Behavioral Ecology Vol. 14 No. 3: 433-437
© 2003 International Society for Behavioral Ecology
Empirical support for a despotic distribution in a California spotted owl population
Department of Fisheries, Wildlife, and Conservation Biology, University of Minnesota, St. Paul, MN 55108, USA
Address correspondence to G.S. Zimmerman. E-mail: gsz{at}fw.umn.edu. W.S. LaHaye is now at P.O. Box 523, Big Bear City, CA 92314, USA.
Received 22 January 2002; revised 9 September 2002; accepted 21 September 2002.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Territorial species, such as the spotted owl (Strix occidentalis), are predicted to follow an ideal despotic distribution. However, debate exists on whether wild populations actually meet the assumptions of an ideal distribution, such as perfect perceptual abilities (i.e., the ability to recognize high- and low-quality sites without error). Because this hypothesis has important life history ramifications for spotted owls, we investigated whether occupancy rates of California spotted owl (S. o. occidentalis) territories in the San Bernardino Mountains of southern California positively correlated with a qualitative "potential fitness" (denoted by
pf) estimated from survival and reproduction of territorial owls. Spotted owls in our study tended to occupy territories with the highest pf, supporting the assumption of ideal perceptual abilities within this population. However, this relationship was noisy, and we suggest that some individuals do not assess site quality accurately because of perceptual limitations, prey dynamics, and large territory sizes. Furthermore, dispersal processes, high survival rates, and long life spans of spotted owls may be other key factors preventing some individuals from selecting sites of the highest quality and, consequently, our ability to precisely estimate pf. Key words: California, California spotted owls, fitness, ideal despotic distribution, Strix occidentalis occidentalis, territory occupancy.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Understanding mechanisms controlling spatial distributions of animals is critical to the conservation and management of wildlife (Kennedy and Gray, 1993
). The ideal free distribution (IFD) and ideal despotic distribution (IDD) are models describing distribution patterns of wildlife populations (Fretwell, 1972). The IFD assumes that individuals will occupy the best sites until, due to crowding effects, average fitness within that habitat declines to equal that in the next best habitat type. Territorial behavior could alter this pattern (Fretwell and Lucas, 1969), which led Fretwell (1972) to formalize the IDD. The IDD model assumes that dominant individuals occupy the best sites, forcing the subdominant individuals to use lower quality sites (Fretwell, 1972). As with all hypotheses in ecology, the IFD and IDD are based on critical assumptions, including an animal's perfect knowledge of available resources. Since their formalization, ecologists have modified the IFD and IDD models in an attempt to address these assumptions, which has improved our ability to predict the spatial distribution of individuals (Tregenza, 1995).
The assumptions that individuals have perfect knowledge of habitat quality, that habitats vary in their suitability, and that individuals incur higher fitness by occupying sites with the most available resources are critical components of both the IFD and IDD (Bernstein et al., 1991; Block and Brennan, 1993; Fretwell and Lucas, 1969; Rodenhouse et al., 1997). Fretwell and Lucas (1969) labeled organisms that maximize survival and reproduction by selecting quality habitats "ideal." They argued that this assumption is valid because ideal individuals are favored through natural selection. However, others contend that this assumption is unrealistic due to perceptual constraints or imperfect knowledge of resources (Gray and Kennedy, 1994; Pyke et al., 1977; Rapport, 1991).
One study of the northern spotted owl (Strix occidentalis caurina), a territorial species, concluded that the owls followed an IDD (Franklin et al., 2000). We have studied the population dynamics and territory occupancy of California spotted owls (S. o. occidentalis) in southern California for 12 years. Some owls settled in areas where they did not reproduce, yet they remained on the unproductive territory. This led us to question the adaptive value of occupying territories that did not confer reproductive potential. Therefore, we investigated whether occupancy rates of spotted owl territories reflected potential fitness (sensu Franklin et al., 2000; see definition below) of those territories. Our goal was to add to the work of Franklin et al. (2000) by exploring whether our field data supported an IDD in another spotted owl population. In addition, we used an analytical technique that incorporated both survival and fecundity into a single fitness estimate to explore the validity of the ideal assumption for this population.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Study area
Our study area was approximately 140 km east of Los Angeles, California, USA (34°15' N; 117°55' E) in the San Bernardino Mountains. This range is oriented east/west with elevations ranging from 800 to 3500 m and is surrounded by desert and chaparral vegetation (Vasek and Barbour, 1988
). The climate is Mediterranean, with most precipitation occurring during the winter in the form of snow above 2000 m and rain at lower elevations. Annual precipitation ranges from 25 to 100 cm and varies with elevation and slope aspect (Minnich et al., 1995). Vegetation grades from Mojave desert scrub and southern coastal scrub at lower elevations to alpine at the highest elevations (Minnich, 1998). Within this continuum, local aspect and topography formed a complex mosaic of forest, chaparral, desert, and wetland vegetation patches. The mountain range was surrounded by vegetation unsuitable for spotted owls (LaHaye et al., 2001); therefore, the population was insular.
