Attach_03_NFNC_Guide.pdf

United States Department of Agriculture

A Land Manager’s Guide to Point Counts of Birds in the Southeast

Forest Service

Paul B. Hamel, Winston Paul Smith, Daniel J. Twedt, James R. Woehr, Eddie Morris, Robert B. Hamilton, and Robert J. Cooper

Southern Research Station

General Technical Report SO-l 20 twoldu Typewritten Text twoldu Typewritten Text twoldu Typewritten Text

AG-4568-S-15-0024

Bird Point Survey NFNC Ranger Districts twoldu Typewritten Text twoldu Typewritten Text twoldu Typewritten Text twoldu Typewritten Text Attachment 03 twoldu Typewritten Text twoldu Typewritten Text

SUMMARY

Current widespread concern for the status of neotropical migratory birds has sparked interest in techniques for inventorying and monitoring populations of these and other birds in southeastern forest habitats. The present guide gives detailed instructions for conducting point counts of birds. It further presents a detailed methodology for the design and conduct of inventorial and monitoring surveys based on point counts, including discussion of sample size determination, distribution of counts among habitats, cooperation among neighboring land managers, vegetation sampling, standard data format, and other topics. Appendices provide additional information, making this guide a stand-alone text for managers interested in developing inventories of bird populations on their lands.

ACKNOWLEDGMENTS

This guide draws on the work of many individuals and committees in the current Partners in Flight ad hoc network, particularly C.J. Ralph, G S. Butcher, W.C.

Hunter, the National Monitoring Working Group, the Southeast Regional Management Working Group, and the Southeast Regional Monitoring Committee. In addition, specific studies conducted in the Mississippi Alluvial Valley are the basis for the quantitative estimates. This manuscript has been improved after review by David Buehler, Jaime Collazo, Ray Franzreb, Glen Gaines, Sid Gauthreaux, Mike Guilfoyle, Wade Harrison, Bob Hatcher, Chuck Hunter, Cherie Irby, Fred Kinard, Terrell Rich, Alan Schultz, and David Wiedenfield. Despite their generous attention, this manuscript may not necessarily reflect their wisdom;

the authors acknowledge responsibility for the approaches and opinions expressed here.

September 1996

Southern Research Station P.O. Box 2680

Asheville, NC 28802

Contents

INTRODUCTION

METHODS PRELIMINARY TO

POINT COUNTS

The Basics More Detail

WHAT IS A POINT COUNT?

The Basics More Detail

1 TRANSCRIBING DATA 18

Transcribing Field Data to a Bird List 18 Summarizing Field Data into Reports 19

2 HAPPY POINT COUNT CENSUSING! 19

WHO CONDUCTS A POINT

COUNT?

The Basics More Detail

CONDUCTING A POINT COUNT 5

The Basics 5

WHAT IS A POINT COUNT SURVEY? 7

Survey Design 7 Determination of Sample Size 1 1 Forming a Group of Cooperators When

Sample Size Exceeds Property Size 15

MEASURING VEGETATION 16

The Basics 16 More Detail 17

LITERATURE CITED 20

Appendix A. Equipment list for conducting point count surveys. 21

Appendix B. Codes for wind speed and sky condition used in conducting point counts. 21

Appendix C. Species codes for birds as found in the North American Bird Banding Manual (USDI 1991). 22

Appendix D. Proposed structure of computer files for point count data. 30

Appendix E. Determining sample size by the power method. 31

Appendix F. Sample vegetation measurements. 35

Appendix G. List and descriptions of computer files and suggested coding standards for Level Three vegetation variables. List follows a general sequence from more general to more specific data items. 38

A Land Manager’s Guide to Point Counts of Birds in the Southeast

Paul B. Hamel, Winston Paul Smith, Daniel J. Twedt, James R. Woehr, Eddie Morris, Robert B. Hamilton, and Robert J. Cooper

INTRODUCTION

Concern over declining numbers of neotropical migratory birds has sparked interest among managers in inventorying and monitoring these and other nongame birds (Hunter and others 1993a).

Although such assessments are time consuming and costly, they are essential in evaluating the status of these species so that informed management will be ensured. This guide is provided to assist managers in the Mississippi Alluvial Valley (MAV) and throughout the Southeast in accumulating inventories and monitoring changes in nongame birds in southeastern habitats.

This guide is consistent with the guidelines of the monitoring group of the Partners in Flight Neotropical Migratory Bird Conservation Program (Ralph and others in press). The standard handbook for monitoring landbirds, Ralph and others (1993), will be useful to readers of this guide. Intended to assist managers in designing, conducting, and maintaining monitoring, this guide is based on work conducted in the Mississippi Alluvial Valley since 1985, and reflects what wilI work on the ground. It is not a complete discussion of the various monitoring and inventory techniques, and cannot claim to provide all the answers; accordingly, additional sources of more technical information are indicated.

Statistical advice beyond that presented in this guide will be useful to managers as they cooperate and pool their data.

The Authors

This guide has been written by a group of land managers and research biologists whose individual opinions are strong and frequently opposed to each other. The group has a common focus, however. It believes that an inventory of nontarget resources, such as birds, is essential for wise land management.

It is equally convinced that monitoring changes in the status of such nontarget resources is both an urgent and an important part of land management activities. Indeed, for many public land managers, such monitoring is a part of legal requirements for maintaining viable populations. The group believes it is also imperative that managers cooperate with their neighbors in accumulating inventory information.

