BARRED OWL HABITAT AND PREY: A REVIEW AND SYNTHESIS OF THE LITERATURE

ABSTRACT Barred Owls (Strix varia) historically inhabited the forests of eastern North America. During the last century, they expanded their range to include forests throughout the southern provinces of Canada, southeastern Alaska, British Columbia, Washington, Oregon, and northern California. To date, there has been no synthesis of the varied habitats or prey used by Barred Owls in their expanded range. Here I review and synthesize studies concerning habitat (N  =  114) and prey (N  =  43) of Barred Owls throughout North America north of Mexico. Barred Owls nested in wilderness areas and national parks as well as small suburban woodlots. They typically preferred old or mature, mixed deciduous/coniferous forests with fairly high canopy closure. Barred Owl nests in natural locations (N  =  341) were in cavities (74.8%), hawk nests (12.9%), tops of hollow trees (8.2%), squirrel nests (2.1%) and other locations (2.1%). Trees used by Barred Owls for nesting (N  =  169) included 22 genera, although one-fourth of all reported nest trees were Populus. Barred Owls used significantly more coniferous nest trees than deciduous nest trees in their expanded range vs. their historic range. They also nested in nest-boxes (N  =  103), in buildings (N  =  6), on nesting platforms (N  =  2), on the ground (N  =  2), in a creek bank (N  =  1), and under a bridge (N  =  1). Prey individuals (N  =  7077) obtained from pellets (N  =  5504) were mammals (74.7%), birds (8.3%), amphibians (6.4%), insects and spiders (5.6%), crayfish (3.0%), fish (1.5%), reptiles, snails and slugs, and earthworms (<1.0% each). The distribution of prey types documented by other means (e.g., observations, prey remains; N  =  1573) was significantly different, due especially to more earthworms in non-pellet samples: mammals (62.0%), birds (13.5%), insects and spiders (9.5%), amphibians (4.8%), earthworms (4.5%), fish (3.3%), reptiles (1.1%), snails, and crayfish (<1.0% each). The distributions of prey types in all samples that could be placed into winter or non-winter months (N  =  4631) differed seasonally, with more mammals in the winter diet (98.4%, vs. 58.9% in non-winter months) and more birds (11.9%), insects or spiders (11.4%), amphibians (10.7%), earthworms (2.3%), fish (2.0%), and crayfish (2.0%) in non-winter months.

[Traducción del equipo editorial] During the past century, Barred Owls (Strix varia) expanded their range from eastern North America to much of central and western North America (Mazur and James 2000), from Saskatchewan (Mazur et al. 1998, Takats 1998, British Columbia (Rand 1944, Hobbs 2005, and southeastern Alaska (D. Gibson pers. comm.) through Washington (Herter andHicks 2000, Gremel 2005), Oregon (Kelly et al. 2003) and central California ( Jensen et al. 2004, Steger et al. 2006. To date, there has been no synthesis of the varied habitats or prey used by Barred Owls in their expanded range. Here I review and synthesize publications, unpublished reports, and personal communications concerning habitat use and prey of Barred Owls throughout North America north of Mexico (1) in total; (2) by ecoregion (Bailey 1998) to present the varied forests used by Barred Owls and to compare use of habitats and prey across ecoregion domains; and (3) by their historic range vs. expanded range to see whether Barred Owls used different habitats or prey in their expanded range. In addition, I compare prey taken by Barred Owls in winter vs. non-winter months to determine seasonal variations in diet. METHODS I compiled publications, theses, dissertations, reports, and personal communications (''studies'') that provided original field data concerning habitat use and food habits of Barred Owls. Other than two studies with regional locations (Fisher 1893, Apfelbaum andSeelbach 1983) and a portion of one study with a state-only location (Bent 1938), I mapped the location(s) of each study relative to the distribution of ecoregions developed by Bailey (1998;R. Bailey pers. comm.). This hierarchical system is composed of ecoregion domains, divisions, and provinces based primarily on climatic conditions and on the prevailing plant formations determined by those conditions. North America north of Mexico has four ecoregion domains: (1) ''Polar'' with tundras and boreal forests; (2) ''Humid Temperate'' with broadleaf deciduous and needleleaf evergreen forests; (3) ''Dry'' with arid deserts, semiarid steppes, montane coniferous forests, and alpine meadows; and (4) ''Humid Tropical'' with savannahs (Bailey 1998). Each domain is divided into divisions which are further subdivided into provinces. Mountains exhibiting altitudinal zonation and having the climatic regime of the adjacent lowlands are distinguished according to the character of the zonation (R. Bailey pers. comm.). I analyzed habitat use and food habits of Barred Owls across ecoregion domains and between their historic range (i.e., prior to about 1900; Manitoba and North Dakota eastward), and their expanded range (Saskatchewan and Montana westward; Mazur and James 2000).
