nebraska game and parks commission · european badger (meles meles, frantz etal. 2003; wilson...

Nebraska Game and Parks Commission2200 N. 33rd St. / r.o. Box 30370 / Lincoln, NE 68503-0370Phone: 402-471-0641/ Fax: 402-471-5528/ www.OutdoorNebraska.org

May 10, 2013

Frederick HullGeneral CounselMountain Lion FoundationP.O. Box 1896Sacramento, CA95812

Dear Mr. Hull:

In response to your request for information regarding mountain lion population data inNebraska:

The population estimation model used by the Nebraska Game and Parks Commission(Commission) to estimate the Pine Ridge mountain lion population in northwest Nebraska isfound in the following paper:

C.R. Miller, P. Joyce and L.P. Waits. 2005. A new method for estimating the size of smallpopulations from genetic mark-recapture data. Molecular Ecology 14:1991-2005.

The model was developed for estimating small populations using noninvasive geneticinformation, and a copy of the paper is enclosed for your review. We used software(lfcapwire") developed bythe senior author in deriving our estimates.

Genetic surveys were conducted in the Pine Ridge in 2010 and 2012 giving population estimatesof 19 and 22 respectively.

Population Estimates With 95%Col.

35

30III

5 25 -1-----::::lc 20

~ 15

:::Io 10:io~ 5

'1:1: 0

2010 2012

The Commission also completed an estimate of suitable habitat for mountain lions in the PineRidge. The estimate is based on a geographic in~ormation system model that is used in North

Printed onrecycledpaper with sqy ink.

Dakota and can be found here:http://www.cougarnet.org!status report of lions in North Dakota.finaI.606.pdf

Density estimates from the nearby Black Hills of South Dakota were used to estimate thenumber of mountain lions that would be expected if density of mountain lions is similar inNebraska. Using 2011 density estimates from South Dakota and the area of suitable habitat inthe Pine Ridge, the expected number of mountain lions would be ""22. Using 2012 densityestimates from South Dakota and the area of suitable habitat in the Pine Ridge, the expectednumber of mountain lions would be ""27.

Historic wildfires burned large areas of the Pine Ridge during 2012, and we assumed this couldaffect mountain lions. In order to estimate the potential effect these fires may have had onhabitat and populations, the suitable habitat within the boundaries of the burned areas wassubtracted from the total suitable habitat in the Pine Ridge. Using 2011 and 2012 densityestimates from South Dakota and the area of unburned suitable habitat in the Pine Ridge afterthe fires, the predicted number of mountain lions the unburned habitat would support in thelong term would be ""15-18. This assumes a worst-case scenario of fire rendering all habitatwithin the burn perimeter unsuitable; the true area of suitable habitat after the wildfires likelylies between the pre-burn and post-burn estimates. Future data collection will help ascertainthe extent to which lions are using habitat within the burn perimeters.

The Commission also assumesthat the Pine Ridge population is an extension of mountain lionpopulations in South Dakota and Wyoming and will work in conjunction with neighboring statesto ensure sound management. We recognize that immigration may be less frequent ifneighboring populations in South Dakota and Wyoming are reduced; however, we assume thatmale and female dispersers will continue to immigrate into the Pine Ridge from neighboringstates at some level.

Regarding documents and studies the Commission referenced in considering the mountain lionseason, the primary source was:

Jenks, J. A., editor. 2011. Managing cougars in North America. Jack H. Berryman Institute,Utah State University, Logan, Utah, USA.

We specifically referred to .Chapter 5 of that document and the papers cited therein. A copy ofthat chapter is enclosed for your review. We also referenced:

Jansen, B. D. 2011. Anthropogenic factors affecting mountain lions in the Black Hills of SouthDakota. Ph.D. Dissertation, South Dakota State University, Brookings.

Jansen's dissertation can be accessed at:http://www.sdstate.edu/nrm/publications!upload/Jansen-Brian-PHD-4-26-11.pdf

Copies of two other documents that staff recently produced are also provided.

If you have any additional questions, please feel free to contact us.

Sincerely,

7::,~Y~Director

Enclosures (4)

Cc: Nebraska Game and ParksCommissionersTim McCoy, Deputy DirectorScott Taylor, Wildlife Division AdministratorSam Wilson, Furbearer and Carnivore Program Manager

Molecular Ecology (2005)14, 1991-2005 doi: 10.1111/ j.1365-294X.2005.02577.x

A new method for estimating the size of small populationsfrom genetic mark-recapture data

CRAIG R. MILLER,*PAUL JOYCEt and LISETTE P. WAITS**Department ofFish & Wildlife, College ofNatural Resources, PO Box44-1136, University ofIdaho, MoscalV, ID 83844-1136,tDepartment ofMathematics, Division ofStatistics, University ofIdaho, Moscow, ID 83844-1103

Abstract

The use of non-invasive genetic sampling to estimate population size in elusive or rarespecies is increasing. The data generated from this sampling differ from traditional markrecapture data in that individuals may be captured multiple times within a session or theremay only be a single sampling event. To accommodate this type of data, we develop a method,named capwire, based on a simple urn model containing individuals of two capture probabilities. The method is evaluated using simulations of an urn and of a more biologicallyrealistic system where individuals occupy space, and display heterogeneous movementand DNA deposition patterns. We also analyse a small number of real data sets. The resultsindicate that when the data contain capture heterogeneity the method provides estimateswith small bias and good coverage, along with high accuracy and precision. Performance isnot as consistent when capture rates are homogeneous and when dealing with populationssubstantially larger than 100. For the few real data sets where N is approximately known,capwire's estimates are very good. We compare capwire's performance to commonly usedrarefaction methods and to two heterogeneity estimators in program CAPTURE: Mh-Chao andMh-jackknife. No method works best in all situations. While less precise, the Chao estimator is very robust. We also examine how large samples should be to achieve a given level ofaccuracy using capwire. We conclude that capwire provides an improved way to estimate Nfor some DNA-based data sets. Capwire is available at www.cnr.uidaho.eduilecgJ.

Keywords: capture heterogeneity, mark-recapture, microsatellites, non-invasive genetic sampling,population estimation

Received 11 October 2004; revision accepted 8 March 2005

Introduction

Estimating the size of wild populations plays a central rolein managing harvested populations and conserving rareand endangered species. One of the most common ways toestimate population size has been to capture, mark, releaseand later recapture (or redetect) individuals. The advancement of genetic techniques has made it possible to capturean individual's DNA rather than the individual itself(Taberlet etal.1999).

Scat and hair are the most common sources of DNAobtained non-invasively. The technique of hair snaring hasbeen used to study populations of the brown bear (Ursusarctos), black bear (Ursus americanus) (e.g.Woods etal.1999;

Correspondence: Craig R. Miller, Fax: 2088859080; E-mail:[email protected]

© 2005 Blackwell Publishing Ltd

Mowat& Strobeck2000), the hairy-nosedwombat tLasorhinus

krefftii, Banks etal.2003) and the marten (Martes americanus,Mowat & Paetkau 2002). Collectingscat to identify individuals and estimate population numbers was first used oncoyotes (Canis latrans, Kahn et al.1999) and has since beenused to study red wolves (Canis rufus,Adams et al., personalcommunications),grey wolves (Canis lupus, Creeletal. 2003),

forest elephants (Loxodonta cyclotis; Eggert et al. 2003), theEuropean badger (Meles meles, Frantz et al. 2003; Wilsonet al. 2003), brown bears in Europe (Bellemain etal. 2005)

and the Scandinavian wolverine (Gulo gulo,Flagstad etal.2004). Population size has also been estimated in humpback whales (Megaptera nooaengiiaei, using sloughed andbiopsied skin as the source of DNA (Palsboll etal. 1997).

There are a number of potential benefits to genetic taggingover physical tagging (Taberlet et al. 1999). These includeincreasingthe number ofobservationsand therebyimproving

1992 C. R. MILLER, P. JOYCE and L. P. WAITS

(N I )( 51 ) TL(N)=' . TI(l/N)Ci

T!(N - T)! C1!C2! •.. cT ! ;=1

probability model (ECM), every individual is equally likelyto be captured on each draw with probability one on thepopulation size (1IN). The likelihood function with respectto N is the multinomial probability distribution,

Taking the natural log and ignoring the combinatorialterms which involve only the data (constants) indicates thatT and the total number of observations are sufficient statisticsfor finding the maximum-likelihood estimator (MLE):

(eqn 1)N N-T T

In L(N) ex; L lnix) - L In(x) + In(l/N)L cj

x=1 x=1 ;=1

It is well known that in real populations, individuals do notdisplay equal capture probability (Burnham & Overton 1979).Several approaches have been proposed in classic markrecapture modelling for dealing with this heterogeneity(e.g. Burnham & Overton 1979; Chao 1988). One suchapproach is to view the population as a mixture of individualswith distinct capture probabilities (Norris & Pollock 1996;Pledger 2000). Our method is based on the simplest of thesemixture models in which there are two types of individuals(the two innate rates model, or TIRM). Let the relativecapture probability of the harder to capture type B individuals be 1 and of the easier to capture type A individualsbe ex (ex > 1). Let the number in each class beNA and NB andnote that NA + NB =N. For now, suppose that each sampledindividuals' type (A vs. B) is observable. Let the number ofsampled type A and type B individuals be TA and TB withtheir individual identities indexed by iA and iB• Then, thelog-likelihood is similar to equation 1 except there are twosubpopulations with differing capture probabilities:

Of course, the types of sampled individuals are notobservable. For computational simplicity and speed, wechose to assign each sampled individual the type thatmaximizes the overall likelihood.

With some data sets a more serious problem occurs whenmaximizing the parameters in equation 2 because thelikelihood surface plateaus as ex and NB simultaneouslyincrease. This specifically plagues 'sparse' data sets wheremany individuals are observed only once. For example, ina tiny data set with two individuals each captured onceand two individuals captured five times each (c = 1,1,5,5),

Methods

Proposed estimator (capwire)

The data are modelled as if they arose from 5 samples ofsize one from the population. We assume that all individualsare correctly and uniquely identified from their genotypes(i.e, there are no undetected genotyping errors and no twocaptured individuals have identical genotypes; Miller et al.

2002; McKelvey & Schwartz 2004). All draws are assumedto be independent and identically distributed (lID). Thusbeing captured does not affect an individual's probabilityof subsequent capture (e.g, there is no trap response). Theresulting data is a multinomial vector of capture counts foreach individual, C= c1' c2' •.. cp where T is the number ofdifferent individuals sampled. In the simple even capture

estimates, reducing stress and mortality, reducing capturebias caused by trap response, and shortening the samplingperiod to better approximate closure.

In traditional trap based mark-recapture studies, anindividual may be captured only once per session. Estimating population size has focused on estimating the probabilityof capture for each individual in each session. An important difference in the data arising from DNA-based markrecapture studies is that sampling is approximately donewith replacement. That is, since an individual is not physically confined at any time, it may leave multiple hair tuftsor scats at multiple locations during a sampling session.One option is to condense all such multiple captures to onecapture per session as has sometimes been done (e.g. Bankset al. 2003; Frantz et al. 2003; Bellemain et al. 2005), but thispotentially wastes information. At the extreme of thismultiple-capture scenario are studies with only a singlesampling session (e.g. Kohn et al. 1999; Eggert et al. 2003).

In this case, a different approach is needed.The purpose of this research is to develop a method for

estimating population size when the data may containmultiple observations of an individual within a session.The performance of the proposed method is evaluated byanalysing a wide variety of simulated data sets where thetrue population size is known. Some of these data sets aregenerated under the same simple model that underlies themethod of analysis, but most of the focus is on data arisingfrom more biologically reasonable models in which individuals move about in space, display differing fidelities toa home range, and deposit DNA at different rates. The samedata are analysed using several of the available methodsincluding rarefaction and estimators within program CAPI'URE

(Otis et al. 1978). We also present analyses of real geneticand traditional data sets using the proposed method. Basedon our results, we make recommendations for current andfuture research on population estimation using non-invasivegenetic sampling.

© 2005 Blackwell Publishing Ltd, Molecular Ecologi}, 14, 1991-2005

a good likelihood score is observed at the reasonableparameter values of Ct =9, NA =2, NB =4 (N =6), while aslightly better score is obtained at the somewhat ridiculousvalues, ex =415, NA =2, NB =167 (N =169). While thisproblem can be circumvented by requiring that there arefew single captures in the data, this would render themethod inappropriate for many real data sets.

Instead, we address the problem by assuming that thesample is large enough that the capture count disparitybetween the seldom-captured and the often-captured individuals provides good information about the value of Ct.

With Ct restricted, we can use equation 2 to estimate N. Inpractice, we accomplish this by initially finding the MLE ofex under the assumption that the all individuals in thepopulation have been sampled (i.e, NA =TN NB =TB, N =NA + NB) · Fixing this as the value of «, the MLEs of NA andNB are obtained. A single bias correction on ex is then madefrom the fact that the expected total number of observations oftype A individuals is equal to [aNAS/(aNA + NB) ].

Solving for ex yields

(eqn3)

Finally, NA and NB are remaximized using this biasadjusted value of Ct. Confidence intervals are estimatedusing the parametric bootstrap. We refer to this proposedestimator as capioire because it is premised on capture withreplacement. The mechanics of the estimator, includinghow sampled individuals are assigned a capture type, areprovided in Appendix 1.

GENETIC MARK-RECAPTURE ESTIMATION 1993

the MLE of N under the ECM, we generated 100 data sets,repeated the maximization process and calculated Asirn '

The P value was the proportion of these 100 where Asirn >Aobs ' When the P value was less than 0.1 we employedTIRM rather than ECM.

Grid simulations. In real mark-recapture studies individualsusually occupy distinct space, display heterogeneous behaviours, and samples are drawn from different locations.Consequently, samples are likely to be neither independentnor identically distributed. For capwire to be useful, it mustbe robust to these violations. Simulations were thereforeconducted to explore how capwire performs on biologicallyreasonable data.

Simulations were designed to emulate a scat-based markrecapture study. A square grid of specified size was populated with a specified number of individuals that wereinitially evenly spaced. Individuals then moved randomlyabout the grid depositing scats. Individuals were of twomovement types (sedentary and transient) and threedeposition rates (seldom, moderate and often) to yield six totaltypes. Combinations of these six types were used to define12types of populations (Table 1). Population compositionswere designed to introduce increasing amounts variancein movement pattern and deposition rate to the population. The details of how grid simulations were conductedare given in Appendix II. All simulations were of the sameduration (4000 steps). After movement and deposition,parallel transects were established at regular, specified,

Table 1 Population composition distributions used in gridsimulation study and corresponding code. Values in the table areprobabilities that an individual is assigned as this type. Individualtypes are abbreviated as follows: first letter represents movementwith 5, sedentary and T, transient; last two letters representdeposition rate with SD, seldom depositor; MD, moderate depositorand OD, often depositor. Hence 'SSD' represents sedentary, seldomdepositor. See Appendix lIBfor details

Type of Individual

0.50.33

0.50.34

0.50.80.25 0.250.167 0.166

Simulation study

Urn simulations. Simulations were conducted to check theperformance of capwire when the model assumptions aremet (i.e.when samples are IID).This was done using an urnmodel (Appendix I, step 10). Nand 5 were both set at 25throughout, and Ctwas varied from 1 to 11 in increments of2. For ex> 1, we considered the scenarios of NA =12,NB =13and NA =3, NB =22. For each scenario we simulated 100replicates. Method performance was based on four criteria.Letting r index the individual replicate and Nthe estimate ofN averaged over 100replicates,these are (i) relative bias =(N N)IN, (ii) mean relative error = 1/[l00(I}~7INr - NI/N)](iii) coverage = proportion of 100 replicates where 95%confidence intervals (CIs) cover N, and (iv) median 95%CIwidth.

In the urn phase of the study we considered the use of alikelihood-ratio test (LRT) to determine if the ECM shouldbe employed for a given data set instead of the TIRM(Pledger 2000).The likelihood was maximized under eachmodel and the ratio of their likelihoods, Aobs' calculated.The distribution of A was obtained by simulation: using

© 2005 Blackwell Publishing Ltd, 14, 1991-2005

Code

G1G2G3G4G5G6G7

G8G9GlOGllG12

SSD TSD SMD

1.0

0.50.8

0.34

0.167

TMD

1.00.50.2

0.33

0.166

SOD

0.50.33

0.250.167

TOD

0.50.330.50.20.250.167


intervals along a cardinal direction. All scats encounteredwere collected and genotyped correctly and uniquely tothe individual. If the number collected was less than thespecified sample size, the simulation was restarted. If morewere collected, observations were removed from the dataat random. The data were summarized as the number ofobservations of each sampled individual and analysed.

An initial set of simulations was conducted for all 12compositions (Table 1)with both Nand 5 =25.Subsequentsimulations focused on four of these (G1, G4, GlO, andG12), providing a gradient from zero (G1) to very high(G12) heterogeneity. These subsequent simulations wereconducted at N = 16,49,100 and 196.Grid size was alwaysscaled to maintain overall density at 1 individual/IO 000steps-. Populations were sampled at round numbersapproximating 1/2N,N, and 2N. Transects were spaced asfollows so that the desired sample size was obtained in mostsimulations: G1 = 40 steps; G4 = 30; G10 = 20; G12 = 30.For each set of conditions 100 replicate simulations wereperformed. Data were analysed with capwire using 95%confidence intervals and only the TIRM (i.e, no LRT formodel selection was conducted, see Results).

A subset of the simulated data was also analysed withseveral other methods that have been applied to geneticmark-recapture data. At N = 16, 49, and 100 we analysedthe grid simulated data using rarefaction. In this techniquethe data are plotted as the number of distinct individualsobserved (y) as a function of sample size (x)and then fit toa curve. The asymptote of the curve provides an estimateof the population size. We employed both the hyperbolicfunctionproposed by Kohnetal. [1999;y = ax/(p +x)] and theexponential equation of Eggert etal. [2003; y = «(I - e~x)].

In both equations a represents the asymptote and pdetermines the rate of increase. For point estimates the'observed' curve was obtained by finding the number ofdistinct individuals expected if the sample was subsampled at size x for x = 1 to 5 (Comps etal. 2001). To obtainconfidence bounds we randomized the order of observations 100 times and fit the equations to each accumulationpattern. After sorting, the 3rd and 98th estimates of a wereused as bounds. Equations were fit using least squaresregression.

For the same subset of simulations, we also examined theperformance of the two closed-population heterogeneityestimators available in program CAPTURE: Mh-jackknife(Burnham & Overton 1979) and Mh-Chao(Chao 1988). Theseestimators assume sampling sessions during which anindividual is either caught once, or not at all. However, thesimulations contained no sampling sessions. The data weretherefore rigged to fit the estimators by defining the numberof sampling sessions for a given data set equal to the largestnumber of times any individual in the data set was captured(i.e, the most caught individual was captured every imaginary session). The rarefaction and CAPTURE estimators

were evaluated using the four criteria described in the urnsimulation section above.

Analysis of real data. The most important test ofan estimatoris its performance on real data, but true population size israrely known in mark-recapture studies. Among DNAbased studies, N was approximately known in only twocases:a population of European badgers (Frantz etal.2003;Wilson et al.2003)and a population of red wolves (Adamset al., personal communication). We analysed both data setsusing capwire. We also analysed DNA-based count datafrom a population of forest elephants (Eggert etal. 2003)and a population of northern hairy-nosed wombats (Bankset al. 2003). In the wombat study (Banks etal. 2003), therewere nightly capture sessions. In their analysis, Banks etal.reduced multiple within-night captures to one in order tomeet CAPTURE assumptions. To explore the effect that suchdata reduction has on estimates and confidence intervals incapwire, we analysed the data both with multiple nightlycaptures included and excluded.

