perspectives from the northern...

REGIONAL LANDBIRD MONITORING

Regional landbird monitoring:

perspectives from the Northern Rocky Mountains

Richard L. Hutto and Jock S. Young

Abstract The Northern Region Landbird Monitoring Program (NRLMP) has been in place for near-

ly a decade. Based on our experience with this program, we offer thoughts on monitoring goals, the need for regional programs, and the components we believe make count-based

monitoring effective. The NRLMP includes a biennial bird survey conducted from permanently marked points for the purposes of long-term population-trend monitoring. Its primary strength, however, emerges from the inclusion of habitat information at each survey point, which allows rapid inference about land-use effects, and from an alternate-year focus on short-term management effects studies. Those who may be planning to initiate a

regional monitoring program may want to incorporate some of the perceived strengths of, and methodology associated with, the NRLMP while avoiding some of the pitfalls we have encountered. The power to detect both population trends and management effects might then emerge not only from the efforts of individual, regionally based programs but also from the collective effort of a variety of similarly designed regional programs.

Key words adaptive management, habitat relationships, landbird monitoring, survey methods

With many state and federal organizations (e.g., United States Forest Service, National Park Service, Bureau of Land Management) currently expressing an interest in starting regionally based monitoring programs, and given the current interest in developing a West-wide Avian Monitoring and Assess- ment Program (WAMAP), we believe it is timely to offer thoughts on the advantages and disadvantages of one region-wide landbird monitoring program that has now been in place for about 10 years. Our objectives are to discuss the overall goal of monitoring, the need for regionally based monitoring efforts, the pros and cons associated with alternative monitoring methods, and accomplishments associated with our own monitoring efforts to date. We hope these comments will serve to focus discussion about the overall goals and methodological issues associated with monitoring.

The overarching goal of wildlife monitoring

Most land management agencies allow a variety of activities to be conducted on their lands, but there is also an overriding constraint that species, ecosystems, and processes be maintained on the landscape. This constraint, within which we must manage public lands, is stated in one form or another in federal legislation such as the National Forest Management Act, National Environmental Policy Act, Fish and Wildlife Conservation Act, Migratory Bird Treaty Act, and National Park Service Organic Act. Various state laws also require that land management agencies not operate in a way that threat- ens the viability of native wildlife populations. While it is true that the narrow goal of bird monitoring is to carry out what those laws require

Authors' address: Division of Biological Sciences and Wildlife Biology, University of Montana, Missoula, MT 59812, USA; e-mail for Hutto: [email protected].

Wildlife Society Bulletin 2002, 30(3):738-750

738

Peer edited

Regional landbird monitoring * Hutto and Young 739

(Salwasser et al. 1983), or maybe to conserve all bird species or detect undesirable impending trends in landbird abundance or distribution, we believe there is considerable merit in stating that the overarching goal of any wildlife monitoring program is to assess whether humans are living sus- tainably. After all, sustainability is the primary conservation goal worldwide. Narrower monitoring goals (e.g., to conserve birds) are well-meaning but are somewhat misdirected because they put too much emphasis on nonhuman organisms for their own sake and fuel a misconception that conserva- tionists worry too much about nonhuman animals and plants and not enough about humans.

The maintenance of birds (and other taxa) would still follow necessarily as a subsumed goal because these taxa provide the necessary ecological servic- es for sustainability. A rich and varied flora and fauna is also essential to the creation of an environment that humans value most (Power 1996). Given this broad context for monitoring, birds can then be appreciated as an effective monitoring tool to determine whether we are indeed living sustain- ably, and if not, how we might operate in a more sustainable fashion. A similar monitoring framework has been advocated elsewhere (Greenwood et al. 1993,Welsh and Venier 1996).

The need for regional monitoring programs

Why set up independent regional monitoring programs when, in the case of birds, national and international efforts such as the Breeding Bird Sur- vey (BBS) are already in place? Simply, long-term monitoring in general, and the BBS in particular, will not enable us to accomplish what we could with supplemental monitoring. This issue has been discussed elsewhere (Downes and Welsh 1997, Sauer 2000), but here we emphasize just two points: 1) land managers want monitoring data that are more regional than national in scope, and the resolution of the BBS is too coarse for regional decision-making; moreover, the numbers of species for which reliable population trends can be generated from regional subsets of the BBS database are fewer than the number for which reliable trend estimates can be provided at the national level; and 2) more importantly, a design that includes habitat information with sample locations will facilitate the dis- covery of nonrandom habitat relationships, as would count durations that are longer than the 3-

minute counts conducted in the BBS. Indeed, we argue below that describing patterns of habitat use will make a much more effective program than one based on monitoring long-term population trends alone.

The selection of response variables Why monitor birds rather than some other taxonomic group?

If the goal of monitoring is to promote sustainable land use by monitoring land-use effects so that we can respond in an adaptive fashion if things are looking suspect, birds can then be seen as a useful tool to detect effects of current land-use practices (Greenwood et al. 1993). The detection of land-use effects will, in turn, allow us to project into the future based on current land-use trends and revise those activities that would otherwise generate negative bird population trends. Thus, bird monitoring fits well within a program of adaptive management. Birds are one of the best tools for monitoring because: 1) they are the most easily and inexpen- sively detected and identified vertebrate animals, 2) a single survey method can cover many species, and 3) accounting for and maintaining many species with different requirements promotes conservation strategies at the landscape scale (see Hutto [1998] for further discussion). We do not suggest that the bird community is the most important or most sensitive part of any functional ecosystem, nor do we believe birds should be the only entity monitored. We do believe, however, that monitoring selected birds across habitats may be the most cost-effective method of assessing a broad- based element of ecosystem integrity.