''Sites'' were single or multiple detections or observations of Barred Owls where neither nests nor fledglings were found, whereas ''nest sites'' included nests or fledglings; both sites and nest sites excluded nests in humanmade structures. I differentiated among: (1) ''preference'' results that were statistical analyses of variables of Barred Owl nest sites or sites relative to the surrounding area or to random plots; (2) ''description'' results that described variables without statistical analysis; and (3) ''comparison'' results that compared, via statistics or descriptions, Barred Owl nest sites or sites to those of other sympatric owl species. Many studies had two or three types of results for different variables. Herein, I used ''preference'' and ''avoidance'' only when reporting preference results that showed use of habitats significantly more or less than expected. I focused primarily on the five habitat variables that were most frequently reported: elevation, slope, proximity to water (streams, rivers, wetlands, lakes), forest composition, and forest age. Because of the relative nature of these variables throughout North America, I used the categories and definitions of these five variables as provided by the studies and did not attempt to standardize them. Because forest age was reported in different classifications in various studies, I simply used two categories for that

SEPTEMBER 2007
BARRED OWL HABITAT AND PREY 179 variable: ''old or mature'' and ''intermediate or young.'' I pooled studies that reported results within the same study area by the same author(s) during either the same year or sequential years (e.g., Devereux 1982 with Devereux andMosher 1984); I reported all studies in the Appendix but only the earliest of pooled studies in the text. I recorded ''3'' for studies that reported sample sizes of ''few'' (Apfelbaum and Seelbach 1983) or ''several'' (Dingle 1926). I included habitat information concerning sites of hybrids between Barred and Northern Spotted Owls (S. occidentalis caurina; Fullerton and Meekins 2004, N 5 2;Seamans et al. 2004, N 5 2). For studies that provided occurrence of taxonomic prey rather than the number of individual prey items per stomach (Mendall 1944) or pellet (Coon 1917, Korschgen and Stuart 1972, Leder and Walters 1980, I attributed one prey item per stomach or pellet, which probably underestimated numbers of some prey. I included probable prey captures of a Northern Spotted Owl (Leskiw and Gutiérrez 1998) and of a Great Gray Owl (S. nebulosa; Graves and Niemi 2006). I divided prey studies into two seasonal categories: ''winter'' (November to February or specifically called ''winter'' data by investigators) and ''non-winter'' (March to October), excluding studies which combined data for all seasons. In one study which combined 54 winter pellets with 14 pellets of undisclosed season (Cahn and Kemp 1930), I assumed all were winter pellets.
When comparing variables, I followed accepted statistical protocols (Conover 1980, Zar 1984, Fowler et al. 1998, avoiding the use of chi-square tests if any expected value for a given category ,1 or if .20% of the expected categorical values ,5. In some cases, in order to decrease the number of low expected values, I grouped similar variables and excluded small-sample categories. Because of low sample sizes, I excluded some studies from some statistical comparisons: studies in the humid tropical domain (N 5 1) from ecoregion domain tests; comparison studies that addressed the five variables relative to nesting season habitat (N 5 4) from historic vs. expanded range tests; and snail and slug (Gastropoda) prey (N 5 16) from seasonal diet analyses.

Habitat.
A total of 114 studies reported habitat use by Barred Owls (Appendix), of which 35 (30.7%) were single-sample description studies. The studies were located in 4 ecoregion domains, 16 divisions, and 22 provinces, ranging from the mixed forests of Nova Scotia to the redwood (Sequoia sempervirens) forests of California, and from the open woodlands, shrubs, and savannahs of southern Florida to the mixed forests of southeast Alaska (Fig. 1, Appendix).