Traditional mark-recapture studies do not sample withreplacement and therefore violate a basic assumption ofcapwire. However, if the number of sessions conducted isnot small, this violation may become unimportant. To analyse these data sets using capwire, all session informationwas ignored, and the data were simply defined as the totalnumber of times each individual was caught in the study.A handful of traditional mark-recapture data sets exist forwhich the true N is known. These include two populationsof cottontail rabbits (Edwards & Eberhardt 1967),a population of eastern chipmunks (Mares et al.1981),two populations of striped skunks (Greenwood etal. 1985) and apopulation of taxicabs in Edinburgh, Scotland (Carothers1973).For all data sets in which N was known (traditionaland DNA-based), the four performance criteria describedin the urn simulation section were used.

Results

Urn simulations

The urn simulations indicate that when samples are independent and all individuals have equal capture probability(a = 1), using the LRTin capwire yields approximately unbiased estimateswith good coverage(Table2).This isbecausethe correct ECMis selected most of the time. Unfortunately,when the population has capture heterogeneity (a> 1),ECM is rejected in favour of the correct TIRM model onlyone-fourth to one-third of the time at this sample size. Aswith ECMs in traditional mark-recapture methods (e.g.Lincoln-Peterson),using ECM in capwire when heterogeneityexists yields underestimates of N (see a values ~ 5). Thus,the low power of small samples can lead to incorrect modelselection and consequent underestimation. This is a serious

© 2005 Blackwell Publishing Ltd, Molecular Ecology, 14, 1991-2005


Table 2 Performanceof capwire on urn simulated data with N =25, 5 =25, 100 replicatesper scenario.When ex> 1, NA =13 and NB =12;when ex =1, capture classesdo not exist

Mod sel" ex N Relbiast CIWidtht Coverageof 95% CIs§ MRE'j[ Cor mod**

LRT 1 26.2 0.05 26 0.96 (0.03,0.01) 0.23 0.893 25.2 0.01 25 0.92 (0.08, 0) 0.27 0.355 21.0 -0.16 15.5 0.75 (0.25,0) 0.25 0.257 19.2 -0.23 14.5 0.65 (0.35, 0) 0.28 0.269 18.0 -0.28 11.5 0.55 (0.45, 0) 0.30 0.22

11 17.4 -0.31 8.5 0.45 (0.55,0) 0.33 0.25

TIRM 1 31.1 0.24 30 0.98 (0, 0.02) 0.31 nla3 28.6 0.14 29 0.99 (0, 0.01) 0.26 nla5 24.5 -0.02 24 0.96 (O.D4, 0) 0.21 nla7 22.4 -0.10 22 0.95 (0.05, 0) 0.20 nla9 21.1 -0.16 19 0.84 (0.16, 0) 0.23 nla

11 19.9 -0.20 18 0.81 (0.19,0) 0.24 nla

"Modelselectionprocess: LRT, use ECMunless rejectedusing LRT at the 0.1 level,and then use TIRM; TIRM, skip LRT and use TIRMregardlessof likelihoodunder ECM.tRelative bias;tmedian width of 95% confidence interval;§numbers in parenthesesare proportions of replicateswhere CIfallsbelow andabove N, respectively; 'j[mean relativeerror.**Correct Model: proportion of replicateswhere correctmodel was selected(ECM when ex = 1, TIRM when ex > 1). Not applicable (n/a)when LRT not used to selecta model.

concern since most real data sets will contain capture heterogeneity. We were therefore compelled to investigate theperformance of capwirewhen no LRTis conducted and TIRMis used irrespective of the consistency of the data with ECM.

Using TIRM when capture probability is even acrossindividuals yields overestimates of N (Table 2). However,it does not reduce coverage when ex = 1; it substantiallyimproves coverage for ex > 1; it reduces mean relative errorfor ex > 1; and it reduces bias for ex> 3. As ex becomes large,N is still underestimated. Similar patterns are observedin simulations with NA =3 and NB =22 (data not shown).Because we view it as more important to improve estimatesin the common event that capture heterogeneity exists andaccept poorer estimates in the rare event that capture probability is even than the converse, we usedTIRM in all subsequent analyses with capwire.

Grid simulations

In the grid simulations, composition distributions Gl to G4represent cases where individual deposition rates are heldconstant and individuals mayor may not differ in their useof space. The results from these scenarios where N =25 andS = 25 are consistent with the urn simulations: coveragesare above 95% and estimates are, on average, about 20%above the true N (Table 3). In compositions G5 through G8,individuals differ in their DNA deposition rates but nottheir use of space. Capwire performs well here with estimateswithin 10%of N and coverages very near 95%.CompositionsG9 to G12 present more complex scenarios where individualsdiffer both in their use of space and their deposition rates.

© 2005 Blackwell PublishingLtd, 14, 1991-2005

Table 3 Resultsofgrid simulated data based onN =25 and 5 =25analysed using capwire. See Table2 for descriptions of columnheaders

Comp ReI CI CoverageofDist* N bias width 95% CIs MRE

Gl 29.7 0.19 30 0.97 (0.02, 0.01) 0.26G2 30.1 0.20 34 0.97 (0.02, 0.01) 0.28G3 31.4 0.25 34 0.96 (0, 0.04) 0.33G4 30.7 0.23 35 0.98 (0, 0.02) 0.32G5 24.1 -0.04 23 0.93 (0.07,0) 0.20G6 27.3 0.09 28 0.97 (0.02, 0.01) 0.24G7 25.2 0.01 24 0.95 (0.05, 0) 0.20G8 26.0 0.04 25 0.94 (0.05,0.01) 0.24G9 24.5 -0.02 23 0.91 (0.09,0) 0.21Gl0 25.2 0.01 25 0.93 (0.07,0) 0.24Gll 25.2 0.01 24 0.99 (0.01,0) 0.19G12 25.6 0.03 25 0.95 (0.05,0) 0.19

"Composition distribution.SeeTable 1.

The results are similar to the uneven deposition simulationsG5 to G8 with estimates of Nbeing approximately unbiasedand coverages in the 90th percentile.

To explore the performance of capwire across a range ofNand S and to compare to other methods, we limited further simulations to compositions Cl, G4, GIO and G12.Several important trends emerge from the results (Table 4).For composition Gl, estimates of N are biased somewhathigh (10-35%), with bias declining as sample size increases.Coverage is generally near 95%. Performance on composition G4 is similar. For both Gl and G4, coverage begins to


Table 4 Analysis of grid simulated data sets for a range of population and sample sizes using capunre. See Tables 2 and 3 for descriptionsof column headers

Comp Dist N S N ReI bias CIwidth Coverage of 95% CIs MRE

G1 16 15 19.9 0.24 25 0.96 (0.03, 0.01) 0.4025 18.7 0.17 17 0.99 (0, 0.01) 0.2335 17.3 0.08 10 1.00 (0, 0) 0.13

25 15 33.8 0.35 74 0.93 (0.04, 0.03) 0.5125 29.7 0.19 30 0.97 (0.02, 0.01) 0.2650 27.9 0.11 17 0.97 (0.01, 0.02) 0.15

49 25 60.1 0.23 92 0.97 (0, 0.03) 0.3450 58.3 0.19 46 0.97 (0, 0.03) 0.22

100 55.3 0.13 25 0.83 (0, 0.17) 0.13100 50 118.9 0.19 105 0.97 (0.01, 0.02) 0.24

100 123.5 0.24 67 0.93 (0,0.07) 0.24200 114.5 0.15 38 0.40 (0, 0.6) 0.15

196 100 234.4 0.20 144 0.98 (0.01,0.01) 0.21200 240.9 0.23 91 0.67 (0, 0.33) 0.23400 223.4 0.14 52 0.07 (0, 0.93) 0.14

G4 16 15 19.3 0.21 28 0.96 (0.04, 0) 0.3725 18.1 0.13 16 0.99 (0.01,0) 0.2535 17.7 0.11 11 0.97 (0.02, 0.01) 0.16

25 15 33.3 0.33 59 0.98 (0.02, 0) 0.4725 30.7 0.23 35 0.98 (0, 0.02) 0.3250 27.7 0.11 17 0.97 (0.01, 0.02) 0.14

49 25 57.6 0.18 73 0.99 (0, 0.01) 0.3150 58.5 0.19 44 0.97 (0, 0.03) 0.21

100 55.4 0.13 25 0.81 (0,0.19) 0.14100 50 121.5 0.21 106 0.97 (0.01, 0.02) 0.28

100 125.4 0.25 68 0.88 (0, 0.12) 0.26200 114.9 0.15 38 0.36 (0, 0.64) 0.15

196 100 223.1 0.14 140 1.00 (0, 0) 0.17200 241.6 0.23 90 0.67 (0, 0.33) 0.23400 223.4 0.14 51 0.Q2(0, 0.98) 0.14

G10 16 15 16.0 0 18 0.83 (0.17, 0) 0.3325 16.1 0.01 14 0.95 (0.05, 0) 0.2235 16.2 0.01 10 0.89 (0.11,0) 0.17

25 100 25.2 0.01 31 0.87 (0.13, 0) 0.3925 25.2 0.01 25 0.93 (0.07,0) 0.2450 24.2 -0.03 14 0.77 (0.23, 0) 0.19

49 25 44.6 -0.09 53 0.90 (0.1,0) 0.2550 52.1 0.06 39 0.98 (0.02,0) 0.17

100 51.5 0.05 26 0.96 (0.02, 0.02) 0.11100 50 92.6 -0.Q7 75 0.84 (0.16, 0) 0.18

100 103.2 0.03 56 0.97 (0.03, 0) 0.11200 106.2 0.06 38 0.95 (0, 0.05) 0.08

196 100 171 -0.13 93 0.64 (0.36, 0) 0.16200 201.1 0.03 79 0.96 (0.04, 0) 0.08400 208.6 0.06 52 0.86 (0, 0.14) 0.Q7

G12 16 15 16.1 0 18 0.92 (0.08, 0) 0.2825 17.1 0.07 16 0.95 (0.05,0) 0.2335 16.0 0 10 0.94 (0.06,0) 0.15

25 15 25.0 0 42 0.93 (0.06, 0.01) 0.3025 25.6 0.03 25 0.95 (0.05,0) 0.1950 24.6 -0.01 14 0.94 (0.06, 0) 0.13

49 25 47.0 -0.04 55 0.89 (0.1, 0.0l) 0.2650 51.1 0.04 37 0.98 (0.01, 0.0l) 0.19

100 50.8 0.04 24 1.00 (0, 0) 0.08100 50 104.9 0.05 89 0.95 (0.04,0.01) 0.19

100 108.1 0.08 58 1.00 (0, 0) 0.12200 104.2 0.04 35 0.93 (0.01,0.06) 0.07

196 100 191.9 -0.02 110 0.91 (0.09, 0) 0.11200 211 0.08 85 0.98 (0.01, 0.01) 0.10400 209.1 0.07 48 0.81 (0,0.19) 0.07