Keeping these complex issues in focus, however, is not easy Strong efforts have been made to maintain a perspective that is neither of entirely practical land management nor of purely research interest. The group sincerely hopes that it has not lost its audience somewhere between the work center and the ivory tower. Thus, readers are urged to follow the guide closely and not avoid the equations.

As a general approach, this guide strives to present a clear statement of a technique or need in the text; it then provides a more technical or extended discussion of the topic in an appendix. For efficiency of use, each topic is divided into “The Basics” and “More Detail.” “The Basics” should be enough for the reader to initiate inventory and monitoring activities. “More Detail” will be available when needed. Still further detail is provided in selected appendices. Some topics, such as the nature of a point count, consist only of “The Basics.”

Paul B. Hamel is a research wildlife biologist at the Southern Hardwoods Laboratory, U.S. Department of Agriculture, Forest Service, Southern Forest Experiment Station, P.0. Box 227, Stoneville, MS 38776; Winston Paul Smith is a research wildlife biologist at the U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station, Tongass Land Management Team, 8465 Old Dairy Road, Juneau, AK 99801; Daniel J. Twedt is station leader at the National Biological Survey, Southern Science Center, Mississippi Valley Research Field Station, 900 Clay Street, Vicksburg, MS 39180; James R. Woehr is senior scientist at The Wildlife Management Institute, 1101 14th St. NW, Suite 801, Washington, DC 20005; Eddie Morris is wildlife biologist at the Chattahoochee-Oconee National Forests, 508 Oak St., Gainesville, GA 30501; Robert B. Hamilton is an associate professor in the School of Forestry, Wildlife & Fisheries, Louisiana State University, Baton Rouge, LA 70803; and Robert J. Cooper is an assistant professor in the Department of Biology, University of Memphis, Memphis, TN 38152.

METHODS PRELIMINARY

TO POINT COUNTS

The Basics

The most basic form of bird inventory is the list of the bird species that have been observed on a particular piece of land. This guide assumes that managers have such lists of birds for their properties. Almost spontaneously, those lists will stimulate the curiosity of managers and of their clients. Questions about the birds on the list will arise, leading managers to desire information about variations in inventory among different kinds of habitats, among different management treatments of those habitats, among adjacent properties with different land uses, and about trends in numbers over time. Answering these questions requires monitoring the birds on the property. Managers can refer to a decision tree developed for counting methods to select an inventory and monitoring approach (table 1).

More Detail

Point counts constitute one of several methods of inventorying and monitoring birds. Other techniques include spot mapping, counts on line transects, capture-recapture with mist nets, and nest monitoring. Point counts are most often conducted in the breeding season, but are suitable for work in other seasons as well (Gutzwiller 1993). Although the recommendation of Ralph and others (1993) that monitoring should involve more than point counting is undoubtedly correct, only point counts are treated in this guide. Other methods are explained in Finch and Stangel (1993), Ralph and others (1993), and Wunderle (1994).

Table 1 .-Decision tree for matching information needs with various approaches to monitoring birds

I. What level of detail is necessary?

1 .

2 .

3 .

Do you want to know about presence/ absence?

NO-Go to 2.

YE!%Simple species list is enough; go to 7.

Do you want to know about numbers and trends?

NO-Go to 3.

YES-Point counts, and this guide, are for you; go to 4.

Do you want to know about population characteristics, such as survival and productivity?

NO-Point counts, and this guide, are for you; go to 4.

YES-Demographic methods, such as nest searches or constant-effort mist netting, are for you; see Ralph and others (1993).

II. Which species are of interest?

4 . Do you want to know about all species?

YES-A general method, like point counts, is for you; go to 7.

NO-Go to 5.

5 . Do you want to know about a particular group of species?

YES-Go to 6.

NO-A single species? In this case your needs are beyond those of a general guide, and you will likely be doing a specific study with methods tailored directly to that species alone.

6 . Which species group interests you?

Game species--These are beyond the scope of this guide; consult a recognized text on the species of interest.

Waterbirds-These are beyond the scope of this guide.

Landbirds-

Raptors, including hawks and owls- Because of low population density, most of these birds are beyond the scope of this guide.

Night birds, including owls, whip-poor-wills, and the like- Most of these require specific nocturnal searches, although nighttime point counts would work for many of them.

Other landbirds, such as flycatchers, woodpeckers, thrushes, warblers, wrens, etc.-Most species belong to this group for which point count monitoring works well; go to 7.

III. What sort of information is needed?

7 . Do you have a list of the birds on your property?

YE!&Go to 8.

NO-Compile one by getting volunteers to share observations with you; go to 8.

8 . Is this enough information to support your monitoring needs?

YES-Continue your existing monitoring scheme.

NO-Go to 9.

9. Do you know the changes in species occurrence from year to year?

YE!GGo to 10.

NO-Begin recording an annual birdlist from checklists supplied by volunteers;

go to 10.

10. Is this enough information to support your monitoring needs?

YE!%Continue your existing monitoring scheme.

NO-Go to 11.

11. Do you know changes in relative abundance in your region from year to year?

YE!&Go to 12.

NO-Participate in the Breeding Bird Survey; go to 12.

12. Do you want to know trends in relative abundance across your property in general?

NO-Continue your existing monitoring scheme.

YES-This guide can help; do an annual set of roadside counts; go to 13.

13. Do you want to know trends in relative abundance by habitat for your property?

NO-Continue your existing monitoring scheme.

YES-This guide can help; do annual set of habitat-specific, off-road counts;

go to 14.