Nesting habitat. Studies documented 466 nest sites and 1411 sites during the nesting season (Appendix). Barred Owls showed preference for areas of low elevation and flatter slope in the deciduous forests of Connecticut (Yannielli 1988) and coniferous forests of Alberta (Piorecky 2003) and Washington (Pearson andLivezey 2003, Singleton et. al 2005 ;   Table 1). However, elevation and slope were nonsignificant variables in the eastern deciduous forests of Maryland (Devereux 1982) and Connecticut (Falk 1990) and the boreal forests of Manitoba (Hinam 2001; Table 1). Two studies (Nicholls andWarner 1972, Fuller 1979) that showed preference for upland oak (Quercus) forests were conducted in the same study area in Minnesota. Some Barred Owls in one of these studies frequently used alder (Alnus spp.) lowland (Fuller 1979; recorded as 0.5 low and 0.5 high elevation in Table 1); the author of this study suggested that these somewhat contradictory results may have been due to differences in sampling intervals and lack of alder lowlands in the home ranges of the Barred Owls in the other Minnesota study.
Of the studies that simply described elevation or slope, 11 of 13 described Barred Owl use of lowelevation areas, and seven of eight studies reported use of forests on terrain with low slope (Table 1). Relative to sites of Northern Spotted Owls, Barred Owl sites were significantly lower in elevation in the western Cascade Mountains of Washington (Pearson and Livezey 2003, Hamer pers. comm.) and were significantly lower in elevation and flatter in slope in the Olympic Mountains of Washington (Gremel 2005; Table 1).
Barred Owls during the nesting season preferred sites near water in seven of 15 of the preference studies that tested that variable (Table 1). These seven studies were conducted in deciduous forests of Connecticut (Yannielli 1988), mixed forests of Michigan (Elody and Sloan 1985), oak and palm (Sabal palmetto) woodlands of Florida (Franz 1992), boreal forests of Manitoba (Hinam 2001), and coniferous forests of Alberta (Olsen 1999, Piorecky 2003, Takats 1998; Table 1). Two studies (Fuller 1979, Mazur et al. 1997a) provided differing preference results for nesting habitat relative to proximity to water: (1) some radio-tagged Barred Owls in the deciduous forests of Minnesota preferred oak uplands, while others preferred mixed deciduous lowlands (Fuller 1979); (2) in the boreal forests of Saskatchewan, 3.0-km-radius circles centered on Barred Owl locations contained significantly more water than those of random locations, but 1.5-km-circles did not (Mazur et al. 1997a). Two preference studies showed avoidance of wetlands because, suggested the researchers, the forested wetlands of those areas were too dense for owl flight (Nicholls andWarner 1972, Van Ael 1996). Proximity to water was not a significant vari-able in four other preference studies (Devereux 1982, Falk 1990, Mazur 1997. In contrast, almost all descriptions that reported proximity of sites to water noted that Barred Owl sites were situated close to water (96.9%, Table 1). In comparison studies, Barred Owl sites were significantly closer to water than those of Great Horned Owls (Bubo virginianus; Bosakowski et al. 1987) and Northern Spotted Owls (Herter and Hicks 2000; Table 1).
Mixed coniferous/deciduous forests and deciduous forests were used by Barred Owls approximately equally during the nesting season in preference studies, while coniferous forests were used less than one-half as often as these other forests (Table 1). Most description studies were located in mixed forests (Table 1). In a comparison study, Barred Owl sites contained more deciduous forest than did sites typically used by Northern Spotted Owls (Herter and Hicks 2000).