Field techniques
In 1989, we began a long-term demographic study of California spotted owls in the San Bernardino Mountains. We annually surveyed all previously occupied spotted owl locations and all forest vegetation that could potentially support spotted owls, which included the entire mountain range from 1989 to 1998. We conducted surveys at night by imitating spotted owl vocalizations for a minimum of 15 min at call points or by calling continuously along designated survey routes (Forsman, 1983). These call points were strategically placed to cover forested vegetation types. For example, we did not establish survey points in areas of desert or chaparral because California spotted owls require at least some forested habitat (Gutiérrez et al., 1992). Therefore, the distribution of survey points reflected the distribution of forests and were not evenly spaced across the landscape. In addition, the number of survey points for each area varied due to landscape features, distribution of forest, and access. Upon receiving a response, we recorded the spotted owl's gender and then attempted to visually locate the bird. Once located, we attempted to capture and band all spotted owls. We placed a locking aluminum band on one leg and a unique combination of plastic color band and colored vinyl tab on the other leg (Franklin et al., 1996). In addition, we attempted to evaluate each owl's social status (paired, single) and estimate its reproductive effort (Franklin et al., 1996). We conducted surveys, capture, and identification of individuals during each breeding season (April–August). If owls were not located at a particular site after three complete surveys within a field season, we classified the territory (previously occupied site) or survey (previously unoccupied, but potential habitat) as vacant for that year.
Specific methods for estimating occupancy and potential fitness
We calculated occupancy rates of individual spotted owl territories as the proportion of years that a territory was occupied by a pair of owls. For example, a site that was occupied by a pair 3 of 10 years had an occupancy rate of 0.30. If we did not conduct a minimum of three surveys in an area within a year, we considered the occupancy of that area for that year as unknown and coded it as missing data. Although we attempted to survey all potential habitats, we did not conduct a complete census of spotted owls in the study area. Further, our estimates of occupancy may have underestimated true occupancy rates because we did not know whether territories without an owl response were unoccupied or if the owls simply did not respond. However, we believe that any bias in occupancy estimates was small because our surveys were designed to maximize detection rates (Reid et al., 1999), and resighting rates of spotted owls using these survey methods are known to be high (Seamans et al., 2001). Areas that were known to contain owls before our study but that were not occupied during our study (i.e., territories with occupancy rates of zero) were not included in our analysis.
Because fitness was a trait of individual organisms, territories could not exhibit fitness (Franklin et al., 2000). Therefore, we calculated a potential fitness value (PF) for each territory from estimates of survival and fecundity of owls that occupied those territories. We calculated PF by incorporating estimates of apparent survival of male and female owls on each territory and estimates of mean annual fecundity of females on each territory into a 3 x 3 Leslie matrix (Franklin et al., 2000) by territory. We then calculated PF as the dominant eigenvalue of the Leslie matrix through matrix eigenanalysis (Danchin et al., 1995; Franklin et al., 2000). This value represented a PF that could be expected for owls occupying that territory. We assumed that PF was influenced by unmeasured attributes of that territory (e.g., prey availability, habitat quality). Thus, the PF estimate represented a relative measure of quality among territories such that owls occupying territories with a higher PF would be expected to have the potential for a higher combination of survival and fecundity (fitness) than owls occupying territories having a lower PF.
Our estimates of PF had large standard errors because in most cases we had very few repetitions (individual birds) per sampling unit (territory) to estimate apparent survival. These large standard errors indicated that the PF estimates were not precise. By using imprecise estimates as continuous variables, we would have implied overly optimistic confidence in our estimates of PF. Therefore, we converted continuous PF estimates to a categorical binary response variable with two categories (high and low) to investigate general trends between occupancy and PF. We converted the PF estimates to a binary variable (denoted as pf) based on biologically meaningful principles rather than statistical principles. Specifically, we assigned territories with estimates of 
into a high-quality group where females were more likely to replace themselves. Thus, these territories were assigned a 1 to represent a success. Territories with PF estimates < 1 were assigned a 0 and represented sites where females were less likely to produce enough offspring to replace themselves.
