. C., Jr.; Smith, C. K.; Miller, D. E.; [and oth-ers]. 1973. Newly developed campsites in the BoundaryWaters Canoe Area: a study of 5 years’ use. Stn.Bull. 511, For. Ser. 14. St. Paul, MN: University ofMinnesota, Agricultural Experiment Station. 27 p.

Mueller-Dombois, Dieter; Ellenberg, Heinz. 1974. Aimsand methods of vegetation ecology. New York: JohnWiley. 547 p.

Parsons, David J. 1986. Campsite impact data as a basisfor determining wilderness use capacities. In: Lucas,Robert C., compiler. Proceedings-national wildernessresearch conference: current research; 1985 July 23-26;Fort Collins, CO. Gen. Tech. Rep. INT-212. Ogden, UT:U.S. Department of Agriculture, Forest Service, Inter-mountain Research Station: 449-455.

Parsons, David J.; MacLeod, Susan A. 1980. Measuringimpacts of wilderness use. Parks. 5(3): 8-12.

Parsons, David J.; Stohlgren, Thomas J. 1987. Impacts ofvisitor use on backcountry campsites in Sequoia andKings Canyon National Parks, California. Tech. Rep.25. Davis, CA: U.S. Department of the Interior, Coop-erative National Park Resources Studies Unit, Univer-sity of California, Institute of Ecology. 79 p.

Schreiner, Edward S.; Moorhead, Bruce B. 1979. Humanimpact inventory and management in the OlympicNational Park backcountry. In: Ittner, Ruth; Potter,Dale R.; Agee, James K.; Anschell, Susie, eds. Proceed-ings, recreational impact on wildlands; 1978 October27-29; Seattle, WA. R-6-001-1979. Portland, OR: U.S.Department of Agriculture, Forest Service, PacificNorthwest Region: 203-212.

Schuster, Ervin G.; Zuuring, Hans R. 1986. Quantifyingthe unquantifiable: or, have you stopped abusing meas-urement scales? Journal of Forestry. 84: 25-30.

34

Stankey, George H.; Cole, David N.; Lucas, Robert C.;Petersen, Margaret E.; Frissell, Sidney S. 1985. Thelimits of acceptable change (LAC) system for wildernessplanning. Gen. Tech. Rep. INT-176. Ogden, UT: U.S.Department of Agriculture, Forest Service, Intermoun-tain Forest and Range Experiment Station. 37 p.

Steele, Brian. 1987. Statistical procedures for the analysisof a campsite monitoring program. Unpublished reporton tile at: Systems for Environmental Management,Missoula, MT. 56 p.

Stohlgren, Thomas J.; Parsons, David J. 1986. Vegetationand soil recovery in wilderness campsites closed tovisitor use. Environmental Management. 10: 375-380.

Sydoriak, Charisse A. [In press]. Yosemite’s wildernesstrail and campsite impacts monitoring system. Paperpresented at the National Park Service Science Confer-ence; 1986 July 13-18; Fort Collins, CO.

van Wagtendonk, Jan W.; Coho, Paul R. 1986. Trailheadquotas: rationing use to keep wilderness wild. Journalof Forestry. 84: 22-24.

35

APPENDIXES A-K: SELECTED PROCEDURES USED TO MONITORWILDERNESS CAMPSITES

Appendix A: Photopoint Photography (Adapted From Brewerand Berrier 1984)

The technique requires a referenced and easily relo-cated camera position from which photographs can betaken periodically for comparison. The first step is toanalyze the subject area carefully. Select a camera posi-tion that provides the most advantageous perspective(with the available equipment) of the expected change.Photographers on successive photo missions may feelcompelled to move the camera slightly to achieve whatthey feel is a better coverage of the subject. This mightresult in a loss of information, which could be avoided byproperly anticipating what coverage will be necessary aschanges occur over time. Documenting the reason for thecamera placement when it is not immediately evidentmay avoid costly changes.

s h e l t e r

B e e c h

Once a location for the photopoint has been determined,a physical marker should be established. Permanentlandmarks such as boulders or other large objects shouldbe utilized when possible. Where landmarks such asthese are not available, some kind of stake can be drivenflush with the ground surface. Size, weight, and durabil-ity are limiting factors. Wood is light, but may deterio-rate faster than desired. Objects as small as nails can beused to permit relocation with a metal detector. (This ismore appropriate in wilderness.) Any marker should beas inconspicuous as possible to avoid vandalism.

Referencing the photopoint is the next step. Twonearby permanent objects can be used as references, butthree are better. Trees are good references and may bemarked with numbered aluminum tags. (Note: Tags arenot appropriate in designated wilderness.) Identificationtag numbers should be recorded along with the bearingand distance from each tree to the photopoint. Sketchmaps should be made showing the azimuth from the ref-erence point to the photopoint, the d.b.h. and species ofthe witness trees, and the general object area in relationto trailheads, shelters, access roads, and so forth (fig. 4).An altimeter reading and slope aspect indication cansometimes help locate the photopoint on topographicmaps. A photograph of the area and camera setup is alsouseful for relocation.

If different cameras are used for successive photos fromthe same point, film format and lens focal length shouldbe the same. Film of the same type, speed, and spectralsensitivity should be used when possible. A change fromblack and white to color film can be made with less loss ofinformation if a set of prints is also made from the colornegatives or slides for the first year of comparison. Thetime of day should be duplicated as closely as possible toavoid shadows in different positions. The photos shouldalso be taken during the same time of year (the size of the“window” of duplication days will vary according to theneeds of the study). Carrying copies of the original photosinto the field can facilitate accurate reproduction.

fireplace

B e e c h

OO

FILM: Kodak Tri-x, 400 ASA.Lens center 4’ above ground.Canon FTB. 28mm Canon FD lens.F-stop, speed f-5.6 @ l/30.DECLINATION: 15-1/2 Ow

Figure 4- sketch map referencing a photopoint.

36

Appendix B: The Frissell Condition Class System (Modified From Frissell 1978)

After locating campsites on a map of the area, assigneach campsite a rating between 1 and 5, using the follow-ing definitions:

Class l-Ground vegetation flattened but not perma-nently injured. Minimal physical change except for possi-bly a simple rock fireplace.

Class 3-Ground vegetation lost on most of the site, buthumus and litter still present in all but a few areas.

Class 4-Bare mineral soil widespread. Tree rootsexposed on the surface.

Class Coil erosion obvious. Trees reduced in vigor ordead.

Class 2- Ground vegetation worn away around fire-place or center of activity.