© 2005 Blackwell Publishing Ltd, Molecular Ecologi}, 14, 1991-2005


~~~ ~t""""l ~ t"""! t"""'! ;::::l ;::::l ;::::l t""""! t"""! 'l""""! t"""! 'l""""! c5

Real data

The capwire estimator performs very well on the real datasets (Table 5). For the red wolf data, the CIs cover N in allthree sessions and there is a dramatic decrease in CI widthas sample effort increases. For the European badger data(Frantz et al. 2003),capwire's estimate of 29 (20-43) is at themidpoint of the range that N is known to lie within (24-34)and is similar to the published estimates of the Chao andjackknifemethods of26 (22-45) and 26 (22-40),respectively.For the forest elephant data (Eggert etal.2003),the captoire

fall well below 95% in larger populations sampled atmoderatetohighintensity(N =100,5 =2NandN =196,5 =Nand 2N). Among the four compositions, GI0 is most problematic for capwire in terms of coverage. Estimates aregenerally unbiased but coverage was sometimes below 90%.For composition G12,bias is generally not greater than 10%and coverage is near the desired 95%. Although bias andcoverage do not always improve with increasing samplesize, the mean relative error (MRE)virtually always does.

The hyperbolic rarefaction curve (y =ax/(13 + x): Kohnet aI. 1999)did not perform well. Summarizing over compositions and sampling intensities, in simulations whereN = 49 the method yielded estimates that were, on average,43%above the true N, MRE values of 46% and an averagecoverage of only 45% (results not shown). The exponentialcurve ofEggertet al.[y = 0.(1 - e~x); 2003], and the two CAPTURE

heterogeneity estimators (Chao and jackknife) performedmuch better.

We compared the performance (bias, coverage, CI width,and MRE) of the latter three methods to capwire across arange of N, compositions, and sampling intensity (Fig. 1).The Chao estimator is the least biased of the four, but italso produces the broadest CIs and the largest MRE. Itscoverage is generally 90%or better. The jackknife estimatorperforms poorly at low sampling intensities with largenegative bias, low coverage and large MRE. The Eggertrarefaction method shows point estimates with excellentMREs,but its coverage declines severely for N = 100.Capwireis the most balanced of the four estimators. Its coverage iscomparable to the Chao method and its CIs are considerablymore narrow at small sample sizes. Its MRE is very good,especially at the smaller N and lower sampling intensities.Simulations at N = 16 and 25 showed that capwire isincreasingly superior to the other estimators in terms ofcoverage, CI width and MRE for N < 50 (data not shown).Capwire does have a positive bias when capture rates areeven as they are in Gl and G4. In the more realistic scenarios where capture rates vary (GlO and G12), it shows thesame smallbias that the Chao method does. The performanceof capwire becomes more sporadic as N grows substantiallyabove 100 and sampling intensity increases. In such casesthe Chao estimator is best overall.



N

Fig. 1 Relative performance of capunre (black bars), jackknife (grey), Chao (white), and Eggert rarefaction (hatched) estimators on gridsimulated data. Two major columns correspond to populations sizes labelled at top. Within each N, each subcolumn corresponds tosampling intensity (5) labelled at bottom. Within each subcolumn, each set of four bars corresponds to population composition (G)labelledat bottom above 5. In the coverage figures, the dashed line represents the 95% objective.


estimate of 214 (143-246) is similar, though more precise,compared with the published estimate based on the jackknifemethod of 225 (173-308) and an estimate based on dungcounts of 228(158-337). For the northern hair-nosed wombat,the capunre estimate based on all observations of 88 (84-98)is smaller and narrower than the published jackknifeestimate of 113 (96-150). When only a single observation isallowed per night, the capwire estimate increases to 98(86-110). Among the five wildlife populations studiedwith traditional mark-recapture techniques where N wasknown, estimates are always within 15% of the true Nandin three cases within 6%. In all cases the CIs cover the trueN. In analysing all 52 data sets published by Carothers(1973) on a taxicab population in Edinburgh, capunreoverestimated N by an average of 13% and produced CIscovering the true N 50 of 52 times (=95%).

Discussion

Capwire anddiffering types ofpopulations

The captoire model treats the population as an urn in whichindividuals are continuously mixing. Animals that maintaincohesive groups such as ungulate herds, primate troupes,canid packs, and marine mammal pods may approximatea mixing urn. When deposition rates are even in such cases,it would be better to use ECM compared to TIRM. WhereasTIRM overestimates N, ECM provides unbiased estimatesof N, produces narrower confidence intervals, appropriatecoverage and smaller MRE (Table 2). If fact, in the urnsimulations with (X = 1, the small bias observed in the LRTmodel selection process comes from the 11% of the timethe incorrect TIRM is selected (Table 2). If ECM is usedexclusively, the mean population estimate is 24.9. Ifa researcher has good biological reason to believe thatdeposition rates are approximately even, the LRT does notsuggest otherwise, and the urn model is reasonable, it maybe preferable to use ECM. On the other hand, if the urnmodel is reasonable but assuming even deposition ratesis not, researchers should impose TIRM. Under TIRM,coverage is good and bias small so long as (X:S; 7, whichshould cover most field applications. This problem ofmodel choice illustrates a reoccurring observation in themark-recapture literature: it is best to know enough aboutthe species and sampling design to have some view of theappropriate model and method of analysis a priori (Otiset al. 1978; Dorazio & Royle 2003; Link 2003; Boulangeret ai. 2004).

More common than a mixing urn-like scenario, individuals will occupy semidiscrete areas. Ursids are a goodexample of a group that has been extensively studied usinggenetic methods where individuals occupy home ranges(e.g.Woods etal. 1999; Bellemainetal. 2005). Grid simulationswere run to determine how robust capwire's urn model is to



spatial segregation. In the event that DNA deposition ratesare approximately even, capunre tends to overestimateN. Theresults from grid simulations with Nand S = 25 (Table 3)change very little with varying proportions of sedentaryand transient individuals (GI-G4) and they are very similar to the analysis of urn data with (X = 1 (i.e, relative bias =20%, coverage = 97%, MRE = 30%; Table 2). This suggeststhe spatial dimension of real data is having a minimaleffect on captotre.

Most often researchers will deal with populations whereindividuals neither mix frequently nor deposit DNA at thesame rate. For example, among the seven real DNA datasets analysed (Table 5), ECM was rejected for all sevenusing the LRTat the 0.05 level (results not shown). The gridsimulations indicate that capuiire generally performs betterwith capture heterogeneity in the data (G5-G12) than without it (GI-G4; Tables 3 and 4 and Fig. 1). For a given Nandsample size, the presence of capture heterogeneity reducesbias, narrows CIs, and lowers MRE. Coverage, however,becomes problematic for capunre as N and sample intensityincrease. Without capture heterogeneity, the drop in coverage to low levels (consistently < 80%) appears somewherein the N = 100-200 range (Table 4). With heterogeneitypresent, the decline is pushed out to greater Ns. Grid simulations at N =400 indicate that the decline is between 200and 400 (data not shown).

Sample sizeconsiderations

To this point we have presented sampling intensity as aproportionofN(e.g. S = 1/2M. BecauseNisnotknown, thismeasure of sampling intensity is not particularly useful forstudy design. One possibility is to use an upper bound for /what N might reasonably be in setting a sample size goalfor a study. An alternative view of sampling effort isto quantify it as the average number of observations persampled individual (obs/ind). Across the range of samplingintensity studied here, there is a clear decrease in MRE as thenumber of obs/ind increases (Fig.2a). In small populations(N:s;25), an average of c. 2.5 obs/ind are necessary toobtain estimates which are c. 15% from N and close to 3.0obs/ind are needed to be c.10% from N. For N in the rangeof 49-100, the news is better: 2.0 obs/ind will provideaverage estimates c. 15% off of Nand 2.5 obs/ind, c. 10%from N. Precision shows a similar improvement with anincreasing number of obs/ind (Fig. 2b). For smaller populations, 2.5 obs/ind will tend to yield interval widths of3/4N and increasing to near 3.0 obs/ind will tend to reduceinterval width to 1/2N. Precision increases substantially asN gets larger. With N = 100, 2.0 and 2.5 obs/ind will tendto produce CIs of width 1/2 and I/3N, respectively. Thesefigures may provide some guidance to researchers indesigning or adjusting studies such that the desired level ofaccuracy can be achieved.


jackknife estimators within program CAPTURE (e.g. Kohnet al. 1999;Banks et al. 2003;Eggert et al. 2003;Frantz etal.2003). Like capunre, these estimators assume a closed population and allow capture heterogeneity. We were interestedin whether they can be used to analyse DNA-basedmark-recapture studies that lack capture sessions. Thegrid simulations indicate that the Chao estimator compareswell with capwire (Fig. 1). It generally shows low bias andgood coverage, though its CIs and MREs tend to be larger.The jackknife estimator generally suffers from high biasand poor coverage when sampling intensity is low. Even athigher sampling intensities, its coverage drops into the20-80% range in larger populations (N = 196and 400;datanot shown).

Which estimator is best depends on the situation and thepriorities of the researcher. The strengths of capwire are thatit displays good coverage, relatively narrow CIs, and ithandles capture heterogeneity well. In simulations withheterogeneity (GI0 and G12), it is consistently as good orbetter than all other methods in all four measures of performance (Fig. 1). Its drawbacks include a tendency tooverestimate N when DNA deposition rates are approximately even and a drop in coverage in larger populationssampled at high intensity. Alternatively, if an accurate pointestimate is of paramount concern and a researcher has goodreason to believe that deposition rates are approximatelyeven, then the exponential rarefaction method of Eggertet al. (2003) is a good option. Using capwire and imposingECM will also provide good estimates in such cases. Unfortunately, coverage for the Eggert method is quite poorunless N is moderate. The strength of the Chao estimator isthat it is robust across a broad range of conditions. It showslow bias, coverage is generally above 90%, and althoughit lacks precision (wide CIs) and accuracy (large MRE) atlow sample intensity, it displays the desirable propertyof providing consistently better estimates as sample sizeincreases. In small populations, it is generally outperformedby other methods (particularly capwire), but as populationsget larger it is increasingly superior. This was confirmed insimulations on N =196and 400 (data not shown).

One potential criticism of this analysis is that both theChao and jackknife estimators have been used in a waythat violates a basic model assumption. They assume thatthe data come from multiple sessions with no more than onecapture per individual per session. In fact, the simulateddata have no sessions. Recall that we arbitrarily created thenumber of sessions for these estimators by defining it as thelargest number of times any individual was captured. Thisis not a trivial concern because these methods estimateN by estimating the per session capture probability, whichdepends directly on the number of capture sessions. Thisbegs the question: would these methods perform better ifsampling really were conducted in temporally distinctsessions with multiple within-session observations reduced

3.0

3.0

¢ N= 16

o N=25

o N=49

A N= 100

O¢§>

2.5

2.5

2.0

1.5

1.5

o +¢

o

0+------.,-----,-------,,------,

1.0

Z 1.5(f.fJ)

.es:'5~(3 0.5

2.0

# Obs/lnd

Fig. 2 Effect of number of observations/sampled individual onmean relative error (a) and precision (b) in grid simulatedpopulations of differing size (indicated in legend). Data frompopulations of different compositions (Cl, G4, GI0 and G12) arecombined.

Comparison toother methods

Rarefaction curve-fitting has been used in a number ofDNA-based studies to estimate population size (e.g. Kohnet al. 1999;Eggert et al. 2003;Wilson et al. 2003; Bellemainet al. 2005). Rarefaction has an intuitive appeal in that itassumes replacement sampling and has no temporal dimension. In grid simulations the hyperbolic curve (y = ax/(~ + x):Kohn et al.1999)overestimates population size nearly 50%on average, shows coverageof lessthan 50%,and large MREs.The exponential curve of Eggert et al.,y = «(l - e~x} (2003),however, performs comparably well (Fig. 1).It produces pointestimates with low MREs. While biased low when capturerates are heterogeneous (GlOand GI2), it is approximatelyunbiased when deposition rates are even (Gl and G4). Formoderate populations (e.g.N =25-49), coverage is generallynear 90% or better, but in small and large populations(i.e. 16:S; N;::: 100), it shows much poorer coverage (datafor small N not shown). These results suggest it might befruitful to explore alternate ways of calculating CIs for thisrarefaction method.

Two of the most commonly used methods for shortduration, mark-recapture studies have been the Chao and

(a) 0.4

0.3

wcc 0.2::;;::

0.1

01.0

(b) 2



Table 6 Comparison of Chao estimator performance on grid simulated data sets with and without sampling divided into temporal capturesessions. For each of the four data sets (G1, G4, G10 and G12), analysis was conducted in two ways: no. of sampling sessions = 4 (andmultiple within-session observations of an individual removed) and no. sample sessions = maximum no. of captures of any individual(with the four real sessions ignored and no observations removed). For all simulations N = 49 and 5 = 100

CompDist No. of samp sessions N ReI bias Clwidth Coverage MRE

G1 4 52.2 0.07 24 0.86 (0.14, 0) 0.06max no. of captures 50.3 0.03 25 0.91 (0.08,0.01) 0.03

G4 4 53.0 0.08 26 0.82 (0.17, 0.01) 0.11max no. of captures 51.1 0.02 24 0.89 (0.10, 0.01) 0.10

GlO 4 52.4 0.07 32 0.91 (0.00, 0) 0.15max no. of captures 50.3 0.03 31 0.95 (0.03, 0.02) 0.13

G12 4 49.5 0.01 27 0.95 (0.05, 0) 0.12max no. of captures 48.8 0.00 30 0.94 (0.04, 0.02) 0.12

to a single observation? When possible, should DNA-basedpopulation estimation studies be designed this way? Weconducted a cursory exploration of this issue by running a small set of simulations (N =49, S =100, compositions G1, G4, GlO and G12) where sampling was dividedinto four sessions. Hence in the 4000 simulated steps,sampling was conducted along transects as before every1000time-steps. The data were then analysed with the Chaoestimator in two ways: (i) in the appropriate manner withthe number of sampling sessions equal to four and multiplewithin-session observations of an individual reducedto one and (ii) as it has been done in this study with thesession data ignored, no observations removed, and thenumber of sessions arbitrarily defined as the maximumnumber of captures.

In this case the results are clear: bias, MRE, coverage andCI width are all good and usually better when the data arepooled into a single sample and not divided into sessions(Table 6). While this result needs to be tested across abroader range of conditions, it suggests that researchers arethrowing away valuable information for estimating N bycondensing multiple captures into one. In fact, the abilityto capture an individual multiple times without expendinggreatly more sampling effort is one of the potential strengthsof DNA-based mark-recapture. Designing sampling so thatit is intense but of short duration is exactly what is neededto minimize violation of the closed-population assumption. Of course, it is important that multiple observationsare not pseudoreplicates from the same time and the sameplace. For example, a bear investigating a sent lure is likelyto leave several hair snares at the same trap on the sameoccasion (entering and leaving). Clearly these multipleobservations should be treated as a single observation andare quite distinct from capturing the individual at twoseparate locations. Capwire was designed to make use ofmultiple observations. This type of DNA-based data canalso be accommodated by the traditional Chao estimator.


Capwire and traditional mark-recapture data

We were additionally interested in whether capwire can beused to analyse traditional mark-recapture data sets. As anexploratory step we analysed five wildlife data sets from theliterature where N was independently known (Table 5).The performance of capwire is remarkably good. In all casesthe true N is contained within the CI and in four of fivecases the MRE is less than 10%. The method also performswell on a population of taxicabs. These data sets have beenanalysed and reanalysed many times in the literature usingdifferent estimators (e.g.Otis etal. 1978;Chao 1988;Tardella2002). Considered across data sets, capwire's performanceis as good as or better than any of the other estimators. Forexample, an analysis of the five wildlife data sets in Table 6using the Chao estimator give very similar bias, MRE values,but substantially wider CIs in four of five cases (results notshown).

Why should capwire perform well on data that violateits sample with replacement model assumption? Forcapwire to perform well, the number of observations of eachindividual must be proportional to the fraction of thepopulation that individual represents. Even when data arecollected as binary events over multiple sessions thisshould still tend to occur, especially as the number ofsample sessions increases. This also suggests that capwiremay be robust to time heterogeneity in capturability. In thesession-based methods, one complexity is that capturabilitymay vary between sessions due to factors such as weatherand season (Otis et al. 1978). Parameters accounting forthese changes in capturability must be estimated. But solong as every individual is affected in the same way (i.e,there are no individual by time interactions), then temporalchanges will not distort the proportional representation ofindividuals in the sample. For this same reason, the lack oftemporal information in many DNA-based mark-recapturestudies (e.g. single sweep scat collection studies) should


not pose a problem. This indication that capwire can beeffectively used on traditional mark-recapture data setsis promising, but needs to be substantiated by furtherstudy.

Improving capwire

Many traditional mark-recapture models view an animal'scapture probability as being drawn from a distribution.Examples include point mixture (or latent class) modelswith two or more types (e.g. Norris & Pollock 1996;Pledger2000),beta-binomial mixture models (e.g, Dorazio & Royle2003),logistical and log-linear models (e.g. Coull & Agresti1999). It may be possible to improve the performance ofcapwire by considering a wider range of underlying capturedistributions. Capwire might also be improved by accountingfor uncertainty in class membership by employing, forexample, the EM algorithm (Coull & Agresti 1999).It alsoseems reasonable to adjust the urn model so that samplesare not independent of one another. In reality, if two samplescome from nearby geographical locations, they will tend tocome from the same individual more often than randomlydrawn samples. Samples from disparate locations will tendto be the same individual less often than random. Thisgeographical component of the data is currently ignored,but it seems that such information could be utilized byadding a spatial dimension to the urn model.

Conclusions

The ability to sample with replacement in DNA-basedmark-recapture studies yields a different type of data thanthat obtained in traditional trap-based studies. Removingmultiple observations from disparate locations withinsessions wastes valuable information. In other DNA-basedstudies, there are no clearly defined sessions. The proposedcapwire method is suited to these types of raw data sets. Thesimulation study conducted here suggests that capwiredoes a particularly good job when dealing with smallerpopulations (N::; 100)and substantial capture heterogeneity.An analysis of a number of real genetic mark-recapturedata sets demonstrates that capture heterogeneity will becommonly encountered.

Acknowledgements

We thank [en Adams, Andrea Taylor and Sam Banks for generously providing raw data for analysis. Funding was provided byNational Science Foundation Experimental Program to StimulateCompetitive Research Grant 9720634and National Science Foundation Grants 0080935 and 9871024.Paul Joyce's research is partiallysponsored by the Initiative in Bioinformatics and EvolutionaryStudies (IBEST)at the University of Idaho; funding was providedby NSF EPSCoR, EPS-0132626and NIH NCRR grant NIH NCRR1P20RR016448-01,and NSF DEB-0089756.

References

Banks SC, Hoyle SD, Horsup A, Sunnucks P, Taylor AC (2003)Demographic monitoring of an entire species (the northernhairy-nosed wombat, Lasiorhinus krefftii) by genetic analysis ofnon-invasively collected material. Animal Conservation, 6, 101107.

Bellemain E,Swenson JE,Tallmon D, Brunberg S,Taberlet P (2005)Estimating population size of elusive animals with DNA fromhunter-collected feces: four methods for brown bears. Conservation Biology, 19, 150-161.

Boulanger J,McLellan BN, Woods JG,Proctor MF, Strobeck C (2004)Sampling design and bias in DNA-based capture-mark-recapturepopulation and density estimates of grizzly bears. Journal ofWildlife Management., 68, 457-469.

Burnham KP, Overton WS (1979)Robust estimation of populationsize when capture probabilities vary among animals. Ecology,60, 927-936.

Carothers AD (1973) Capture-recapture methods applied to apopulation with known parameters. Journal of Animal Ecology,42,125-146.

Chao A (1988)Estimating animal abundance with capture frequencydata. Journal of Wildlife Management, 52, 295-300.

Comps B, Comory D, Letouzey J, Thiebaut B, Petit RJ (2001)Diverging trends between heterozygosity and allelic richnessduring postglacial colonization in the European beech. Genetics,157,389-397.

Coull BA,Agresti A (1999)The use of mixed logit models to reflectheterogeneity in capture-recapture studies. Biometrics, 55, 294301.

Creel S, Spong G, Sands JL et al. (2003)Population size estimationin Yellowstone wolves with error-prone noninvasive microsatellite genotypes. Molecular Ecology, 12, 2003-2009.

Dorazio RM, Royle JA (2003) Mixture models for estimating thesize of a closed population when capture rates vary among individuals. Biometrics, 59, 351-364.

Edwards WR, Eberhardt L (1967)Estimating cottontail abundancefrom live-trapping data. Journal of Wildlife Management, 31, 8796.

Eggert LS, Eggert JA, Woodruff DS (2003) Estimating populationsizes for elusive animals: the forest elephants of KakumNational Park, Ghana. Molecular Ecology, 12, 1389-1402.

Flagstad 0, Hedmark E, Landa A et al.(2004)Colonization historyand noninvasive monitoring of a reestablished wolverinepopulation. Conservation Biology, 18, 676-688.