We estimated apparent survival of owls on individual territories with a matrix of individual capture histories using the program MARK (White and Burnham, 1999). These estimates could have been biased if dispersal rates from the study area were high. However, field data on dispersal indicated that immigration and emigration rates for this population were low if they occurred at all (LaHaye et al., 2001). Previous survival models for this population indicated that gender did not influence survival estimates (LaHaye et al., 1994). Therefore, we did not include gender effects when modeling survival. Our estimates of pf incorporated survival estimates of territorial adults and subadults only. Previous modeling indicated annual survival rates varied between adults and subadults (LaHaye, unpublished data). However, the number of territorial subadults in the population was so small relative to adults that we excluded age effects from survival models. We assumed that the probability of recapture was constant among territories. Apparent survival estimates for each territory represented the mean annual probability of survival for male and female spotted owls that occupied that territory. Owls moving among territories could have artificially reduced apparent survival estimates. We accounted for the movement of owls among territories by considering them "losses on capture" (Franklin et al., 2000; Jolly, 1965). Thus, owls that moved among territories were removed from the analysis on the original territories once they moved, rather than treating them as deaths. These individuals were then included in the analysis for the new territories.
We defined fecundity as the number of female young fledged per territorial female. We used mixed modeling (Proc Mixed: SAS Institute, 2000; see also Littell et al., 1996) with a repeated-measures design to model fecundity. Territory was the sampling unit, and years were repeated within territories. We considered territory a random effect because we probably did not detect all members of the population even though we surveyed all potential owl habitat in the study area every year. Therefore, we technically monitored a sample of the spotted owl population in the San Bernardino Mountains.
We assessed normality with normal probability plots and the assumption of constant variance with a modified Levene test (Neter et al., 1996). These tests indicated non-normality and nonconstant variance in our fecundity data. Therefore, we used a normal-based mixed-model analysis to estimate fecundity because it was more robust to departures from the assumptions of normality and constant variance than nonlinear models were to departures from their assumed distributions for count data (White and Bennetts, 1996). Further, mixed modeling allowed the error variance matrix to be structured for nonconstant sampling variances among years (Franklin et al., 2000) using the global model (Burnham and Anderson, 1998) and restricted maximum likelihood (Littell et al., 1996). We compared compound symmetric, autoregressive, and standard variance components covariance structures (Littell et al., 1996). Covariance structures were ranked by their ability to fit the data with a small sample size adjustment of Akaike's information criterion (AICc; Hurvich and Tsai, 1989; Burnham and Anderson, 1998). We used standard maximum likelihood and the highest ranked covariance structure to estimate territory-specific fecundity. Similar to survival, we eliminated age effects in modeling fecundity because we had few repetitions per territory. Final estimates represented the mean annual fecundity of the females on that territory during the years that the territory was occupied and that reproduction could be assessed.
After estimating an occupancy rate and pf for each territory, we used logistic regression (Agresti, 1996) to assess whether there was a positive correlation between occupancy rate and the probability that 
. Therefore, we performed what is analogous to a one-tailed test with the null hypothesis that occupancy rate and pf of territories were correlated using an information-theoretic approach (see below). We used the G2 goodness-of-fit test and the Pearson statistic (Agresti, 1996) to assess whether the data fit a logistic distribution. We adjusted variances and covariances for overdispersion from a binary distribution by using an overdispersion parameter (Burnham and Anderson, 1998).