Appendix C: The Sequoia-Kings Canyon Campsite Class System(Adapted From Parsons and MacLeod 1980; Parsons and Stohlgren 1987)

Campsites are located on an area sketch map (fig. 5).Then each campsite is rated on the basis of eight criteria(fig. 6). A rating between 1 and 5 is assigned to each fac-tor that applies. These ratings are summed and dividedby the number of factors used. For example, if there areno trees on the site, this criterion is ignored and the sumof rankings is divided by 7, instead of 8. This mean,rounded to the nearest integer, is the campsite class.Figure 7 shows the field inventory form used for eight ofthe campsites around Lake Reflection.

Additional instructions that might not be self-explanatory from the criteria and rating factors infigure 6 include:

1. Density of vegetation is evaluated by comparing theextent of vegetative ground cover on the campsite withthat on environmentally similar but unimpacted areas offthe site.

2. Composition of vegetation also involves a comparisonwith an undisturbed area.

3. Total area of the campsite is an estimate of the areaaffected by trampling on the site.

4. Barren core area is an estimate of the area on whichtrampling has removed all vegetation; organic horizonsmay or may not still be present.

5. Social trails are the informal trails that develop be-tween the campsite and the trail, water, and other camp-sites. It will be necessary to define what constitutes awell-developed, as opposed to a discernible, trail.

6. Mutilations refer to trees; this factor will not applyin nonforested areas. Mutilations include carvings, axmarks, and nails. More than one mutilation can occur onone tree. A definition should be developed for what con-stitutes a highly obtrusive mutilation and a decision mustbe made about how far offsite to count trees.

In addition to the campsite impact class, descriptiveinformation on the local environment is recorded on theinventory form (fig. 7). The distance to water and thenumber of class 3,4, and 5 sites within 100 ft (30 m)-ameasure of campsite crowding-are also recorded. If the

site is not acceptable, a potential large group site, orneed of obliteration, this is noted, as are any recom-mended management actions.

in

Figure 5-Example of afield map showing thecampsites around LakeReflection.

37

Density of Vegetation(With respect to surrounding vegetation):

1 - same as surroundings3 - moderately less dense than surroundings5 - considerably less dense than surroundings

Composition of Vegetation(With respect to surrounding vegetation):

1 - same as surroundings3 - moderately dissimilar5 - significantly dissimilar

Total Area of Campsite

1 - less than or equal to 20 ft2 (2 m2)

32 - 21 to 100 ft2 (2 to 9.3 m2)

- 101 to 500 ft2 (9.4 to 46 m2)4 - 501 to 1,000 ft2 (46.1 to 93 m2)5 - greater than 1,001 ft2 (93 m2)

Barren Core Area

1- absent2 - 5 to 50 ft2 (0.5 to 4.6 m2)3 - 51 to 200 ft2 (4.7 to 18.6 m2)4 - 201 to 500 ft2 (18.7 to 46 m2)

5 - greater than 501 ft2 (46 m2)

Campsite Development

1 - windbreaks and paraphernalia absent; trash andseats minimal; firerings absent or scarce

2 - trash, windbreaks, seats, and firerings minimal;paraphernalia absent

3 - trash, windbreaks, seats mostly moderate; fire-rings mostly minimal; paraphernalia minimal

4 - trash, windbreaks, seats, firerings, and parapher-nalia mostly moderate; some heavy

5 - trash, windbreaks, seats, firerings, paraphernaliamostly heavily developed

Litter and Duff

1 - trampling barely discernible: some needles bro-ken; scattered cones

2 - moderately trampled; needles broken, compacted;few cones

3 - heavily trampled, clumped, pulverized; conesabsent

4 - litter +/- absent, pulverized, ground into soil5 - litter, cones, and duff completely absent

Social Trails

1 - none2 - 1 trail discernible3 - 2 trails discernible4 - 1 to 2 trails well developed, or 3 or more trails +/-

discernible5 - 3+ trails well developed

Mutilations

1- none2 - 1 t o 23 - 3 to 54 - 6 to 10 or 1 to 2 highly obtrusive5 - 11+ or 3 +/- highly obtrusive

Figure 6-The criteria and rating factors used to inventory campsites in Sequoia and Kings Canyon National Parks.

38

CAMPSITE FIELD INVENTORY FORM

Date 9 / 3 / 7 7

Management Area LAKE Ref lect ion Zone 67 Elevation 10,005Landform L a k e B a s i n

Campabil i ty: Potential 2 0 % Currently Used 2 0 %

Overs tory /Cover L P - F T / 5 0 %

M e a d o w s Not Significant Fuels Rating s c a r c e

C o m m e n t s N o E n d L P - F T F o r e s t 5 0 % c o v e r ;

CampsiteNumber Campsite Ecological Type Site(on map)

Dis t .Class Overstory

Crowding Understory Pot. to H2O 3,4,5’s

I

Comments

Application of Rating Factors for Campsite Class Determination:

Camp Site

Density

Composition

Total Area

Barren Core

Camp Development

Lit ter & Duff

Social Trai ls

Muti lat ions

Mean Rating orCampsite Class

Figure 7-Example of the data collection form used to inventory campsites around LakeReflection.

39

Appendix D: The Eagle Cap Method of Measurements on PermanentSampling Units (Modified From Cole 1982)

Campsite Measurements-Locate a center point in aposition that will permit easy measurement of the site.Mark it, for later relocation, with a large buried nail.Reference the location of the center point, noting azimuthand distance from two (or preferably three) landmarks.(See discussion of referencing in appendix A)

Measure the distance for 16 azimuths (N, NNE, NE,ENE, E, ESE, SE, SSE, S, SSW, SW, WSW, W, WNW,NW, and NNW) from the center point to the first signifi-cant amount of vegetation (defined, for the Eagle Capstudy, as at least 15 percent cover in a 1.09-by 3.28-ft[0.33-by l-m] quadrat oriented perpendicular to andbisected by the measuring tape). Be sure to note whethertrue or magnetic north was used. These intercepts definethe boundary of the devegetated central core of the camp-site (the bare area in fig. 8). Also measure the distance tothe edge of the campsite (where trampling is no longerevident) and record these 16 transect intercepts. If anuntrampled “island” is encountered in any direction, notethe distance to the “island” and the reentry onto thecampsite, as well as the campsite boundary. These inter-cepts define the campsite boundary, as well as the area ofthe “island” which is to be subtracted from the area of thecampsite (fig. 8).