Abundance versus demography Point counts and other monitoring techniques

that provide only an index of abundance are often criticized for not including information on the reproductive success of each bird species present (DeSante and Rosenberg 1998). We agree with Mar- zluff et al. (2000) that any comprehensive monitoring program must have both count and demographic data. Relative abundance (count) data are needed to expose interesting habitat association patterns or land-use effects, and mechanistic studies based on demographic data are necessary to provide probable explanations for those abundance patterns. Nonetheless, because we currently lack adequate knowledge of conditions under

740 Wildlife Society Bulletin 2002, 30(3):738-750

Birds are one of the best tools for monitoring because a single survey method (point counts) can be used to easily and inex- pensively detect a relatively large number of species.

which each species occurs, let alone how well they are doing, we suggest that the best first step may be to gather count data to provide a rough approxi- mation of habitat relationships and population trends for a suite of species as ecologically diverse as possible. The demographics behind these issues can then be investigated more thoroughly later

using nesting, netting, banding, or other kinds of

study as needed. In addition, most regions have

already identified target or priority species based on known count or trend data. For those species, we should already be moving toward mechanistic studies to gain some insight into the causes behind demonstrated population-level differences or population declines. Because most of our experience with monitoring is based on count data, and such data are integral to monitoring efforts as argued above, the remainder of this paper deals exclusive-

ly with how to best generate count rather than

demographic data in association with monitoring efforts.

Which species should we monitor? Our experience suggests that we should focus

regional monitoring efforts on the large number of

species common enough to monitor easily through the use of one or a few simple field methods, and on those that are not well monitored by those methods but whose patterns of habitat association or biological considerations (other than rarity per se) suggest a potential management issue that needs a more intensive, focused monitoring effort. In a sense, we advocate using relatively common landbird species as an "indicator group." Threat- ened and endangered species will receive separate

monitoring or research attention in any event. No

generic, region-wide monitoring program will be able to generate useful trend information on rare

species; there will always be some limits to species coverage associated with monitoring due to financial and logistic constraints.

Although numerous landbird species can be monitored using a single field method, we do not advocate using summary statistics such as species diversity or guild totals as response variables just so rare species can be included in the data analysis. This is because, although members of guilds may use the environment similarly, they do not respond similarly to environmental influences (Mannan et al. 1984, Szaro 1986, Paige 1990). There are still sub- tle but important differences in the way any two

species use the environment on a year-round basis; no two species occupy the same niche. In addition, the response of rare species will be statistically swamped by the more common species within the same guild. Thus, guilds and diversity indices will remain useful for heuristic purposes in ecology, but will be ineffective and even misleading if used as

monitoring tools (Hutto 1989, Baker and Lacki

1997).

Long-term population trends versus other types of monitoring data

Long-term trend monitoring is essential because, while local activities may not be affecting organisms negatively, it may very well be that activities outside the jurisdiction of a given agency cause declines in the organisms that reside therein. For example, land-use patterns in Mexico could affect populations of bird species that breed in Montana. The

only way to detect the presence of a problem lying outside the area that is actively managed would be

through long-term population-trend monitoring. Another value of monitoring long-term population trends is that it might allow us to detect negative effects of local activities (e.g., recreational activity or

pesticide use) that operate independently of local land-use changes. However, a program that relies

entirely on the monitoring of long-term population trends will always be reactive. The monitoring of

long-term population trends is useful for discover-

ing whether populations are in decline but not very useful for discovering the reason behind such declines. Population-trend data generally allow only retrospective analyses, which are great for generating hypotheses but much weaker than the use of

experimentation and adaptive management to


discern causes behind trends (Nichols 2000). Unfortunately, most monitoring efforts seem to focus primarily on the tracking of population trends.

So how can we make monitoring programs more

proactive? The only information we need in order to anticipate declines in bird species due to simple habitat loss are data on patterns of bird distribution in relation to vegetation types (including disturbed and relatively undisturbed categories), coupled with information on trends in coverage of those vegetation types. We refer to the distribution of organisms in relation to vegetation types as "habitat association" information. Habitat association data can help us move beyond long-term population- trend monitoring, which most of us equate with the term "monitoring." Perhaps most importantly, habitat association data can be used to anticipate problems, thereby allowing an agency to modify its activities as a result of anticipated effects instead of waiting to react to a long-term trend that looks bad. As a hypothetical example, habitat association data might tell us that a particular species is restricted in distribution to forest type A. If forest type A is dis- appearing from the landscape, that should be adequate warning that trouble is in the works, even if the species is too rare for population-trend monitoring or if the trend in vegetation change is not yet far enough along to yield a population trend that reflects current change in vegetation coverage. Pro-

jection requires knowledge of trends in coverage of vegetation types, of course, but these trends are increasingly available from various sources.

If we include both habitat associations and management effects studies in our monitoring programs, it will become apparent that it does not take a 10- to 20-year commitment to do meaningful monitoring. Monitoring can and should be much more than the effort to track population trends; it can be a much more proactive effort to understand the effects of human activities that will persist or increase in intensity or extent over time. Those, after all, are the activities over which we have some control. Indeed, with respect to landbird monitoring, the current Partners in Flight (PIF) landbird monitoring strategy (Bart and Ralph 2001) lists the integration of long-term monitoring programs and short-term assessments as one of its main recom- mendations. In summary, monitoring should involve both the tracking of long-term population trends and the description of habitat associations and land-use effects. We wish to highlight and

emphasize the power of habitat association data, however, because those data do not appear to be commonly used or highly valued in monitoring programs, as perhaps they should be.

The Northern Region Landbird Monitoring Program

The Northern Region Landbird Monitoring Pro-

gram (NRLMP) involves the breeding season monitoring of all diurnal (primarily forest) landbird

species that can be detected through a single (point-count) methodology. The program began in 1990 with preliminary studies of the potential bias associated with roadside counts (Hutto et al. 1995) and of the logistical problems associated with visit- ing randomly placed points. This led, in 1994, to the

development of a feasible monitoring protocol (Hutto et al. 1998), and a full-scale monitoring effort using about 3,500 permanently marked points. The

program is currently region-wide, but primarily within United States Forest Service lands (Northern Region), with additional participation by Plum Creek Timber Company; Potlatch Corporation; Bureau of Land Management; United States Fish and Wildlife Service; National Park Service; Montana Department of Fish,Wildlife and Parks; and the Con- federated Salish and Kootenai Tribes. These partners comprised the kernel of what we believe was to become the first statewide landbird monitoring working group in the nation. This working group subsequently evolved into the Montana Partners in Flight working committee.

The regional program serves to coordinate monitoring activities conducted at a regional level, and this coordination provides a sharing of expertise as well as a consistency of design and methods. Each local unit (district, forest) benefits from the increased power of the regional data set, while sav- ing time and money that would otherwise be spent planning and conducting their own monitoring programs.