14. Do you want to know how relative abundance responds to specific management actions?

NO-Continue your existing monitoring scheme.

YES-This guide can help; arrange off-road counts for before-after comparisons;

go to 15.

15. Do you want to know population (demography) responses rather than how relative abundance responds to management treatments?

NO-Stop; continue your existing monitoring scheme.

YES-Your needs exceed what this guide can offer; institute demographic monitoring technique such as nest searching or constant-effort mist netting;

see Ralph and others (1993).

WHAT IS A POINT COUNT?

The Basics

A point count is a tally of all birds detected visually or aurally by a single observer from a fixed station during a specified period (e.g., 5 minutes). Counts are made in the morning (e.g., before 10:00 a.m.

daylight time [09:00 a.m. standard time]) typically during the breeding season (usually May and June) under acceptable weather conditions (e.g., winds less than 20 km/h [12 mi/h] and no rain). In the tally, the birds are identified by species and, where desired, their sex and age are recorded. Their distance from the observer is listed in four categories: from 0 to 25 m (0 to 82 ft), 25 to 50 m (82 to 164 ft), beyond 50 m in the habitat, or as flyovers of the habitat, regardless of distance. The time of first detection is listed in two categories: within the first 3 minutes of a count period or during the last 2 minutes. Where time to travel from point to point exceeds 15 minutes, longer counts of 10 minutes are warranted (Ralph and others, in press). When lo-minute counts are conducted, a third time category is listed for the final 5 minutes. This protocol is almost identical to the standard proposed by Ralph and others (1993, in press).

More Detail WHO CONDUCTS A POINT COUNT?

Contrary to the proposed national standard, separate, additional listing of birds within 25 m (82 ft) is recommended. This recommendation is more specific than that of Ralph and others (1993). In the Southeast, three distance bands are desirable because of the relatively dense populations of birds in the forests and the complexity of the vegetation. Recording the field data in a more detailed manner is useful because the distance bands within 50 m (164 ft) can be later collapsed into a single value, whereas birds recorded in the field as only within 50 m (164 ft) can never later be separated into those within 25 m (82 ft) and those between 25 and 50 m (82 and 164 ft).

Because most detections are aural, usually of singing males, singing behavior must be taken into account in designing the surveys. Male song is strongly influenced by breeding status; thus, the timing of surveys is important. Singing frequencies are usually greatest when males are establishing territories and attempting to attract mates; usually, the singing frequencies decline after the eggs are laid. This decline is very noticeable in some species. Thus, the singing frequencies are often greatest in unmated males. Furthermore, nesting dates vary among species; consequently, optimal dates for sampling some species may be inadequate for others.

Generally, there exists a 4- to 5-week period that will be optimal for simultaneously counting most species.

In the southernmost parts of the Southeast, point count sampling can start around 1 May In the northernmost parts, sampling should not start before 21 May because not all migrant species will have arrived before that date. Consult local experts or State bird guides for optimal survey times. A useful rule of thumb, in any case, is to make all counts on one piece of property during as few calendar days as possible.

The Basics

Unavailability of suitable personnel is a serious problem for a manager designing a point count survey. Identifying birds found on point counts requires great skill. Most of the birds encountered in the field, usually more than 90 percent of them, are heard and not seen. Personnel conducting point counts must be able to identify the species present by sight and sound and be able to estimate the distance to the birds detected. Ideally they will have some prior experience doing point counts. Finding people with the proper skills to conduct such counts is not easy. Local chapters of the Audubon Society and State ornithological societies are a reliable source of volunteer counters, many of whom are eager to assist in other tasks involving birds and land management.

More Detail

Many people find potential workers, both volunteer and paid, through “The Ornithological Newsletter,” a bimonthly publication of the Ornithological Societies of North America (OSNA). The address is “The Ornithological Newsletter,” c/o American Ornithologists’ Union, c/o OSNA, 810 E. 10th Street, Lawrence, KS 66044-8897, U.S.A.

In the long run, it may be more useful for managers, especially those in charge of larger properties, to train their own staffs to conduct point counts.

Training sessions can be arranged through several universities in the Southeast. Approaching schools that currently offer courses in ornithology is one method of determining the availability of such training. It will likely be necessary to use volunteers for some of the work or the training; each State in the Southeast has active local birdwatching groups.

Because trained personnel are the most limited resource in a point count survey, managers will do well to use the counters to the fullest extent in conducting as many counts as possible during the brief morning counting period in the short breeding season.

CONDUCTING A POINT COUNT

These procedures are suggested as standards for counts in the Southeast; they are based upon the standard protocol of Ralph and others (in press) as modified by the experience of Smith and others (1993) in the Mississippi Alluvial Valley The instructions of the Breeding Bird Survey (U. S.

Department of the Interior, Fish & Wildlife Service n.d.) are useful as well. A checklist of field equipment is in appendix A.

The Basics

Locate Each Station with Explicit Geographic Coordinates.-Locate each counting station with the greatest practical precision and with an explicit geographic reference, such as Universal Transverse Mercator (UTM) coordinates or latitude-longitude using a Geographic Positioning System (GE’S) device, or determine the location in advance on a U. S.

Geological Survey 7.5-minute topographic map.

Some GPS devices are more difficult to use and less accurate under a dense forest canopy It will be useful to locate and permanently mark the counting stations in advance, including reference distance markers at 25 m (82 ft) and 50 m (164 ft) from the counting point in the cardinal directions.

