factors associated with habitat use by Timberline Sparrows (Spizella (breweri) taverneri) the southern end of their range?

Suzanne A. Cox

The University of Montana IBS-CORE Program

Back to Abstract

Introduction

While most high-elevation habitats in the United States are not threatened by direct human encroachment, anthropogenic effects on alpine and sub-alpine regions can be expected to be dramatic in the coming years. Vegetation at ecotones can show a dramatic and rapid response to climate change (Gates 1990, Peteet 2000). High elevation sites in the Northern Rockies are already experiencing warming temperatures (Inouye et al. 2000) and tree-ring dendrochronologies show a corresponding increase in recruitment of young trees at and above tree-line (Szeicz and MacDonald 1996, reviewed in Lloyd and Graumlich 1997), along with a tendency for trees that were previously growing as krummholz to develop an upright growth form (reviewed in Lloyd and Graumlich 1997). Any upward shift in treeline will result in a reduction of area available to organisms that depend on regions above the limit of upright timber. Additionally, white-pine blister-rust (Cronartium ribicola) is eliminating whitebark pine from large areas of the sub-alpine forest (Arno 1990); this may have effects on recruitment of subalpine fir and other species that live at the upper limit of tree growth. Species at the edges of their ranges, and those with very restricted ranges, can be expected to suffer from the apparently inevitable habitat alteration resulting from climate change and disease.

Currently, efforts are underway to form a continuous corridor of protected wildlands along the backbone of the Rocky Mountains from Yellowstone National Park in Wyoming to the Yukon region of northern Canada (Y2Y). While political, social and economic realities will play a large part in determining the final form of this corridor, biological considerations, including the maintenance of species and ecosystems in the face of these changes, are strongly driving this project. It is vitally important, therefore, that as much information as possible be gathered concerning the habitat requirements and distribution of organisms which might be affected by the Y2Y corridor.

The Timberline Sparrow (Spizella [breweri] taverneii) is one species that depends on high-elevation habitats in the Northern Rockies. The habitat requirements of this bird remain poorly understood and the full extent of its distribution is almost certainly unknown.

The Timberline Sparrow is known to breed at scattered high-elevation sites between southeastern Alberta and Alaska, but it is not known whether this distribution is continuous, as few surveys have been conducted in the northern Canadian Rockies (Rotenberry et al. 1999). In 1998, Walker first documented breeding south of Canada. His reports of Timberline Sparrows breeding at numerous locations on the east side of Glacier National Park (Walker 2000), and Cox’s (2000) confirmation of Timberline Sparrow presence at five sites in the Lewis and Clark National Forest, south of Glacier, indicate that the southern limit of the range of this bird could extend much farther south. It is likely that this bird has been overlooked in the past, as general patterns of range expansions since the end of the Pleistocene (i.e. Sage and Wolff 1986 in Luikart and Allendorf 1996) provide evidence against a recent southward range expansion.

The breeding habitat of this bird is known to vary somewhat throughout its range. It is generally reported to breed in patchy dwarf or krummholtz vegetation on east-, south- and west-facing slopes (McTaggart-Cowan 1946, Walker 2000). It has been reported to use subalpine fir (Abies lasiocarpa) (Swarth 1926, McTaggart-Cowan 1946, B. McGillivray, in Doyle 1997, Doyle 1997, Walker 2000), dwarf spruce (Picea spp.), dwarf birch (Betula nana) (Swarth 1926, B. McGillivray, in Doyle 1997), stunted willow (Salix spp.) (Doyle 1997), and alder (Alnus spp.) (Cox et al. 2000). While Doyle (1997) and Walker (2000) indicate that they consistently found Timberline Sparrows at all sites contianing apparently suitable habitat, this was not true at Cox’s sites. Numerous reasons for this discrepancy can be suggested (Cox et al. 2000), but habitat preferences that are narrower than general vegetation types are perhaps the most likely.

The first goal of this research was to clarify the local-scale vegetation structure and composition used by Timberline Sparrows at the southern end of their known range. They are known to breed only in krummholz and other shrubby vegetation (Swarth 1926, McTaggart-Cowan 1946, B. McGillivray, in Doyle 1997, Doyle 1997, Cox et al. 2000, Walker 2000), but are not consistently found in this habitat type, so I specifically wanted to determine what vegetation characteristics they are associated with within the krummholz. The second goal is to develop a predictive GIS model as an aid in determining the full extent of the range of this bird. As a by-product of the fieldwork, I hoped to locate additional local populations of Timberline Sparrows south of Glacier National Park. In this paper, I report on the results of the analysis of vegetation characteristics associated with Timberline Sparrows.