The dual focus of the NRLMP is to produce short- term results on habitat relationships and management effects as well as long-term population-trend data. In the short term, the NRLMP collects habitat data in association with permanent monitoring points so that habitat relationships can be deter- mined, and it also conducts short-term studies of the effects of specific management practices on selected bird species. This dual monitoring focus is unique to the NRLMP and other more recently


developed programs (e.g., Huff et al. 2000) and, we

believe, strengthen it beyond what might be achieved through a long-term, population-trend monitoring program alone.

Overall design The NRLMP consists of a series of more than 300

transects located along roads or trails, each transect

consisting of 10 permanently marked points, at which bird point counts are conducted according to a standard point-count protocol (Ralph et al.

1995, Hamel et al. 1996). Transects are located in a

geographically stratified random fashion across the Forest Service Northern Region on Forest Service lands and lands of several cooperating agencies. The points provide a representative sample of all cover types available, including managed vegetation types. The inclusion of managed lands is the key to

gaining inference about land-use effects from a ret-

rospective, observational data set. The permanently marked transect points are visited every other

year, and the alternate years are devoted to gathering quasi-experimental data on the effects of various land-use practices (e.g., grazing, timber harvest-

ing, prescribed fires).

Long-term population-trend monitoring design

Region-wide, long-term trends in population abundance can be achieved by sampling in a geographically stratified but otherwise random and unbiased manner using population-based monitor-

ing designs. A population-based approach to bird

monitoring spreads survey locations randomly across a region, irrespective of habitat. The philos- ophy is to determine an overall population trend for the entire region, with points that fall wherever

they may in various cover types, landscapes, man-

aged habitats, and heterogeneous mosaics. Exam-

ples of such monitoring efforts include the BBS and our own monitoring program. As long as the same

points are sampled over a specified period of time, overall population trends are relatively simple to calculate and are robust.

Efficiency generally requires modifications to a

purely random placement of points, however, and that will introduce some bias. The need to use roads is one such modification. The desire to strat-

ify by habitat is another. Although randomly located points might provide a good overall picture of bird populations in a region and would include all

types of landscapes, such a design does not sample

Including the full range of managed conditions within each vegetation cover type, such as this seed-tree cut within a western larch (Larix occidentalis) forest, is the key to gaining inference about land-use effects from a retrospective, observational data set.

rare but species-rich (e.g., riparian) habitats suffi-

ciently enough to generate population trends for

many species that are restricted to those habitat

types. Thus, the most efficient method of moni-

toring the greatest variety of species will include stratification not only by geography but by habitat as well (Margules and Austin 1994, Howe et al.

1995). The stratification of point locations by habitat typically requires more planning and effort than simple random placement of transects, however. Survey points need to be established in pre- selected habitats according to set criteria, and

although tools for locating specific habitat patch- es (e.g., GIS) are available, they also often require field verification. These logistic problems often lead to fewer than the anticipated numbers of

independent sites being used, which decreases

power to detect habitat effects.


While some habitat stratification seems warrant-

ed, entirely habitat-based monitoring designs (Huff et al. 2000, Leukering et al. 2000), which stratify all

point locations by habitat to assure that various habitats and habitat specialists are adequately covered, will produce biased estimates of overall population trends because they spread the sampling effort non-

randomly (evenly) over habitats of interest. Thus, the

population characteristics resulting from such a

design will not be representative of those within the

larger landscape mosaic. Another drawback to a

purely habitat-based monitoring design is the associated requirement that the habitat surrounding each

point or transect is extensive enough to adequately represent the target habitat type. This will further bias samples away from heterogeneous landscapes, edges, and other early-successional habitats (Brath- waite 1991, Howe et al. 1995) where some of the most seriously declining species may occur (Hunter et al. 2001).

The current design of the NRLMP incorporates transects that are positioned randomly along roads and trails in Forest Service lands within geographically stratified blocks. The points are then placed systematically within transects at 300-m intervals. The random component allows a general survey of

lands, including altered cover types, edges, small

stands, and other heterogeneous areas. About half the points end up in relatively homogeneous vegetation types. Perhaps the best design for population-trend monitoring sticks as closely as possible to the ideal model for population-based monitor-

ing, which includes geographically stratified random transects but also includes relatively few additional points to increase coverage in otherwise

undersampled habitat types (Howe et al. 1995). We also recommend the use of permanently

marked points for long-term population-trend mon-

itoring. This will guarantee that the same points are

sampled every time, ensuring that variation among years is due to variation in bird occupancy or activ-

ity patterns at the point, and not to variation in

point location. Monitoring programs that allow

year-to-year changes in point locations will add considerable variation to the long-term monitoring data, and this increased variation will reduce the

power to detect trends. Having permanent points will also aid in habitat analyses, permitting flexibili-

ty in repeating or changing vegetation data collection methods, and tracking habitat changes over time. A prime example of this benefit involves the more than 100 survey points within the Bitterroot

Valley area that burned in 2000. We were able to

capitalize on the fire event by conducting surveys at precisely the same points after the fires that were visited before the fires, and will be able to use the data in a highly valued before-after-control-impact (BACI) design to evaluate effects of stand-replace- ment fires on bird populations.

Habitat association monitoring design The key to generating meaningful habitat rela-

tionships involves sampling from fairly homogeneous habitat categories. If we use the random

points that fall within relatively homogeneous vegetation types, and care is taken to add points in less common but relatively important habitats, that

design should also be sufficient to generate information for each of the habitat or vegetation types used for stratification purposes. The inclusion of habitat data with our long-term monitoring points allowed us to build meaningful habitat-relationship models for more than 50 bird species in only three

years (Hutto and Young 1999). Moreover, a retro-

spective analysis of data from our long-term moni-

toring points revealed the same effects of partial- cut timber harvesting as were revealed through a

separate experimental effort (Young and Hutto

2002b). To understand habitat relationships, there must

be a concerted effort to sample sufficiently (>30 points as recommended in Ralph et al. 1995) from each category of interest, and this will normally require samples above and beyond those permanently marked points that are associated with a

long-term monitoring program. This is especially true if we wish to explore habitat associations at a

meaningful resolution. We have found the most useful habitat-based information comes from sam-

pling a variety of successional stages and management categories (e.g., burned, young, old, recently cut) within the general cover types (e.g., pine for-

est) that typify most habitat-based monitoring programs. Indeed, even heterogeneous edges of various sorts should be included as vegetation categories that might be favored by some species.