We created two models to investigate whether there was a correlation between pf and occupancy. First, we calculated parameter estimates from a means (intercept only) model [expressed as 
, where 
is the probability that ]. Then we calculated parameter estimates from a model containing occupancy rates as a predictor variable [expressed as 
]. We then estimated quasi-likelihood estimates of QAICc (Burnham and Anderson, 1998) to rank the two models because we included an overdispersion parameter when comparing models. If the model containing occupancy was ranked higher (lower value of QAICc) than the means model, we concluded that the slope parameter for occupancy was not equal to zero and that occupancy positively correlated with the probability that (i.e., sites where females were more likely to replace themselves). The sign of the estimate indicated whether the correlation was positive or negative. Further, we used a 95% likelihood ratio confidence interval of the estimate to assess the strength of evidence that a parameter estimate was different from zero. Confidence intervals that did not include zero provided strong evidence that a parameter estimate was different from zero. Logistic regression analyses were done using SAS software (Proc GENMOD: SAS Institute, 2000).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
We identified and monitored 142 spotted owl territories between 1989 and 1998. We eliminated 14 territories from the analysis because of missing data. The overall occupancy rate of territories declined from 89% in 1989 to approximately 50% in 1998, but the trend was not constant over time. Occupancy rate decreased between 1989 and 1992, remained stable until 1996, then declined to 50% by 1998 (Figure 1). Occupancy rates within territories ranged from 10% to 100%. Annual apparent survival estimates averaged across years, and individuals on each territory ranged from 0 to 1. Although reproductive output was high for some individuals over the course of their lives (maximum = 16), these values were halved (to adjust reproductive output so that it represented fecundity) and averaged across years for each territory, resulting in annual fecundity estimates that ranged from 0 to 1. Standard errors of apparent survival estimates ranged from 0.00 (territories where the same owls were observed on the site throughout the study period) to 0.36. Standard errors of fecundity ranged from 0.25 to 0.78.
|
All three estimates of model fit indicated that the data fit a logistic distribution (G2 goodness-of-fit test:
; Pearson statistic: 
; 
). The model that included occupancy was ranked substantially higher (
) than the means model (
). The parameter estimate for occupancy (
, 95% likelihood ratio 
) indicated that the correlation was positive, adding additional support that occupancy correlated with pf (Figure 2). Thus, spotted owls in our study tended to occupy sites where they were more likely to at least replace themselves during this study. Although occupancy generally correlated with pf in this study, we detected considerable scatter from a smooth relationship (Figure 2).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Many previous studies of spotted owl population dynamics have estimated temporal variation in survival and fecundity to describe variability in these parameters (e.g., Franklin et al., 2000
; Seamans et al., 1999). These authors were interested in estimates of temporal variation for their respective populations, whereas we investigated spatial variation in survival and fecundity within our study population. We suspected that the high spatial variation in survival and fecundity in our study resulted from a combination of habitat (Franklin et al., 2000; Gutiérrez et al., 1992), social factors (Forsman et al., 1984; LaHaye et al., 2001), individual characteristics of resident owls (
omnicki, 1988), and stochastic differences among territories. However, additional data collection and analyses would be necessary to explain trends in spatial variation of demographic parameters in this population.
The IFD model (Fretwell and Lucas, 1969) was one of the first mechanisms proposed to conceptualize spatial distribution and settlement patterns in animals within their preferred habitat. The territorial behavior of spotted owls (Gutiérrez et al., 1995) led us to predict that they would follow an IDD rather than an IFD because dominant individuals should gain access to higher quality territories. Ens et al. (1995) described six criteria that needed to be met under an IDD. We did not have data to test four of the six criteria. However, we explored differences in fitness potential among territories and whether competitors occupied territories with the highest fitness potential.
Franklin et al. (2000) investigated whether spotted owls in their study area followed an IDD or IFD by calculating habitat fitness potential values for each territory. Specifically, they used habitat covariates to model survival and fecundity of individual owls. The regression equation developed from their modeling was then used to back-transform to territory-specific estimates of survival and fecundity. These estimates represented values that would be expected of an individual owl if it settled in a territory based on the habitat characteristics of that territory. These back-transformed survival and fecundity estimates were then incorporated into a single estimate of habitat fitness potential for each territory. Because Franklin et al. (2000) documented spatial process variation among habitat fitness potential estimates of territories, they predicted that owls in their study area followed an IDD. We did not use habitat in our estimates of fitness potential because preliminary modeling indicated that habitat was not correlated with survival and fecundity in our study area (LaHaye, unpublished data). Nevertheless, we also observed spatial variation in overall fitness potential estimates among territories. In addition, spotted owls in the San Bernardino Mountains appeared to recognize site-quality differences among territories because occupancy rates correlated with estimates of pf.