Mature or old forests were used by Barred Owls during the nesting season significantly more than young forests in all 13 preference studies that analyzed that variable (Table 1). In addition, mature or old forests were used by Barred Owls in 27 of 30 (90.0%) descriptions that mentioned forest age (Table 1). Of the three studies in which Barred Owls used younger forests, two described large remnant individual trees or stands of trees present (Duncan and Kearns 1997, J. Buchanan pers. comm.); in the third study, nest boxes in young forests were used (Elderkin 1987). In comparisons, Barred Owl sites in the Washington Cascades contained significantly less old forest within 0.8 km of site-centers than did those of Northern Spotted Owls (Herter and Hicks 2000) and site-centers of Barred Owls were located in forest stands that were significantly younger than those of Northern Spotted Owls (Pearson and Livezey 2003). Also, Barred Owl sites in the mixed forests of New Jersey contained significantly more mature forest than did those of Eastern Screech-Owls (Otus asio; Bosakowski et al. 1987). Low expected values prohibited comparisons of elevation, slope, proximity to water, forest composition, and forest age (Table 1) among the three ecoregion domains and between eastern and western studies in preference and description studies.
close to openings, Devereux 1982; more edge, more forest patches, more small forest patches, Olsen 1999), negative in two studies (less edge, Yannielli 1988, Van Ael 1996, and not significant in two studies (amount of fragmentation, Hinam 2001, Hamer pers. comm.). Analyses of proximity to human development were both positive (closer to trails) and negative (fewer close buildings) in one study (Yannielli 1988) and not significant in another (distance to farmland, McGarigal and Fraser 1984). Barred Owls nested in areas ranging from small suburban woodlots (Yannielli 1988, Harrold 2003, Mason 2004 to wilderness areas (Wright and Hayward 1998) and national parks (North Cascades National Park, Kuntz and Christophersen 1996;Olympic National Park, Gremel 2005; Mt. Rainier National Park, J. Schaberl pers. comm.). In a comparison study, Barred Owl sites in New Jersey had significantly fewer clearings and were significantly farther from human habitations than were those of Great Horned Owls and Eastern Screech-Owls (Bosakowski et al. 1987). Similarly, another comparison study indicated that when numbers of large deciduous woodlots with hollow trees decreased in southeastern Michigan, Great Horned Owls replaced Barred Owls (Craighead and Craighead 1969). Canopy closures of nest sites and sites during the nesting season were significantly higher than expected in three of the five preference studies that tested that variable (67-80% in Takats 1998;$70% in Hinam 2001;$70% in Singleton et. al 2005); those in the other two preference studies were 68% (Devereux 1982) and 51-70% (Piorecky 2003). In descriptions, canopy closures of sites of the deciduous forests of Manitoba were .71% (Duncan and Kearns 1997), those of the marineclimate mountains of British Columbia and Washington were 50-60% (J. Hobbs pers. comm.), 50-70% (R. Christophersen pers. comm.), 70.6% (Buchanan et al. 2004), .70% (J. Schaberl pers. comm.), and 86.0 6 1.7% (Gremel 2005), and those of 30 of 34 sites in the redwood forests of California were 85-90% (L. Diller pers. comm.).
Barred Owl nests in natural locations (N 5 341) were in cavities and other locations (Table 2). Stick nests occupied by Barred Owls were attributed to Red-shouldered Hawks (Buteo lineatus) or Cooper's Hawks (Accipter cooperii, N 5 18), Red-shouldered Hawks (N 5 8), Red-tailed Hawks (Buteo jamaicensis, N 5 4), Northern Goshawks (Accipiter gentilis, N 5 3), Broad-winged Hawk (Buteo platypterus, N 5 1), and Cooper's Hawk (N 5 1; Bent 1938, Johnson and Follen 1984, Shupe 1985, Postupalsky et al. 1997, Buchanan et al. 2004, J. Hobbs pers. comm.; nine of 44 nests in Table 4 were not attributed to any species). Two pairs of Barred Owls nested on the ground (Robertson 1959, Postupalsky et al. 1997, and one pair nested in a creek bank (Shackelford and Earley 1996). Comparison of seven locations of natural nests (Table 2) across ecoregion domains had prohibitively low expected values; after excluding two small-sample categories (mistletoe clump; on ground or in streambank hole) and combining two sets of similar categories (top of hollow or broken-topped tree with tree fork or split limb; hawk nest with squirrel nest), expected values were still too low. Further exclusion of the domain (Polar) with the smallest number of nests (5.7%) yielded adequate expected values, but no significant difference in the distribution of the three combined groups of natural nest locations in humid temperate vs. dry domains (x 2 2 5 3.17, P . 0.10). Between eastern and western studies, low expected values prohibited testing of the seven locations of natural nests. After excluding and grouping categories as described above, there were no significant differences among the three groups of natural nest locations (x 2 2 5 3.16, P . 0.10).