Frantz AC, Pope LC, Carpenter PJetal.(2003)Reliable microsatellitegenotyping of the Eurasian badger (Meles meles) using faecalDNA. Molecular Ecology, 12, 649-1661.

Greenwood RJ, Sargeant AB, Johnson DH (1985) Evaluation ofmark-recapture for estimating striped skunk abundance. Journalof Wildlife Management, 49, 332-340.

Kohn MH, York EC, Kamradt DA, Haught G, Sauvajot RM,Wayne RK (1999) Estimating population size by genotypingfaeces. Proceedings oftheRoyal Society ofLondon. Series B,BiologicalSciences, 266, 657-663.

Link WA (2003) Nonidentifiability of population size fromcapture-recapture data with heterogeneous detection probabilities.Biometrics, 59, 1123-1130.

Mares MA, Streilein KE, Willig MR (1981) Experimental assessmentof several population estimation techniques on an introducedpopulation of eastern chipmunks. Journalof Mammalogy, 62,315-328.


McKelvey KS,Schwartz MK (2004)Genetic errors associated withpopulation estimation using non-invasive molecular tagging:problems and new solutions. Journal of Wildlife Management, 68,439-448.

Miller CR, Joyce P, Waits LP (2002)Assessing allelic dropout andgenotype reliability using maximum likelihood. Genetics, 160,357-366.

Mowat G, Paetkau D (2002)Estimating marten Martes americanuspopulatin size using hair capture and genetic tagging. WildlifeBiology, 8, 201-208.

Mowat G, Strobeck C (2000)Estimating population size of grizzlybears using hair capture, DNA profiling, and mark-recaptureanalysis. Journal of Wildlife Management, 64, 183-193.

Norris III JL, Pollock KH (1996) Nonparametric MLE under twoclosed capture-recapture models with heterogeneity. Biometrics,52,639-646.

Otis DL, Burnham KP, White Gc, Anderson DR (1978)statisticalinference from capture data on closed animal populations. WildlifeMonographs, 62.

Palsbell PJ, Allen J, Berube M et al. (1997) Genetic tagging ofhumpback whales. Nature, 388, 767-769.

Pledger S (2000) Unified maximum likelihood estimation forclosed capture-recapture models using mixtures. Biometrics, 56,434-442.



Taberlet P, Waits LP,Luikart G (1999) Noninvasive genetic sampling:look before you leap. Trends in Ecology & Evolution, 14, 323-327.

Tardella L (2002) A new Bayesian method for nonparametriccapture-recapture models in presence of heterogeneity. Biometrika,89,807-817.

Wilson GJ, Frantz AC, Pope LC et al. (2003) Estimation of badgerabundance using faecal DNA typing. Journal ofApplied Ecology,40,658-666.

Woods JG, Paetkau D, Lewis D, McLellan BN, Proctor M, StrobeckC (1999)Genetic tagging of free-ranging black and brown bears.Wildlife Society Bulletin, 27, 616-627.

Craig Miller is a post-doc at University of Idaho working onstatistical problems in conservation genetics. Paul Joyce is aProfessor of Mathematics, Statistics and Bioinformatics at theUniversity of Idaho. His interdisciplinary work involves mathematical modeling and statistical theory in population genetics,experimental evolution and Systematic Biology. Lisette Waits isthe co-director of the Laboratory for Ecological and ConservationGenetics and the Center for Research on invasive species andsmall populations. Her research program focuses on the conservation genetics of a variety of wildlife species.

2004 C. R. MILLER, P. JOYCE and 1. P. WAITS

Appendix I

Capwire algorithm used to estimate population size

1. Initialize a: Calculate mean no. of captures/sampled individual. Calculate mean no. of captures for individuals captured more than thisaverage (average above) and less than this average (average below). Define initial U= (average abover/Iaverage below).

2. Findexpected capture counts: Assuming N =T, NA =N/2, NB=N/2, and a =U,calculate the expected number of captures an individualof each type: E(cA) =S[a/(aNA + NB)J and E(cB) =S[1/(aNA + NB)].

3. Assign capture classes: Assign each sampled individual a capture class based on the following rules. If it is observed one time, assign itto capture class B. Otherwise calculate the absolute difference between the observed no. of captures and that expected for each class:I cj - E(cA) 1and 1ci - E(cB) I.Assign the individual to whichever absolute difference is smaller. This defines TA and TB as well.

4. MLE estimation of N: Given this vector of capture class assignments, TAl TB, and a, find the MLE of NA and NB. Do this by initiallycalculating likelihood (equation 2) assuming NA = TAl NB= TB'Then begin incrementing NBup by one and calculating the likelihood.Continue until the likelihood begins to decline; the largest likelihood defines the MLE of NB" MLE of NA is TA' (NA is not incrementedbecause adding the easier to capture type A individuals yields a smaller likelihood than adding type B individuals).

5. Bias adjust a using equation 3 to obtain Uadjusted'

6. Repeat MLE estimation: Repeat step 4 using Uadjusted to update estimates of NA and NB.

7. Repeat capture class assignment: Repeat steps 2 and 3 based on Uadjusted' and update estimates of TA and TB.

8. Check forconvergence: If any of the capture class assignments in the sample changed in step 7, return to step 4. If capture assignmentsdo not change, go to step 9.

9. EstimateN:N=NA +NB•

10. Bootstrap to obtainconfidence intervals: Generate many data sets by drawing with replacement from an urn in which there are NA typeA balls, NB type B balls, and the balls have weights of 1 and a, respectively. For each conduct parameter estimation (steps 1-9). Leta ts be the test size specified by the user. Define the lower and upper confidence bounds as the a tj 2 and 1- a t/ 2 quantiles of thebootstrap estimates.

© 2005 Blackwell Publishing Ltd, Molecular Ecologv, 14, 1991-2005


Appendix II

Grid simulations were conducted using the algorithm given in section A. The types of individuals composing populations are listed andparameterized in section BA. Grid Simulation algorithm. Note that individuals are independent; their movements and depositions have no effect on each other.

1 Specify size of square grid and population size, N.2 Subdivide grid into N non-overlapping squares and place an individual in centre of each square. Designate this as an individual's horne

region centre (HRC).3 Randomly assign each individual a type according to specified composition distribution (Table 1). This defines movement type

(sedentary or transient) and deposition rate (seldom, moderate, or often) for each individual as given in Appendix lIB.4 Determine deposition schedule for each individual. For each individual draw number of steps until 1st depositing a scat from an

exponential distribution with rate Adeposit (Appendix lIB). Determine number of additional steps until 2nd scat by drawing again.Continue until sum of steps exceeds 4000 (the duration of the simulation). While simulating movement of each individual (steps 5-10),deposit scat at that grid location according to this deposition schedule.

5 For each individual chose one of the Cartesian directions to move with equal probabilities.6 For each individual, determine number of steps to move before potentially changing direction by drawing from an exponential

distribution with rate Atum (Appendix lIB).7 For each individual, move this number of steps. Deposit scats as specified by deposition schedule (step 4).8 For each individual, determine new direction to move. With specified probability P(H), the individual moves in the direction that takes

it most directly toward its HRC and in one of other three directions with probability 1- P(H)/3. Hence P(H) describes fidelity to hornewith P(H) > 0.25 causing sedentary behaviour and P(H) < 0.25 causing transient behaviour. When two directions tie for directness tohorne, they are each assigned probability P(H) /2 and the other two directions are assigned probability (1 - P(H) /2) /2. When individualsencountered boundaries, the same rules were employed except that there are only three (at sides) or two (at corners) potential directionsto choose from.

9 For each individual repeat steps 6, 7, and 8 until total number of steps is 4000.

B.Parameterization of six types of individuals in grid simulations and resulting deposition count and horne region size. Based on simulationof 4000 step duration. Deposition rate, draw to horne, and turning rate defined in Appendix lIA

Ind Type Deposition Draw to Turning Exp. no. of 90% hornecode Type description rate (Adeposit) horne = P(H) rate (Adeposit) scats on grid region*

SOD sedentary, often depositor 1/40 0.35 5 100 66 (± 21)SMD sedentary, moderate depositor 1/80 0.35 5 50 66 (±21)SSD sedentary, seldom depositor 1/160 0.35 5 25 66 (± 21)TOD transient, often depositor 1/40 0.15 5 100 300 (±96)TMD transient, moderate depositor 1/80 0.15 5 50 300 (±96)TSD transient, seldom depositor 1/160 0.15 5 25 300 (±96)

*Distance to horne region centre average individual spent 90%of time within (± SD) based on 100simulations on grid of size 500 x 500 steps.


gars•m rica

Editor

Technical Editors

HILARY S. COOLEY AND MICHAEL R. CONOVER

JONATHAN A. JENKS

•ana Inin N rth

Western Association of Fish and Wildlife AgenciesCougar Working Group

andJack H. Berryman Institute

UtahState MISSISSIPPI STATEUniversity UNIVERSITY

TM

Berryman Institute Press, Logan, Utah, 84322, USA

Western Association ofFish and Wildlife AgenciesJack H. Berryman Institute, Utah State University and Mississippi State University

© 2011 by Western Association of Fish and Wildlife Agencies and Jack H. Berryman Institute. All rights reserved. This book or any part thereof may not be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, microfilming, or by any information storage or retrieval system,without prior written permission from the publisher.

This book contains information obtained from authentic and highly-regarded sources. Reprinted material is quoted with permission, and sources are indicated. A wide variety of references are listed. Reasonable efforts havebeen made to publish reliable data and information, but the authors and the publishers cannot assume responsibility for the validity of all materials or for the consequences of their use.

ISBN (paper) 0-9742415-20-0 Berryman Institute Press

Library of Congress Control Number: 2011925643

For inquiries or to order copies ofthis publication, please write to the Jack H. Berryman Institute Press, UtahState University, 5270 Old Main Hill, Logan, UT 84321-5270; or e-mail [email protected].

Publisher's Cataloging-in-Publication InformationManaging cougars in North America / edited by Jonathan A. Jenks; technical editors, Hilary S.

Cooley, Michael R. Conover.200p. em.Includes bibliographical references and photographs.ISBN 0-9742415-20-01. Puma-North America. 2. Wildlife management-North America. I. Jenks, Jonathan Alden,1954-

II. Cooley, Hilary S. III. Conover, Michael R.QL737.C23 M362 2011

Cover photo: Daniel J. ThompsonBack cover photos: Cougar (top) by Eric York, National Park Service (deceased); cougar in snow (middle) by

David Stoner; cougar (bottom) by Brandon Holton, National Park Service, Wldlife Biologist

Suggested citation formats:Jenks, J. A., editor. 2011. Managing cougars in North America. Jack H. Berryman Institute, Utah State

University, Logan, Utah, USA.

Apker, J. A., D. Updike, and D. Ho1dermann. 2011. Strategies to manage cougar-human interactions.Pages 145-164 in J. A. Jenks, editor. Managing cougars in North America. Jack H. Berryman Institute,Utah State University, Logan, Utah, USA.

Jack H. Berryman Institute PressUtah State University

Logan, Utah 84322-5230

abolishment of bounties, depredation policies started targeting cougars associated withlivestock losses. By the early 1970s, cougarswere managed as a game species across moststates and Canadian provinces (median establishment was 1971), which represented thefirst form ofprotection for cougar populations(CMGWG 2005).

Hunting accounts for the majority ofmortalities in most cougar populations (Lindzey1987, Logan and Sweanor 2001, Pierce andBleich 2003). Although cougar management varies by state, plans usually containone or more of the following objectives: (1)to provide recreational hunting opportunitiesto sportsmen; (2) to reduce cougar numbersaround urban areas; (3) to reduce cougar numbers in areas where livestock production isimportant; (4) to balance cougar populationswith ungulate populations; and (5) to maintainsustainable cougar populations. In this chapter, we discuss these objectives, the strategiesmanagers typically use to regulate hunting(i.e., general seasons, limited entry programs,or quota systems), and some of the variablesthat affect hunter success (e.g., snow conditions, hound hunting, nonconventional hunting methods, and road densities). In addition,

anagement: Cougar Huntingpulation

,L.I..lJLJftl'U. s. COOLEY Wildlife Biologist, Idaho Department ofFish and Game, Nampa, Idaho,USA

D. BUNNELL Mammals Program Coordinator, Utah Division of Wildlife, Salt LakeCity, Utah, USA

C. STONER Ph.D. Candidate, Utah State University, Logan, Utah, USAH,L.l'-'.ll.i~L L. WOLFE Professor, Department of Wildland Resources, Utah State University,

Logan, Utah, USA

Unregulated hunting, poisoning, trapping, and habitat alterations that affectedprey numbers caused the near extinction ofcougars in eastern North America. By 1900,cougars had largely been extirpated east ofthe Rocky Mountains, with the exception ofFlorida (Young and Goldman 1946, Nowak1976). Through the first half of the twentiethcentury, management emphasized eradication. Bounties were paid as an incentive toremove cougars to protect wild ungulates anddomestic livestock. The payments were significant for the time. In 1905, bounties paidwere between $5 and $20 for most westernstates, with higher bounties often paid forfemales in an attempt to reduce populations(US. Department ofAgriculture 1905). In the1920s, California paid bounties of $30 for afemale cougar and $20 for a male; in 1945,bounties increased to $60 for females and $50for males (Cougar Management GuidelinesWorking Group [CMGWGJ 2005). Throughout the western states, bounties continued intothe 1960s when cougar management shiftedto a briefnon-bountied but nonprotected period. At a minimum, 62,932 cougars (1,338 peryear) were killed in western North America(Table 1; Nowak 1976). Coincident with the

L

--------------·4112 Managing Cougars in North America

Table 1: Approximate numbers of cougars harvested inNorth America, 1910-1974, based on the compilation ofNowak (1976).

Management Objectives

3815343017638619

3822

1261693

138

ed to harvest cougars under thisstrategy currently is unattainable(Sinclair et aI. 2006). Statewidepopulation estimates often areextrapolated from study-specificdensity estimates and amount ofcougar habitat (Table 2). Giventhe low confidence in such methods, some states (e.g., Utah andWyoming) provide no estimatesof cougar population numbers.Thus, wildlife managers are leftto estimate the level of sustainable harvest by tracking the demographics ofcougars in the harvest.For example, in a lightly-huntedpopulation, male and subadultfemale cougars comprise the majority of harvest, and populationsgenerally maintain older animals.In contrast, heavily hunted popu-lations typically experience an

increasing presence of adult females in theharvest and an overall younger age structure(Anderson and Lindzey 2005, Cooley et aI.2009a). As a result of this retrospective approach, it is difficult to know when harvestrates are exceeding recruitment until after thefact. Assuming that there are adequate sourcepopulations to provide dispersal immigration,combined with in situ recruitment, then, cougar populations can be resilient when harvestrates exceed recruitment. For example, onMonroe Mountain in southern Utah, cougardensities were reduced by >60% over 6 yearsof heavy harvest, but recovered to pretreatment densities after 5 years of reduced hunting pressure (Stoner et aI. 2006). In the SanAndres Mountains in southern New Mexico,53% of the estimated adult cougars were removed in 6.5 months. The adult segment ofthat population recovered in 31 months due to

Averagenumber of

cougarsharvestedper year

8858,257

15,97410,9832,0262,791

5022,0241,2836,841

776

5,3375,253

Minimumnumber of

cougarsharvested

Period

1942-19641917-19731910-19551907-19731917-19741928-19731921-19741917-19731917-19731918-19731925-19731913-19781936-1973

State or province

Alberta

Ariz.

British Columbia

Calif.

Colo.

Ida.

Mont.

Nev.

N.M.

Oreg.

Tex.

Ut.

Wash.

Providing Hunting OpportunitiesFor most western states and provinces,

the primary objective of hunting is to providerecreational opportunities for hunters whilemaintaining sustainable wildlife populations.State agency records indicate that >62,000cougars (an average of2,264 cougars per year)were harvested through sport hunting over a36-year period (Figure 1). In theory, harveststrategies based on the maximum sustainedyield (i.e., harvest rate = net recruitment rate)should produce the largest sustainable harvestopportunities. However, the precision need-

we include a summary of studies that haveinvestigated the effects of hunting on cougarpopulations, along with a list of managementimplications.

r

Removal Program is aimed at removing cougars in areas where the number of complaintsis high. However, recent research suggeststhat targeting an area (rather than an individualcougar) for removals may not be effective inreducing cougar numbers because ofrapid immigration into vacated home ranges (Robinsonetal. 2008, Cooley et al. 2009a). Whileremovals may not achieve a desired population effect, removals targeted at areas with a propensity of cougar-livestock damage or concernsfor human safety can ease social and politicalconcerns. Predator suppression (mortality =

16 to 20% of independent cougars) in 4 Colorado management areas with historic livestockdamage problems and in the management areaimmediately around metropolitan Denver isprimarily aimed at achieving desired sociopolitical goals (J. Apker, Colorado Division ofWildlife, personal communication). Alterna-

I '"".••• • • • e 0.

~l:\600+-----------------i--Jv-------~--+-------I::::s

8 500+----------------I--+---5'---';--'\--'---';--------j

'0; 400+------~---------#-_,i'_~~_<:_:<'_;_~-~-r__I

.Q

E 300 +-------~------.._!__-¥-__h,~_:ir_-----=-------j"--'";;;--_\_"~::--1::::sZ

200t--;...-:-."----~-~-",;-~_::;:.I:,~~~~fA:1~~~~~__j

Population Reductions around Urbanand Livestock Areas

Cougar-human conflicts have increasedin recent years due to fragmentation and urbanization of traditional cougar habitat (Torres et al. 1996). All western states and provinces allow the killing of individual cougarsthat threaten public safety or private property.Another population reduction managementoption is to target specific areas that have hadnumerous cougar-human-livestock conflicts(Aune 1991, Halfpenny et al. 1991, Torres etal. 1996). Washington's Public Safety Cougar

Chapter 5: Population Management: Cougar Hunting 1 13

900.,----------------------------,Sport cougar harvest

700+--------------------rl-------l~.___------I

800+-------------------..-;:,..----------{

Figure 1. Number of cougars harvested from recreational hunting in the western United States, 1972- 2008.Data from Texas and California are not included because hunter data are reported voluntarily (Texas) and sporthunting is absent in California.

local recruitment and immigrants from an adjacent population (Logan and Sweanor 2001).A Wyoming cougar population reduced by43% under 2 years ofheavy harvest recoveredfollowing 3 years of reduced pressure (18%reduction; Anderson and Lindzey 2005).

114 Managing Cougars in North America

1 Habitat and associated densities defined as: core (2.0-3.0 cougars/lOO km-), patch-dispersal(0.89-1.2 cougars/l 00 krrr'), and poor/marginal (0.2-0.3 adults/l 00 krrr').2 Density estimates from preliminary studies in various regions from 1995-2005.3 Density estimates from 2 ecoregions and are 10 years old.

Table 2. Statewide estimates of cougar habitat, cougar density, cougar population size, and extrapo-lation method used to derive population size. Estimates current in February 2009. Oregon estimatesfrom Oregon Cougar Management Plan (2006).

Cougar Density (cou- Population Extrapolation methodhabitat gars/l 00 knr') estimate(knr')

Ariz. 291,374 0.7 cougars 1,500-2,000 Amount of habitat and densitycougars range.

Calif. 250,000 3-9 cougars 4,000-6,000 cou- Densities and vegetation types.gars

Colo. 141,879 2-2.5 cougars 3,000-3,500 inde- Habitat quality and an associatedpendents density.

Ida. N/A N/A 1,500-2,500 Back-calculations from harvest,life tables, and known cougarproductivity.

Mont. 118,137 1.3-3.2 adults and 1,583-3,744 adults Amount ofhabitat and densitysubadults and subadults range.

N.Dak. 4,637 N/A 45-74 potential Habitat quality and density esti-cougars mates from other populations.

N.M, 289,507 0.2-3.0 adults! 2,041-3,043 cou- Habitat quality and density esti-gars mates from past research.

Nev. 146,311 0.82-1.23 average 2,400 cougars Life table model using harvest,adults recruitment, prey availability.

Oreg. N/A 1.7-6.2 yearlings 1,284-7,644 cou- Prey density.and adults" gars

S.Dak. 4,856 5.2 cougars 220-280 cougars Research, mark/recapture, andin Black Hills modeling.

Tex. 198,654 0.3-0.7 adults' 500-1,300 cougars Amount ofhabitat and high-lowdensity estimates.

Vt. 111,249 3.1-3.3 cougars N/A

Wash. 88,497 3.3-3.5 total cou- 1,900-2,100 sub- Density estimates from pastgars adults and adults studies.

Wyo. N/A Variable N/A

Balance Cougar Populations withUngulate Populations

The effect that cougars have on biggame populations is a common concern

tive methods to deal with nuisance cougarsinclude relocation, aversive conditioning,public education and outreach, and fines forattracting wildlife (Beausoleil et al. 2008).

Chapter 5: Population Management: Cougar Hunting 115

among wildlife managers and public stakeholders. Yet, experimental research examining predator-prey relationships is limited.Most work on this subject has focused on effects of cougar predation on small, isolatedpopulations of rare prey, particularly bighornsheep (Ovis canadensis). Research resultshave been equivocal; some studies report thatcougar predation can limit growth or evenextirpate small populations of bighorn sheep(Wehausen 1996, Hayes et al. 2000, Schaeferet al. 2000, Kamler et al. 2002, Rominger etal. 2004, McKinney et al. 2006), pronghorn(Antilocapra americana; Ockenfels 1994),and feral horses (Equus caballus; Turner andMorrison 2001). Others, however, have foundimpacts to be inconsequential (Homocker1970, Rominger et al. 2004). More importantly, intensive monitoring of individual cougarshas revealed complex behavioral patterns ofpredation. Several studies have suggested thata single cougar can cause sporadically highrates of predation in small prey populations,such as bighorn sheep (Logan et al. 1996,Ross et al.1997, Ernest et al. 2003) and mountain caribou (Rangifer tarandus; Katnick2002). In these cases, removal of the offending animal may prove successful. The presence of alternate prey can threaten sympatric,declining prey species because it increases thenumber ofcougars that can be supported in anarea. In some cases, the presence of alternateprey has resulted in higher predation rates onthe less abundant prey species, including bighorn sheep (Kamler et al. 2002, Rominger etal. .2004), mule deer (Odocoileus hemionus;Robinson et al. 2002, Cooley et al. 2008), feral horses (Turner et al. 1992), and livestock(Linnell et al. 1999). In situations where numerous alternative prey exist, cougar population reductions may be temporary due to rapid