- Edge of actual camp area---- Edge of bare area - Edge of camp area,

defined by transect end points

x Recorded transect intercepts

Untrampled "island"

In the case of both camp and bare area, boundaries areapproximated by drawing straight lines between adjacentintercepts. Note in figure 8 that while this boundarydiffers from the actual campsite boundary, the total camparea is about the same. To calculate area, intercepts andconnecting lines are plotted on a radial map. Figure 9shows the campsite and “island” boundaries on such amap. Use a planimeter to calculate the area of totalcampsite, "islands,” and bare area. Subtract the "island”area from the total campsite area to obtain the camp area.A simpler method is to simply calculate the area of each ofthe 16 triangles defined by adjacent transects (fig. 8) andsum these. The area of each triangle is the length of eachof the two transects times 0.383 (the sine of the 22.5 de-gree angle) divided by 2. In this case, ignore the “island”in calculating these areas of triangles, estimate the areaof the “island” in the field, and subtract this value fromthe sum of the triangles. Refer to appendix J for furtherdiscussion of this technique.

Place flagging temporarily at each of the 16 pointsalong the edge of the campsite. Straight lines drawn be-tween these points define the campsite on which repli-cable measurements will be taken. Count all tree repro-duction, defined for the Eagle Cap study as trees more

Figure 8-Illustration of the radial transect technique for estimating thearea of the campsite and the devegetated central portion of the camp-site (bare area).

40

Figure 9--Example of a radial map used to calculate thearea of the campsite in figure 8. Concentric circles are 1 mapart (0.5 m apart within 5 m of the center point). Eachsquare is 1 m2.

than 6 inches (15 cm) and less than 4.5 ft (140 cm) tall,within this area. Exclude reproduction in untrampled‘islands.”

Within this same area, count all trees and then notehow many are damaged. The types of damage noted inthe Eagle Cap study were felled trees, exposed roots,trunk scars, cut branches, nails, and other minor injuries.Another approach would be to use damage categories, asMarion and Merriam (1985) did (see section on tree dam-age in step 2).

Take additional measurements in quadrats establishedalong four transects that originate at the center point andextend to the edge of the site. Randomly select the azi-muth of the first transect (from random numbers between1 and 90). The azimuth of each successive transect is 90degrees greater than the azimuth of the previous transect.Bury nails at the end (campsite boundary) of eachtransect.

Locate about 15 quadrats along these transects. Thenumber on each transect should be roughly proportionalto the relative length of each transect. The distance be-tween quadrats on any transect should decrease with

distance from the center point (fig. 10). This avoids sam-pling more heavily toward the center of the site. Measureonly quadrats that fall entirely within the campsite.

Within each quadrat estimate the percentage cover ofunderstory vegetation, exposed mineral soil, exposed rockand tree roots, and trunks. Estimate the cover of organiclitter, whether it is under vegetation or not (this is animprovement over the technique used in the Eagle Cap).Finally, estimate the cover of all plant species. In theEagle Cap study, all mosses and all lichens were esti-mated as a group. Coverage was estimated in 10 percentcoverage classes between 10 and 100 percent or to thenearest percentage if cover was 10 percent or less.

Within each quadrat, measure the depth of the organichorizons and take a reading of penetration resistance,with a pocket soil penetrometer. In the Eagle Cap study,four sets of soil samples, bulk density, and infiltrationrate measurements were taken. Given the variability ofresults, this number of samples was probably too small.Read the section in step 2 on impacts to the mineral soilfor a discussion of these techniques.

41

Figure 10-Location of quadrats for sampling groundcover parameters on the campsite in figure 8.

Control Sites- Take a set of comparable measure-ments on control sites; amount of impact can then beestimated as the difference between conditions on thecampsite and on an undisturbed control. Locate controlsas close to the campsite as possible, in places that areundisturbed but where the topography, rockiness, treecanopy cover, and understory species are similar to thecampsite. Often the understory composition has to becompared to what is surviving in protected places on thecampsite.

Bury a nail at the center point of the control and refer-ence it to landmarks, as well as to the campsite. In theEagle Cap study, controls were generally circular, with anarea of 1,000 to 2,000 ft2 (100 to 200 m2). Estimate thepercent cover of understory vegetation, exposed mineralsoil, exposed rock and tree roots and trunks, organic lit-ter, whether it is under vegetation or not, and of all plant

species. In the Eagle Cap study, a single cover estimatefor the entire control was made, rather than estimatingcover in quadrats, as was done on the campsite. This wasmore rapid and seemed justified because precision wasless of a concern on controls. As on campsites, cover wasestimated in 10 percent coverage classes between 10 and100 percent or to the nearest percent if cover was 10 per-cent or less.

Take measures of penetration resistance, organic hori-zon thickness, bulk density, and infiltration rates in regu-larly distributed locations on the control. The number ofsamples should be the same as the number on campsites.

Finally, count tree reproduction in a circle, centered atthe center point, with an area of 538 ft2 (50 m2). Appro-priate areas for control plots will vary between regionsand impact parameters.

42

Appendix E: The Sequoia Method of Measurements on Permanent Plots(Modified From Stohlgren and Parsons 1986)

Establish a 32.8- by 32.8-ft (10- by 10-m) sampling unit,aligned along compass directions and located such thatmost of the campsite is included. Place permanent mark-ers (such as buried nails) at each corner and reference atleast one corner. (Refer to the appendix section on photo-points for a discussion of referencing.) Place temporarystakes at 3.28-ft (1-m) intervals along each side. Connectstakes with string to form a 100-cell grid of 10.76-ftz (1-m2)sections.

Subdivide each section mentally into four 2.69-ft2

(0.25-m2) plots. Stratify each of these plots subjectivelyinto core, intermediate, and periphery (essentially control)plots. Core plots are generally in the center of the site andshow nearly complete loss of vegetation and organic mat-ter and continuous disturbance of the mineral soil. Inter-mediate plots show notable but less substantial damage(more vegetation cover, less litter and duff pulverization,and pockets of intact sod). Periphery plots appear to beunimpacted and border the site. Map each zone (see fig.11) and take a subsample of five to 10 plots randomly fromeach zone.

In each plot, estimate the foliar cover of each plant spe-cies to the nearest 5 percent (to the nearest 1 percent ifcover is less than 5 percent). Collect five to 10 soil samplesfrom each zone to analyze bulk density, soil moisture, soiltexture, organic matter content, pH, and chemistry.

C CORE

I INTERMEDIATE

P PERIPHERY

R ROCK

A ACCESS TRAIL. TREE

SCALEN NON- HOMOGENEOUS

AREA

Figure 11-Map of zones within the 1,076-ft2

(100-m2) sampling area on campsites.

