Power of track surveys to detect changes in cougar populations

Paul Beier, Stanley C. Cunningham

Research output: Contribution to journalArticle

63 Citations (Scopus)

Abstract

Little is known about the ability, or statistical power, of track surveys to detect a change in abundance of cougars (Puma concolor). We examined monitoring schemes that would have 80% power to detect a 30% or 50% change in track abundance between 2 survey periods. We used data from track transects in southeastern Arizona to evaluate survey designs for 8-km transects in first- and second order dry washes. Track density (number of 0.5 km segments with tracks along an 8-km transect) followed a Poisson distribution, with no serial correlation between consecutive surveys of a given transect. We used simulated Poisson data to determine how power varied in response to number of 8-km transects, risk or Type 1 error, direction of change (increase or decrease), magnitude of change (30% or 50%), and whether track density between surveys changed uniformly or patchily across transects. Power decreased only slightly when change in track density was patchy. Track transects bad low power to detect increases in track density (e.g., about 190 transects would be needed to detect a 30% increase with 80% power and α = 0.05), but somewhat more power to detect decreases (about 1.40 transects would detect a 30% decrease with 80% power at α = 0.05). Managers can increase the power of surveys (or decrease the number of transects) if a 10- 20% risk of Type 1 error is acceptable, i.e., about 140 transects would be needed to detect a 30% decrease in track density with 80% power at α = 0.05, 110 transects at α = 0.10, and 85 transects at α = 0.20. If surveys need to detect only large decreases (50%), track surveys are more powerful, with only about 50 transects needed for 80% power at α = 0.05, and 30 transects at α = 0.20. Thus, track surveys usually will not detect small annual changes, but may reveal large changes more efficiently than other methods.

Original languageEnglish (US)
Pages (from-to)540-546
Number of pages7
JournalWildlife Society Bulletin
Volume24
Issue number3
StatePublished - 1996

Fingerprint

Puma concolor
transect
autocorrelation
managers
survey design
monitoring

Keywords

  • cougar
  • mountain lion
  • Poisson distribution
  • population monitoring
  • Puma concolor
  • statistical power
  • track surveys

ASJC Scopus subject areas

  • Animal Science and Zoology
  • Ecology

Cite this

Power of track surveys to detect changes in cougar populations. / Beier, Paul; Cunningham, Stanley C.

In: Wildlife Society Bulletin, Vol. 24, No. 3, 1996, p. 540-546.

Research output: Contribution to journalArticle

Beier, Paul ; Cunningham, Stanley C. / Power of track surveys to detect changes in cougar populations. In: Wildlife Society Bulletin. 1996 ; Vol. 24, No. 3. pp. 540-546.
@article{47e6aecdeddc4b84bce817ffd168bc37,
title = "Power of track surveys to detect changes in cougar populations",
abstract = "Little is known about the ability, or statistical power, of track surveys to detect a change in abundance of cougars (Puma concolor). We examined monitoring schemes that would have 80{\%} power to detect a 30{\%} or 50{\%} change in track abundance between 2 survey periods. We used data from track transects in southeastern Arizona to evaluate survey designs for 8-km transects in first- and second order dry washes. Track density (number of 0.5 km segments with tracks along an 8-km transect) followed a Poisson distribution, with no serial correlation between consecutive surveys of a given transect. We used simulated Poisson data to determine how power varied in response to number of 8-km transects, risk or Type 1 error, direction of change (increase or decrease), magnitude of change (30{\%} or 50{\%}), and whether track density between surveys changed uniformly or patchily across transects. Power decreased only slightly when change in track density was patchy. Track transects bad low power to detect increases in track density (e.g., about 190 transects would be needed to detect a 30{\%} increase with 80{\%} power and α = 0.05), but somewhat more power to detect decreases (about 1.40 transects would detect a 30{\%} decrease with 80{\%} power at α = 0.05). Managers can increase the power of surveys (or decrease the number of transects) if a 10- 20{\%} risk of Type 1 error is acceptable, i.e., about 140 transects would be needed to detect a 30{\%} decrease in track density with 80{\%} power at α = 0.05, 110 transects at α = 0.10, and 85 transects at α = 0.20. If surveys need to detect only large decreases (50{\%}), track surveys are more powerful, with only about 50 transects needed for 80{\%} power at α = 0.05, and 30 transects at α = 0.20. Thus, track surveys usually will not detect small annual changes, but may reveal large changes more efficiently than other methods.",
keywords = "cougar, mountain lion, Poisson distribution, population monitoring, Puma concolor, statistical power, track surveys",
author = "Paul Beier and Cunningham, {Stanley C.}",
year = "1996",
language = "English (US)",
volume = "24",
pages = "540--546",
journal = "Wildlife Society Bulletin",
issn = "0091-7648",
publisher = "Wiley-Blackwell",
number = "3",