Methods

Study Area: The study area was located on the east side of the Continental Divide in Glacier National Park and the Lewis and Clark National Forest (47˚ 54’ - 48˚ 55’N, 112˚ 41’ -113˚ 39’W). This is the only region in the contiguous United States that Timberline Sparrows have been confirmed breeding.

Site Selection: Seven pairs of sites were chosen from Timberline Sparrow locations identified in 1998 and 1999, with three additional pairs located by surveying areas chosen from topographical maps. A pair of sites included a used (presence) site and an unused (absence) site with roughly similar abiotic parameters. The intent of these pairings was to control for the effect of abiotic factors such as climate, aspect and elevation on vegetation, so pairs were sampled within three days of each other. Wherever possible these were located adjacent to each other or in the same drainage; in drainages where Timberline Sparrows were found in virtually all potential habitat, unused habitat in another drainage was used to complete the pair. At the end of the season, one unpaired used site was sampled. The birds were singing so infrequently by then that I could not have confirmed absence at an unused sight.

Sites were chosen in an effort to sample the full known range of latitude, elevation, slope, aspect, and vegetation structure in which Timberline Sparrows have been found in Montana. Potential habitat was defined as west-, south-, or east-facing slopes, above 1700 m, with a non-continuous cover of stunted, skirted or krummholtz trees, generally not exceeding 3 m in height. I recognize that this non-random method of site selection limits any inferences to the sites sampled. However, due to the rugged nature of the terrain, the remoteness of many sites, and the limited number of documented sites, it was not possible to select sites randomly from all potential Timberline Sparrow habitat.

Definition of used and unused areas: The boundaries of each site were visually assessed and encompassed the area that met the requirements as stated above. The UTM coordinates of a point within 10 m of the perch of any Timberline Sparrow heard singing or seen on the site was recorded with a Garmin 12XL GPS unit. Playbacks of male Timberline Sparrow long songs were used throughout any parts of the site where Timberline Sparrows were not heard singing. Playback of a male Timberline Sparrow on another’s territory will often cause the possessor of the territory to sing or perch in view. The locations of any birds thus located were marked as well.

Used areas were defined as all areas within 50 m of a point where a Timberline Sparrow had been recorded (fig. 2). Towards the end of the field season, as the birds sang more infrequently and were less responsive to playback, birds were sometimes missed in the original survey; if heard singing later in the day, the used area was extended. Unused areas were defined as habitat that met the above requirements, but where Timberline Sparrows were not found. The unused site was bounded at least 150 m from any used point, and playbacks were performed in unused areas repeatedly during vegetation sampling as well to be absolutely certain that there were no Timberline Sparrows present. The field season was ended when detection of the birds became so difficult that there was a reasonable probability of missing birds, and thus, misclassifying a used area as unused.

Vegetation sampling: In each used and unused area the vegetation was sampled from points spaced at 50 m (70 m in very large sites) intervals on a systematic grid (fig. 1). An 11.3-m-radius circular plot (0.04-ha), divided into four quadrants, was laid out at each point. One of the four radii was laid out directly toward the sun to avoid any human induced bias in the positioning of the lines. The average height of the trees in the plot was estimated (see table 3 for variable names to be referred to hereafter). In each quadrant, the distance to the nearest tree over .5 m high, and its height was recorded. The width of the tree, or cluster of overlapping trees, at the longest axis was recorded, as well as the length of the axis perpendicular to the first axis. These were multiplied to get an approximate area for the cluster. This was repeated for shrubs, with a minimum shrub height set at .3 m. Trees were defined as woody plants that are single-stemmed and grow to over 10 m in height under normal conditions; and shrubs were defined as woody plants that are generally multi-stemmed .5 m above the ground. Ground cover was recorded at 1 m intervals along each radius for a total of forty points per plot. Ground cover categories included eight genera of shrubs, six species of tree, herbaceous plants (included forbs, grasses, and arctostaphyllus), bare ground, rock or scree, and debris (primarily dead wood). Additionally, each plot location was marked in the GPS-unit to allow addition of slope, aspect, elevation and other information to the field data, as well as for use in future GIS modeling.