43

Appendix F: The Olympic Bare Ground Technique (AdaptedFrom Schreiner and Moorhead 1979)

The first step is to map all individual campsites withingroups of campsites. For each site, fill out a human im-pact inventory form (fig. 12). The impact data on the formare found in items 31, 32, and 35 through 38, estimated asfollows:

Item 31: Note whether or not horse feces are present.

Item 32: Note the number of horse trample areas(trampled depressions around trees where horses havebeen tethered) within 100 ft (30 m) of the site.

Item 35: Count the number of social (informal access)trails that enter the site.

Item 36: Note whether or not erosion is obvious on thesite.

Item 37: Measure the distance from a temporary centerpoint to the first vegetation (this must be defined in anagreed-upon manner) in eight directions (N, NE, E, SE, S,SW, W, NW). Note the mean of these eight radii in theseven classes provided-no bare ground; 1-2 ft (0.3-0.6 m);2.1-4 ft (0.7-1.2 m); 4.1-6 ft (1.3-1.8 m); 6.1-8 ft (1.9-2.4 m);8.1-10 ft (2.5-3.0 m); and 10.1 ft (3.1 m) and longer.

Item 38: Record each tree on the site, by species, notingdiameter and the extent of damage.

Finally, draw a sketch map of the site, to scale. Themap is drawn on either a l- by l-m grid or a 2-by 2-mgrid. This map serves as a baseline for the size of thebare area, the location of social trails, downed logs, andthe approximate location of the center point used to deter-mine mean bare radius.

Figure 12-Front side of a completed campsite inventory edge-punched card used to monitor campsitesat Olympic National Park.

45

Appendix G: The Great Smoky Mountains Area1 MeasurementTechnique (Adapted From Bratton and Others 1978)

Little of the information on the campsite monitoring pling), trampled vegetation, firewood clearing, tree dam-form (fig. 13) concerns impacts. Most of it provides infor- age, and trash dispersal. Quantify trash dispersal, treemation on environmental characteristics, attractions, de- damage, and firewood clearing in terms of maximumvelopments, and the water source. The primary impact distances from the center of the site. Quantify other dis-information is contained in the section on site dimensions. turbances by summing the areas of disturbed patches.For each type of disturbance, measure two dimensions Total disturbance is the maximum area1 extent of humanand then multiply these to obtain an area measurement. impact at the site. In addition, trash level, sanitation,The disturbances measured are bare rock, mud, slope and vegetation damage are rated in classes from good orerosion, bare soil, leaf litter (vegetation removed by tram- no damage to bad or high damage.

# Name

LocationT y p e

Forest type

Understory

Quad Coordinates

Opencanopy?

Exotics?

Site dimensions 1 2 Area

Attractions:Fruit plantsWild flowersBig treesBaldsViewsWaterfallsFishingPoachingHorse campTowerShelterNear cmpgrdNear viscenMajor accessNear roadAT nearRemoteFrivateDryOther:

Developments:ShelterBear fenceShelter frameTent spacePrivyFireplacePicnic tablesBear barrelsFirepitsHitchracksCamp circleSign campSign waterSign trailOther:

General comments:

Last rainLeaf fallObserverD a t e

Rating: Frequency useCarry capacityTrash levelFirewood levelMud and dirtSanitationVegetation damPlacementDrainageMaintenance

Suggested improvementsand hazard reduction:

Site future, why?

Figure 13-Form used to monitor campsites at Great Smoky Mountains National Park.

46

Appendix H: The Bob Marshall Rapid Estimation Procedure (AdaptedFrom Cole 1983a, 1984)

The information on one side of the form consists of loca-tional and environmental information (fig. 14a). Theimpact data are included on the other side of the form (fig.14b). Instructions for filling out this side are as follows:

Item 19: Using the five coverage classes on the form,estimate the percent coverage of the live understory vege-tation. Do not include dead vegetation, duff, trees, treeseedlings, or shrubs taller than a person. Estimate coverfor the entire campsite.

With a large campsite, it may help to divide the siteinto equal quarters; estimate the percentage cover of eachquarter and take the average. It may also help to visuallycluster all vegetation into one part of the site and esti-mate what percentage of the site would be covered. Try toselect one coverage class decisively. If you cannot, circleyour best estimate and note the other coverage class itmight be.

Make the same estimate of vegetation cover on a nearlyunused site similar-except for the impact-to the camp-site. The idea here is to select a site that is similar towhat the campsite probably looked like before it was used.Choose a site that is similar to the campsite in terms ofrockiness, slope, aspect, overstory composition and cover,and understory species composition. Protected plantsaround the base of trees or rocks can provide hints aboutspecies composition.

Item 20: Using the same five coverage classes, estimatethe percentage of the campsite without either live vegeta-tion or duff-the percentage on which mineral soil is ex-posed. In many cases, a thin layer of disturbed needles,leaves, or wood chips is scattered about with mineral soilshowing through. Consider these areas to be exposed soil.

Make the same estimate on the comparative area. Inpractice it will be easiest to estimate both vegetationcover and mineral soil exposure on the campsite, selectthe comparative area, and make the same estimatesthere.

Item 21: Using the information in item 19, record thedifference in vegetation cover class between campsite andcomparative area. If there is no difference (for example, ifboth campsite and comparative area are class 4, 51-75percent), circle rating 1. If coverage on the campsite isone class less than on the comparative area (for example,if the campsite is class 3, 26-50 percent, and the compara-tive area is class 4, 51-75 percent), circle 2. If the differ-ence is greater, circle 3.

Item 22: Using the information in item 20, record thedifference in mineral soil coverage class between thecampsite and comparative area. In this case, ratings of 2and 3 are given when mineral soil is one, or more thanone class higher on the campsite, respectively.

Item 23: Count the total number of damaged trees onthe campsite, the area visible from the campsite, and anystock holding areas. Never count the same tree on morethan one site. Damaged trees include stumps that show

cut marks, scarred trees, and trees with nails in them.Trees with lower branches cut off for firewood are notincluded. (Ignore the estimate of percentage of trees; thisinformation is not necessary.) If no trees were damaged,rate the site 1. If one to eight trees were damaged or ifone to three trees were felled or had bad scars (scarslarger than 1 ft2 [929 cm2]), rate the site 2. If more treesare damaged, badly scarred, or felled, rate the site 3.

Item 24: Count the number of trees with exposed rootson the same area as for tree damage. Exposure should bepronounced, extending at least 1 ft (0.3 m) from the treetrunk. It should also be the result of trampling-not theresult of a root running over a rock, for example. Assigna rating of 1 (no trees with exposed roots), 2 (one tosix trees), or 3 (more than six trees).