immigration from neighboring areas, makingpredation-related reductions ineffective.

Beyond the quantification of predation rates (e.g., Murphy 1998, Anderson andLindzey 2003, Cooley et al. 2008, Laundre2008), little research has been conducted atthe population level to understand the effectsofcougar predation on big game species, suchas mule deer and elk (Cervus elaphus). Logan and Sweanor (2001) argued that cougarpredation, when combined with poor forageconditions resulting from drought, had thegreatest effect on mule deer and desert bighorn sheep populations. In another study thatmonitored mule deer for 14 years and cougarsfor 6 years, Bowyer et al. (2005) showed thatmule deer numbers were influenced primarily by range conditions and that the cougarpopulation actually tracked mule deer with an8-year lag. This suggests that mule deer andcougar abundance are intertwined, and determining l-way effects may be extraordinarilydifficult without well-replicated experimentalmanipulations conducted under a range of environmental conditions (see Chapter 3 for afurther review of cougar-prey relationships).

Arizona currently uses multiple cougarbag limits for areas identified with decliningpopulations of pronghorn or bighorn sheep,a recently translocated population of bighornsheep, or pronghorn or bighorn sheep populations below management objectives. Multiplebag limits did not increase statewide cougarharvest nor harvest within areas specificallymanaged under multiple bag limits. But multiple bag limits did allow hunters to participate in efforts to reduce predation in specificareas to meet management objectives. Idahoand Utah also allow higher harvest in areaswhere ungulate populations are a concern.Idaho reported improved elk calf:cow ratios


after cougar and bear removal (M. Nadeau,Idaho Department of Fish and Game, unpublished data). Utah reported beneficial effectsfor bighorn sheep, but not for mule deer (K.Bunnell, Utah Division of Wildlife Resources, unpublished data). Colorado currently isplanning a predator control program to improve bighorn sheep populations that are limited by predation and to increase success oftranslocation efforts. Control actions will beconducted as management experiments withattendant a priori hypotheses and appropriate monitoring to determine success or failureof the efforts (1. Apker, Colorado division ofWildlife, personal communication; Colorado Bighorn Sheep Management Plan 2009).

Predator Control: Targeted orWidespread?

Predator control has long been an accepted game-restoration tool (Leopold 1933),and it is used today to protect endangered orthreatened species, support ungulate reintroductions, and facilitate recovery of ungulatepopulations following catastrophic events(Hornocker 2007). Two competing management paradigms exist relative to game restoration. The first states that predator numbersshould be reduced to the extent possible so thatprey mortality is minimized and the fledglingpopulation can reach the critical mass necessary to achieve exponential growth. The second perspective states that cougar predationshould be addressed on an individual basis.

In some areas, the overall reduction ofcougar densities from hunting has resulted ingreater prey survival and population growth(Sawyer and Lindzey 2002). Yet, the strengthof this cause-and-effect relationship is confounded by numerous factors, including habitat characteristics and the abundance of alternative prey species, such as deer and livestock

(Rominger et al. 2004). Studies of cougarpredatory behavior suggest that hunting habits are individualistic, and particular cougarsor social classes may specialize on certainprey species. This has been demonstratedwith free-ranging horses (Turner et al. 1992),moose (Alces americanus; Ross and Jalkotzy1996), bighorn sheep (Ross et al. 1997, Linnell et al. 1999, Logan and Sweanor 2001),and caribou (Katnik 2002). This suggests thatan individualistic approach for predator control may be an effective method for minimizing predation losses (Hoban 1990, Linnell etal. 1999, Sawyer and Lindzey 2002, Ernest etal. 2003).

In some instances, social or politicalpressures may force managers into reducingcougar densities to address perceived impactsof cougars on big game populations. In thesecases, managers need to monitor the controlefforts in relation to objectives and identifywhen cougar reduction efforts will be stoppedif big game populations fail to respond to reduced cougar populations. For example, inUtah, predator management efforts, which arefocused on cougars, are required to identifywhen the efforts will stop, based on 3 scenarios: (1) recovery of the game population backto a specified level; (2) lack ofrecovery ofthegame population despite cougar density reductions; and (3) status of the cougar population. If cougar predation is not the root causeof struggling game populations, and managers fail to identify a priori criteria to stopcougar reduction efforts based on the statusof the cougar population or a lack of responsein the prey population, managers can findthemselves perpetually stuck in these efforts.

Maintain Sustainable CougarPopulations

In all western states, except Texas, cou-

0.9.---------------------------.

0.8+-------------------------+---1

immigrants may inhibit population recoveryuntil harvest is reduced. Additionally, overexploitation of adult females can depress internal recruitment, leading to population decline (Stoner et al. 2006, Robinson et al. 2008,Cooley et al. 2009a). For these reasons, moststates limit female hunting mortality to <50%

I of the total harvest (Figure 2). Conversely, ifsuch replacement or immigration occur, localized hunting pressure may be ineffective forpopulation control (Robinson et al. 2008).

According to Levins (1969), a metapopulation is a group of spatially-separatedpopulations of the same species that interactthrough dispersal. Managing cougars as interacting metapopulations can work to counteract the risks of population decline whilemaintaining hunting opportunities. Two related methods that acknowledge the need fora landscape approach to cougar managementhave been developed from long-term popu-

-InCD 0.7 ~

i::«l.c 0.6.5In 0.5CD

CiSE 0.4.!....o ._ 0.3cCDe 0.2CD

Q. 0.1

Sport female cougar harvest

Figure 2. Percentage of female cougars harvested from recreational hunting in the western United States,1972-2008. Data from Texas and California are not included because hunter data are reported voluntarily inTexas, and sport hunting is absent in California.


gars are a protected game animal. As such,agencies are charged with "sound stewardship," "long-term viability," and "preservation, protection, and perpetuation ofwildlife,"or similar phrases found in agency missionstatements or in statutory authority establishing the agency. These charges can be formidable because long-term research that capturesthe range of variation in populations of bothcougars and their primary prey are nonexistent, and affordable methods for enumeratingcougar populations have not been reported inthe literature (Choate et al. 2006).

Cougar populations have been documented to recover relatively quickly fromoverexploitation; this resiliency is thought torely on juvenile immigration from neighboring areas and philopatric recruitment of female offspring (Lindzey et al. 1992, Sweanoret al. 2000). If heavy hunting encompassesseveral neighboring populations, mortality of


lation studies: (1) zone management (Loganand Sweanor 2001) and (2) the source-sinkmetapopulation approach (Laundre and Clark2003). Both of these methods use patterns ofanimal behavior combined with metapopulation theory and principles of landscape ecology to compensate for unknowns associatedwith cougar abundance. These approachesmay contrast with current methodology thatmanages small-scale cougar management areas independently of one another.

Zone management is an adaptive management approach whereby large regionaltracts of land are partitioned into zones withdifferent population management objectives.Logan and Sweanor (2001) recommend 3types of zones: (1) control zones, where reduction ofcougar numbers is the objective; (2)hunting zones, where management providessport hunting opportunities while maintainingsustainable populations; and (3) refuge zones,where hunting is not allowed. The third category allows for reliable source populationsand natural selection. Refuge zones also areintended both to dampen effects of management mistakes that might negatively impactcougar populations in hunting zones and toserve as reference areas in cougar research ormonitoring programs.

The source-sink approach proposed byLaundre and Clark (2003) is similar to zonemanagement, except that it includes only 2zone categories: source areas (closed to hunting) and sink areas (open to hunting). Bothapproaches are designed from a statewide ormultistate perspective, which is important because the species occurs at low densities andexhibits extensive dispersal capabilities (Stoner et al. 2008). Thus, the scale ofmanagementshould approximate the scale ofanimal movement patterns. Additionally, both strategiesrequire the explicit consideration of several

important variables, including size, spatialorientation, habitat quality of sources (e.g.,refuges), or sinks (e.g., control zones), anddistances between zones and their connectivity. Research indicates that some hunted populations also can act as source areas by producing a net emigration rate. An annual net emigration rate of 12 to 15% was documented ina lightly hunted Washington population (11%hunting mortality rate; Cooley et al. 2009b).Observations also indicate that a hunted areain Utah (Monroe Mountain) may serve as asource population (Wolfe and Stoner 2008).During a 5-year period in which <20% of thepopulation was removed annually throughharvest (female cougars comprised only 17%of harvest), growth rates and reproductiverates increased. Given these examples, it follows that areas with harvest rates <11%, suchas dryland areas where hound hunting is lesseffective, may be serving as de facto refugia.

The duration of a given regimen also isan important consideration in management.Under a rest-rotation schedule, it is importantto determine a priori how long each population would be managed as either a source or asink and what metrics would be used to assessthe effectiveness of a particular strategy. Forexample, because cougar birth schedules tendto operate on approximate 2-year intervals, itis important to consider behavioral characteristics of the population, such as age of firstbirth, extent of maternal care, and dispersalpatterns of males and females. All of thesevariables influence how long a given management regimen should be scheduled and ensure that putative sources are indeed acting assuch.

A rest-rotation strategy implies that agiven unit can be managed as either a sourceor a sink. As such, a manager can select a unitand manage it for one of 2 qualities: maxi-


I

l ..I11111111111111........IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII ................. _

mum harvest or as a source of immigrants.One of the advantages of this design is that arest-rotation form of management can serveas a structure for adaptive: multi-species management. For example, parallel objectivessuch as monitoring mule deer survival canbe assessed simultaneously. Well-conceived,relevant, and unbiased response variables arean important component ofmonitoring the efficacy of this system. Another considerationis that the rest phase is critical for units slatedto shift from sink to source. Recovery timeswill vary widely with habitat productivity andconnectivity. However, 1 case example fromsouthern Utah suggests the need for a recovery period of 5 to 6 years under a light harvest (1 permit/280 km2 of habitat; Wolfe andStoner, Utah State University, unpublisheddata). Lightly-hunted areas can serve as restunits, provided that they produce a net emigration rate. Maintaining high female survivalwith quotas may improve the effectivenessof a rest phase because females rely heavilyon progeny recruitment to replace harvestedanimals rather than relying on immigration,which is more typical ofmale cougars (Loganand Sweanor 2001).

The static source-sink model is predicated on the same demographic mechanismsas the rest-rotation source-sink, but the differences lie in the delineation of respectivesources and sinks. Each source is based on alegally defined refuge area, such as a nationalpark, military installation, or some other areaofsuitable habitat that does not allow hunting.One of the problems with these types of refuge areas is that they mayor may not functionas a source population. Effectiveness of a permanently closed area depends on the amount,shape, and quality of habitat in the refuge, aswell as the distance between and connectivity to potential sinks. For example, Zion Na-

tional Park in southern Utah has most of thecharacteristics of a functional refuge. That is,it is large (>500 km"), the shape approximatesa circle, thus, minimizing the perimeter-arearatio, and it encompasses both summer andwinter ranges of ungulates. A source area ofthis size is probably large enough to addressdispersal movements within source and sinks,but not large enough to provide effective areas in which ecological processes functionnaturally. Deterministic sinks exist, as well. Inmost jurisdictions, habitat quality varies, and,thus, areas cannot be treated equally. Somepatches may act as sinks because of relatively low prey densities or high hunter successrates (see Delibes et al. [2001] for a discussion of management-induced sinks). In thesesituations, a manager may want to incorporatethese de facto sources and sinks into a management plan. However, these are essentiallypermanent characteristics of particular localities and are generally not amenable to manipulation.

Washington, Wyoming, Idaho, Colorado,and New Mexico have recently adopted someform of zone management. New Mexico andWashington established zones based on acombination of biogeographical and politicalboundaries. Idaho and Colorado manage areas of de facto refugia (e.g., roadless areas,wilderness) under low to moderate harvestrates. Utah is currently investigating the roleof landscape heterogeneity in cougar vulnerability to harvest with the objective of identifying de facto refugia and population sinks. Additionally, Wyoming is planning to update itssource-sink adaptive management strategy byevaluating (1) density of human-caused mortalities, (2) sex and age composition ofcougarharvest focusing on relative proportion of'harvest of adult females, and (3) relative age ofharvested adult females.

Table 3. Cougar hunting stategies, season lengths, and percentage ofthe harvest that iscomposed of females, by state for the 2008-2009 season.

State Harvest strategy Season length Hound hunt Boot hunt % female

Ariz. General Sept I-May 31 Yes Yes 38

Colo. Quota (T) Nov 15-Mar 31 Yes Yes 39

Ida. General Aug 30-Mar 31 Yes Yes 46

Quota (T) Aug 30-Mar 31 Yes Yes 65

Mont. Limited entry Oct 26-April 14 Yes Yes

Quota (T) Oct 26-April 14 Yes Yes

Nev. Quota (T) Year-round Yes Yes 45

N.M. General Oct I-Mar 31 Yes Yes 25

Quota (T) Oct I-Mar 31 Yes Yes 20

N.Dak. Quota (T) Aug 29-Mar 31 Yes Yes 67

Oreg. Quota (T) Aug I-May 31 No Yes

S. Dale Quota (T) Jan I-Mar 31 No Yes 64

Tex. Open season Year-round

Ut. Limited entry Nov 19-June 1 Yes Yes 37

Quota (T) Year-round Yes Yes 40

SplitLE and Nov 19-Feb 08 Yes YesQuota (T)

Wash. General Aug I-Mar 31 No Yes

Quota (T + F) Dec I-Mar 31 Yes Yes

Wyo. Quota (T) Sept I-Mar 31 Yes Yes 44T = Total; F = Female; LE = Limited entry


Harvest StrategiesGeneral Seasons

General harvest seasons allow an unlimited number of cougars of either sex to be removed from the population during a specifichunting season. No restriction is placed on thenumber of hunters allowed, and primary control over harvest is the time and length of theseason. General seasons often are prescribedin areas where difficult access, low harvest,and difficult habitat conditions would reduceimpacts of hunting pressure. General seasonsprovide the greatest hunting opportunities tothe largest number of hunters. However, theygive managers the least control over harvest

level, composition, and distribution. For example, vagaries of weather that may bring either adequate snow for tracking or deep snowthat prevents access could result in annualvariation in harvest even if season timing andlength remain constant. Additionally, huntingpressure and harvests might not be distributed evenly over the hunt area. Rather, areasthat are most easily accessible could be heavily hunted, and others hunted little, if at all.Furthermore, topography and land ownershippatterns could create small refuges within alarger hunt area that contribute young to support the segment of the population that is exposed to harvest (CMGWG 2005). Currently,Arizona is the only state that manages cougars


using general seasons exclusively (Table 3).Idaho, New Mexico, and Washington employgeneral seasons in addition to other harveststrategies. New Mexico employs a generalseason from October 1 to March 31, combined with a management objective strategy.Current management objectives allow up toa 20% harvest of the total estimated population in any cougar management zone, withno more than a 25% female harvest of the total. Seasons in any zone end with a harvest10% below the estimated total sustainablemortality for any zone, or when the femalesub-limit is reached, whichever comes first(R. Winslow, New Mexico Department ofGame and Fish, personal communication).

limited EntryLimited entry programs control the num

ber of hunters allowed in an area by limitingthe number of licenses sold. Licenses are soldon a first-come, first-served basis, until thesupply is exhausted, or through a lottery. Limited-entry programs can be used to reduce anddistribute hunters, thereby relieving pressurein areas where easy access increases the number of hunters (Murphy 1983). The strategyalso may be used where lower hunter densitiesare desirable, such as in areas used heavily bythe public for other activities. Hunts designedusing limited-entry programs ensure that harvest prescriptions that are set biologically arenever exceeded. They also improve the quality of the hunting experience by eliminating"hunter race" situations associated with anopen quota hunt.

If the primary reason for initiating alimited-entry season is biological, the methodrequires that the general relationship betweennumber oflicenses sold and number ofcougarsharvested be understood and used to arrive atthe number of licenses available to meet pop-

ulation objectives. This could be difficult todetermine because the number of cougars harvested could vary annually even when licensenumbers and limited-entry permits are heldconstant because weather patterns, methodsof take and hunter effectiveness, and hunterselectivity vary annually (CMGWG 2005).Montana and Utah currently use a limitedentry strategy, along with a quota system tomanage cougars (Table 3). Utah uses limitedentry to provide a quality hunting opportunity,a quota system to deal with urban issues andpredator management plans, and a split harvest strategy (both limited-entry and quotasystem) to provide up-front quality followedby more quota-driven results.

Quota Systems (Harvest Limit). Quota systems limit the number of cou

gars that may be harvested in a season or ina particular area, but do not limit the numberof hunters that can participate in the harvest.Harvest quotas are set either for the totalnumber of cougars, total number of females,or less frequently, total number of males. Under total-number quota seasons, hunting endswhen a specified number of cougars is killed,regardless of gender (and regardless of age insome states) or when the season ends. Underfemale sub-quotas, the season closes when thefemale quota has been reached or at the endof the legal season, which ever occurs first.In some cases, the season may remain openfor male-only harvest after the female quotahas been reached. Colorado uses a total quotasystem, but manages female composition inthe harvest by monitoring female mortalityand reducing overall quotas when necessaryuntil the female population returns to an acceptable level. This strategy, along with activecommunication with hunters, on-line reportsofquota status and percentage of female com-


position, and a mandatory cougar hunter education program has shown initial success forachieving management objectives for femaleand total harvest. Quota systems require thatagencies develop a system whereby hunterscan be notified when the quota is met. It isnot uncommon for the quota to be exceededbecause hunters harvest animals during thegrace period. Total quota systems can producea less selective harvest because hunters oftenare eager to harvest at their first opportunitybecause the season could close before anotheropportunity presents itself. However, femaleand age subquotas can effectively accomplishsex and age composition management goals.The quota system currently is the most popular cougar harvest strategy used by state agencies (Table 3).

Variables AffectingHunter Success

Increasing harvests do not necessarily represent increasing cougar populations;they may simply reflect the level of hunting effort or changes in a number of variables that affect hunter success. Favorablesnow conditions, use of trailing hounds,and high road densities improve the hunter's ability to locate cougars (Dawn 2002).

Snow ConditionsSnow improves harvest success dra

matically, particularly when coupled with theuse of hounds, which is why agencies oftenschedule cougar hunting seasons during winter months when snow conditions improvetracking (Lindzey 1987). Hunting seasons in8 of 10 states begin in the fall or early winterand extend through the winter or early spring.For general seasons, frequent snowfall maytranslate into higher harvests. Conversely,hunter success may be lower in areas that re-

ceive little or no snowfall; accordingly, hunting regulations and seasons can be less restrictive. Arizona and Nevada, 2 of the 3 westernstates receiving the lowest average snowfallemploy year-round seasons, presumably tocompensate for inferior tracking conditions.

Hound HuntingHounds are widely considered the most

effective tool for hunting cougars (CMGWG2005). Additionally, hound hunters can bemore selective ofspecific sex and age classes,increasing the potential of achieving harvestobjectives. In Wyoming, hound hunters whoreported being selective took 17% fewer females than hunters who were not selective (C.Anderson, Colorado Division ofWildlife, unpublished data). Females comprised 43% ofthe hound hunt harvest and 53% of the bootharvest (i.e., harvest without the use ofhoundsor traps) in Idaho (S. Nadeau, Idaho DivisionofFish and Game, unpublished data). Arizonahound hunters were significantly more selective than boot hunters, with hound huntersharvesting more males (~60%) than females,and boot hunters harvesting more females thanmales (>60%; Zornes et al. 2006). In 1996,hound hunting became illegal in Washington.In anticipation oflower hunter success, Washington increased the season length and baglimit and reduced the price of a cougar tag (license), which resulted in higher harvests. Additionally, median age of harvested cougarsdeclined (from 4.7 to 2.7 years for females and5.8 to 2.9 years for males), and the percentageof females in the harvest increased from 43%with hounds to 58% without hounds (Martorello and Beausoleil 2003). Similar changesin hunter success and harvest selectivity occurred during 1994 in Oregon, which alsobanned the use of hounds (Whittaker 2005).