One way to accomplish the task of bolstering sample sizes within each of a large number of habitat categories is to conduct long-term monitoring at

permanent points on a biennial basis, which leaves

every other year free not only to conduct quasi- experiments on management effects but also to pur- sue the acquisition of additional data from non-permanently marked points located within otherwise

744 Widlife Society Bulletin 2002, 30(3):738-750

undersampled habitat categories. Because habitat association data can be derived not only from permanently marked points but also from single visits to non-permanently marked points in alternate years, the habitat-association database will grow continuously. Moreover, other researchers can make use of the NRLMP habitat-based data for their own analytical studies (for use of data from the NRLMP see White and Bennetts 1996; Thornton 1997; Karl et al. 2000, 2002) and can contribute to the habitat database as well. One difficulty with sharing data is that no two studies or monitoring programs use habitat or vegetation classification schemes that are exactly the same, so coordination with similar efforts elsewhere is problematic. To make matters worse, managed lands are becoming more and more difficult to pigeonhole because treatment styles are ever changing and seemingly infinite in number. This means the collection of local-scale vegetation information (e.g., canopy cover, or percent coverage of various plant species) may be critical to understanding bird distribution patterns in the future.

In addition to categorizing the vegetation cover type at a point, we measure various local-scale vegetation characteristics surrounding every point (see Hutto et al. 1998 for details), and have found that acquiring accurate vegetation data requires considerable training. We now have four estimates by different field observers in separate years, an independent sample by experienced forestry crews from additional plots at most points, and 2-4 photographs taken at every point. These independent sources of information have allowed us to discover that errors in even something as simple as the cat- egorization of vegetation type at a point are sur- prisingly common, and that photographs may be necessary to reconcile any discrepancies among observers. We highly recommend photo documen- tation for one-time visits to points that might be used to bolster a habitat-relationships database.

We also build regression models to expose finer- resolution patterns of occurrence due to the continuously variable nature of vegetation features within cover types (Young and Hutto 2002a). We need to know what these vegetation features are if we are to manage bird species successfully. We also need to know whether there are important variables that cannot be detected and mapped through remote sensing. If so, predicting suitabili- ty of sites on the basis of remotely sensed data alone will be difficult at best.

To look more closely at landscape influences on occurrence patterns, the precise locations of all monitoring points are also geo-referenced. We used these data to explore the distribution of brown- headed cowbirds (Molothrus ater) throughout the region and discovered that proximity to agriculture was the single most important influence on cowbird presence at a point (Young and Hutto 1999). We have also shown probable area effects for a variety of additional bird species, including Townsend's warbler (Dendroica townsendi) and golden- crowned kinglet (Regulus satrapa); (Hutto 1997).

Effects of management activities By conducting permanent, long-term monitoring

transects on an alternate-year basis, we are free to use the intervening years, not only to add to our undersampled vegetation cover types but also to study the effects of management activities as well. Coordination and funding are in place to efficiently survey a large number of sites across the region- so many, in fact, that the level of true site-level replication (30-40 sites per treatment category) is almost unheard of in published studies (see summary of the usual levels of replication in Sallabanks et al. 2000). These alternate-year monitoring efforts have the potential to be a very powerful way to provide managers with management-oriented results in the short term, so that planning can be improved before long-term trends might reveal a problem. We have already conducted 2 management-effects studies (one on the effects of partial- cut logging and one on the effects of grazing in grasslands), and 2 additional studies were conducted in 2001 (the effects of ecological restoration in dry forests and grazing in willow riparian strips).

Use of roads and trails The placement of long-term points along roads

greatly increases one's ability to relocate permanently marked points, reduces travel time between points, reduces the chance of injury to field workers, yet still allows one to detect all species known to occur within a region (Hanowski and Niemi 1995, Hutto et al. 1995, Keller and Fuller 1995). Nonetheless, there are several types of potential biases when transects are placed along roads: 1) some bird species may use particular cover types only because of the presence of the road per se; 2) some bird species may be associated with particular cover types only because of the presence or density of roadside vegetation that is not normally


associated with the surrounding cover type; 3) roads are more likely to occur in some cover types and landscapes than others, thus biasing trends toward what happens in those types rather than what happens region-wide; and 4) long-term changes in habitat composition or changes in the abundance of potential nest predators near roads

may differ from those in unroaded areas, producing a biased picture of long-term population trends. As long as roads are not wide enough to create major habitat changes nearby, the relative distribution of a bird species among habitats should not be biased because of the use of roads as sampling corridors (Hutto et al. 1995). It is for this reason that we prefer to use smaller tertiary roads and trails rather than well-traveled roads as transect routes.

Even if survey points are located off-road, access constraints will almost certainly require them to be near roads, so a road bias will persist. Counts in roaded landscapes, whether the points are on roads or not, will be biased because of the last two issues outlined above. Only the direct effects of roads themselves (numbers 1 and 2 above) can be ame- liorated by off-road surveys near roads. Therefore, roadside counts will surely bias population-trend estimates by some unknown amount, but will probably not bias habitat relationships significantly. This tradeoff of some bias for a gain in logistical sim- plicity is generally considered worthwhile (Ralph et al. 1995,Johnson 2000), but the issue merits con- tinued attention.

Distance sampling: index versus density We agree with Thompson (2002) that unadjusted

counts are biased by some unknown amount, but we question the assertion (e.g., Burnham 1981, Thompson et al. 1998, Nelson and Fancy 1999, Rosenstock et al. 2002) that one simply cannot use unadjusted point-count data for monitoring purposes. We have yet to be convinced that density estimates somehow improve the quality of data to an extent that real biological patterns would not have been exposed otherwise. The single study (Nelson and Fancy 1999) showing that a density estimate from variable-circular plots (VCPs) was comparable to the known density of a single species at one point in time was also off by 244% at another time. In any event, whether adjusted counts provide an accurate estimate of density is irrelevant because the goal of most studies is not to determine the exact density of a species but to determine whether its abundance differs among

conditions, or whether its abundance changes through time. We suspect that conversions to den-

sity estimates may be as likely to hide as they are to reveal actual patterns because virtually every additional assumption needed to calculate density through the use of detectability profiles is clearly violated. The most critical violations are 1) birds move in response to observers (even to within detection range from outside that range during the count), which leads to biased detectability profiles; 2) observers cannot provide sufficiently precise estimates of distances to sources of bird sounds; 3) detectability profiles are biased estimates because

they are constructed from non-independent obser- vations taken repeatedly from the same points or transects; and 4) the lateral distance at the inflec- tion point on a detectability profile is not necessar-

ily equivalent to that special distance across which there is no net flow of birds toward or away from the observer. Moreover, even if assumptions were met, computer programs such as DISTANCE (Laake et al. 1994) solve only the lateral detectability issue; among-habitat variation in basal detectability is not dealt with at all.