Foraging habitat. Foraging sites of Barred Owls in two studies conducted in the same study area in Minnesota were higher-elevation oak forests due, apparently, to the openness of these stands that permitted flight between trees, the many available hunting perches, and the concealing cover (Nicholls andWarner 1972, Fuller 1979; Table 1). Barred Owls in the earlier study did not routinely hunt in lower-elevation forested swamps and ponds in deciduous forests, concluded the researchers, because the dense understory and herbaceous plants on the forest floor there limited their ability to fly between trees and hunt prey in swamps (Nicholls and Warner 1972). Some Barred Owls in the later study frequently hunted in alder lowlands and along field-forest edges (Fuller 1979). In Alberta, foraging stands had lower mean canopy closure than that in nesting or roosting stands, slightly lower mean dbh than surrounding stands, and significantly less shrub/herb cover than surrounding stands (Takats 1998). Nine of 11 foraging perches were in Populus, one was in white spruce snag, and one was a humanmade post (Takats 1998).
Home-range sizes and site densities. Nine studies reported home-range data (Table 5). Four studies estimated home-range sizes using the same metric and recorded data throughout a full breeding season, nonbreeding season, or year. Mean home-range sizes in these studies, all of which were in the west, were 256.7 ha for breeding-season males (N 5 24), 297.8 ha for breeding-season females (N 5 25), 900.4 ha for nonbreeding-season males (N 5 12), and 536.2 ha for nonbreeding-season females (N 5 17; Table 5 The nest-cavity heights were significantly higher than (3 m higher than) random cavities in Devereux (1982) and Devereux and Mosher (1984 Blakemore (1940) also reported prey in 81 stomachs by volume (not by individual prey): rodents (44.0%), insects (35.4%), passerines (9.8%), arthropods (7.2%), amphibians and reptiles (2.4%), and poultry (1.2%). c Errington and McDonald (1937) also described by season, but did not quantify, the types of prey found in pellets (N 5 305).
documented 7077 individual prey items via direct observations, camera-documented prey deliveries to a nest, intact prey at nests, prey remains at nests, stomach contents, and pellets (Table 7). More than 75% of all samples were pellets (Table 8). The distribution of prey items by taxonomic group ob-tained from pellets differed significantly from those obtained by other means (x 2 8 5 414.76, P , 0.001, Table 7, 8), due especially to more earthworms in non-pellet samples than in pellet samples. Among ecoregion domains, the distributions of the six types of prey samples (Table 7) differed sig- nificantly (x 2 10 5 521.14, P , 0.001) largely due to the high number of prey remains at nests in the humid temperate domain (17.6%) vs. polar (0%) and dry (0.8%) domains. The distributions of the eight taxonomic groups of prey (Table 7) also differed significantly among ecoregion domains (x 2 14 5 1824.39, P , 0.001) as a result of greater numbers of amphibians in the polar domain (62.8%, primarily from a single study, Mazur et al. 1997c) than in the humid temperate (4.5%) and dry (2.8%) domains, and correspondingly fewer mammals in polar studies.
Comparing the historic vs. expanded ranges, the distributions of the six types of prey samples differed significantly (x 2 5 5 702.16, P , 0.001) due to high numbers of pellets in the west (97.9%) vs. the east (70.1%) and high number of prey remains at nests in the east (18.2%) vs. the west (0.7%). The distributions of taxonomic groups of prey also differed between eastern and western studies (x 2 7 5 378.36, P , 0.001), with more amphibians (10.5%) in the western regions than in the eastern (4.4%) and fewer crayfish in the western region (0% vs. 3.4% in the east).