Count the %&.-Steps for conducting the actual point counts in the field are as follows:

1. Approach the location, noting any birds within 50 m (164 ft) of the counting station that are flushed, fly away, or retreat. Mark these birds in the appropriate distance band on the bulls-eye data sheet. Circles on the data sheet (fig. 1) indicate distances of 25 m (82 ft) and 50 m (164 ft).

2. Orient the bulls-eye data sheet to a fixed direction, record the wind and sky conditions (appendix B), date, time, and observer.

3. Position the GPS device and start it recording. Set the thermometer in the shade.

4. As soon as possible, start the count. Use a pocket timer or watch to keep track of the time.

5. Record each bird seen or heard with the appropriate species code (appendix C). Count family groups of juveniles with a single adult as a single bird. It may be useful to note the sex and any other indicator of behavior or breeding status, and keep track of number of juveniles separately from number of adults. The standard protocol does not require this information, but other workers have found that recording only 0 brown-headed cowbirds (Molothrus ater) provides more accurate data related to the intensity of nest parasitism than does the total number of cowbirds.’

6. Mark birds on the data sheet in the appropriate distance band and appropriate azimuth; i.e., use the bulls-eye data sheet as a map of the count station and a compass to determine the approximate azimuth. A set of standard coding symbols is included on the actual bulls-eye data sheet (fig. 2) to aid in separating individuals.

7. Record data for the different time intervals of the count in different ways. Some people use a pen with multiple colors and record data for the first 3 minutes in one color only, using a second color for minutes 4 and 5, etc. Others choose to underline those birds observed in the second time interval and circle those observed in the third. Be sure to put a legend of the chosen coding scheme on the bulls-eye data sheet.

8. Holding the sheet in a fixed position, spend part of the time facing in each of the cardinal directions in order to better detect birds in each.

9. Mark each bird once, using the mapped location to judge whether subsequent songs are from new or already mapped individuals. All flyovers are recorded outside the second circle, underneath the word “Flyover”.

10. Record birds observed in the different time intervals separately Mark the birds encountered in the final 2 minutes separately, such as by using a second color, underlining those observations, or otherwise distinguishing them. Be sure to note the chosen method of distinguishing these different time periods on the bulls-eye data sheet. Where longer times are indicated because travel time between points is long, count no longer than 10 minutes, and use a third color or other designation for the 6th to 10th minutes.

11. Do not record any birds believed to have been counted at previous stations.

* S. Robinson, Illinois Natural History Survey, and F. Thompson, North Central Research Station, personal communication, 1993.

ZGFz -_--- - - - - - - - - - -_- - - - -

Region station Point t v&lit I Year i&zii Day mne

Observer Temp g: W%d S??y H%itat Forest type S-W isite-

-me - - - --- N-S cootd%rtes E-W coord%;ltes

Geographic coding: Zone-0 fo r lat-lOW, else U’fM Zone

08 C

Flyovers: Remarks

W;NG SYMBOLS

s URW0 Position d singing m&r

,a--.

S e l

Pair together P - auumed mdod obswc4 sax O - unkrorm

Figure L-Blank bull’.+eye data sheet.

12. At the end of 10 minutes, stop recording bird observations. Do not record any new birds seen or heard after the 10 minutes have passed.

13. Discontinue logging data on the GPS unit. From the display on the GPS unit, transfer the latitude and longitude to the bull’s-eye data sheet.

Although these are uncorrected data, they can serve as the back-up in case the electronic data are inadvertently lost from the GPS unit before they can be dumped to the PC for correction.

14. Record the temperature.

15. Mark the location to facilitate a return, at a later time, for appropriate vegetation measurements.

Field notations from the bull’s-eye data sheets will be transcribed to bird count data forms and entered into a computer format (appendix D) compatible with the Southeastern standard. A sample completed bull’s-eye data sheet appears in figure 2, figure 3 is a blank bird count data form, figure 4 is a sample completed bird count data form for the count in figure 2, and figure 5 is the sample completed bird count data form entered into the Southeastern standard computer format. A blank bulls-eye data sheet and a blank bird count data form are inserted into the back of this book.

WHAT IS A POINT COUNT SURVEY?

A point count survey is a group of point counts, strategically distributed across a property such as a management area, that are visited in a predetermined order at a particular time. The tasks involved in conducting a point count survey or inventory are:

1. Preparation of a survey design, determining where the counts will be conducted;

2. Identifying personnel;

3. Counting the birds, and assessing vegetation appropriate to your needs;

4. Transcribing the field data into a standard form, usually a computer format; and

5. Summarizing the data into reports about the birds.

This section deals with the first stage in the process;

separate sections treat the others.

Survey Design

The Basics.-An inventory or monitoring project is designed to find out what bird species occur on a property, where they are, and how their numbers are changing through time. The single most important part of the process is to decide upon the objective of the project. Carefully considering the kind of information wanted, the level of detail required, and the frequency of need to evaluate change will aid greatly in making this decision; the decision tree in table 1 can help.

Point counts are an efficient way to carry out these activities. No inventory or monitoring survey can be done without a map of the property of interest, preferably marked with different forest stands, compartments, ecological classification units, or other areas that the manager wishes to consider as single management types. Management units such as forest types, successional stages, and stand condition classes are the habitat units that monitoring is designed to sample. Results of the activity will then be expressed as numbers of birds in the different management units.

Guidance for monitoring design comes from the studies of Smith and others (1993), from the recommendations of Ralph and others (1993), and from standard statistical considerations of precision.