Density calculations might help control for observer bias, but the only way one can control for observer bias using VCPs is to calculate lateral detectability profiles for each habitat by each observer separately, and that is impossible to accomplish because the data are too sparse. It has been estimated that from 75 to 100 detections are necessary to generate a detectability profile that is meaningful from point-count data (Thompson et al. 1998). If one then lumps non-independent data from all observers to build a reliable profile, it will defeat the purpose of using the technique to control for observer bias. Density calculations might also help control for habitat-based detectability bias, but the same problem persists because one needs a separate profile for each habitat. We believe the observer bias can be minimized through training, and the habitat bias lessened by using data from within a fixed radius (Hutto et al. 1986) without having to deal with additional unmet assumptions. Again, we agree that unadjusted counts are biased by some unknown amount, but we question whether the conversion of those data to density estimates solves the problem. We also note that the use of adjusted counts tends to promote undue confidence in results, even though the number of violated assumptions is greater than for unadjusted counts. Fortunately,


distance estimates are required whether one cal- culates density with the information or not, so differences of opinion on this issue should not ham- per the ability to share data.

Count duration We prefer 10-minute counts to those of shorter

duration for the following reasons:

* The study of relatively fine-scale habitat relationships necessitates analysis at individual points. When single points are used as sample units, counts of at least 10 minutes are recommended (Ralph et al. 1995, Drapeau et al. 1999). This improves the accuracy of habitat association analyses by increas- ing the probability of detecting an individual when it is present (Buskirk and McDonald 1995,Thomp- son and Schwalbach 1995);

* Our experience with breeding bird counts and especially with winter counts in Mexico (Hutto et al. 1986) has revealed that observers often require more than 5 minutes to record all birds heard at a point and then take time to refocus on songs they may have overlooked;

* Longer (10-minute) counts should help reduce sources of variation due to 1) differences in detectability among species and individuals (e.g., with breeding status), 2) changes in activity levels through the morning and the season (allowing extension of useful sampling time), and 3) bias and error due to observer differences (Buskirk and McDonald 1995). Longer counts will also produce greater effectiveness in sampling less abundant or less conspicuous species on a per-point basis (Buskirk and McDonald 1995). Because we are try- ing to emphasize those uncommon and difficult-to- monitor species, counts of at least 10 minutes seem essential. For example, data from the NRLMP showed that only 50% of brown creepers (Certhia americana) were recorded in the first 5 minutes, even though 80-90% of many of the more common species were detected during that period. Other studies have shown that species were still accumu- lating at 10 minutes (e.g.,Verner 1988, Smith et al. 1993);

* Ten-minute counts have been recommended as a compromise between efficiency and the risk of double counting when travel time is significant (Verner 1988, Ralph et al. 1995,Welsh 1995);

* A 10-minute duration is commonly used in research involving point counts (Hutto et al. 1986) and in many already existing monitoring programs

(e.g., Nicolet National Forest Bird Survey, Vermont Forest Bird Monitoring Program, Ontario Forest Bird Monitoring Program, Ontario Wildlife Assess- ment Program), so the use of 10-minute counts should enable meaningful comparisons with most other point-count studies.

Intra-regional cooperation The most important step toward a regional moni-

toring program is, of course, securing participation by all landowners within the region of interest, including private landowners and government agencies. Unfortunately, it is usually difficult to get agree- ment from all parties on the need for a monitoring program or on the methodology. These problems are

diminishing, however, because there now appears to be great interest in the development of regional monitoring programs and a good deal of conver-

gence in methodologies among regional programs now in existence. Full financial participation in

regional monitoring by all partners is much more difficult to achieve. For example, financial participation by a broad cross-section of partners in the Northern

Rocky Mountain Region was probably hampered from the start because federally earmarked (United States Forest Service) funds were used to get the NRLMP up and running. This was key within the

agency but led to the perception that landbird mon-

itoring was exclusively a Forest Service program. The Rocky Mountain Bird Observatory (formerly

the Colorado Bird Observatory) has been the model of success in achieving the necessary level of cooperation and financial participation for

region-wide monitoring. This has apparently been achieved due to the economic efficiency of the program, its assignment of habitats to partners as a basis of cost assessment, and its coordination from a nongovernmental organization (NGO). Using an

independent NGO contractor may also help agencies deal with the potential problem of using agency personnel on non-agency lands, and probably results in a greater consistency in field skills

among field workers. Hiring field personnel from within various agencies gives them more owner-

ship in the program, however. Whichever model one selects, an NGO should probably be responsi- ble for selecting the pool of experienced techni-

cians, conducting training sessions, summarizing and analyzing data, and making data and results available to participants and to the public at large. This represents a considerable cost over and above that of collecting field data, but the cost must be


included and justified to financial partners if mean-

ingful products are to be part of the monitoring program. How the personnel load and financial burden are divided among the participants is probably best left to the participants in each region to decide. We believe that success in achieving full

participation lies with clearly defined roles and costs for each partner. However cooperation is achieved, it seems clear that monitoring requires long-term commitments of resources from a broad base of organizations (Huff et al. 2000). This is the single most difficult task associated with regional monitoring.

Integration into management and planning processes

The integration of monitoring and management is the essence of adaptive management (Holling 1978, Walters 1986, Walters and Holling 1990). Indeed, any monitoring program fits well within a scheme of adaptive management (Holling 1978, Sal- wasser et al. 1983,Walters 1986). Long-term trend monitoring and habitat-relationships monitoring (both of which accumulate data over time) allow us to bring findings at any time to planning meetings. Management plans can then be altered on the basis of exposed patterns. Indeed, partly because of the habitat relationships exposed by the NRLMP, the extent and intensity of post-fire salvage logging in the Northern Region is much less than it was a decade ago. Program results have also spawned new studies designed to look at whether timber- harvesting activities might actually create ecological traps for those species that appear to be nowhere more abundant than in harvested forest types.