Although occupancy correlates with our binary, categorical measure of pf (higher fitness, lower fitness), our data indicate that there is still considerable variation in the relationship. This variation may be due to (1) an inherent inability to rate territory quality (Levin et al., 2000), (2) the inability of some owls to correctly assess site quality because large home range sizes limit the number of territories an individual can sample (Pulliam and Danielson, 1991), (3) inability of individuals to predict high-quality sites when resources are temporally variable, (4) high costs of movement (Kennedy and Gray, 1997; Levin et al., 2000), (5) population status relative to carrying capacity (i.e., a declining population versus an increasing population; Bautista et al., 1995), (6) competition (Spencer et al., 1995), (7) individual differences in ability to secure resources that are present (Spencer et al., 1995), and (8) dispersal processes (e.g., the need to secure a mate; LaHaye et al., 2001). In addition, the low precision of our territory-specific demographic estimates could introduce considerable variation from a smooth relationship between occupancy and fitness in this analysis due to high standard errors in PF estimates and associated difficulties in converting values near 1 to a binary response variable. Further, our inability to include individual covariates (e.g., age of territorial owls) implies that individual characteristics of owls on territories are confounded with PF estimates.
Densities of woodrats, the main prey species in our study area (Smith et al., 1999), are temporally and spatially dynamic (Ward et al., 1998; Williams et al., 1992), which may complicate settlement decisions. In addition, dispersing owls may discern high-quality sites from those that can be occupied, but are of low quality, partially by the presence of conspecifics (LaHaye et al., 2001; Stamps, 1988). However, site selection based on conspecifics assumes birds already occupying territories are in the highest quality sites. Further, large territories may be difficult to assess by dispersing spotted owls. For example, Pulliam and Danielson (1991) argued that spotted owls should not conform to ideal selection because their home ranges are large and dispersal mortality is high, which reduces the number of breeding sites that a dispersing owl can sample. In addition, once an owl settles in a territory that provides enough resources to survive and reproduce, it may not be adaptive to move and gamble that it would find both a better territory and a mate (Levin et al., 2000). Hirons (1985) suggested that tawny owls (S. aluco), and perhaps other owls, must learn the best hunting sites in a territory to successfully breed. Accumulating such knowledge about the territory in a short time period would be nearly impossible for spotted owls because they have large home ranges (Gutiérrez et al., 1995). Thus, the cost of abandoning a territory in which an owl has invested significant exploration time would need to be balanced against the risks of moving to a territory of equal or lower quality. It may not be worth risking the loss of accumulated knowledge of a territory, even if it is not the best territory, given the longevity of spotted owls. For example, some owls may live long enough to contribute to the next generation even when occupying a lower quality territory simply because they could experience a range of prey densities during their lifetime, allowing them to reproduce at irregular intervals. In summary, we suspect that the variability in our data largely resulted from a combination of the ecological factors discussed and the low precision of our demographic estimates.
Our results support conclusions from previous work suggesting that wild populations of spotted owls appear to follow IDD rather than an IFD (Franklin et al., 2000). In addition, we add additional support for the IDD by addressing a second assumption: that spotted owls can recognize and do occupy high-quality sites. However, we found an imperfect fit between occupancy and pf when we used a binary response variable as a measure of territory quality. This relationship may have been even weaker if the response variable was continuous. This led us to explore potential causes for the deviation from ideal selection in this analysis. It may be unreasonable to expect ideal site selection at the scale of owl home ranges (Pulliam and Danielson, 1991), especially if this assessment is influenced by an imperfect ability to rate the quality of sites, social stimuli for settling, the need to acquire a mate, and stochastic opportunities associated with site and mate availability in a natural population.
Because spotted owls may not always occupy sites with the highest potential fitness, demographic rates may be lower than expected in areas that have limited amounts of suitable owl habitat. This suggests that either reserve size for spotted owl conservation or amount of suitable habitat needed to maintain a viable population of spotted owls may be larger than predicted by vital rates alone due to factors that influence territory selection (Greene and Stamps, 2001).
|
ACKNOWLEDGEMENTS |
|---|
We thank our many field assistants. We also thank personnel of the San Bernardino National Forest and Big Bear Ranger District for their cooperation and assistance during this study. G. Gould, J. Palmer, J. Verner, D. Solis, and J. Robinson provided additional support. The California Department of Fish and Game, California Department of Natural Resources, Southern California Edison, Region 5 USDA Forest Service (contract #FS/53–91S8-00-EC14 to R.J.G.), Pacific Southwest Forest and Range Experiment Station, Snow Summit Mountain Resort, Big Bear Mountain Resort, Salad King, Inc., and the University of Minnesota provided funding for this project. R. Ydenberg and two anonymous reviewers provided comments that greatly improved the quality of the manuscript.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Agresti A, 1996. An introduction to categorical data analysis. New York: John Wiley and Sons.