When a point count survey is set up, points are located far enough apart to avoid encountering the same birds; e.g., 250 m (820 ft) apart for off-road counts and approximately 1 km (0.5 mi) apart for roadside counts. As a general rule, point count stations are distributed among habitats in proportion to the number of stands in the different habitats on the property. Counts within habitats are distributed one per stand, wherever practicable. Placing more than one count in a stand will probably constitute a waste of resources. Counts intended to sample particular habitats are located at least 100 m (328 ft) from the edge of the habitat except when habitat patches are too small to permit such positioning. In that case, a counting station should be located at least 50 m (164 ft) from the edge.

Ideally, counts are conducted once a year, at as near the same time as possible; and at the exact same points each year. Where possible, the same personnel conduct the counts each year.

- - 2-3 h - - - Observer Tap =C iii-and Sky tiiitat Forest type S&e

- - -em- - - - --m---w Zone N-S coordinates E-W coordinates Geographic coding: Gme-0 for lat-low, d.sa UIN Zoa*

Flyovers: R e m a r k s

Figure 2.-Sample of point count made with bull%-eye data sheet.

State Region

Eint I

Statlon VTsit t Year Month Day TTmii- - -

Observes Temp T

Wxd STy Hxitat Siege - - - - Forest type Zone

N - S coordmates E-W coordinates

Geographic coding: Zone-O for lat-long, else UTM Zone

Species Alpha Code

Count < 25 m

O-3 min 4-5 min 6-10 min

Flyovers

O-3 min 4-5 min 6-10 min

25-50 m > 50 m

O-3 min 4-5 min 6-10 min O-3 min 4-5 min 6-10 min

I I--~ - -l

I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I

--~I I I

I I I I I I I I I I I I I I I I I I I I I I I I I I

I I

I -____ I I- - - - __ - I I-____ --~ I

I

I- - - -

I--~ --~ I l- - ______ --

I I-___ -___ I I

I I- - - -

Submitter Remarks

Figure 3.-Blank bird count data form.

eBd-- 3 1 -- Obsekvsr Tamp 1: Wind S)ry Eabitst Foresttype Sta6e

/ 6 3f_dh_S2o 24re_53-0-- - -zone N-S coordinates E-W coordinates Geographic codins: Zone=0 for lat-lon&, else UIM Zone

Species Alpha count < 25 m

0;3 min 4-5 min 6-10 min

1- -

I- - -

2- -

--P

25-50 m > 50 m Flyovers

O-3 min 4-5 min 6-10 min O-3 q in 4-5 min 6-10 min O-3 min 4-5 min 6-10 min l- - -

1--P

I /- - - I--P

--P

-* 2- -

2 , “ 2 :- 3 “ 2

‘I

‘I

Submitter Remarks ml4 1 234-57 8 9 (0

. . - L un4Ja

Figure 4.-Data transcribed from bull’s-eye data sheet to bird count data form.

1 0

TNCHICKAIIAR CLRRKB 1199306150943pBN

TNCHICKABAR CKRRKB 1199306150643PBN

TNCHICKABAR CBRUKB 1199306150843DBH

TNCHICKABAR CKRRKB 1199306150643PBH

TNCHICMBAR CBRKKR 1199306150643PBH

TNCHICMSAI CCRRKB 1199306150B43PBH

TNCHICKAMR CCRUKB 1199306150643PBH

TNCHICKAAIAR CERRKB 1199306150B43PBH

TNCHICKAIJAW CBRUKB 1199306150B43PBN

TNCHICMBAU CBRRKB 119930615OB43PBH

TNCHICKABAR CBRRKB 1199306150843PBH

TNCHICKABAR CERRKB 1199306150643PBH

TNCHICKAIAW CIRIKB 1199306lSOB43PBH

TNCHICKAABAR CRRUKB 1199306150643PBH

TNCHICKABAR CBRWKB 1199306150843PBH

TNCHICKABAR CBRWKB 1199306150843PBH

TNCYICKABAR CBRRKB 1199306150843PBH

TNCHICKABAR CBRRKB 1199306150943PBH

TNCHICKABAR CBRRKB 11993061SOB43PBH

311 nEfl163966.520 260530BGGN 1--_ __-_--.----- ---~~~~~

31? BBB163966520 260530RTRU 1

3 1 1 BB1163966520 260530NDCA 2 1 2

319 801163966520 26053OTUTI

311 808163966520 26053OACFL :.l

311 881163966520 26053ODOWO

311 666163966520 260530RBVI

31? 881163966520 260530RRRO

3lp 682163966520 260530YTUA 3lC 669163966520 260530BHCO 3ll 661163966520 26053OCARR

311 886163966520 26053OYBCU

3lp 881163966520 26053OBUPB

31? 881163966520 26053OWBNU

319 881163966520 26053ORSHA

31Y 881163966520 260530NOPA

311 081163966520 260530RBRO

311 808163966520 260530~~01

31F 988163966520 260530KEIA

3 2

1 1 1 1

P B H

PBH

PBH

PBH

PBH

PBN

PBH

PEN

PBH

PBH

PBH

PBH

PBH

PBH

PBH

PBH

PBH

PBH

P B H

RIND BEGAN AT 2

UIND BEGAN AT 2

UIND BEGAN AT 2

UIND BEGAN AT 2

RIND BCGAN AT 2

UIND BBGAN AT 2

RIND SKGAN AT 2

UIND BEGAN AT 2

KIND BEGAN AT 2

RIND BEGAN AT 2

UIND BLGAN AT 2

RIND BBGAN AT 2

RIND BEGAN AT 2

UIND BEGAN AT 2

UIND BBGAN AT 2

UIND BEGAN AT 2

UIND BEGAN AT 2

RIND BEGAN AT 2

RIND BBGAN AT 2

Figure S.-Sample point count transcribed into standard ASCII fiat file format.