Item 25: Assign the site a rating of 1 if there are nofacilities-not even a firering. Afire site is considered aring only if the ring of stones is there; if they have beenscattered, it is a fire scar (see item 26) but not a tirering.If there is only one firering, primitive log seats (withoutsawed off ends), or both, assign the site a 2. If there ismore than one firering, or if there are any more elaboratefacilities, such as constructed seats, shelves, hitchrails,corrals, toilets, and so forth, assign the site a 3. If thefacilities are to be removed, rate the site as it was foundand then note in item 31 what actions were taken.

Item 26: Count the number of fire scars on the site,including any firerings as fire scars. Assign the site a 1 ifthere is only one fire scar and essentially no evident litter,stock manure, or human waste on the campsite. Assignthe site a 2 if there is more than one fire scar or if litter orstock manure is evident. If litter or stock manure is “allover the place,” or if there is any evident human waste,assign the site a 3.

Item 27: Social trails are the informal trails that leadfrom the site to water, the main trail, other campsites, orsatellite sites. Discernible trails are trails that you cansee but that are still mostly vegetated. Well-worn trailsare mostly devegetated. Count the total number of trails,regardless of whether they are discernible or well worn.Assign the site a 1 if there is only one discernible trail andno well-worn trails. Assign a 2 if there are two or threediscernible trails or one well-worn trail. Assign a 3 ifthere are more than three discernible trails or more thanone well-worn trail.

Item 28: Estimate the square footage of the disturbedcampsite and any satellite or stock holding areas. Thedisturbed area can usually be identified by either shorteror no vegetation in comparison to the periphery of thesite. Where there is no vegetation naturally and no otherevidence of disturbance to identify the edge of the site,place an N/A in the estimated area space and assign arating of 1. This may also be necessary on lightly usedsites where little vegetation loss is evident.

47

BOB MARSHALL - GREAT BEAR-

GENERAL SITE DESCRIPTION

(1)

(2)

(3)

(4)

(5)

(6)

(7)

SITE NUMBER:

UTM COORDINATES: 1 2 0 1 5 3 E 0 2 3 6 5 N

USGS QUADRANGLE: P e n t a g o n M t n .

DATE CODED: _ _07 (Month) 1 5 (Day) 1 9 88 (Year)- - 1 -

CODED BY: ( N a m e ) C o l e

ELEVATION: (To nearest 100 ft) 5 9 0 0

VEGETATION: (Circle one)

1 - Closed forest 3 - Nonforested. densely vegetated

4 - Nonforested, sparsely vegetated

Dominant species D o u g l a s f i r

H a b i t a t t y p e , i f k n o w n P S M E / C A R U

(8) LANDFORM: (Circle one)

1 - Floodplain Other valley bottom‘! 3 - Cirque basin_

4 - Sideslope 5 - Ridgetop 6 - Other

SCAPEGOAT WILDERNESS COMPLEX - CAMPSITE INVENTORY

Do in off ice(9) DISTANCE TO CLOSEST TRAILHEAD: (miles)

(10) DISTANCE TO CONSTRUCTED TRAIL: 1 0 0 (feet)

Screening : 1 - Complete Maintained:

2 - P a r t i a l 2 - No

3 - None

(11) DISTANCE TO WATER: 2 0 0 ( f e e t )

Type: 1- River/Creek 3- Spring

2 - Lake 4 - Other

(12) DISTANCE TO CLOSEST CAMPSITE:

Screening: 1 - Complete(circle o n e 2 - P a r t i a l

3 - None

1 5 0 (feet)

(Do In office) (13) NUMBER OF OTHER CAMPSITES WITHIN l/4 MILE:

(14) MAXIMUM PARTY SIZE ACCOMMODATED: (Circle one)

1 - 1-2 5 - more than 15

2 - 3-6 4 - 11-15

(15) TYPE OF USE: (Circle as many as apply)

1- Foot 3- River

2 - S t o c k 4 - O u t f i t t e r

(16) CLOSEST FIREWOOD SOURCE: (Circle one)

1 - One-site 3 - 100-300 feet 5 - <1/4 mile

2 - <100 feet

(17) CLOSEST FORAGE SUPPLY: (Circle one)

1 - On-site 5 - >l/4 mile

2 - <100 feet 4 - 300 ft- l /4 mile

( 1 8 ) F A C I L I T I E S : P r e s e n t & A b s e n t(If present, write number of each type in blank.)1 - Fire ring 2 6 - H i tchra i l 12 - P r i m i t i v e s e a t 4 7 - Corral3 - Constructed seat 8 - Toi let4 - Table/shelf/counter 9 - Other- - a5 - Meat rack

Figure 14-The front (a) and rear (b) sides of a completed form used to inventory campsites in the BobMarshall Wilderness complex.

IMPACT EVALUATION ON CAMPSITE ON UNUSED COMPARATIVE AREA

(19) VEGETATION COVER:(Be sure to compare similar 1 - 0-5% 3 - 26-50% 5 - 76-100%areas. same species, slope,rockiness. and canopy cover)

2- 6-25% 4 - 51-75%1 - 0-5% 3 - 26-50%2 - 6-25% 4 - 51-75%

(20) MINERAL SOIL EXPOSURE:(Percent of area that isbare mineral soil)

_________________________________

1 - 0 - 5 % 3 - 2 6 - 5 0 % 5 - 7 6 - 1 0 0 % 5 - 76-100%

(21) VEGETATION LOSS:1

(no differencein coverage)

(22) MINERAL SOIL INCREASE: (No differencein coverage)

(23) TREE DAMAGE:No. of t rees scarred or fe l led 10

(No more than broken

% of trees scarred or felled 5 (est.)lower branches)

(24) ROOT EXPOSURE:No. of trees with roots exposed 3

(None)

% of trees with roots exposed 25

(25) DEVELOPMENT:1 x 1 = 2

(26) CLEANLINESS:No. of f ire scars 2

(27) SOCIAL TRAILS: N o . o f t r a i l s 4

(28) CAMP AREAEstimated area 1600 (ft 2)

(29) BARREN CORE CAMP AREA:Estimated area 600 (ft 2)

(30) PHOTO RECORD 2 - 16

(No more thanscattered charcoalf rom 1 f i re r ing)(No more than 1d i s c e r n i b l e t r a i l )

( < 500 ft2 )

( < 50 ft2 )

Rating (Circle one category)

(Difference two ormore coverage classes)

(1-B scarred trees, orl-3 badly scarred orf e l l ed )

(1 f i re r ing wi th orwi thout pr imi t ivelog seat)

(2-3 discernible.max. 1 well-worn

(>6 t rees wi th rootsexposed)

( > 1 f i r e r i n g o r o t h e rmajor development)

(Human waste, muchl i t ter or manure)

(>3 discernible or morethan 1 well-worn)

( > 2 0 0 0 f t2 )

Calculation ofImpact indexdo in of f ice)

2 x 3 = 6

3 x 2 = 6

2 x 3 = G

3 x 2 = 6

1 x 2 = 2

2 x 2 = 4

4 x 2 = 8

2 x 3 = 6

4 5(32) IMPACT INDEX

bFigure 14 (Con.)