Bautista LM, Alonso JC, Alonso JA, 1995. A field test of ideal free distribution in flock-feeding common cranes. J Anim Ecol 64:747-757.[CrossRef]
Bernstein C, Krebs JR, Kacelnik A, 1991. Distribution of birds amongst habitats: theory and relevance to conservation. In: Bird population studies. Relevance to conservation and management (Perrins CM, Lebreton J-D, Hirons GJM, eds). Oxford: Oxford University Press; 317–345.
Block WM, Brennan LA, 1993. The habitat concept in ornithology. Curr Ornithol 11:35-91.
Burnham KP, Anderson DR, 1998. Model selection and inference: a practical information-theoretic approach. New York: Springer-Verlag.
Danchin E, González-Dávila G, Lebreton JD, 1995. Estimating bird fitness correctly by using demographic models. J Avian Biol 26:67-75.
Ens BJ, Weissing FJ, Drent RH, 1995. The despotic distribution and deferred maturity: two sides of the same coin. Am Nat 146:625-650.[CrossRef]
Forsman ED, 1983. Methods and materials for locating and studying spotted owls. PNW-GTR 162. Portland, Oregon: USDA Forest Service.
Forsman ED, Meslow EC, Wight HM, 1984. Distribution and biology of the spotted owl in Oregon. Wildl Monogr 87:1-64.
Franklin AB, Anderson DR, Forsman ED, Burnham KP, Wagner FW, 1996. Methods for collecting and analyzing demographic data on the northern spotted owl. Stud Avian Biol 17:12-20.
Franklin AB, Anderson DR, Gutiérrez RJ, Burnham KP, 2000. Climate, habitat quality, and fitness in northern spotted owl populations in northwestern California. Ecol Monogr 70:539-590.[CrossRef]
Fretwell SD, 1972. Populations in a seasonal environment. Monographs in population biology 5. Princeton, New Jersey: Princeton University Press.
Fretwell SD, Lucas HL, Jr, 1969. On territorial behavior and other factors influencing habitat distribution in birds. I. Theoretical Development. Acta Biother 19:16-36.
Gray RD, Kennedy M, 1994. Perceptual constraints on optimal foraging: a reason for departures from the ideal free distribution? Anim Behav 47:469-471.[CrossRef]
Greene CM, Stamps JA., 2001. Habitat selection at low population densities. Ecology 82:2091-2100.[CrossRef]
Gutiérrez RJ, Verner J, McKelvey KS, Noon BR, Steger GN, Call DR, LaHaye WS, Bingham BB, Senser JS, 1992. Habitat relations of the California spotted owl. In: The California spotted owl: a technical assessment of its current status. PSW-GTR 133 (Verner J, McKelvey KS, Noon BR, Gutiérrez RJ, Gould GI Jr, Beck TW, eds). Albany, New York: USDA Forest Service; 79–98.
Gutiérrez RJ, Franklin AB, LaHaye WS, 1995. Spotted owl (Strix occidentalis). In: The birds of North America, no. 179 (Poole A, Gill F, eds). Philadelphia: The Academy of Natural Sciences; 1–28.
Hirons GJM, 1985. The effects of territorial behaviour on the stability and dispersion of tawny owl (Strix aluco) populations. J Zool 1:21-48.
Hurvich CM, Tsai C-L, 1989. Regression and time series model selection in small samples. Biometrika 76:297-307.
Jolly GM, 1965. Explicit estimates from capture-recapture data with both death and immigration-stochastic model. Biometrika 52:225-247.
Kennedy M, Gray RD, 1993. Can ecological theory predict the distribution of foraging animals? A critical analysis of experiments on the ideal free distribution. Oikos 68:158-166.
Kennedy M, Gray RD, 1997. Habitat choice, habitat matching and the effect of travel distance. Behaviour 134:905-920.
LaHaye WS, Gutiérrez RJ, Akcakaya HR, 1994. Spotted owl metapopulation dynamics in southern California. J Anim Ecol 63:775-785.[CrossRef]
LaHaye WS, Gutiérrez RJ, Dunk JR, 2001. Natal dispersal of the spotted owl in southern California: dispersal profile of an insular population. Condor 103:691-700.[CrossRef]
Levin PS, Tolimieri N, Nicklin M, Sale PF, 2000. Integrating individual behavior and population ecology: the potential for habitat-dependent population regulation in a reef fish. Behav Ecol 11:565-571.