Barred Owls in some areas during some years ate large amounts of fish, salamanders, frogs, crayfish, and earthworms (e.g., Errington and McDonald 1937, Bosakowski and Smith 1992, Hamer et al. 2001. For example, of the 231 prey items identified by pellets or prey remains in a study in Saskatchewan, 147 (63.6%) were frogs (Mazur et al. 1997c). During a 20-yr study in Missouri, annual diet volume of crayfish was 0-31.1%, that of fish was 0-8.4%, and that of insects was 0-7.6%; during the only yr in which salamanders were eaten, they comprised 4.3% of diet volume (Korschgen and Stuart 1972). In Nova Scotia, a female Barred Owl fed earthworms to her young at least once every 10 min during rainy nights (Elderkin 1987). DISCUSSION Although this review is not an attempt to map the complete distribution of Barred Owls, the locations of the studies included herein (Fig. 1) do approximate the entire distribution of the many thousands of mapped Barred Owl detections throughout their historic and expanded ranges (T. Fleming and K. Livezey unpubl. data). This synthesis is not a result of a stratified random sampling of habitats or prey throughout the range of the Barred Owl. However, from the data presented here, it is evident that Barred Owls survive and reproduce in most of the mature or older deciduous, coniferous, or mixed forests of North America, from those in remote, undisturbed areas to those close to human development with high amounts of fragmentation. Barred Owls use similar locations for natural nests throughout North America, and prey almost exclusively on small mammals in the winter and on a wide variety of animals including mammals, birds, reptiles, amphibians, and invertebrates during non-winter months.
In the west, available evidence indicates that Barred Owls prefer forests located in lower, flatter areas. In the mountainous forests of southern Alberta, southern British Columbia, western Washington, western Oregon, and northern California, Barred Owls used sites during the nesting season in low and/or flat areas in all three preference studies, all eight description studies, and all three comparison studies (relative to Northern Spotted Owls) that addressed these variables. A long-term study in Olympic National Park, an area largely free from human influences including timber harvest, demonstrated this preference (Gremel 2005). Early sightings of Barred Owls in various portions of the Olympic peninsula and park from 1985-1990 were in floodplain forests, near lakes, or in riparian areas (Gremel 2005). Through time, Barred Owls increased their numbers while extending their distribution farther up valleys and higher in elevation (Gremel 2005). Mean elevation of Barred Owl sitecenters significantly increased from 1987-2003 (r 2 5 0.57, N 5 17, P , 0.001, S. Gremel pers. comm.). Number of Barred Owl site-centers below 343 m in elevation increased from two in 1987 to 13 in 2006 whereas those above that elevation increased from zero in to 20 in 2006.
The Barred Owl studies presented mixed results concerning proximity to water. The reason for this may be that definitions of ''water'' or ''riparian areas'' and distances required to be ''near'' these areas differed among studies, and what meaningfully constitutes a wetland for Barred Owls might differ from that for other species and possibly differs for Barred Owls across the range of habitats in North America. Consequently, it seems reasonable for investigators to undertake exploratory analyses of the statistically explanatory effects of ''water'' by analyzing the power of alternative definitions (e.g., ephemeral, seasonal, or permanent bodies) using residuals after removing variation attributable to directly measurable variables. Such explorations in which ''water'' was treated a posteriori could then identify optimally informative definitions of wetlands for Barred Owls in various parts of the species' range and be used in subsequent analyses no differently than directly measured variables.
Use of wetland or lowland areas by Barred Owls in some areas may be related to availability of amphibian and fish prey (Yannielli 1988, Buchanan et al. 2004, and to larger nest trees that are often found in these areas because of decreased frequency of fires (Hinam 2001, Hinam andDuncan 2002) and faster growth rates of trees. Frequent use of black cottonwood (Populus trichocarpa) and balsam poplar as nest trees may be due to the presence of these species in moist, lowland areas. Cottonwoods are the fastest-growing native tree in North America (Elias 1980, Burns andHonkala 1990); they are early successional species that rapidly grow large trunks (Millet et al. 1999), are relatively short-lived, and have weak wood, so they frequently produce cavities and broken limbs that provide nesting opportunities for Barred Owls.