grown over the years. We have summarizedthese studies to help facilitate the incorporation of research into management (Figure 3;Tables 4 and 5). It is important to note thatthese studies essentially are snapshots intime and may not represent current conditions. Additionally, given the problems involved in enumeration, they should not beextrapolated to ecoregional or statewidescales; however, they may help managersdevelop population strategies and objectivesand assess impacts of harvest and control.

Managing for MetapopulationsCougar populations interact through a

source-sink metapopulation structure. Therefore, management may be most effectivewhen conducted at a regional level. Metapopulation management requires a demographicand spatial assessment of habitat connectivity between sources and sinks, and the reliance on functional source populations, mostlymanaged through the establishment ofrefugia(Murphy 1983, Spreadbury et al. 1996, Loganand Sweanor 2001, Stoner et al. 2006, Robinson et al. 2008, Cooley et al. 2009b).

In states where there are large contiguous tracts of cougar habitat, heavy harvestconducted at small «1000 knr') scales maybe ineffective in short-term population suppression because hunted animals are quicklyreplaced by immigrants from neighboringareas (Robinson et al. 2008, Cooley et al.2009a). Population reductions may be moreeffectively achieved through prolonged heavyhunting at larger scales.

Heavy harvest maintained over a prolonged period may reduce local densities aswell as densities in adjacent areas becauseof a lack of available dispersers that couldmove among populations as potential recruits(Logan et al. 1996, Spreadbury et al. 1996,

Impacts of Hunting:Case Studies

The body of scientific information onhow hunting affects cougar populations has

Nonconventional Hunting MethodsAlthough houndhunting is the mostwide

ly used method to hunt cougars, boot huntingand predator calling also have been used successfully. All states allow boot hunting ofcougars. South Dakota, Oregon, and until recently,Washington, have relied exclusively on boothunting to manage cougar numbers (Table 3).State agencies frequently open boot seasonsearlier than hound seasons so that they coincide with big game seasons. Big game huntersoften have the option of buying a big gamepackage that includes a cougar tag. In NorthDakota, most cougars are killed by boot hunters while they are hunting big game. Becausethe hound season opens later (December 1versus August 29), most quotas have beenfilled by boot hunters before the hound seasonopens (D. Fecske, North Dakota Game andFish Department, personal communication).

Reed DensitiesPosition and density ofroads can strong

ly influence hunter success and harvest distribution. Hunters often cluster in areas withhigh road densities, resulting in increased harvest rates in these areas. Hemker et al. (1984)also found that high road densities exposedfemales with young to harvest. Alternatively,low road densities may reduce hunting pressure and harvest for some cougar populations.Harvest may be low enough that these areasfunction as sources and act as de facto refugiafor cougar populations. These refuges mayhelp to maintain harvest levels if progeny arerecruited into adjacent hunted areas (Anderson and Lindzey 2005).

124 Managing Cougars in North America 1

Figure 3. Map of studies investigating the effectsof hunting on cougar populations in western NorthAmerica.

and greater philopatry. Sensitivity analysesshow that cougar populations are most sensitive to survival of adult females. Reducedharvest pressure on adult females will allowfor population resiliency (Lindzey 1992, Rossand Jalkotzy 1992, Logan and Sweanor 2001,Martorello and Beausoleil 2003, Andersonand Lindzey 2005, Stoner et al. 2006, Robinson et al. 2008, Cooley et al. 2009a).

Percentage of adult females in cougarharvests may provide an index to populationstatus and harvest impact (Lindzey et al. 1992,Anderson and Lindzey 2005). However, harvest composition should be monitored overa number of years before inferences aboutpopulation trends are drawn (Anderson andLindzey 2005).

Prolonged (2:4 years), heavy harvest (2:40% ofadultpopulation) canpromote ayoung-

Study, researchers

MFWP, unpublished data 2009

Cooley et al. 2009a

Robinson et al. 2008

Stoner et al. 2006

Lambert et al. 2006

Anderson and Lindzey 2005

Logan and Sweanor 2001

Cunningham et al. 2001

Lindzey et al. 1992

Ross and Jalkotzy 1992

Logan et al. 1986

Murphy 1983

IDno.

1

2

3

4

5

6

7

8

9

10

11

12

Table 4. Key for studies investigating the effects ofhunting on cougar populations (Figure3).

Harvest Impacts on Sex and AgeStructure

Harvest of adult females impacts populations more severely than harvest of bothmales and subadult females because adult females provide recruits and are more difficultto replace due to shorter dispersal distances

Cunningham et al. 2001, Logan and Sweanor2001, Robinson et al. 2008).

Resiliency of cougar populations relieson immigration from neighboring populations and recruitment of philopatric femaleoffspring (Lindzey et al. 1992, Sweanor et al.2000, Logan and Sweanor 2001, Robinson etal. 2008, Cooley et al. 2009b)

Hunting mortalities are not compensatedby a reduction in natural deaths or increasesin vital rates (reproduction, kitten survival, female growth). Instead, hunting is compensatedby immigration, suggesting that assumptionsbased onclosedpopulations arenot appropriate(Lindzey et al. 1992, Ross and Jalkotzy 1992,Logan and Sweanor2001, Cooley et al.2009a).

Table5. Summary of studies investigating the effects of hunting on cougar popualtions in western North America.

ID Duration Area size Design Prey species # Cougars Harvest strategy Hunting mortality Density method

1 1998-2006 915 km 2 Treatment Elk, white- 117 "Heavy harvest 1998- T92% oftotal mortality No. of cougars/l 00 km 2 withintailed deer, 2000 (>24 month) study areamule deer,moose

CReduced harvest2001-2006

Control C33% of total mortality (24month)

2 2002-2007 735lan2 Treatment TWhite-tailed T57 "Heavy harvest "Mortality rate due to hunt- Proportion of cougars/l 00 km 2

deer, mule ing 0.24" within 95% F composite homedeer, elk range ()

2002-2007 594 km 2 Control cMule deer, c46 CLight harvest "Mortality rate due to hunt-::>Q

elk ing=O.l1" -0....CD

3 2001-2006 735 km 2 Treatment White-tailed 34 Heavy harvest Mortality rate due to hunt- No. of cougarsll 00 km?within...,01

deer, mule ing = 0.24" study area ..

"deer, elk, 0-0

moose cQ....0'

4 1997- 2004 1,300 Treatment "Mule deer, T60 "Heavy harvest 1996- T60% of total mortality No. of cougarsll 00 km- within ~

km 2 elk 2001 study area ~Q

"Reduced harvest, T17.6-54.5 % of adult~

Q

2002-2004 population c.nCD

\:;3 1997- 2004 950 km 2 Control cMule deer, c50 CNohunting C24% of total mortality3

<::l-' CD(i) elk

~

tr('")

()Cl1998-2003 32,800 Treatment White-tailed 52 Heavy harvest Mortality rate due to hunt- No. of cougars/l00 km 2 within~ 5 0

S· km 2 deer, mule ing = 0.38" 95% female kernel home range c~

c.n(I:l deer, elk, (minimum density) QI:l.. ...,Cl moose, I~ c~ caribou ~

~....... 5'

~ c.n

~ "Treatment; cControl; a Mortality rate calculated as a function of radio days; Harvestable animals> 1 yr old.

t-..>01

Table5 (continued). Summary of studies investigating the effects of hunting on cougar popualtions in western North America.l-V0-

lD Duration Area size Design Prey species # Cougars Harvest strategy Hunting mortality Density method I~Q::l

6 1997- 2003 2760 km2 Treatment 61 T2yr heavy harvest, 3 yr 50% reduction of total No. of cougars/1 00 km2 within Qea

light harvest population in yr 1 and 2 suitable cougar habitat 5'ea

1997-2003 2960 km2 Control CStaticharvest ()

7 1985- 1995 2059 km2 Treatment Mule deer, 107 adults "Experimental reduction T47% reduction oftotal Proportion of cougar 0c:

bighorn population locations/luukm? within study eaQ

sheep, area ;;j

pronghorn, 5'oryx Z

0174 kittens TNo hunting Tl3 cougars removed in 6.5

.....-:.-months »

Control CNohunting 3CD.....

8 1991-1994 4035 km2 Treatment 24 "Depredation control Mortality rate due to sport o'TSport hunting = 0.06a Q

Mortality rate due to

Control CNohunting depredation = 0.20a

9 1979-1981 4500 km", Treatment Mule deer, elk Experimental reduction 27% reduction ofharvestable

No hunting 6 cougars removed in 6 mo.

10 1981-1989 780 knr' Treatment Bighorn 71 Moderate to heavy 63% oftotal mortality No. of cougarsllOO km- within

\:;3 sheep, elk, harvest in east and north, 21% reduction of study area0- moose, mule minimal in southwest harvestable~

deer, white- 11% reduction oftotal pop.v,(J tailed deera;:s

925 km2 Mule 46 1 yr moderate harvest, 1 19% reduction of total No. of cougars/Iuukm" whoseg. 11 1981-1983;;:: deer, elk, yr no harvest population first study yr home ranges largely overlappedtil~ pronghorn the study area during wintera;:s

12 1979-1982 741lan2 White-tailed 8 Moderate harvest 22% of available F No. of cougarsll 00 km2 within;:s

9. winter deer, mule 48% of available M study area'"\j deer, elk~ TTreatment; CControl; a Mortality rate calculated as a function of radio days; Harvestable = animals> 1 yr. old; M = male; F = female.~

Table 5 (continued). Summary of studies investigating the effects ofhunting on cougar popualtions in western North America.

ID Cougar density Population Survival rates Reproduction Age structure (yr) Compensatory Recovery?growth versus additive

"Iotal = 3.06 TAS= 0.73 ± 0.01 TAdult F = 0.60 "Matemity = 1.4 T>12 month(range = 4.04- = 2.86 (M = 2.37, F =2.18) 3.19)

CTotal= 3.09 cAS = 1.08 ± 0.11 CAdult F = 0.97 CMaternity=1.7 C>12 month = 3.55(M =(range = 2.30- 2.83, F 3.87)3.61)

2 "Adult = 1.6 (M = TAS= 0.78 ± 0.19 TX = 0.56 "Kitten survival = "Adult = 4.0 (M = 3.5, Additive mortality: TM recovery0.2, F = 1.5) 0.32 F = 4.3)

"Total = 3.5 (M = TAa = 0.91 (M = 0.45, F = 0.64) "Litter size = 2.63 TTotal 2.3 (M = 2.0, F male density static TMimmigration0.6, F = 2.8) =2.8)

Total and F de- (M= 1.02, F = "Matemity = 1.15 No increase in ():::>

cline, M static 0.86) maternity 0

cAdult cAS = 1.10 ± 0.12 cAdult = 5.1 (M-0

1.9 (M = Cx = 0.71 CKitten surval = 4.9, No increase in kit- CEmigration -+C1>

0.8, F = 1.1) 0.58 F = 5.7) ten survival ""''Total = 3.6 (M = cAa = 0.98 CTotal = 3.2 (M = 3.3,

Ot(M = 0.60, F = 0.83) CLitter size = 2.47 No reduction in ,.

1.3, F = 2.3) F = 3.3) natural mortality -00

All static (M 0.96, F = CMaternity = 1.12 Decline in female-0c

0.97) growth 0-+

3 Adult = 2.32 - AS = 0.84 ± 0.21 X = 0.76 Maternity = 1.2 Total = 2.2 yr, decline Mrecovery 0'::::l

5.03~

Total = 5.03 Aa= 1.00 (M = 0.47, F = 0.73) Adult M = 3.4 yr M immigration 0::::l

All static Adult F = 3.8 yr 0(Q

"Subadult and "Decline > 60% "Litter size = 1.7 "Adult = 3.4 yrC1>

1:;34 TX = 0.64 Aditive mortality: T52% recovery after 3 3

adult 1996-2001 yr light hunting C1>~ ::::l~ =3.2 trv,

"Subadult and "High = 1.0, Low = 0.36 "Litters/Fzyr = 0.24 Lack of starvation ()

'" 0C) adult 2002-2004 c~ =2.0 (Q

Si' 0l:; "Decline, then TIl kittens TM immigration ""'(I:l Static prey II:l... static orphaned populations cC) ::::l;:;:

cSubadult and CStatic CLitter size = 1.9 cAdult = 4.6 yr No increase in CEmigration-+

;:;: Cx = 0.76 3'~ adult 1996-2004 reproduction (Q..... =2.8'1;;j

~ CStatic CHigh = 0.63, Low = CLitters/F/yr = 0.340.91

"Treatment, CControl; a Mortality rate, calculated as a function of radio days; Harvestable = animals> 1 yr old; M = male; F female. I~

Table 5 (continued). Summary of studies investigating the effects ofhunting on cougar popualtions in western North America.'I\,,)

coID Cougar density Population growth Survival rates Reproduction Age structure (yr.) Compensatory Recovery? Is:versus additive

Q

AS = 0.80 ± 0.11::l

5 Adult = 0.46 X =0.59 Litter size = 2.53 Q(Q

Total cougars = Adult F = 0.77 Maternity = 1.26 5'(Q

1.09, decline ()

Decline AdultM 0.33 5 kittens orphaned 0c:

"Total cougars = 4 F with young T,CF Decline from 6-8 yr Recovery within 2 yr(Q

6 Q3.42 and 2.35 harvested to 3-4 yr following first ;A

treatment yr 5'Immigration and re- Z

0cruitment .,

3=-7 "Adults pre- TAa = 1.23 (pre- T,cAdultM = 0.91, F = T,CLitter size = 3.0 "Adult> 5.6 Additive mortality: Recovery after 31 »

treatment treatment) 0.82 month :31.36-2.01 Cl>.,"Adult post- TAa = 1.32 (post- T,cSA M = 0.56, F = 0.88 "Maternity = 1.6, (M = 6.1, F = 5.3) human mortalities o'

Qtrreatment = treatment) 2.0 suppressed adult1.09-1.87 density

"Kitten 0.63 cAdult = 4.6

CAdult = 1.13- cAa= 1.12 CKitten = 0.59 CMaternity = 1.4, (M = 5.2, F = 4.3)1.79 1.3

8 TX annual 0.44 "Captured = 3.2 Immigration

(M = 2.7, F = 4.1)

Cx annual = 0.53 cDepredations = 4.5

~ (M = 4.5, F = 3.9)<::l-'(i) 9 Additive mortality: Recovery after 1 yrv,

" no reduction in Male immigrationCl;::: natural mortalityS-

10 Adult = 1.7 Overall increase High kitten and juvenile Litter size = 2.2 Immigration~(I:l (1984),2.1 survival throughi::l...Cl (1989) independence (> 97%);:::;::: Total = 2.7-3.3 Females: Nondispersalof$'i (1984),4.2 increase independent females....

'\::$ (1989)ci%

51> Males: static

TTreatment; CControl; a Mortality rate, calculated as a function of radio days; Harvestable = animals>1 yr old; M = male; F female.

Table 5 (continued). Summary of studies investigating the effects of hunting on cougar popualtions in western North America.

"Treatment; C Control; "Mortality rate, calculated as a function of radio days; Harvestable = animals>1 yr old.

()oc

(Qo..,

10-0

:cc::l-+

:5"(Q

01

d''"0co-+

c5"::l

~o::lc

(QCD3CD::ltr

():::rc

'"0-+CD..,

Immigrationcompensated formortalities

Recovery?Compensatoryversus additive

Age structure (yr)ReproductionSurvival ratesPopulationgrowth

11 Adult = 1.6 (pre- Litter size = 2.9 Young structuretreatment), 1.4(post-treatment)

Juvenile andadult 3.5 (pre-treatment), 3.0(post-treatment)

12 Adult = 4.3 Static Litter size = 2.6(winter)

Total = 7.1(winter)

ID Cougar density


er age structure and social instability in cougarpopulations (Lambert et al. 2006, Stoner et al.2006,Robinsonetal. 2008,Cooleyetal. 2009a).

Indirect Impacts of Harvest onSurvival

Frequent removal of males and the consequent immigration of transient males maydecrease survival of males because of direct and exploitative competition for matesand territory (Lo,gan et al. 1986, Logan andSweanor 2001).

Frequent removal of adult males mayresult in increased kitten mortality throughinfanticide by new immigrant males (Rossand Ja1kotzy 1992, Logan and Sweanor 2001,Cooley et al. 2009a).

Hunters can inadvertently orphan dependent kittens. To reduce harvest pressure on females, management agenciesmay consider hunter education programsdirected at identification of track characteristics of different sex and age classes ofcougars in conjunction with a quota systemthat regionally limits female harvest (Rossand Jalkotzy 1992, Spreadbury et al. 1996).

Management Based on ComplaintsIncreased cougar-human conflicts may

not correspond with increasing cougar populations; therefore, hunting regulations shouldbe based on demographic trends obtainedthrough reliable census, survey, or index techniques, rather than on the number of complaints received (Lambert et al. 2006).

Literature CitedAune, K. E. 1991. Increasing mountain lion popu

lations and human-mountain lion interactionsin Montana. Pages 86-94 in C. E. Braun, editor. Mountain lion-human interactions: symposium and workshop. Colorado Division ofWildlife, Denver, Colorado, USA.

Anderson, C. R., Jr., and F. G. Lindzey. 2003. Estimating cougar predation rates from GPS location clusters. Journal of Wildlife Management67:307-316.

Anderson, C. R, Jr., and F. G. Lindzey. 2005. Experimental evaluation of population trend andharvest composition in a Wyoming cougar population. Wildlife Society Bulletin 33: 179-188.

Beausoleil, R A., D. Dawn, D. A. Martorello, andC. P. Morgan. 2008. Cougar management protocols: a survey of wildlife agencies in NorthAmerica. Pages 205-231 in D. E. Toweill, S.Nadeau, and D. Smith, editors. Proceedings ofthe ninth mountain lion workshop. Idaho Gameand Fish Department, Boise, Idaho, USA.

Bowyer, R T., D. K. Person, and B. M. Pierce.2005. Detecting top-down versus bottom-upregulation of ungulates by large carnivores:implications for conservation of biodiversity.Pages 342-361 in 1. C. Ray, K. H. Redford, RS. Steneck, and J. Berger, editors. Large carnivores and the conservation of biodiversity.Island Press, Washington, D.C., USA.

Choate, D. M., M. L. Wolfe, and D. C. Stoner.2006. An evaluation of the accuracy and efficacy of cougar population estimators. WildlifeSociety Bulletin 34:782-799.

Colorado Division of Wildlife. 2009. Coloradosheep and goat hunting regulations. ColoradoDepartment of Natural Resources, Denver,Colorado, USA.

Cooley, H. S., R. B. Wielgus, G. M. Koehler, H.S. Robinson, and B. T. Maletzke. 2009a. Doeshunting regulate cougar populations? A test ofthe compensatory mortality hypothesis. Ecology 90:2913-2921.

Cooley, H. S., R B. Wiegus, G. M. Koehler, andB. T. Maletzke. 2009b. Source populations incarnivore management: cougar demographyand emigration in a lightly hunted population.Animal Conservation 12:321-328.

Cooley, H. S., R B. Wielgus, H. S. Robinson, and


Hornocker, M. G. 1970. An analysis of mountainlion predation upon mule deer and elk in theIdaho Primitive Area. Wildlife Monographs21:3-39.

Hornocker, M. G. 2007. Conserving the cats,cougar as a model: a review. Pages 111-122in T. E. Fullbright and D. G. Hewitt, editors.Wildlife science: linking ecological theory andmanagement applications. CRC, Boca Raton,Florida, USA.

Kamler, l F., R. M. Lee, l C. deVos Jr., W. B.Ballard, and H. A. Whitlaw. 2002. Survival andcougar predation of translocated bighorn sheepin Arizona. Journal of Wildlife Management66:1267-1272.

Katnik, D. D. 2002. Predation and habitat ecology of mountain lions (Puma concolor) inthe southern Selkirk Mountains. Dissertation,Washington State University, Pullman, Washington, USA.

Lambert, C. M., R. B. Wielgus, H. S. Robinson,H. S. Cruickshank, R. Clarke, and J. Almack.2006. Cougar population dynamics and viability in the Pacific Northwest. Journal ofWildlifeManagement 70:246-254.

Laundre, J. W. 2008. Summer predation rates onungulate prey by a large keystone predator:how many ungulates does a large predator kill?Journal of Zoology, London 275: 341-348.

Laundre, L, and T. W. Clark. 2003. Managingpuma hunting in the western United States:through a metapopulation approach. AnimalConservation 6:159-170.

Leopold, A. 1933. Game management. Scribner's, New York, New York, USA.

Levins, R. 1969. Some demographic and geneticconsequences of environmental heterogeneityfor biological control. Bulletin of the Entomological Society ofAmerica 15:237-240.

Lindzey, F. G. 1987. Mountain lion. Pages 656668 in M. Novak, J. A. Baker, M. E. Obbard,and B. Mallock, editors. Wild furbearer man-

C. S. Lambert. 2008. Cougar prey selection ina white-tailed deer and mule deer community.Journal of Wildlife Management 72:99-106.

Cougar Management Guidelines Working Group.2005. Cougar management guidelines. WildFutures, Bainbridge Island, Washington, USA.

Cunningham, S. C., W.B. Ballard, and H. A. Whitlaw. 2001. Age structure, survival, and mortality of mountain lions in southeastern Arizona.Southwestern Naturalist 46:76-80.

Dawn, D. 2002. Management of cougars (Pumaconcolor) in the western United States. Thesis,San Jose State University, San Jose, California,USA.

Delibes, M., P. Gaona, and P. Ferreras. 2001. Effects of an attractive sink leading into maladaptive habitat selection. American Naturalist158:277-285.

Ernest, H. B., W. M. Boyce, V. C. Bleich, B. May,S. J. Stiver, and S. G. Torres. 2003. Geneticstructure of mountain lion (Puma concolor)populations in California. Conservation Genetics 4:353-366.

Halfpenny, J. C., M. R. Sanders, and K. A.McGrath. 1991. Human-cougar interactionsin Boulder County, Colorado: past, present,and future. Pages 86-94 in C. S. Braun, editor.Mountain lion-human interaction: symposiumand workshop. Colorado Division of Wildlife,Denver, Colorado, USA.

Hayes, C. L., E. S. Rubin, M. C. Jorgensen, R. A.Botta, and W. M. Boyce. 2000. Mountain lionpredation of bighorn sheep in the PeninsularRanges, California. Journal of Wildlife Management 64:954-959.

Hemker, T. P., F. G. Lindzey, and B. B. Ackerman.1984. Population characteristics and movementpatterns ofcougars in southern Utah. Journal ofWildlife Management 48:1275-1284.

Hoban, P.A. 1990. A review of desert sheep in theSan Andreas Mountains, New Mexico. DesertBighorn Council Transactions 34:14-22.


agement and conservation in North America.Ontario Ministry of Natural Resources, Toron

to, Canada.Lindzey, F. G., W. D. VanSickle, S. P. Laing, and

C. S. Mecham. 1992. Cougar population response to manipulation in southern Utah. Wild

life Society Bulletin 20:222-227.Linnell, J. D., J. Odden, M. E. Smith, R. Aanes,

and J. E. Swenson. 1999. Large carnivores thatkill livestock: do "problem individuals" really

exist? Wildlife Society Bulletin 27:698-705.Logan, K. A., and L. L. Sweanor. 1998. Cougar

management in the West: New Mexico as a template. Proceedings of the Western Association

ofFish and Wildlife Agencies 78:101-111.Logan, K. A., and L. L. Sweanor. 2000. Puma.

Pages 347-377 in S. Demarias and P. R. Krausman, editors. Ecology and management oflargemammals in North America. Prentice Hall, Up

per Saddle River, New Jersey, USA.Logan, K. A., and L. L. Sweanor. 2001. Desert

puma: evolutionary ecology and conservationof an enduring carnivore. Island Press, Wash

ington, D.C., USA.Logan, K. A., L. L. Irwin, and R. Skinner. 1986.

Characteristics of a hunted mountain lion population in Wyoming. Journal of Wildlife Man

agement 50:648-654.Logan, K. A., L. L. Sweanor, T. K. Ruth, and M.

G. Hornocker. 1996. Cougars ofthe SanAndresMountains, New Mexico. New Mexico Department of Game and Fish, Final Report, FederalAid in Wildlife Restoration, Project W-128-R,

Santa Fe, New Mexico, USA.Martorello, D. A., and R. A. Beausoleil. 2003.

Characteristics of cougar harvest with andwithout the use ofdogs. Pages 129-135 in S. A.Becker, D. D. Bjornlie, F. G. Lindzey, and D.S. Moody, editors. Proceedings of the seventhmountain lion workshop. Wyoming Game andFish Department. Lander, Wyoming, USA.

McKinney, T, T W. Smith, J. C. deVos Jr. 2006.

Evaluation of factors potentially influencinga desert bighorn sheep population. WildlifeMonographs164:1-36.

Murphy, K. M. 1983. Characteristics of a huntedpopulation of mountain lions in Western Mon

tana. Federal Aid to Wildlife Restoration, Montana Game and Fish Department, Final Report.Project W-120-R-13 and 14. Helena, Montana,USA.

Murphy, K. M. 1998. Ecology ofthe cougar (Pumaconcolor) in the northern Yellowstone ecosystem: interactions with prey, bears, and humans.Dissertation, University of Idaho, Moscow,Idaho, USA.