Some individuals might argue that long-term monitoring programs cannot be part of an adaptive management scheme because adaptive management is an object-driven activity whereby management is conducted, results are monitored to see whether objectives are attained, and then management activity is adjusted if goals are not attained. This is only one of many ways to manage adaptive- ly, however. In another form, one could conduct baseline monitoring as an ongoing program to learn about management consequences that might not have been exposed otherwise. Indeed, most land- use activities are not monitored, as they should be, so a well-distributed ongoing monitoring program that is set up irrespective of any particular activity could be a useful tool in exposing cumulative man-

agement effects. This information could then be

incorporated in the planning process for future activities. That having been said, however, there is no reason we could not include object-driven adaptive management in a monitoring program by using every other year (in a program designed to conduct long-term monitoring on a biennial basis) to conduct fieldwork specifically designed to examine the effects of selected management activities. For

example, we used a BACI design involving a large number of treated and untreated plots within each of several forests during the 2001 field season to investigate the effects of forest restoration activities (thinning and burning) on birds. And although we have not yet conducted any demographic followup studies as part of our own program, we are aware of a number of studies that have been stimulated by results from this program. Our own program could easily accommodate such studies during the alternate years as well, using methods described elsewhere (e.g., Martin and Geupel 1993, Ralph et al. 1993, Nur et al. 1999). Such ongoing management- oriented aspects are essential to a successful, integrated monitoring program (DeSante and Rosen- berg 1998).

If there is one weakness associated with adaptive management in practice, it is the lack of a formal involvement of monitoring participants in the adaptive management loop, where participants have a chance to present results that might bear on future land-use plans. Although monitoring is at the core of adaptive management and "...essentially synony- mous with effective decision-making" (Mulder and Palmer 1999:6), monitoring results need to be integrated into the decision-making process. Monitor- ing should be part of a management planning cycle that involves 1) gathering data on long-term population trends and short-term land-use effects, 2) informing planners of results, and 3) discussion of whether results merit consideration of changes in land-use plans.

Most of our findings that have influenced policy have done so because the information filtered infor- mally into management circles. Once agencies fully appreciate that birds can be used as effective tools to accomplish their monitoring goals, the formal involvement of monitoring information in the planning process should follow closely behind. Until then, we suggest that an integral part of any effective monitoring program design should include formal agreements with land management agencies that consideration of monitoring results be part of


their planning process. "Communication needs to combine effective evaluation with regular report- ing to meet the needs of biodiversity conservation" (Welsh and Venier 1996: 57).

Inter-regional coordination The potential benefit associated with the opera-

tion of similarly designed but independently run monitoring programs cannot be overstated. Not only is there political power associated with becoming part of a larger effort, but there is also considerable statistical power through the combi- nation of data sets. We encourage funding for the development and maintenance of formal regional coordination centers that could house data, produce reports, disseminate information, facilitate coordination among projects within a region, and train and certify field workers so that we can be assured that they have the skills necessary to conduct reliable work.

We hope these thoughts will stimulate discussion and, more importantly, stimulate the development of well-coordinated regional monitoring efforts.

Acknowledgments. We would like to acknowl- edge the efforts of those at the National Fish and Wildlife Foundation and all the other participants at the 1990 Atlanta meeting, which signaled the beginning of Partners in Flight and resulted in the availability of federally earmarked funds that were used to develop the NRLMP. We especially thank A. Christensen for consulting with the senior author and S. Hejl about how to best use those earmarked funds in the Northern Region and to S. Kowalski for seeing to it that the monitoring effort was sustained past the end of that funding period. Finally, we thank S. Kowalski, P. Hendricks, and S. Ritter for comments on an earlier draft of the manuscript.

Literature cited BAKER, M. D., ANI) M. J. LACKI. 1997. Short-term changes in bird

communities in response to silvicultural prescriptions. For- est Ecology and Management 96:27-36.

BART, J. AND C. J. RALPH. 2001. The Partners In Flight landbird monitoring strategy. United States Fish and Wildlife Service, Office of Migratory Bird Management,Washington, D.C., USA.

BRATHWAITE, R. W 1991. Fauna and habitat surveys as ecological pathfinders. Pages 23-28 in C. R. Margules and M. P. Austin, editors. Nature conservation: cost effective biological surveys and data analysis. Commonwealth Scientific and Indus- trial Research Organization, Melbourne, Australia.

BURNHAM, K. P. 1981. Summarizing remarks: environmental influences. Studies in Avian Biology 6: 324-325.

BItSKIRK, W H., AND J. L. MCDONALD. 1995. Comparison of point

count sampling regimes for monitoring forest birds. Pages 25-34 in C.J. Ralph,J. R. Sauer, and S. Droege, editors. Moni-

toring bird populations by point counts. United States Forest Service General Technical Report PSW-GTR-149.

DESANTE, D. F, AND D. K. ROSENBERG. 1998. What do we need to monitor in order to manage landbirds? Pages 93-106 in J. M. Marzluff and R. Sallabanks, editors. Avian conservation: research and management. Island, Covello, California, USA.

DOWNES, C. M., AND D. A. WELSH. 1997. Monitoring Canada's landbirds: an integrated approach. Page 62 in E. H. Dunn, M. D. Cadman, and J. B. Falls, editors. Monitoring bird populations: the Canadian experience. Canadian Wildlife Service Occasional Paper 95.

DRAPEAU, P., A. LEDUC, AND R. MCNEIL. 1999. Refining the use of

point counts at the scale of individual points in studies of bird-habitat relationships. Journal of Avian Biology 30: 367-382.

GREENWOOD, J. J. D., S. R. BAILLIE, H. Q. P. CRICK, J. H. MARCHANT, AND W. J. PEACH. 1993. Integrated population monitoring: detecting the effects of diverse changes. Pages 267-342 in R. W Furness and J. J. D. Greenwood, editors. Birds as moni- tors of environmental change. Chapman and Hall, London, United Kingdom.

HAMEL, P. B., W P. SMITH, D. J. TWEDT, J. R. WOEHR, E. MORRIS, R. B.

HAMILTON, AND R. J. COOPER. 1996. A land manager's guide to

point counts of birds in the Southeast. United States Depart- ment of Agriculture Forest Service General Technical Report SO-120.