More DetuiL-Stands are selected at random, either irrespective of habitat type or by habitat type, so that each stand has an equal chance of being selected at the beginning of the process. Counts are distributed one per stand, wherever practicable. Summaries that managers make of the counts and analyses that managers will ask researchers to conduct on their point count data will be much simpler if this general guideline is followed. Statistical considerations of independence of individual counts are also important in understanding the meaning of results of a point count survey. Generally speaking, counts at least 250 m (820 ft) apart are independent of each other because they count different birds. If, however, a manager wishes to compare habitats but all counts from one habitat were made on one property or one stand, whereas those from another habitat were made from multiple stands, deciding whether differences have to do with that one stand or with the habitat will be difficult. Consulting a statistician for help in designing the study will be useful, particularly if the policy implications of the work are great or the differences in bird numbers among various habitats or years are small.

For practical purposes, only one visit per point count station will be necessary to compare populations from one year to the next. This approach will provide a single estimate of the abundance of birds on the property at that time. The only time that points should be revisited within a year will be when managers wish to examine how birds change during a single season on their properties. The results of such work can show how the populations change during the breeding season.

Determination of Sample Size

The Basics.-No magic formula for determining number of counts is available. The actual number of point counts required to answer a particular question will vary with the question, with the species, with the variability in numbers of birds in each physiographic area, with the habitat, as well as with the desired precision. The two most common sorts of questions relate to changes in bird populations over time and to responses of bird populations to management treatments. Ralph (1993) suggests a minimum total of 250 counts to enable monitoring annual changes in bird populations on a property by roadside counts.

Smith and others (1993) showed that, for any given number of counts, visiting a larger number of counting stations once yielded more information about the bird community than did visiting a smaller number of stations more than once. They found that, in any case, no more than five points per stand were necessary for adequate sampling of the bird community. They also found that point counts longer than 10 minutes are not justified because they are less efficient at sampling the bird community and produce more variable results than do counts of 10 or fewer minutes.

A practical method of sample size estimation for land managers is presented in this section. A more detailed and rigorous approach to sample size determination appears in appendix E. The following discussion assumes that individual forest stands are the sampling units’and that each stand is sampled by the same amount of effort, usually one point count. It also assumes that a pilot sample of counts has been conducted, such as the 50 to 100 counts that a volunteer observer could gather by doing a roadside survey of a property over a single weekend in late May according to the standard methodology of the U.S. Department of the Interior, National Biological Service, Breeding Bird Survey (Butcher and others 1993).

Determining sample size starts with the question of interest. Land managers may commonly wish to answer one or both of two basic questions: “Are bird populations on my property changing over the years?” or “How did bird populations respond to my management activity in particular habitats?”

A manager who wants to answer the first of these questions should allocate the points for bird counts among the different habitats so that the percentage of points in each roughly corresponds to the percentage of the property in that habitat. This strategy enables the manager to detect trends in population numbers over several years, as long as the points are visited each year. A manager seeking to answer the second question faces a slightly more complicated task in deciding how many points to allocate and where to put them.

More Detail: A “Quick-and-Dirty” Method of Determining Sample S&z-The next issue is, how many points are needed? The answer to that question depends upon the specific question the manager wants to answer. In general, the number of points needed increases as questions get more specific, as one attempts to detect smaller population changes, as the desired confidence level increases, and as the population of interest gets less abundant or more variable. This trend accentuates the importance of deciding in advance exactly what is the objective of the monitoring.

How these factors affect needed sample size can be illustrated by using a set of pilot data from bottomlands in the Mississippi Alluvial Valley (Smith and others 1993) to demonstrate how the required sample size is affected by:

1. The nature of the question (from most general to most specific)

a. Has the total number of birds or species changed?

b. Has the total number of neotropical migratory birds changed?

c. Has the total number of birds with high

Concern Scores (Hunter and others 1993a) changed?

d. Has the total number changed of whatever species is being counted; e.g., wood thrushes (Hylocichla mustelina)?

2. The magnitude of population change one desires to detect

a. 20-percent population change.

b. lo-percent population change.

3. a = The desired confidence level with which a change is detected

a. a = 0.20, lower confidence.

b. a= 0.10.

c. a = 0.05, higher confidence.

The likelihood that answers of “No” to questions (a) through (d) in 1 are really “No,” which is the power of the test (l-b), will be discussed in appendix E.

From their pilot data for these determinations, Smith and others (1993) present in detail the required sample sizes for a selection of species in the Mississippi Alluvial Valley; these are summarized in appendix E, table E.l.

Managers who want to detect population changes between years will compare populations in one year with populations in another year on the same site.

Those who want to determine how birds responded to a management activity will need to compare changes in populations on the managed (treatment) sites to changes in populations on nearby untreated (control) sites. In either case, the hypothesis to be tested will be that the average number of birds recorded per count in one year or habitat type or management treatment (pi) does not differ from the average in another year, habitat type, management treatment, or control area (@:

H, (the null hypothesis): ui = uLz, and the test will be to compare that hypothesis with the alternative that the average number of birds per count in the 2 years, management treatments, or habitat types differ:

H, (the alternative hypothesis): pi z uz.