Nowak, M. C. 1976. The cougar in the UnitedStates and Canada. USDI, Fish and Wildlife

Service and New York Zoological Society,New York, New York, USA.

ockenfels, R. A. 1994. Factors affecting adult

pronghorn mortality rates in central Arizona.Wildlife Digest 16, Arizona Game and FishDepartment, Phoenix, Arizona, USA.

Oregon Department of Fish and Wildlife. 2006.Oregon Cougar Management Plan. Salem, Oregon, USA.

Pierce, B. M., and V. C. Bleich. 2003. Mountainlion. Pages 744-757 in G. A. Feldhamer, B. C.Thompson, and J. A. Chapman, editors. Wild

mammals of North America. Johns HopkinsUniversity Press, Baltimore, Maryland, USA.

Robinson,H. S.,R. B. Wielgus, andJ. C. Gwilliam.2002. Cougar predation and population growthof sympatric mule deer and white-tailed deer.Canadian Journal ofZoology 80:556-568.

Robinson, H. S., R. B. Wielgus, H. S. Cooley, andS. W. Cooley. 2008. Sink populations in car

nivore management: cougar demography andimmigration in a hunted population. EcologicalApplications 18:1028-1037.

Rominger, E. M., H. A. Whitlaw, D. K. Weybright,W. C. Dunn, and W. B. Ballard. 2004. Influenceof mountain lion predation on bighorn sheep


Sweanor, L. L., K. A. Logan, and M. G. Hornocker. 2000. Cougar dispersal patterns, metapopulation dynamics, and conservation. Conservation Biology 14:798-808.

Torres, S. G., T. M. Mansfield, J. E. Foley, T. Lupo,

andA. Brinkhaus. 1996. Mountain lion and human activity in California. May-June:22-23.

Turner, J. W., Jr., and M. L. Morrison. 2001. Influence of predation by mountain lions on numbers and survivorship of a feral horse population. Southwestern Naturalist 46:183-190.

Turner, J. W., M. L. Wolfe, and J. F. Kirkpatrick.1992. Seasonal mountain lion predation ona feral horse population. Canadian Journal ofZoology 70:929-934.

U.S. Department ofAgriculture. 1906. 1905 year

book of the United States. U.S. GovernmentPrinting Office, Washington, D.C., USA.

Wehausen, J. D. 1996. Effects of mountain lion

predation on bighorn sheep in the Sierra Nevada and Granite Mountains, California. Wildlife Society Bulletin 24:471-479.

Whittaker, D. G. 2005. Oregon mountain lionstatus report. Pages 11-16 in R. A. Beasoleiland D. A. Martorello, editors. Proceedings of

the eighth mountain lion workshop. Olympia,Washington, USA.

Young, S. P., and E. A. Goldman. 1946. The puma:

mysterious American cat. American WildlifeInstitute, Washington, D.C., USA.

Zornes, M. L., S. P. Barber, and B. F. Wakeling.

2006. Harvest methods and hunter selectivity ofmountain lions in Arizona. Pages 85-89 in J.W.Cain III and P.R. Krausman, editors. Managingwildlife in the Southwest. Southwest Section ofthe Wildlife Society, Tucson, Arizona, USA.

translocations. Journal of Wildlife Management 68:993-999.

Ross,1. P., and M. G. Jalkotzy. 1992. Characteristics of a hunted population of cougars in southwestern Alberta. Journal of Wildlife Management 56:417-426.

Ross, P. 1., and M. G. Jalkotzy. 1996. Cougar pre

dation on moose in southwestern Alberta. A1ces 32:1-8.

Ross, P. 1.,M. G. Jalkotzy, and M. Festa-Bianchet,1997. Cougar predation on bighorn sheep insouthwestern Alberta during winter. CanadianJournal ofZoology 74:771-775.

Sawyer, H., and F. G. Lindzey. 2002. A review ofpredation on bighorn sheep (Ovis canadensis).Special Report, Wyoming Cooperative Fish

and Wildlife Research Unit, University ofWyoming, Laramie, Wyoming, USA.

Schaefer, R. L., S. G. Torres, and V. C. Bleich.

2000. Survivorship and cause-specific mortality in sympatric populations of mountainsheep and mule deer. California Fish and Game86:127-135.

Sinclair, A. R. E., J. M. Fryxell, and G. Caugh

ley. 2006. Wildlife ecology, conservation andmanagement. Wiley-Blackwell, Malden, Massachusetts, USA.

Spreadbury, B. R., R. K. Musil, J. Musil, C. Kaisner, and J. Kovak. 1996. Cougar populationcharacteristics in Southeastern British Columbia. Journal of Wildlife Management 60:962

969.Stoner, D. c., M. L. Wolfe, and D. M. Choate.

2006. Cougar exploitation levels in Utah: impli

cations for demographic structure, populationrecovery, and metapopulation dynamics. Journal ofWildlife Management 70:1588-1600.

Stoner, D. C., W. R. Rieth, M. L. Wolfe, M. B.

Mecham, and A. Neville. 2008. Long distancedispersal of a female cougar in a basin andrange landscape. Journal of Wildlife Management 72:933-939.

Mountain Lion Hunting Season RecommendationsMay 24,2013

Chadron, Nebraska

Background

Mountain lions (Puma conca/or) are native to Nebraska but were extirpated by the early1900s due to unregulated hunting, trapping, poisoning and decimation of prey species.Prey populations recovered throughout the 20th century due to protection offered bygame laws. Mountain lion populations recovered throughout the Mountain West overthis time period due to the elimination of bounties and management of mountain lions asa big game species. Mountain lion populations expanded in the nearby Black Hills ofSouth Dakota during the 1990s and 2000s. Juvenile male dispersers were documentedthroughout the Prairie Plains region during this time period. In 1991 mountain liontracks were found and a female was legally shot in the Pine Ridge region ofnorthwestern Nebraska marking the first documented presence in modern times.

In 1995 the Nebraska Legislature added mountain lions to the statutory list of gameanimals, thereby affording protection for mountain lions under the Game Law. TheNebraska Game and Parks Commission (hereafter, Commission) began formallyinvestigating observations of mountain lion presence after the original confirmation in1991. In 2004 the Commission adopted a Mountain Lion Response Plan that detailedthe agency's response to various situations regarding mountain lions and identifiescriteria for confirming presence of mountain lions in Nebraska. Mountain lion presencewas documented on 24 occasions between 1991 and the beginning of 2006. Duringthis timeframe, all mountain lions for which age and gender could be determined wereyoung males that fit the profile of dispersing animals. No evidence of resident animalsor females was documented until 2006, when a female ear-tagged in South Dakota wasphotographed in the Pine Ridge area of northwestern Nebraska. A female mountainlion with a litter of kittens was documented the following year in 2007, which providedthe first evidence of a resident population. Based on this evidence, mountain lionsappear to have recolonized the Pine Ridge during the mid-2000s. Kittens have beendocumented in the Pine Ridge every year since 2007. Female mountain lions have alsobeen documented in Scotts Bluff County in 2009, 2010, and 2013, and in Cherry Countyin 2012 and 2013. Dispersing mountain lions have been found throughout the state andare typically young males. The Commission has documented 92 instances of mountainlion presence outside of the Pine Ridge population since 1991.

1

Confirmed Mountain Lion Presence in Nebraska1991- Present

legend<) Female

• 1996·2006

o 2007 ·2011

• 2012·2013

PineRk:lge Population

-1st modem confirmation in Pine Ridge in 1991-1st kittens documented in Pine Ridge in 2007- The Pine Ridge is the only area with evidence of a resident population- 92 Confirmations outside of Pine Ridge population- Most outside of Panhandle appear to be dispersing males (not resident)..Individual confirmations are not mapped in Dawes, Sheridan,or Sioux counties where an established population exists

The number of instances of confirmed mountain lion presence in Nebraska hasincreased each year since 2003. Since confirmations can consist of tracks, DNAevidence, and/or photos, one animal can be responsible for multiple confirmedpresences.

Mountain Lion Research

Mountain lions have recently recolonized the Pine Ridge in northwest Nebraska andmay be recolonizing other areas, such as the Wildcat Hills and Niobrara River Valley.The Commission began conducting research in 2010 to assess the number of mountainlions in the Pine Ridge and to estimate suitable habitat throughout the state.

Population Estimates

Surveys utilizing scat detector dogs and genetic analyses were conducted in the PineRidge in 2010 and 2012. During the 2010 survey, eight males and five females weredetected, and during the 2012 survey six males and eight females were detected. The

2

program CAPWIRE was used to estimate the size of the Pine Ridge mountain lionpopulation, providing maximum likelihood population estimates of 19 and 22,respectively. This model was specifically developed for estimating small populations ofelusive species utilizing information gathered through collection of genetic samples (e.g.scat, hair, urine, and blood).

Population Estimates With 95% Confidence Intervals40

35 3S

30

VIC 25!C'iii

20-C:::J0:IE 15-0- 10

5

0

2010

Suitable Habitat Estimates

31

2012

The Commission also created a habitat-based estimate of potential population size formountain lions in the Pine Ridge. The estimate is based on a geographic informationsystem model of habitat suitability developed by the North Dakota Game and FishDepartment; model details are available at:http://www.cougarnet.org/status report of lions in North DakotaJinaI.606.pdf. Themodel identifies areas of suitable mountain lion habitat using three primary landscapecriteria: concealment and stalking cover (woody cover/forest/shrubs), topographicconcealment and stalking cover (steep terrain/slopes), and proximity to water. Areasthat are steep, forested, and have available water are considered most suitable. Themodel identified about 96% of Nebraska as unsuitable habitat for mountain lions. Themodel estimated -664 km2 of suitable habitat in the Pine Ridge, which is the largestnon-riparian area of suitable habitat in the state. The Niobrara River Valley is thesecond largest area of contiguous suitable habitat with -351 krrr'. Some other river

3

valleys were identified as suitable habitat by the model, but it is unknown if these thinlinear strips of habitat can be used as a home range or if they only serve as dispersalcorridors.

Density estimates from research in the nearby Black Hills of South Dakota were used toestimate the number of mountain lions that would be expected if density of mountainlions in suitable habitat is similar in Nebraska. Using 2011 density estimates from SouthDakota and the area of suitable habitat in the Pine Ridge, the expected number ofmountain lions would be -22. Using 2012 density estimates from South Dakota theexpected number of mountain lions would be -27. These habitat-based estimatesprovided a second measure that was similar to the genetic estimates, thus increasingour confidence that the genetic-based estimates are valid.

# of Mountain Lions Estimated by GeneticSurvey and Habitat Analysis

30

25

20

15

10

5

o2012 Genetic Survey Estimate Habitat Based Estimate Habitat Based Estimate

(Assuming2011 SO Density) (Assuming 2012 SO Density)

Estimating the Impact of Wildfires

Historic wildfires burned large areas of the Pine Ridge during 2012. In order to estimatethe potential effect these fires may have had on habitat and populations, the suitablehabitat within the boundaries of the burned areas was subtracted from the total suitablehabitat in the Pine Ridge. This provided a lower estimate of 442 km2 of suitable habitatin the Pine Ridge. Using 2011 and 2012 density estimates from South Dakota and thearea of unburned suitable habitat in the Pine Ridge after the fires, the expected number

4

of mountain lions the unburned habitat would support would be .....15-18. This assumesa worst-case scenario of fire rendering all habitat within the burn perimeter unsuitable;the true area of suitable habitat after the wildfires likely lies between 442-664 km2

.

Future data collection will help ascertain the extent to which lions are using habitatwithin the burn perimeters.

Suitable Habitat for Mountain lions and WildfireBurn Areas (from 2012) in the Pine Ridge

• Suitable Mountain Lion Habitat

• 2012 Wildfire Burn Areas

Proposed Mountain lion Harvest

When the Nebraska Legislature classified the mountain lion as a game animal in 1995,it signaled to the Commission that hunting of the species should be allowed if thepopulation was large enough to sustain a harvest. Population information gatheredthrough genetic surveys and habitat estimates in the Pine Ridge indicates thepopulation has reached that level of abundance. In order to provide huntingopportunities for this species in Nebraska, Commission staff recommends a limitedhunting season in the Pine Ridge.

5

The proposed objective of the regional hunting season is to provide a harvestopportunity for mountain lions that is sustainable over time and will allow the populationto remain at or near its current level of abundance. Staff recognizes that otherpopulation objectives (e.g., increasing or decreasing the population) may have meritand different harvest alternatives would be better suited to meet those goals.

To achieve the proposed objective, staff recommends a quota system for controlling theharvest, with the hunting season ending as soon as the quota is reached or the seasonclosing date occurs, whichever happens first. Research elsewhere in the western U.S.regarding hunting effects on mountain lion populations suggests a harvest rate between10-20% to maintain a stable population trend (i.e., 10-20% of the population may beremoved by hunting each year without driving the population into decline) if immigrationfrom source populations is limited. Given a likely post-fire population size in the PineRidge of between 15 and 22 animals, this suggests a quota of three lions is appropriateto meet the management objective and corresponds to a maximum harvest rate of 1420%.

Staff recommends restricting the number of females that be taken under the quotasystem to help ensure a stable population size. Small mountain lion populations rely onimmigration, particularly by males, and recruitment of female offspring from within thepopulation of resident adult females. Since female mountain lions are more likely tostay in the area they were born and less likely to disperse long distances than males,harvest of females has a greater chance of affecting subsequent population productivity.Female immigration probabilities may be reduced in the future if neighboring statesmanage for further reductions in mountain lion populations below present levels.Present mountain lion management objectives of South Dakota Game, Fish and Parksinclude a reduction of the Black Hills population and the reduction of dispersingmountain lions, so a reduction in female immigrants is a realistic possibility.

Assuming population demographics similar to those described by research in SouthDakota during 2012, the post-fire Pine Ridge population likely includes 4-6 adultfemales. Given the harvest objective and considering the estimated population size,area of suitable habitat, and average number of non-hunting human caused mortalitiesin the Pine Ridge, Commission staff recommends a harvest sub-quota of one femalewithin the overall quota of three total mountain lions.

The proposed harvest season is divided into two periods: (Period 1: January 1 throughFebruary 9, and Period 2: February 15 through March 31). The season would closeprior to March 31 if either the female sub-quota or the total quota was harvested prior tothat date. The season would be held in January - March because this is the time ofyear females are least likely to have dependant young that may be orphaned if thefemale is harvested (per research in South Dakota). In order to maximize the harvest

6

opportunities for hunters, 101 permittees (100 by lottery and 1 by auction) would beallowed to hunt during Period 1 without the aid of dogs. During Period 2 all permitteeswould be allowed to hunt, with the auction permittee and the first four lottery permitteesbeing allowed to hunt with dogs. The allowance of dogs during Period 2 may provideopportunities for people who prefer to hunt with dogs, as well as increase the chance ofharvest success if snow is not available for tracking mountain lions during Period 1.The number of permittees and limited use of dogs is designed to allow the harvestquota to be met before the end of the season, while decreasing the likelihood ofexceeding the harvest quota.

Harvest Unit

The proposed harvest unit will have the same boundaries as the Pine Ridge firearmdeer unit (Le. the area north of the Niobrara River and west of Nebraska Highway 27).Other areas of the state such as the Wildcat Hills and Niobrara River Valleydownstream from the Pine Ridge possess suitable habitat for mountain lions. Theseareas have some recent indications of resident populations, including the presence offemales and an adult male in the case of the Niobrara River Valley; however, nopopulation information presently exists that can be used to estimate sustainable harvestlevels. The Commission presently plans to survey both of these areas and createpopulation estimates as part of the 2014 genetic survey.

Summary

Mountain lions have returned to Nebraska through natural expansion from adjacentstates. Commission staff have estimated the size of the mountain lion population andarea of suitable habitat in the Pine Ridge and determined that the population couldsupport a harvest of 1-3 mountain lions. A hunting season has been proposed with theobjective of providing a harvest opportunity for mountain lions in Nebraska whileallowing the population to remain stable. A quota of three mountain lions with a subquota of one female is proposed for the hunting season which would begin on January1, 2014. The sub-quota for female mountain lions provides assurance that this nativegame species will not be eliminated through hunter harvest. The number of permitsissued and use of dogs in Period 2 will allow the harvest quota to be met whiledecreasing the likelihood of exceeding the harvest quota. The Commission recognizesthe Pine Ridge population is connected by immigration and emigration to mountain lionpopulations in South Dakota and Wyoming. We will continue to communicate withneighboring states to ensure sound management. The Commission intends to managemountain lion populations over time with consideration given to social acceptance,effects on prey populations, depredation on pets and livestock, and human safety.

7

Answers to Questions Posed by Commissioner Marshall

What would be involved with federal government putting lions on protected or endangered status?

Before a species can receive the protection provided by the Endangered SpeciesAct, it must first beadded to the Federal lists of threatened and endangered wildlife and plants. A species is added to thelist when it is determined to be endangered or threatened because of any of the following factors:

• the present or threatened destruction, modification, or curtailment of its habitat or range;• overutilization for commercial, recreational, scientific, or educational purposes;• disease or predation;• the inadequacy of existing regulatory mechanisms;• the natural or manmade factors affecting its survival

The initial listing process can happen two different ways: through the petition process or through thecandidate assessment process. The ESA provides that any interested person may petition the Secretaryof the Interior to add a species to, or to remove a species from, the list of endangered and threatenedspecies. Through the candidate assessment process, FWS biologists identify species as listing candidates.

Under the petition process, the USFWS must make a finding within 90 days of receiving a petition as towhether or not there is "substantial information" indicating that the petitioned listing may bewarranted. If this preliminary finding is positive, a status review is conducted. Within one year of receiptof the petition, USFWS must make a further finding that the listing either is or is not warranted.

Through the candidate assessment process, USFWS identifies species for which the best scientific andcommercial data available indicates that a proposal for listing is appropriate, using the listing factors insection 4 of the ESA. A species assessment document, prepared by the Candidate Conservation staff, isprovided to the Service Director, who makes the final decision as to whether a species should beelevated or removed from candidate status or have its listing priority number changed.

USFWS Website detailing overview of listing process: http://www.fws.gov/endangered/what-wedo/listing-overview.html

USFWS Fact Sheet: http://www.fws.gov/endangered/esa-Iibrarv/pdf/listing.pdf andhttp://www.fws.gov/endangered/esa-Iibrary/pdf/candidate species.pdf

Does pasteurella come from stress and do lions cause stress?

Bacteria, primarily Mannheimia spp. and Pasteurella spp., have led to massive, all-age respiratorydisease and die-offs of bighorn sheep. Bighorn sheep die offs often follow incidents of contact withdomestic sheep. Although domestic sheep are resistant to strains of wild sheep bacteria, bighorn sheepare highly susceptible to strains of Pasteurella spp. carried by domestic sheep. Pasteurella spp. issomewhat common in the majority of animal species throughout the world, but not all species ofPasteurella lead to respiratory disease. Because Pasteurella spp. is sometimes found naturally inungulates, some studies suggest that continued high stress levels could lead to suppressed immunesystems, which may explain sudden die-offs of apparently healthy populations. Pasteurella strains canalso remain in bighorn sheep for months or years after contact with domestic sheep, without presentingsymptoms. According to previous research, there is a possibility that respiratory problems related to

Pasteurella may be exacerbated by stress. However, the sheep would have to have the bacteria alreadypresent in their system (either from contact with infected other animals or from natural bacteria sheepcommonly carry).

Continued high levels of predation pressure can lead to high stress levels in a variety of species, whichcan then cause reduced immune function. However, in most studies, there has been constant contactbetween prey and predator, with frequent attacks or kills. This scenario may cause chronic stress in theprey, leading to suppressed immune function, however, we have not observed frequent attacks or killsof bighorn sheep by mountain lions in Nebraska.

There are no known studies examining stress-levels in bighorn sheep associated with mountain lioncaused predation pressure. However, predation is only one of the many factors (along with inclementweather, sudden changes in social status, habitat degradation, decreased food resources, and disease)that have been associated with increased stress levels.

Additional details:J. Malmberg Thesis: http://nlcsl.nlc.state.ne.us/epubs/G1000/B129-2008.pdf; USDA:http://www.bighornsheep.org/article rmrs gtr209.pdf

Mountain lion master plan?

We will complete a statewide mountain lion management plan for Nebraska after completion of

population surveys in the Wildcat Hills and central Niobrara River Valley. The plan should include

population objectives and preferred harvest options for various regions throughout the state.

-What is range?

Mountain lions are distributed across the western U.S. and more recently have moved into western NO,

NE, and South Dakota. The species is widely distributed outside ofthe U.S. as well with populations