According to the studies reviewed here, Barred Owls apparently preferred mature or older forests, possibly because of greater availability of nest sites, lower stem densities in the understory that allow unimpeded visibility and travelways for foraging, and dense canopies that provide thermally neutral microclimates and protection from mobbing birds (Haney 1997). Barred Owls used young forests when nest boxes were provided (Elderkin 1987). Because Barred Owls are generalist predators, habitat selection may be influenced more by prey availability than by strong affinity for any specific type of forested habitat (Olsen 1999). The greater use of coniferous trees for nests in the west probably merely reflects the greater availability of coniferous forest there, and demonstrates the ability of Barred Owls to use different forests in their expanded range. Additional research in fragmented, younger forests, such as those done in suburban studies in North Carolina (Harrold 2003, Mason 2004, would provide a more complete understanding of the use of such habitats by Barred Owls. The lack of significant differences among the natural nest locations in studies in humid temperate vs. dry domains and eastern vs. western regions suggested that Barred Owls used similar nest locations throughout North America. The vast majority of the natural nests (.75%) were placed in the most productive locations: cavities. In Michigan, Barred Owls that nested in cavities (N 5 49) and boxes (N 5 52) produced young 80% of the time, whereas only 31% of breeding attempts in hawk nests and other open sites (N 5 13) were successful (Postupalsky et al. 1997).
Breeding-season home ranges of adult Barred Owls averaged only 250-300 ha, and densities ranged as high as 0.21-0.25 sites/km 2 . Annual home ranges of sympatric Northern Spotted Owls were 3-4 times larger than those of Barred Owls in the western Cascade Mountains of Washington (Hamer pers. comm.), probably due to the more-varied prey base of Barred Owls (Hamer et al. 2001, Forsman et al. 2004. Barred Owls' diet consisted almost exclusively (98-99%) of small mammals during winter months, but included other types of prey (41-43%) as available during other times of the year. The differences in distributions of prey types among ecoregion domains and between east and west were significant primarily because of a few studies with large numbers of amphibians in pellets (Mazur et al. 1997c), crayfish in pellets (Korschgen and Stuart 1972), or earthworms documented by camera (Elderkin 1987) or observations ( J. Hobbs. pers. comm.). These studies provided important data on the varied diet of Barred Owls and underscored the need for additional research to thoroughly understand the Barred Owl diet in North America throughout the year.
Different methods used to document Barred Owl diet yielded markedly different results. For example, 97.3% of the 73 records of Barred Owls eating earthworms were from camera-documented food deliveries or visual observations; only 2.7% were from pellets. The amount of soft-bodied prey such as earthworms eaten by Barred Owls probably was greatly underestimated in the pellet studies presented here due to the rapidity with which such prey decomposes and because such prey do not contain durable materials such as bones, feathers, hair, and exoskeletons necessary in pellet analysis. Because diet information gathered through pellet analysis can be biased toward some prey groups, such data should be used with data obtained through other, complementary methods (Yom-Tov and Wool 1997, Marchesi et al. 2002, Bonvicino and Bezerra 2003, Torre et al. 2004. Barred Owls began their range expansion westward beginning more than 100 yr ago according to early reports (T. Fleming and K. Livezey unpubl. data). The range expansion apparently was facilitated by increased distribution of trees in the northern Great Plains (Maini 1960, Moore 1972, Bragg and Hulbert 1976, Grant and Murphy 2005 brought about by exclusion of fires historically set by Native Americans (Sauer 1950, Lewis 1980, Pyne 1983, Higgins 1986, Kimmerer and Lake 2001, suppression of fires (Arno 1980, Pyne 1982, Williams 1989, planting of trees (Droze 1977, Williams 1989, and other factors (T. Fleming and K. Livezey unpubl. data). Barred Owls have approximately doubled the size of their range during this time, and are now living in forests and preying on wildlife of central and western North America. Thus, there are abundant opportunities to study their use of various habitats and prey, as well as their effects on other raptors including Northern Spotted Owls (Kelly et al. 2003, Pearson and Livezey 2003, Gremel 2005, Olson et al. 2005 and Western Screech-Owls (Otus kennicottii; Elliott 2006), especially if they increase their densities in newly colonized areas and further expand their range. Appendix. Continued.