Visualize the site as a circle, a rectangle, or some combi-nation of these geometric figures. Pace off the appropri-ate dimensions. Calculate area and assign a rating of 1(<500 ft2 [<46 m2]), 2 (500-2,000 ft2 [46-186 m2]), or 3(>2,000 ft2 [>186 m2]).

Item 29: Using geometric areas and pacing, estimatethe area without any vegetation. Bare area may or maynot be covered with duff. Areas with scattered vegetationare not counted as bare area. Lump together in one meas-ure all bare areas on the campsite, including the areaaround the fire, as well as any bare tent areas, if appli-cable. If the bare area extends off the campsite intoneighboring undisturbed areas-in other words, if thearea is devoid of vegetation naturally-write N/A in theestimated area space and assign a rating of 1. If the barearea is less than 50 ft2 (5 m2), 50-500 ft2 (5-46 m2), ormore than 500 ft2 (>46 m2), assign ratings of 1, 2, or 3,respectively.

Item 32: The impact index is either the sum of theratings of each of these parameters or the sum ofweighted ratings. The weights assigned in the BobMarshall were as follows: vegetation loss (2), mineral soilincrease (3), tree damage (2), root exposure (3), develop-ment (l), cleanliness (l), social trails (2), camp area (4),and barren core camp area (2). Individual ratings aremultiplied by these weights and then these products aresummed to obtain the impact index. In the Bob Marshallthis index could vary from 20 (least impact) to 60 (mostimpact). In figure 14b, the first column of values, under“calculation of impact index” is the weights; the secondcolumn consists of the ratings. Other weighting valueshave been used to reflect different opinions about themost critical types of impact. If you do not use weights,you are implicitly stating that each of these types of im-pact is equal in importance.

50

Appendix I: The Canyonlands Rapid Estimation Procedure(Adapted From Kitchell and Connor 1984)

This procedure is similar in many ways to the proce-dure used in the Bob Marshall; however, more informa-tion is collected and impact parameters have beenadapted to desert environments. They also use slightlydifferent forms to monitor sites used primarily by threedifferent types of use: backpackers, river floaters, andpeople on four-wheel drive. Information on site character-istics is collected; the site is quickly mapped, photopointsare established; and an impact rating form is filled out.

The form (fig. 15) provides ratings for 24 parameters.The ratings include weights; some vary from 1.5 to 6,

while others vary from 0.5 to 2. These ratings aresummed. Then the condition of each site is considered tobe excellent if this sum is between 25 and 37. It is consid-ered good, fair, or poor if the sum is 38 to 62,63 to 87, or88 to 100, respectively.

Many of the ratings involve comparisons between thecampsite and an adjacent undisturbed area, as describedfor the Bob Marshall procedure (appendix H). Most oth-ers should be self-explanatory from the form (fig. 15),although many definitions need to be agreed on by differ-ent field workers. For example, for tree and shrub dam-age, how much damage must occur for it to be counted?

51

I. VEGETATION COVER

a. % cover

b. Composition

c. Distribution

<10% reduction when 1.5 10-30% reduction. 3 30-60% reduction. 4.5 >60% reduction. 6compared with adjacentundisturbed area.No exotic or disturbance 1 10-20% of vegetation 2 20-50% exotics and/or 3 >50% exotics and 4species present. composed of exotics/ disturbance species. disturbance species.

disturbance species.Vegetation evenly dis- 0.5 Faint appearance of 1 Up to 30% of vegetation 1.5 >30% of vegetation built 2tributed throughout isolated "islands" of built up around shrubs up around shrubs andsite. vegetation and “islands’ of ‘islands’ of vegetation.

vegetation.

2. SOIL DISTURBANCE

a. Cryptogamiccrust

b. Compaction/loosening/erosion

c. Excavationsand trenches

3. LlTTER

No disturbance: stillintact in appropriatehabitat.None apparent.

None apparent.

1 <30% reduction of crust 2 30-60% reduction of 3 >60% reduction of crust. 4when compared to adja- crust.cent/undisturbed area.

1 <30% of soil in site 2 30-60% of soil shows 3 >60% of soil shows corn- 4shows compaction (fine compaction or loosening; paction or loosening;soils) or loosening signs of erosion or signs of erosion in 2(coarse soils). gullying in 2 locations. locations.

1 1 or 2 small trenches 2 2-4 excavations or 3 >4 excavations or 4or excavations. trenches: a few may trenches: some show

show slight erosion. erosion and gullylng.

a. % cover

b. Distribution

c. Condition

<10% disturbed.

Evenly distributed.

No obvious signs ofbroken and crushedlitter.

1 10-35% reduction in con- 2 35-70% reduction corn- 3 >70% reduction com- 4trast to adjacent/undis- pared to adjacent/ pared to adjacent

turbed areas undisturbed areas undisturbed areas.1 50% of litter around 2 50-80% around edge and 3 >80% of litter around 4

edge of site and stable stable objects. edges and stableobjects objects.

1 Slight appearance of 2 <60% appears crushed or 3 >60% appears crushed 4crushed and broken broken or broken.litter.

4. SIDE TRAILS

a Number

b. Width

c. Depth

Only 1 present: not veryobvious from main trailto or through site; nospur trails, and only afew isolated footprintspresent.

Average width <12".

Trail at same level asadjacent area.

1 2 distinct trails from 2 3 distinct trails from 3 3 distinct trails from 4main trail to site or main trail to site or trail to site: 3 sidebetween attraction site between attraction site; or spur trails develop-(arch site or spring); 3 side trails or spurs ing; trails have begunno spurs; few isolated developing; footprints to merge; numerous foot-footprints. apparent. prints in and around

trail and site.1 Average width of 1 trail 2 2 trails wider than 12". 3 >2 trails wider than 12"; 4

>12". trails merging.1 1 trail-wearing below 2 At least 2 trails deeper 3 All trails deeper than 4

level of adjacent area. than adjacent ground adjacent ground level.

level.