}

TY - JOUR

T1 - Power of track surveys to detect changes in cougar populations

AU - Beier, Paul

AU - Cunningham, Stanley C.

PY - 1996

Y1 - 1996

N2 - Little is known about the ability, or statistical power, of track surveys to detect a change in abundance of cougars (Puma concolor). We examined monitoring schemes that would have 80% power to detect a 30% or 50% change in track abundance between 2 survey periods. We used data from track transects in southeastern Arizona to evaluate survey designs for 8-km transects in first- and second order dry washes. Track density (number of 0.5 km segments with tracks along an 8-km transect) followed a Poisson distribution, with no serial correlation between consecutive surveys of a given transect. We used simulated Poisson data to determine how power varied in response to number of 8-km transects, risk or Type 1 error, direction of change (increase or decrease), magnitude of change (30% or 50%), and whether track density between surveys changed uniformly or patchily across transects. Power decreased only slightly when change in track density was patchy. Track transects bad low power to detect increases in track density (e.g., about 190 transects would be needed to detect a 30% increase with 80% power and α = 0.05), but somewhat more power to detect decreases (about 1.40 transects would detect a 30% decrease with 80% power at α = 0.05). Managers can increase the power of surveys (or decrease the number of transects) if a 10- 20% risk of Type 1 error is acceptable, i.e., about 140 transects would be needed to detect a 30% decrease in track density with 80% power at α = 0.05, 110 transects at α = 0.10, and 85 transects at α = 0.20. If surveys need to detect only large decreases (50%), track surveys are more powerful, with only about 50 transects needed for 80% power at α = 0.05, and 30 transects at α = 0.20. Thus, track surveys usually will not detect small annual changes, but may reveal large changes more efficiently than other methods.

AB - Little is known about the ability, or statistical power, of track surveys to detect a change in abundance of cougars (Puma concolor). We examined monitoring schemes that would have 80% power to detect a 30% or 50% change in track abundance between 2 survey periods. We used data from track transects in southeastern Arizona to evaluate survey designs for 8-km transects in first- and second order dry washes. Track density (number of 0.5 km segments with tracks along an 8-km transect) followed a Poisson distribution, with no serial correlation between consecutive surveys of a given transect. We used simulated Poisson data to determine how power varied in response to number of 8-km transects, risk or Type 1 error, direction of change (increase or decrease), magnitude of change (30% or 50%), and whether track density between surveys changed uniformly or patchily across transects. Power decreased only slightly when change in track density was patchy. Track transects bad low power to detect increases in track density (e.g., about 190 transects would be needed to detect a 30% increase with 80% power and α = 0.05), but somewhat more power to detect decreases (about 1.40 transects would detect a 30% decrease with 80% power at α = 0.05). Managers can increase the power of surveys (or decrease the number of transects) if a 10- 20% risk of Type 1 error is acceptable, i.e., about 140 transects would be needed to detect a 30% decrease in track density with 80% power at α = 0.05, 110 transects at α = 0.10, and 85 transects at α = 0.20. If surveys need to detect only large decreases (50%), track surveys are more powerful, with only about 50 transects needed for 80% power at α = 0.05, and 30 transects at α = 0.20. Thus, track surveys usually will not detect small annual changes, but may reveal large changes more efficiently than other methods.

KW - cougar

KW - mountain lion

KW - Poisson distribution

KW - population monitoring

KW - Puma concolor

KW - statistical power

KW - track surveys

UR - http://www.scopus.com/inward/record.url?scp=0029660526&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0029660526&partnerID=8YFLogxK

M3 - Article

VL - 24

SP - 540

EP - 546

JO - Wildlife Society Bulletin

JF - Wildlife Society Bulletin

SN - 0091-7648

IS - 3

ER -