found from Canada to southern South America. In Nebraska a mountain lion population (resident

animals with proof of reproduction) has been documented in the Pine Ridge. Some evidence exists for

populations in the Wildcat Hills and Niobrara River Valley (females, an adult male in the caseofthe

Niobrara River Valley, continuous confirmations of presence) but the Commission has not yet

documented kittens in these areas (the final criteria for population designation).

-Mix of public and private land?

The Pine Ridge population is distributed over both public and private land. This is the largest block of

suitable habitat in the state and it also possesses the largest tracts of suitable public land (primarily

Commission and U.S. Forest Service properties).

Where are ideal terrain areas for lions?

Mountain lions are ambush predators and require stalking cover and adequate levels of prey. Generally

open pasture, short grass prairie, harvested crop fields do not provide usable habitat for mountain lions.

Steep areas with heavy woody cover like the Pine Ridge and Niobrara River Valley represent the best

habitat in Nebraska. The model Commission staff used to identify suitable mountain lion habitat uses

three primary landscape criteria: concealment and stalking cover (woody cover/forest/shrubs),

topographic concealment and stalking cover (steep terrain/slopes), and proximity to water. Areas that

are steep, forested, and have available water are considered SUitable. Approximately 96% of Nebraska is

identified as unsuitable to mountain lions. The model estimated ~664 km2 of suitable habitat in the Pine

Ridge, which is the largest block of suitable habitat in the state (habitat for 15-27 mountain lions

depending on effects of the 2012 wildfires). The Niobrara River Valley is the second largest block of

suitable habitat with ~351 km2 (habitat for 9-14 mountain lions depending on effects ofthe 2012

wildfires). The Wildcat Hills also appear contain some suitable habitat although the area is much smaller

and may only support a few individuals. Other areas of the state may have some suitable habitat but it is

not presently known if there is enough habitat to support populations or if they are only stopover sites

for dispersers.

Un,utubl"

Mod./atelyUnsuitable

III Suitable

• Highly $u'tabl"

Is mountain lion management art and science?

Mountain lion management is similar to managing other game species in that we set objectives for

populations and associated recreational opportunities, then prescribe hunting regulations and other

programs (e.g., hunter access and habitat improvements) to achieve those objectives. Biological science

does not totally dictate what the objectives "should" be, as this is the realm of social science and

policymaking. This part ofthe process is sometimes referred to as "art," because it mostly involves

weighing human values and preferences rather than biological facts. Biological science is also often

fairly irrelevant in deciding how to distribute limited hunting opportunities or what hunting methods

should be legal- again, these decisions mostly depend on public opinions and values.

Once a management objective is decided, biologists can use scientific facts and methods to recommend

one or several courses of action to achieve that objective based on the biological capacities of the

species involved. There is also "art" involved in this step, because creativity is required to come up with

the broadest spectrum of potentially successful alternatives. Also, biological factors used in making

decisions often have a range of possible values, so biologists must deal with uncertainty in suggesting

options. There is some "art" in weighing uncertainties; some people and institutions are more risk

averse than others.

In the caseof mountain lions, staff proposes a management objective of keeping the population at least

large enough to allow some hunting opportunity year after year, and to maximize the hunting

opportunity afforded by this population. This was based not on mountain lion biology, but on staffs

presumption of what the intent of the legislature was when they classified the species as a game animal

in 1995. The harvest quotas staff recommended to achieve that objective are based on the biological

capacities of mountain lions to withstand harvest as documented in other states. The recommended

inclusion of an auction permit and some allowance for hunting with dogs was based on the desire for

those measures expressed by some Commissioners at the March Information Meeting in Kearney.

How reliable are lion population estimates?

Many state game agencies to not provide population estimates due to statewide distribution of

mountain lions and extreme costs to conduct statewide surveys. With the recent development of

genetic surveys like those we conducted in the Pine Ridge, cost effective population estimates of small

semi-isolated populations are possible. In judging the value of our population estimate we decided not

to put all of our eggs in one basket and look at only one parameter. We have three science-based

measures of the population to look at: 1) the 2010 genetic survey, 2) the 2012 genetic survey, and 3) the

number of mountain lions we would expect if our mountain lion population is as densely packed into

our area of suitable habitat as neighboring populations (e.g., how many mountain lions we would expect

in a Pine Ridge-sized area of similar habitat in Wyoming or the Black Hills of South Dakota). The 2012

genetic survey was greatly improved over the 2010 survey as indicated by the larger area we surveyed

and the larger number of mountain lion scats collected. The information from this survey provides a

maximum likelihood population estimate of 22 (best estimate) and 95% confidence intervals of 16-37

(meaning that if we surveyed this population over and over many times, 95% of the resulting best

population estimates would be between 16 and 37). Below is a chart showing three ways to determine

the likely population. We consider the 2012 genetic results to be the best estimate of our population.

The 2010 survey and the habitat based estimates are very similar to this best estimate, which increases

our confidence in it.

# of Mountain Lions Estimated by GeneticSurvey and Habitat Analysis

2012 Genetic Survey Estimate Habitat BasedEstimate Habitat BasedEstimate(Assuming2011 SDDensity) (Assuming2012 SDDensity)

How many lions can our range support?

From a purely biological standpoint, this depends on a number of factors that may change over time

based on climatic effects on habitat/prey species (wildfire, drought, etc.), disease/harvest impacts on

prey speciesand other factors. The graph above provides two estimates (22-27) using density of

mountain lions from South Dakota. This is appears to be a reasonable estimate given known densities

from research in other states. However, the carrying capacity could be lower or higher in a given

year/time period. We will learn more about carrying capacity of our habitat over time as we gather

more data regarding our population. We necropsy mountain lions that are killed in Nebraska and look

for signs of the population exceeding carrying capacity such as emaciation, evidence of intraspecific

strife (scarring on males from fighting), and decrease in body condition (compared to earlier necropsies).

We have not yet seen any evidence that the population has exceeded carrying capacity; most have good

to excellent body condition. As an example, the most recent mountain lion killed in the Pine Ridge was a

male that weighed 160 Ibs. The largest mountain lion out of the 61 harvested in the Black Hills this year

was 138Ibs.

What is estimated mix of males and females?

We confirmed six males and eight females in the Pine Ridge during the 2012 genetic survey. If we

assume sex and age structures are similar to those in the Black Hills and extrapolate from the total

population estimate, we likely have 4-6 adult females, 2-3 subadult females, 2-3 adult males, 2 subadult

males, and 5-8 kittens (~50% each gender).

Do lions eat deer, sheep, elk, cattle, dogs, cats, small mammals, and do we know how many are

eaten?

Mountain lions primarily eat deer but they are capable of eating a variety of prey species. The few

Nebraska mountain lions we have been able to necropsy have almost all fed on porcupines. Deer,

raccoons, turkeys, rabbits, and mice are the other species we commonly find. We have investigated

>100 claims of mountain lions killing or attacking pets and livestock but evidence of mountain lions has

not been found during any of the investigations. We do have an internal policy (Mountain Lion Response

Plan) in place as well as statutory authority to respond to any depredation events with lethal control of

the offending animal. Domestic sheep are the primary domestic species killed by mountain lions in

Wyoming. Other domestic livestock can be killed but larger species such as cattle and horses are not

killed as frequently by mountain lions. Depredation problems with cattle are greatest in the desert

southwest where prey species are less abundant.

We can estimate the number of each prey species killed by our mountain lion population (assuming a

population of 22) if we assume similar prey selection as the 1,400 kill sites investigated in South

Dakota's Black Hills. For these calculations we assumed the standard of about one big game sized animal

killed per week (SO/year) for all independent animals (~14 of our 22 lions would be independent sub

adults and adults). The South Dakota study identified ~20% of prey items as scavenged (roadkills,

EHD/CWD-killed animals, shot but unrecovered, etc.). So the estimated number of big game species

killed is ~560/year (14*(50*0.8)) for a population of 22 lions. The number of each species killed is

estimated as 560 multiplied by the proportion of each species of prey identified in the South Dakota

study. This approach produces estimates of 342 white-tailed deer (l.e., 560*0.61=342)' 95 deer of

unknown species, 39 elk, and 17 mule deer killed per year in the Pine Ridge by mountain lions.

Primary Prey Species Percent of Total SD KillsWhite-tailed Deer 61%Unknown Deer 17%Elk 7%Mule Deer 3%

We expect the number of mule deer killed may be slightly higher in Nebraska given they make up a

larger portion of the deer herd in the Pine Ridge than the Black Hills. Further, one percent of the kills in

the Black Hills were bighorns, but we expect the number of bighorn sheep killed in Nebraska would be

lower given the high level of effort put forth monitoring Nebraska's bighorn population with radio/GPS

collars/technician observations and the lack of evidence of predation events (only 1 event documented).

Do lions have 2+ offspring?

Mountain lion females typically begin breeding at 2-3 years of age. They have a litter approximately

every 1.5-2 years. Litter sizes range from 1-6 with 3 being the most common.

Do offspring survive on their own after 6-10 months?

Mountain lions can survive on their own at 6-10 months but typically stay with their mothers for 1-1.5

years. Typically the younger a mountain lion is when it is separated, orphaned, or disperses, the less

likely it is to survive.

What is acceptable mortality rate?

Research elsewhere in the western U.S. regarding hunting effects on mountain lion populations suggests

a harvest rate between 10-20% to maintain a stable population trend (Le., 10-20% of the population

may be removed by hunting each year without driving the population into decline) if immigration from

source populations is limited. Higher harvest rates may be sustainable in areas where mountain lion

populations and habitat surround the harvest area (thus providing source animals to fill voids created by

high harvest).

Small semi-isolated mountain lion populations rely on immigration, particularly by males, and

recruitment of female offspring from within the population of resident adult females. Since female

mountain lions are more likely to stay in the area they were born and less likely to disperse long

distances than males, harvest of females has a greater chance of affecting subsequent population

productivity. Female immigration probabilities may be reduced in the future if neighboring states

manage for further reductions in mountain lion populations below present levels. Present mountain lion

management objectives of South Dakota Game, Fish and Parks include a reduction of the Black Hills

population and the reduction of dispersing mountain lions, so a reduction in female immigrants is a

realistic possibility and is something we are considering in our proposal.

AZ and CA had 60+% (bighorn) sheep mortality?

In general population-level effects for bighorns due to predation by mountain lions is greatest in isolated

populations in desert environments. Part of this is attributed to these areas having only two primary

prey species (mule deer and bighorn sheep). The primary prey species is mule deer but as mule deer

populations decline due to drought or disease, mountain lions respond by increasing predation on

bighorns (prey switching). This may be why the largest negative effects of predation on bighorns are

typically reported in the desert southwest. Areas with multiple prey species like the Black Hills (white

tailed deer, mule deer, elk, turkeys, porcupines, raccoons, etc.) are less influence by prey switching due

to the variety of prey to choose from if one population declines. The availability of alternate prey species

may partially explain the 1% predation rate on bighorn sheep during their prey selection study.

Specific individual mountain lions can learn to specialize at preying on bighorn sheep and can have a

disproportionately large influence on sheep populations. If this is discovered in Nebraska, these

individuals can be targeted by trapping at carcasses where predation is discovered, or by tracking with

hounds.

California:III In an effort to rebuild the sheep herd in the Sierra Nevada population (in Yosemite National

Park), 38 sheep were introduced near Tioga Pass and named the Yosemite Herd in the late1980's. The population grew to nearly 100 animals by 1993, then unknown factors resulted in apopulation crash with a loss of more than 60% and a continued decline over the next severalyear. By 2000, only 20 bighorns remained in the Yosemite Herd. Other herds in the SierraNevada were experiencing similar plummets in population for unknown reasons. Populations inother locations in California did not apparently suffer the same declines as this population in the1990's. http://www.nps.gov/yose/naturescience/sheep.htm

A study conducted from 1992-1998 followed 113 radio-collared sheep in the Peninsular Ranges ofsouthern California. Overall mortality of the sheep totaled 54%, most of which was due to predation bymountain lions (69%), other causes such as disease (16%L and unknown causes (15%).https:/Iwww.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=5&cad=rja&ved=OCEQQFjAE&url=http%3A%2F%2Fnrm.dfg.ca.gov%2FFileHandler.ashx%3FDocumentVersionID%3D75565&ei=QZGBUaucL6TC2QXxmYHACw&usg=AFQjCNHVv9-U8w9UfF-IHOKTL10SQsTcg&bvm=bv.45921128.d.b21

Arizona:III Researchers tracked 395 translocated sheep throughout Arizona from 1979-1997. Mortality for

this period totaled 48%, most of which was due to predation (66%L disease (14%L naturalcauses and accidents (9%), hunting (2%), and other causes (9%).

o Kamler, J.F., R. Lee, J. C. deVos, Jr., W. B. Ballard, and H. A. Whitlaw. 2002. Survival andCougar Predation of Translocated Bighorn Sheep in Arizona. The Journal of WildlifeManagement 66: 1267-1272.

III Bighorn sheep in the Silver Bell Mountains in Arizona contracted IKC and Contagious Ecthymafrom domestic goats in 2003 (Jansen et al. 2007). Thirty-nine percent of the population duringthe epizootic went blind with 50% of those recovering sight and 50% dying. Primary cause ofdeath in affected animals was predation and secondarily, starvation.http://fwpiis.mt.gov/content/getltem.aspx?id=39746

III Arizona officials documents a 52% decline in the sheep population in 1988-89 in Arizona due tolivestock viral disease and nutritional stress:http://library.sandiegozoo.org/factsheets/bighornsheep/bighorn.htm

10 people killed in last 20 years by lions?

III 10 human deaths have been reported from 1890-1990. Beier, P. 1991. Cougar attacks onhumans in the United States and Canada.The Wildlife Society Bulletin 19:403-412.http://www.jstor.org/stable/3782149

III Half of the 20th century's 14 known deaths from cougar attacks in North America occurred inthe 1990s: http://news.nationalgeographic.com/news/2001/08/0827wirepredators2.html

III There are 185 instances of mountain lions attacking humans in the U.S. and Canada from 18902000:http://www.ingentaconnect.com/content/bloomsbury/azoos/2009/00000022/00000001/artOO008

How many lions killed by autos in NE?

We have documented eight mountain lions that have been killed by trains or automobiles over the last

22 years.

Do males cover larger ranges?

Male home ranges (~77-300 square miles) are larger than those of females (~39-115 square miles). On

average, males also disperse much farther as females often stay in the area they were born and

establish home ranges that overlap their mothers'.

How many reported sightings outside our primary lion range annually over the last five years?

We have confirmed mountain lion presence on 63 occasions outside of the Pine Ridge over the last five

years. The advent of inexpensive digital trail cameras provides a greater opportunity to detect dispersing

mountain lions and a greater chance of multiple detections of individuals. The recent increase is partially

driven by multiple detections over the last three years in the Wildcat Hills and Niobrara River Valley

typical of resident populations. These two areas account for 38% of the total over the last five years.

Year # Confirmed Outside the Pine Ridge

2007 12008 22009 72010 172011 142012 23

Does use of dogs increase take rate of males vs. females?

The use of dogs to harvest mountain lions increases the rate of males harvested by 10-17%.

Do males range farther and without breeding females ultimately interact with humans?

Young males that disperse to the east of established mountain lion populations are believed to be

searching for females. This explains the lack of resident males establishing home ranges in suitable

habitat they encounter as they disperse. Some of these young males do interact with people as is

evidenced by the young males killed in South Sioux City, Kearney, and Chicago. Others will die of natural

causesor turn back toward populations in the west. Young male dispersers cause a disproportionate

amount of negative interactions with humans. The State of Washington is limiting total harvest rates

throughout the state to 14%to allow more dominant males to survive so they can kill or displace the

more troublesome sub-adult males.

What do other states allow for shooting lions in sheep range?

Arizona• There is no specific information in Arizona regulations regarding mountain lion shooting within

bighorn sheep areas

• AZ bighorn sheep auction/raffle: http://www.azgfd.gov/w c/bhsheep/hunting.shtmlhttp://www.azgfd.gov/h f/game bighorn.shtml, and mountain lion hunting:http://www.azgfd.gov/regs/mainregs.pdf

California• With the passage of Proposition 117 in 1990, mountain lions became a "specially protected

species," making mountain lion hunting illegal in California. This status and other statutesprohibit the California Department of Fish and Wildlife from recommending a hunting season forlions, and it is illegal to take, injure, possess} transport, import, or sell any mountain lion or partof a mountain lion. Mountain lions may be killed only 1) if a depredation permit is issued to takea specific lion killing livestock or pets; 2) to preserve public safety; or 3) to protect listed bighornsheep. - http://www.dfg.ca.gov/wildlife/lion/lionfag.html

• CAdoes, however allow opportunity for hunting Nelson bighorn sheep:http://www.fgc.ca.gov/regulations/current!mammalregs.aspx#362,http://www.dfg.ca.gov/wildlife/hunting/sheep/dates.htmI

Colorado• There is no specific information in Colorado regulations regarding mountain lion shooting within

bighorn sheep areas.

• Colorado allows hunting of both mountain lions and bighorn sheep.http://wildlife.state.co.us/SiteCoIlectionDocuments/DOW/RulesRegs/BrochureISheepandGoat.mit, http://d27vj430nutdmd.c1oudfront.net/14470/123801/123801.4.pdf.

Idaho• There is no specific information in Idaho regulations regarding mountain lion shooting within

bighorn sheep areas.

• Idaho allows hunting of mountain lions and bighorn sheep:http://fishandgame.idaho.gov/public/docs/rules/bgMtnLion.pdf,http://fishandgame.idaho.gov/public/docs/rules/mgsSheeplnfo.pdf

Montana

• There is no specific information in Montana regulations regarding mountain lion shooting within'bighorn sheep areas.

• Montana allows hunting of mountain lions and bighorn sheep:http://fwp.mt.gov/fishAndWildlife/management/bighorn/defauIt.htm I,http://fwp.mt.gov/hunting/planahunt/huntingGuides/lion/

North Dakota• There is no specific information in North Dakota regulations regarding mountain lion shooting

within bighorn sheep areas.

• North Dakota allows hunting of mountain lions and bighorn sheep:http://gf.nd.gov/gnf/regulations/docs/bgh/bighorn-elk-moose.pdf,http://gf.nd.gov/gnf/regulations/docs/furbearer/furbearer-guide.pdf

Utah

• UTallows mountain lion and bighorn sheep hunting:http://wildlife.utah.gov/dwr/hunting/hunting-information/cougar.htmlandhttp://wildlife.utah.gov/guidebooks/2012 pdfs/2012-13 cougar low.pdf ,http://wildlife.utah.gov/maps/public/details species.php?feature item=12

• UT manages for higher mountain lion harvests within their bighorn sheep range through the useof minimum harvest quotas.

Washington

• There is no specific information in Washington regulations regarding mountain lion shootingwithin bighorn sheep areas.

• WA allows mountain lion and bighorn sheep hunting: http://wdfw.wa.gov/living/cougars.html.http://wdfw.wa.gov/hunting/permits/raffles/auctions.htmI

Wyoming

• There is no specific information in Wyoming regulations regarding mountain lion shooting withinbighorn sheep areas.

• WY allows hunting of mountain lions and bighorn sheep:http://wgfd.wyo.gov/web2011/hunting-l001123.aspx, http://wgfd.wyo.gov/web2011/hunting1001234.aspx

-80% in SO support season if healthy pop exists

-62% favor a hunting season

-45% want decrease in population and 10% want increase

-56% of pop think healthy viable pop lions is good

-57% of pop think there should be hunting season

The above statements appear to come from South Dakota's report on their public input regarding

mountain lions. The report can be found at: http:Ugfp.sd.gov/wildlife/critters/mammals/docs/mt-lion

management-meetings-2010.pdf

Additional comments:

Staff would like to clarify the tooth that is proposed for removal to allow aging of harvested mountain

lions. Below is a figure showing the premolar (P2) that agencies typically remove for aging by cementum

annuli counts. The tooth is the small peg-like premolar just behind the upper canine.

ftCURt 6 t.",tG:aJ. view o)f a mountain lion $kull with lett~~1numbe.r

des:1snatious for p~n:tan~,mt. dentition.nl'~ing by M. A1IJe.~50U.