HANOWSKI, J. M., AND GJ. NIEMI. 1995. A comparison of on- and off-road bird counts: do you need to go off road to count birds accurately? Journal of Field Ornithology 66:469-483.

HOLLING, C. S. 1978. Adaptive environmental assessment and management. Wiley, New York, New York, USA.

HOWE, R. W, A. T. WOLF, AND T. RINALDI. 1995. Monitoring birds in a regional landscape: lessons from the Nicolet National forest bird survey. Pages 83-92 in C.J. Ralph, J. R. Sauer, and S. Droege, editors. Monitoring bird populations by point counts. United States Forest Service General Technical

Report PSW-GTR-149. HUFF, M. H., K. A. BETTINGER, H. L. FERGUSON, M. J. BROWN, AND B.

ALTMAN. 2000. A habitat-based point-count protocol for ter- restrial birds, emphasizing Washington and Oregon. United States Forest Service General Technical Report PNW-GTR- 501.

HUNTER, W C., D. A. BUEHLER, R. A. CANTERBURY, J. L. CONFER, AND P. B. HAMEL. 2001. Conservation of disturbance-dependent birds in eastern North America. Wildlife Society Bulletin 29: 440-455.

HUTTO, R. L. 1989. The effect of habitat alteration on migratory landbirds in a west Mexican tropical deciduous forest: a conservation perspective. Conservation Biology 3:138-148.

HUTTO, R. L. 1997. An investigation of the potential threat of habitat fragmentation on some Northern Region bird species. Final Report, United States Forest Service Northern Region, Contract 53-0343-2-0007, Missoula, Montana, USA.

HUTTO, R. L. 1998. Using landbirds as an indicator species group. Pages 75-92 in J. M. Marzluff and R. Sallabanks, editors. Avian conservation: research and management. Island, Washington, D.C., USA.

HUTTO, R. L., S. J. HEJL, J. F KELLY, AND S. M. PLETSCHET. 1995. A

comparison of bird detection rates derived from on-road vs. off-road point counts in northern Montana. Pages 103-110 in C.J. Ralph, J. R. Sauer, and S. Droege, editors. Monitoring


bird populations by point counts. United States Forest Ser- vice General Technical Report PSW-GTR-149.

HLTTO, R. L., S. M. PLETSCHET, P. HENDRICKS. 1986. A fixed-radius

point count method for nonbreeding and breeding season use. Auk 103:593-602.

Hi TTO, R. L., AND J. S. YOUNG. 1999. Habitat relationships of landbirds in the Northern Region. United States Forest Service General Technical Report RMRS-GTR-32.

Htrro, R. L., J. S. YOUNG, AND J. HOFFIAND. 1998. United States

Department of Agriculture Forest Service Northern Region Landbird Monitoring Project: field methods. Interim Report. United States Forest Service Northern Region, Contract 53-0343-7-0007, Missoula, Montana, USA.

JOVINSON, D. H. 2000. Statistical considerations in monitoring birds over large areas. Pages 115-120 in R. Bonney, D. N. Pashley, R. J. Cooper, and L. Niles, editors. Strategies for bird conservation: the Partners in Flight planning process. United States Department of Agriculture, Forest Service, Proceed-

ings RMRS-P-16, Ogden, Utah, USA. KARL, J. W, P J. HEDGLUJND, E. O. GARTON, J. M. ScoTt, N. M. WRIGHT,

ANI) R. L. HTUTTO. 2000. Sensitivity of species habitat-rela-

tionship model performance to factors of scale. Ecological Applications 10: 1690-1705.

KARL, J. W., L. K. BOMAR, P J. HEGLUND, N. M. WRIGHT, AND J. M.

SCOTT. 2002. Species commonness and the accuracy of habi-

tat-relationship models. Pages 573-580 in J. M. Scott, P. J. Heglund, M. L. Morrison,J. B. Haufler, M. G. Raphael,W. A.Wall, and F B. Samson, editors. Predicting species occurrences: issues of accuracy and scale. Island, Covello, California, USA.

KELLER, C. M. E. AND M. R. FULLER. 1995. Comparison of birds detected from roadside and off-road point counts in the Shenandoah National Park. Pages 111-116 in C.J. Ralph,J. R. Sauer, and S. Droege, editors. Monitoring bird populations by point counts. United States Forest Service General Technical

Report PSW-GTR-149. LAAKE, J. L., S. T. BiCKLANI), D. R. ANDERSON, AND K. P. BURNAM.

1994. DISTANCE. Colorado Cooperative Fish and Wildlife Research Unit, Colorado State University, Fort Collins, USA.

LEIIKERING, T., M. F CARTER, A. PANJABI, D. FAULKNER, AND R. LEVAD.

2000. Monitoring Colorado's birds: the plan for count-based

monitoring. Rocky Mountain Bird Observatory, Denver, Col- orado, USA.

MANNAN, R. W, M. L. MORRISON, AND E. C. MESLOW. 1984. Com-

ment: the use of guilds in forest bird management. Wildlife Society Bulletin 12:426-430.

MAR(ILI.ES, C. R., ANI) M. P AUJSTIN. 1994. Biological models for

monitoring species decline: the construction and use of data- bases. Philosophical Transactions of the Royal Society of London Biological Sciences 344(1307):69-75.

MARTIN, T. E., AND G. R. GElUPEL. 1993. Nest-monitoring plots: methods for locating nests and monitoring success. Journal of Field Ornithology 64:507-519.

MARZ.LUFIF, J. M., M. G. RAPHAEI, AND R. SALIABANKS. 2000. Under-

standing the effects of forest management on avian species. Wildlife Society Bulletin 28:1132-1143.

MIIDI)ER, B. S., AND C. J. PALMER. 1999. Introduction to effectiveness monitoring. Pages 1-19 in B. S. Mulder, B. R. Noon,T. A.

Spies, M. G. Raphael, C.J. Palmer, A. R. Olson, G. H. Reeves, and H. H.Welsh, editors. The strategy and design of the effectiveness monitoring program for the Northwest Forest Plan. United States Forest Service General Technical Report PNW- GTR-437.

NEI.SON, J. T., AND S. G. FANCY. 1999. A test of the variable circu-

lar-plot method where exact density of a bird population was known. Pacific Conservation Biology 5:139-143.