5. SHRUB DAMAGE

a % damaged None show any damage. 1.5 <10% of shrubs show 3 10-30% of shrubs show 4.5 >30% of shrubs show 6reduced vigor damage (such as broken damage: 1 or 2 show damage; 2 show

limbs, crushed reduced vigor as a reduced vigor; dead orappearance result of damage. dying shrubs present.

b. Root exposure No roots exposed. 1.5 Exposed roots on 1 3 Exposed roots on 2 4.5 Exposed roots on 3 6shrub. shrubs. shrubs.

6. TREE DAMAGE

a. Broken limbs, No damage; or no trees 1 <10% of trees have 2 10-35% of trees have 3 >35% of trees have 4gashes, present. broken limbs, gashes, broken limbs, gashes, broken limbs, gashes,damage or other damage. or other damage. or other damage.

b. Root exposure No roots exposed: or 1 1 root exposed in site. 2 2 roots exposed in site. 3 3 or more roots exposed 4no trees present. in site.

Figure 15-The impact rating form used on backpacker campsites at Canyonlands National Park. For eachparameter, circle the number to the right of the appropriate category and then sum all of these ratings.

52

7. HUMAN WASTE

a. Toilet paper None present. 1 l-2 pieces of toilet 2 3-4 pieces of toilet 3 4 pieces of toilet 4paper present. paper. paper.

b. Fecal matter None present. 1. 1 pile of feces 2 2 piles of feces. 3 >2 piles of feces 4encountered. encountered.

8. FIREPITS

a. Number None present. 1 Sign of 1 small firering 2 1 firering >2' 3 >l firering. 4(<2’ diameter) diameter.

b. Rock scarring None. 1 <25% of rocks show fire 2 26-50% of rocks show 3 >50% show fire scars. 4scars. fire scars.

c. Charcoal None present. 1 Small trace of charcoal 2 Concentrated pile of 3 Charcoal and ash 4and ash and ash concentrated in charcoal and ash in scattered throughout

1 pile; site can be easily obvious pile. site, mixing into soil.returned to natural orundisturbed condition.

9. ROCK DISPLACEMENT

None. 1 1-5 small rocks (6" 2 >5 rocks moved; no 3 >5 rocks moved; tables, 4diameter) moved; no tables or seats seats. and other itemstables or seats constructed constructed.constructed.

10. TRASH

None present. 1 <4 pieces of trash, 2 4-6 pieces of trash. 3 >6 pieces of trash. 4biodegradable or non-biodegradable.

11. PESTS AND INSECTS

None. 1 1 small ant colony in 2 1 ant colony; ants in 3 >l ant colony; ants 4or at edge of site. <50% of site; few throughout site; numer-

scattered signs of ous signs of rodents:rodents within 20’ of tracks, burrows, nestssite. within 20' of site.

Excellent (E) = 25-37Good (G) = 36-62Fair(F) = 63-87Poor (P) = 88-100

Figure 15 (Con.)

53

Appendix J: The Delaware Water Gap Rapid Estimation Procedure

This procedure was initially quite similar to the BobMarshall procedure. With practice, it evolved into a pro-cedure that collects only interval level data and is nowmost similar to the Eagle Cap method of measurements

on permanent sampling units. Figure 16 shows the twopages of a completed field form. Impact parameters re-quiring explanation are as follows:

CAMPSITE INVENTORY AND IMPACT ASSESSMENT FIELD FORM

1)

2)

3 1

3)

4)

5)

6)

7)

8)

9)

10)

11)

12)

13)

14)

15)

16)

17)

18)

19)

20)

21)

22)

Site Number:

Site Name: 1=Calestini 2=Minisink 3=Namanock 24=Sandyston 5=Hornbeck 6=Shapnack 7=Buck Bar8=Tom’s Creek 9=Valley View 10-Peter’s 11-Quinn12=Decker’s 13=Sambo 14=Freeman 15=Hamilton 16=Depew17=Poxono 18=Hialeah 19=Schellenberger 20=Unnamed Site

Site Designation: 1-designated 2=undesignated I

River Segment: l=N-Milford 2:Milford-Dingmans3=Ding.-Bushkill 4=Bush.-Smithfield 5-Smith.-S 2

Site Location: 1=island 2=PA shore 3=NJ shore 1

Substrate of Landing Area: 1-bedrock 2=cobble3=sand 4=soil 2

Length of Shoreline Disturbance (m): 1 9

Distance to River (m): 1 3

NO. of Other Sites Visible from Campsite: 0

*************** Do Campsite Map Before Proceedinq *************

No. of 8x10 ft. Tent Pads: 5

Vegetative Ground Cover Onsite: 1=0-25%2=26-50% 3=51-75% 4=76-95% 5=96-100% 1

Vegetative Ground Cover Offsite: l=0-25%2=26-50% 3=51-75% 4:76-95% 5=96-100%

Type of Ground Cover Onsite: l=grass2=herbaceous 3=ferns 4=moss

5

1

Type of Ground Cover Offsite: l=grass2= herbaceous 3=f erns 4=moss 2

Tree Canopy Cover Over Site: 1=0-25%2=26-50% 3=51-75% 4=76-95% 5=96-100% 4

No. of Trees Within and On Site Boundaries: 1 4

NO. of Trees With Moderate-Severe Damage: 4

NO. of Tree Stumps Within and On Site Boundaries: 1

Total No. of Trails: 6

Type of Fire Site: l=stone 2=cement 3=steel 4=fire scar 1

No. of Fire Sites: 2

Toilet: l=clivus 2=pit toilet 3=no toilet present 3 a

Figure 16-The front (a) and rear (b) sides of a completed form used toinventory campsites at Delaware Water Gap National Recreation Area.

54

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -23) NO. of Garbage bags of Litter Present: - 0 . 2

24) No. of Human Waste Sites: a- -

25) Date (Day/Mo/Yr): 7 / 1 9 / 8 8- - - -- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Compass Bearings:0 22 45 67 90 112 135 157 180 202 225 247 270 292 315 337

x: 6.2| 6.0| 6.0|6.9| 6 .5 |4 .8 | 3.9| 4.0|:4.0| 4 . 2 | 4 . 5 | 6.0|7.3| 8.5|8.6|7.0

o:11.0|10.1|125|14.0|14.5|12.0|10.0|8.9|8.8|8.9|10.7|13.5|16.0|17.0|17.5|16.5|

26) Devegetated (X) Area: (sq. m.)27) Campsite (0) Area (computed in DBASE) + Satellite Area 27.7

- Island Area 10.2 (sq. m.)- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

28) Coded by: (names)

29) Comments:

bFigure 16 (Con.)