The test of this hypothesis is the Student’s t-test. The estimate of some average count of birds obtained from sampling will be in the form of the mean * a confidence interval, indicating that there is a specific confidence level that the value of the mean falls conservative estimate for a = 0.10, and t = 1.33 is a within the interval: conservative estimate for a = 0.20.

where n = the number of counting points in a pilot sample from a particular habitat or year, etc., s2 = the sample variance of the variable counted in the pilot data, t = Student’s t for a given confidence level, X = the sample mean of the variable counted in pilot data.

The basic problem of sampling design is to limit the size of the confidence interval to the proportional change (P) in a population which one desires to detect (Snedecor and Co&ran 1967:516). The objective is to ensure that

Solving this inequality for n yields s2t2 ?lL-p2 x2 where

IZ = the required number of counting points in a particular habitat or year, etc.

P = the proportional population change one desires to detect and s2, t, and X are the same as above.

The manager will choose the value of P desired; e.g., P = 0.10 to detect a lo-percent population change, P =

0.20 to detect a 20-percent population change, etc.

Since t-values are taken from a standard t-table based on the number of degrees of freedom (sample size minus l), and the sample size is unknown, how does one choose an appropriate t-value? Nearly every point count sampling for birds will exceed 20 points, the equivalent of two mornings in the field. Hence, a conservative estimate of t (one that will ensure a large enough n) can be obtained by selecting t-values for 20 degrees of freedom and the desired confidence level. From appendix E, table E-2, t = 2.1 is a conservative estimate for a = 0.05, t = 1.73 is a

Now, all that remains is to obtain estimates of X and s2, the sample mean and sample variance. These estimates can be obtained from pilot study sampling;

a separate pilot sample will be required for each habitat. The question now becomes, “How many points are needed to get adequate estimates of Z and s2 from pilot data?”

A graphical technique is useful to answer that question. The counts of all birds combined from the pilot data of Smith and others (1993) can be used to illustrate how the sample mean (fig. 6a) and sample variance (fig. 6b) can be estimated. These graphs are obtained by calculating and plotting a new sample mean and sample variance each time the data from an additional counting station is added to the data set. Such graphs of how sample means and sample variances change as additional point counts are added to the sample set typically show the depicted pattern. They tend to have large oscillations initially, and then either approach an asymptote or reach a condition of steady-state, low-amplitude oscillations around a fixed value (see figs. 6c, d). In either case, estimates can be obtained by projecting graphically to the ordinate axis. Because the formula for calculating n has s2 in the numerator X and in the denominator, use of a slightly low estimate for !i and a slightly high estimate for s2 is recommended. Such usage will ensure a conservative (probably large enough) estimate for the sample size needed.

Now that estimates of X and s2 have been obtained from pilot data, the question could be asked, “How many points are needed to detect a lo-percent change in the total number of birds with a = O.lO?” If P = 0.10 is used to represent the proportional population change in question and if t = 1.73 while s2 is estimated from the graphs at 25 and X at 16.4, then s2t2 25(1.73)2 ” p2x2 = (0.g2 (16.4)2 - 28

Now suppose a more specific question is asked; i.e., How many points are needed to detect a change in the number of neotropical migratory birds? Figures

All Birds Combined Neotropical Migratory Birds

Smple Mean So

2 6 sample Mean

2 .-

0 !

0 6 10 16 20 26 60 26 40

‘0 L 6 1 0 16 20 26 60 56 40

Number of Counts

Sample Varhnce

Number of Counts

(b)

8.

r

Sample variance (d)

1 5 10 15 2 0 2 5 3 0 3 5 4 0 Number of Counts Number of Counts

Figure 6. -Selected variables as functions of the total number of point counts made in the pilot data of Smith and others (1993): a, mean number of birds, drawn from all birds combined; b, variance of number of birds, drawn from all birds combined; c, mean number of neotropical migrate y birds; d, variance of number of neotropical migrate y birds.

6c and 6d, which were derived from a subset of the pilot data consisting of counts of neotropical migratory birds only, yield estimates of X = 5.8 (fig. 6c) and s2 = 10 (fig. 6d).

If a lo-percent change in the number of neotropical migratory birds needs to be detected at a = 0.10, the number of points needed is

22 10(1.73)2 ” +& = (0.1)2 (5~)2 = 89

If the same values for the sample means and variances and the same t- values are used, substituting P = 0.2 will give the sample size necessary to detect a 20-percent population difference in neotropical migratory birds-or n = 23 for cx = 0.10 and n = 33 for a = 0.05.

From the same pilot data, another subset of data points consisting only of counts of birds of species with high Concern Scores (2 24) in the MAV (Hunter and others 1993a) was used to estimate s2 and X graphically in the same way, producing estimates of X = 2.6 and s2 = 6. These values yielded a required sample size of n = 266 points to detect a lo-percent population difference at a = 0.10 and n = 392 points to detect a lo-percent population difference at a = 0.05.

Detecting a 20-percent difference in species with high Concern Scores required sample sizes of n = 67 and n = 98 points for = 0.10 and 0.05, respectively

The final and most specific question asked was “How many points are needed to detect lo-percent and 20-percent differences in wood thrush populations at a levels of 0.10 and 0.05?” The pilot data and graphical estimates for only wood thrushes yielded estimates of X = 0.4 and s2 = 0.8. To detect a lo-percent difference in wood thrush numbers at = 0.10 would require a sample size of n = 1,497 points. To detect the same difference at a = 0.05 would require n = 2,205 points. Detecting a 20-percent difference in wood thrush populations would require n = 375 points for a = 0.10 and n = 552 points for a = 0.05.