NICHOLS, J. D. 2000. Monitoring is not enough: on the need for a model-based approach to migratory bird management. Pages 121-123 in R. Bonney, D. N. Pashley, R.J. Cooper, and L. Niles, editors. Strategies for bird conservation: the Partners in Flight planning process. United States Department of Agri- culture, Forest Service, Proceedings RMRS-P- 16, Ogden, Utah, USA.

NUR, N., S. L. JONES, AND G. R. GEUPEL. 1999. Statistical guide to

data analysis of avian monitoring programs. United States Fish and Wildlife Service Biological Techniques Publication BTP-R6001-1999.

PAIGE, L. C. 1990. Population trends of songbirds in western North America. Thesis, University of Montana, Missoula, USA.

POWER, T. M. 1996. Lost landscapes and failed economies: the search for a value of place. Island,Washington, D.C., USA.

RALPH, C. J., G. R. GEUPEL, P. PYLE, T. E. MARTIN, AND D. F DESANTE.

1993. Handbook of field methods for monitoring landbirds. United States Forest Service General Technical Report PSW- GTR-144.

RALPH, C.J., S. DROEGE, ANDJ. R. SAUER. 1995. Managing and mon-

itoring birds using point counts: standards and applications. Pages 161-168 in C.J. Ralph,J. R. Sauer, and S. Droege, editors.

Monitoring bird populations by point counts. United States Forest Service General Technical Report PSW-GTR-149.

ROSENSTOCK, S. S., D. R. ANDERSON, K. M. GIESEN, T. LEUKERING, AND

M. F CARTER. 2002. Landbird counting techniques: current

practices and an alternative. Auk 119:46-53. SALLABANKS, R., E. B. ARNETT, AND J. M. MARZLUFF. 2000. An evalu-

ation of research on the effects of timber harvest on bird

populations. Wildlife Society Bulletin 28:1144-1155. SALWASSER, H., C. K. HAMILTON, W. B. KROHN, J. F LIPSCOMB, AND C.

H. THOMAS. 1983. Monitoring wildlife and fish: mandates and their implications. Transactions of the North American Wildlife and Natural Resources Conference 48:297-307.

SAUER, J. R. 2000. Combining information from monitoring programs: complications associated with indices and geograph- ic scale. Pages 124-126 in R. Bonney, D. N. Pashley, R.J. Coop- er, and L. Niles, editors. Strategies for bird conservation: the Partners in Flight planning process. United States Depart- ment of Agriculture, Forest Service, Proceedings RMRS-P-16, Ogden, Utah, USA.

SMITH, W. P., D.J. TWEDT, D. A. WIEDENFELD, P. B. HAMEL, R. P FORD,

AND R. J. COOPER. 1993. Point counts of birds in bottomland hardwood forests of the Mississippi Alluvial Valley: duration, minimum sample size, and points versus visits. IUnited States Forest Service Research Paper SO-274.

SZARO, R. C. 1986. Guild management: an evaluation of avian

guilds as a predictive tool. Environmental Management 10: 681-688.

THOMPSON, F R., AND M. J. SCHWALBACH. 1995. Analysis of sample size, counting time, and plot size from an avian point count survey on Hoosier National Forest, Indiana. Pages 45-48 in

C.J. Ralph,J. R. Sauer, and S. Droege, editors. Monitoring bird

populations by point counts. United States Forest Service General Technical Report PSW-GTR-149.

THOMPSON, W. L., G. C. WHITE, ANI) C. GOWAN. 1998. Monitoring vertebrate populations. Academic, NewYork, NewYork, USA.

THOMPSON, W. L. 2002. Towards reliable bird surveys: accounting for individuals present but not detected. Auk 119:18-25.

THORNTON, PC. 1997. An investigation of the potential threat of habitat fragmentation on some Northern Region bird


species. Thesis, University of Montana, Missoula, USA. VERNER, J. 1988. Optimizing the duration of point counts for

monitoring trends in bird populations. United States Forest Service Research Note PSW-395.

WALTERS, C. J. 1986. Adaptive management of renewable resources. MacMillan, New York, New York, USA.

WALTERS, C. J., AND C. S. HOLLING. 1990. Large-scale management experiments and learning by doing. Ecology 71:2060-2068.

WELSH, D. A. 1995. An overview of the Forest Bird Monitoring Program in Ontario, Canada. Pages 93-97 in C.J. Ralph,J. R. Sauer, and S. Droege, editors. Monitoring bird populations by point counts. United States Forest Service General Technical Report PSW-GTR-149.

WELSH, D. A., AND L. A. VENIER. 1996. Binoculars and satellites:

developing a conservation framework for boreal forest wildlife at varying scales. Forest Ecology and Management 85:53-65.

WHITE, G. C., AND R. E. BENNETrS. 1996. Analysis of frequency count data using the negative binomial distribution. Ecology 77:2549-2557.

YOUNG, J. S., AND R. L. HUTTO. 1999. Habitat and landscape factors affecting cowbird distribution in the northern Rockies. Studies in Avian Biology 18:41-51.

YOUNG, J. S., AND R. L. HUTTO. 2002a. Use of regional-scale exploratory studies to determine bird-habitat relationships. Pages 107-119 in J. M. Scott, P.J. Heglund, M. L. Morrison,J. B. Haufler, M. G. Raphael, W. A. Wall, and E B. Samson, editors. Predicting species occurrences: issues of accuracy and scale. Island, Covello, California, USA.

YOUNG, J. S., AND HUTrO, R. L. 2002b. Use of a landbird monitoring database to explore effects of partial-cut timber harvesting. Forest Science 48:373-378.

Richard Hutto (left) is a professor of biology and wildlife biology at the University of Montana, where he teaches ornithology, Montana wildlife, and field ecology. Hutto serves as advisor and supervisor of the Northern Region Landbird Monitoring Program and has conducted research on habitat selection in landbirds in the northern Rocky Mountains during the breeding season, in the southwestern U.S. during the migratory seasons, and in western Mexico during winter. He received his M.S. from Northern Arizona University and Ph.D. from UCLA. Jock Young (right) is currently the coordinator of the Northern Region Landbird Monitoring Program. He received his B.S. from Oregon State University and M.S. degrees from the Uni- versity of California at San Diego and the University of Montana. His interests include bird monitoring and conservation, and the modeling of habitat relationships.

perspectives from the northern...

Documents