55

7. Length of Shoreline Disturbance: Distance (to thenearest meter) of shoreline where vegetation is absent orobviously disturbed by trampling. This judgment must bemade by comparing the site to undisturbed shoreline. Ifthe landing area is naturally barren (bedrock, for ex-ample), simply enter 1 m for the path width.

17. Number of Trees With Moderate-Severe Damage:A count of the number of trees (>l inch [2.5 cm] d.b.h.)within or on campsite boundaries with large branches cutor broken off and/or large or extensive knife or ax scars.Include trees within undisturbed “islands” and disturbed“satellite” areas. Multiple tree stems that are joined atthe base at or above ground level should be counted as onetree when assessing damage on any of its stems. If amultiple-stemmed tree has one of its stems cut, this wouldbe assessed as tree damage, not as a stump. Do not counttree stumps as tree damage.

18. Number of Tree Stumps: A count of the total num-ber of tree stumps (>1 inch [2.5 cm] diameter) within oron campsite boundaries. Due to the difficulty of differen-tiating stumps cut by humans from those created natu-rally, count all stumps regardless of origin.

23. Number of Garbage Bags of Litter Present: An esti-mate of amount of litter within the campsite and 100 ft(30 m) from the campsite boundaries, expressed as thenumber of 40-gallon garbage bags that could be filled withlitter and tied at top. Use decimals to indicate fractions ofa bag. Use zero if the site has only a handful of smallitems.

24. Number of Human Waste Sites: A count of thenumber of places with evident human waste and/or toiletpaper, within 100 ft (30 m) of campsite boundaries.

26. Devegetated Area: Area (in square meters) of thedevegetated central core of the campsite. Calculate inoffice, using procedures described in section on campsitemap.

27. Campsite Area: Area (in square meters) of anyground showing clear evidence of human disturbance.For procedures, refer to the following section.

Campsite Map: Draw a map of the campsite by connect-ing points on campsite boundaries along 16 transectsradiating from a center point. Begin by locating a centerpoint and referencing it to three permanent features,usually trees. Standing at the center point, consecutivelyestablish 16 transects, radiating from the center point,along the following bearings (not corrected for declina-tion): 0, 22,45, 67,90, 112, 135, 157, 180, 202, 225, 247,270,292,315, and 337 degrees. Measure the distancealong each transect to the first significant amount of vege-tation, defined as the first location where a 0.3- by 0.3-ft(1- by1-m) quadrat centered on the transect line wouldhave more than 25 percent vegetative cover. Measurefrom the center point to the closest edge of this imaginaryquadrat and record this distance for the appropriate bear-ing in the row labeled "X.” Also place an "X" on the mapat the measured distance (map intervals are equal to 1 meach). These will define the devegetated area, which willbe calculated in the office, using basic trigonometry. Atthe same time, measure the distance along each transectto the campsite boundary, indicated by a pronouncedchange in vegetation cover, height or composition, or insurface litter. Note this distance in the row labeled “0”and place an “0” on the map. These will define the maincampsite area, which will be calculated in the office usingbasic trigonometry. Finally, estimate the area of un-trampled “islands” of vegetation within campsite bounda-ries (to be subtracted from the total campsite area) andthe area of any disturbed “satellite” areas outside camp-site boundaries (to be added to the total campsite area).The size of these areas can be estimated by superimposingan appropriate geometric figure over the area and takingthe requisite linear measurements.

56

Appendix K: The Hardware and Software Used to Collect Data atYosemite (Source: Sydoriak in press)

Hardware- The HP71B microcomputer weighs only 12oz (340 g) and runs on four AAA batteries. With additionof an HP82162A thermal printer for hard copies, and theHP82401A HP-IL interface, which facilitates communica-tion between the two, the package weighs less than 3 lb(1.4 kg). The computer is small-l inch (2.5 cm) by 4inches (10 cm) by 8 inches (20 cm)-making it very easy totransport.

The HP7IB comes with only 16K RAM (random accessmemory). The risk of memory loss is significant in theunadorned HP71B. Therefore, two 64K RAM memorymodules from Firmware, Inc., complete with their ownbattery backup, were added. These retain their memoryeven when removed from the computer. When onebecomes full, the second can be installed in seconds.Programs and data are stored in the module in case ofmemory loss.

An HP82161A digital cassette drive is used to down-load data from the memory module to tape at the end ofeach collection period. It protects data and program filesfrom accidental memory loss.

The cassettes are unloaded to an IBM-compatible per-sonal computer via an HP821643A RS-232-C interface.Programs on the personal computer check the data forerrors not detectable in the field and prepare the data foranalyses.

Although watertight cases are available, the cost andweight are excessive. Heavy-duty Ziploc bags, though notwaterproof due to the need to run cables to the printer,keep dust and rain out, and permit keypunching. Thecomputer and printer are worn around the waist in acustomized carrying pouch. This arrangement leaves thehands free for keypunching.

Software-The HP71B computer has a powerful set ofBASIC functions. Customized data collection programswere developed to be as automatic as possible so thatuntrained individuals could learn to work with them eas-ily. As an added protection, a data entry and trouble-shooting guide is carried with the HP71B.

On power-up, a menu appears for the three main proce-dures: trails data input, campsite data input, or storage tocassette. After the operator indicates the desired pro-gram, it is automatically loaded. The most recent dataare displayed to aid in determining where the operatorleft off.

As each data field is entered, the program advances tothe next required field. Cursor keys allow access to anyfield in the current campsite or trail segment data group.Unfortunately, due to the absence of an editor (takes uptoo much memory), previous data groups are not acces-sible, and corrections must be recorded on paper and reen-tered on the office personal computer.

Each data item is checked for errors upon entry.Battery power is checked after each entry. When a lowbattery is detected, the anunciator signal is activated andthe machine locks up until new batteries are installed.According to Hewlett-Packard, a low-battery signal stillallows up to 15 minutes of program time. The use of alow-power state command, as in the programs, extendsbattery time.

When a new campsite is encountered, the previous dataare stored as text in a sequential file and variables arecleared for the new data set. Other options suspend pro-gram operation indefinitely to permit battery replacementor to interrupt the camp program to run the trails pro-gram. Total program, lex file, and associated file bytesneeded for the inventory program are about 15K Dataare stored in the 64K memory module. Only 32K at atime are accessible to a file, so checks have to be madefrequently to determine when a 32K file partition is be-coming filled. Firmware, Inc., states that this limitationmay soon be overcome.

57