Thus, for the u = 0.10 confidence level, the number of separate points needed in each group to detect a lo-percent population difference between 2 groups of counts increases from 28 for all birds combined to 89 for neotropical migratory birds alone, to 266 for birds with high Concern Scores, to 1,497 for a single species (wood thrush).

In general, increasing the confidence level from a =

0.10 to a = 0.05 will require about 50 percent more counting points, and it will require four times the number of counting points to detect a lo-percent population difference as are needed to detect one of 20 percent. Hence, detecting a Xl-percent population difference at a = 0.05 will require roughly six times the number of counting points needed to detect a 20-percent population difference at a = 0.10. Note especially that increasing the confidence level from a = 0.10 to a = 0.05 is far less costly than reducing the size of the detectable population change from 20 to 10 percent.

Furthermore, because the usual annual or seasonal variation of bird numbers on particular properties is not well known, the amount of annual change that a manager should worry about is not well known either, As more data accumulate from actual point count surveys, this issue will clarify itself. Managers interested in monitoring individual bird species should note that the more common and abundant species (relatively high Y) and the more evenly distributed species (relatively low s2, known as habitat generalists) will require far fewer counting points than relatively scarce (low X) species with clumped distributions (high s2, known as habitat specialists). Unfortunately, the relatively scarce habitat specialists are precisely the species most often in greatest need of monitoring and management.

Forming a Group of Cooperators When Sample Size Exceeds Property Size

Monitoring the scarce, habitat-specialist species requires changing the monitoring scale. First, scale down by conducting point counts at the habitat-specific as well as the propertywide scale. For example, if point counts are distributed proportionally across all habitats on a property, counts of species such as the prothonotary warbler (Protonoturiu citreu), northern parula (Purulu americana), and Acadian flycatcher (Empidonslx virescens) will likely produce a low X and a high s2 (yielding a large n), because these species are confined to corridors of riparian habitat where relatively few counting points may have been placed.

If a large number of points are placed entirely within riparian habitat, counts of these species will produce a larger X and a smaller s2 (yielding a much lower n) than counts at the propertywide scale. Thus, habitat-specialist species will probably require a separate monitoring program focused upon their specific habitats; this program may be quite different from one in which the habitats are limited in area.

Next, scale up by forming a network of cooperating bird monitors on other properties across the landscape and then pooling the data. This method can increase sample sizes enough to permit the detection of population changes among habitat-specialist species on a landscape scale, even when sample sizes on individual properties may be too small to detect population changes. The Southeast Management Working Group of Partners in Flight, organized along physiographic-area lines, will help with this task (Hunter and others 1993b) and will help managers obtain the proper statistical advice as they cooperate and pool their data. It will be important to ensure that the habitat definitions used by the various cooperators in a physiographic area are clear enough to allow managers to know how to pool their data.

Because of the large sample sizes needed to detect population changes in individual species, especially scarce species, most managers should not expect to detect small population changes in individual species on their own properties and may only be able to detect changes of 30 percent or greater at the = 0.10 level. Moreover, most habitats on an individual property will be too small to contain enough counting points to detect differences in bird populations among habitats. Nevertheless, a propertywide monitoring program is vital to the detection of such population changes on the larger landscape scale.

If point counts are distributed proportionally among habitats on each monitored property, then a cumulative total of 2,000 point counts taken from at least 10 properties spread across a physiographic area should suffice to detect population changes of most individual bird species within that area. The contribution of each property to this total is consistent with the number of point counts, 200 to 250, which can be comfortably completed by one observer on a single property in a sampling season.

If 10 properties large enough to contain 200 point counts each cannot be located or monitored within a physiographic area, then at least 1,000 cumulative point counts should be taken at the propertywide scale from as many properties as possible. In this case, each property manager should also conduct a separate monitoring program at the habitat-specific scale by placing an additional 40 point counts exclusively within those habitats of highest priority that contain the bird species of highest priority in the physiographic area. If the habitats of highest priority on a property are too small to contain 40 counting points, then a manager should take as many point counts as possible in those habitats.

Thus, land managers can best contribute to regionwide bird monitoring by combining a propertywide monitoring program for all bird species with a separate habitat-specific program designed to monitor particular bird species in habitats of special interest and by sharing their data with other bird monitors across the landscape.

The pilot data used in this illustration (figs. 6a, b, c,

d) showed that 30 to 40 points were sufficient in the habitats sampled to yield adequate graphical estimates for X and 9, regardless of whether the data graphed were for all birds combined, just neotropical migrants, just birds with high Concern Scores, or just the wood thrush. Thus, pilot study data sufficient to custom-design a monitoring program for an individual property can be obtained by one person in just two to three mornings of preliminary sampling.

Using only a hand-held calculator with statistical functions, the necessary calculations to plot the graphs and to determine required sample sizes can be completed in one additional day. As a result, as few as three observer-days of effort are required to design a site-specific bird monitoring program consistent with a land manager’s goals, budgets, and staffing limitations.

MEASURING VEGETATION

The Basics

Vegetation measurements of some kind are an essential part of the inventory and monitoring task.

How involved those measurements must be is a source of concern for those who pay for and carry out monitoring activities and a source of…