Littell RC, Milliken GA, Stroup WW, Wolfinger RD, 1996. SAS system for mixed models. Cary, North Carolina: SAS Institute.
omnicki A, 1988. Population ecology of individuals. Monographs in population ecology 25. Princeton, New Jersey: Princeton University Press.
Minnich RA, 1998. Vegetation, fire regimes, and forest dynamics. In: Oxidant air pollution impacts in the montane forests of southern California: a case study of the San Bernardino Mountains (Miller PR, McBride JR, eds). New York: Springer-Verlag; 44–80.
Minnich RA, Barbour MG, Burk JH, Fernau RF, 1995. Sixty years of change in Californian conifer forests of the San Bernardino Mountains. Conserv Biol 9:902-914.[CrossRef]
Neter J, Kutner MH, Nachtsheim CJ, Wasserman W, 1996. Applied linear statistical models, 4th ed. Chicago: Irwin.
Pulliam HR, Danielson BJ, 1991. Sources, sinks, and habitat selection: a landscape perspective on population dynamics. Am Nat 137:(suppl): S50-S66.[CrossRef]
Pyke GH, Pulliam HR, Charnov EL, 1977. Optimal foraging: a selective review of theory and tests. Q Rev Biol 52:137-154.[CrossRef]
Rapport DJ, 1991. Myths in the foundation of economics and ecology. Biol J Linn Soc 44:185-202.
Reid JA, Horn RB, Forsman ED, 1999. Detection rates of spotted owls based on acoustic-lure and live-lure surveys. Wildl Soc Bull 27:986-990.
Rodenhouse NL, Sherry TW, Holmes RT, 1997. Site-dependent regulation of population size: a new synthesis. Ecology 78:2025-2042.[CrossRef]
SAS Institute,, 2000. SAS/STAT. Version 8.01. Cary, North Carolina: SAS Institute.
Seamans ME, Gutiérrez RJ, May CA, Peery MZ, 1999. Demography of two Mexican spotted owl populations. Conserv Biol 13:744-754.[CrossRef]
Seamans ME, Gutiérrez RJ, Moen CA, Peery MZ, 2001. Spotted owl demography in the central Sierra Nevada. J Wildl Manage 65:425-431.
Smith RB, Peery MZ, Gutiérrez RJ, LaHaye WS, 1999. The relationship between spotted owl diet and reproductive success in the San Bernardino Mountains, California. Wilson Bull 111:22-29.
Spencer HG, Kennedy M, Gray RD, 1995. Patch choice with competitive asymmetries and perceptual limits: the importance of history. Anim Behav 50:497-508.
Stamps JA, 1988. Conspecific attraction and aggregation in territorial species. Am Nat 131:329-347.[CrossRef]
Tregenza T, 1995. Building on the ideal free distribution. Adv Ecol Res 26:253-307.
Vasek FC, Barbour MG, 1988. Mojave desert scrub vegetation. In: Terrestrial vegetation of California, 2nd ed (Barbour MG, Major J, eds). Sacramento: California Native Plant Society; 835–867.
Ward JP, Jr, Gutiérrez RJ, Noon BR, 1998. Habitat selection by northern spotted owls: the consequences of prey selection and distribution. Condor 100:79-92.
White GC, Bennetts RE, 1996. Analysis of frequency count data using the negative binomial distribution. Ecology 77:2549-2557.[CrossRef]
White GC, Burnham KP, 1999. Program MARK: survival estimation from populations of marked animals. Bird Stud 46:(suppl): S120-S139.
Williams DF, Verner J, Sakai HF, Waters JR, 1992. General biology of major prey species of the California spotted owl. In: The California spotted owl: a technical assessment of its current status. PSW-GTR 133 (Verner J, McKelvey KS, Noon BR, Gutiérrez RJ, Gould GI Jr, Beck TW, eds). Albany, New York: USDA Forest Service; 207–221.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  F. Sergio, J. Blas, M. G. Forero, J. A. Donazar, and F. Hiraldo Sequential settlement and site dependence in a migratory raptor Behav. Ecol., September 1, 2007; 18(5): 811 - 821. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||