Analysis: All analyses were conducted using SPSS version 10. For each plot, means for the four quadrants were calculated for the TRDST, TRHT, and TRAREA. This was repeated for shrubs (SHDST, SHST, and SHAREA). From the ground cover points, a percent of ground cover was calculated for 18 categories (table 3) on each plot. Two additional variables (FIR and PINE) were created by combining appropriate variables (table 3). Values of each variable were averaged for each site.

A three-step process was used to eliminate non-significant variables. First, any variable that did not occur on at least 5 used or 5 unused sites was considered to be biologically insignificant and omitted from further analysis. Second, Pearson’s correlation coefficients between all the remaining variables were examined. Based on the results and biological interpretation, I dropped some highly correlated variables. Finally, I fit individual logistic regressions to all remaining variables (table 4), and dropped variables that had significance level > 0.02 for exp(β).

A principal components analysis (PCA) was run on the nine remaining variables to look for broad patterns. Finally, I fit a logistic regression function with all remaining variables entered using both the forward likelihood ratio and the backwards likelihood ratio methods of variable entry. Because the number of cases used to build the model was so small, I used very conservative probabilities to enter (P = 0.05) and remove variables (P = 0.10). The resulting model with the fewest parameters was cross-validated by withholding 2 randomly selected cases (sites) and refitting the model with the same variables to the remaining 19 cases. This was repeated 100 times.

RESULTS

Sites: I sampled a total of ten unused and eleven used sites (table 1). I also survey three sites where birds had been found in the past (Sheep Creek, Crazy Creek; Cox et al. 2000 and Peigan Pass; Walker 2000) but where no birds were found this year. These were not sampled or used in the analysis. Birds were found at three new sites that had not previously been surveyed (fm, mw, pg; see table 1 for site abbreviations). The 21 sites that were sampled spanned the full latitude and longitude of the known range of the Timberline Sparrow south of Canada (figure 2). The distribution of sites with respect to elevation, slope and aspect was very similar for used and unused sites (table 2). Elevation, slope and aspect were uncorrelated (r < 0.3, P > 0.2 for all two-way correlations). Because the loose rock and scree made slopes much above 70% dangerous to work on, I was unable to sample the full range of slopes occupied by Timberline Sparrows. It also is possible that my choice of unused habitat was biased towards shorter trees, as the value for TREE_HEI was lower for every unused site than for the corresponding used site.

Analysis: Ten variables (ALDER, AMAL, ARTR, LONIC, PAMY, VAME, PIFL, PICO, POTR, and SNOW) were removed from analysis because they occurred on plots at <5 used or <5 unused sites. TREE_HEI and TRHT were highly correlated (r = 0.895, P < 0.0005), as were ABLA and FIR (r = 0.780, P < 0.0005) and ABLA and PSME (r = -0.605, P = 0.004), so TRHT, ABLA and PSME were dropped from the analysis. TREE_HEI was felt to be more reflective of the average height of the trees because trees at the edge of a cluster were often younger and smaller than the average tree on a plot. ABLA and PSME were combined into into FIR as the two are similar in structure and Timberline Sparrows use both for nesting (Cox et al. 2000, Walker 2000). Several other variables were somewhat strongly correlated but the choice of which to eliminate was less obvious, so I entered them all into the univariate logistic regressions. Nine variables had significance levels for exp(β) < 0.02 (table 4). Of these, PERTR and FIR were highly correlated (r = 0.869, P < 0.0005), apparently because fir is so dominate at the sites that it is the primary determinate of tree cover. FIR was less highly correlated with other variables than PERTR, so I kept FIR and dropped PERTR.

The 1^st principal component (figure 2) accounted for 34.5% of the variation in the data and was heavily weighted on one end by TRDST, SHDST and HERB, and the other by TREE_HEI and FIR. The 2^nd principal component accounted for 25.2% of the variation and was weighted on one end by PIEN and the other by TRAREA. A logistic regression model fit using the forward likelihood ratio method of variable entry produced a model with three parameters, PIEN, HERB and FIR, which correctly classified 100% of the data (table 5). The error rate remained 0% with cross-validation. Entering the variables using a backward likelihood ratio method produced a model with four different parameters (FIR, TRAREA, TRDST, and SHDST) that also correctly classified 100% of the data.

DISCUSSION

Although there are limitations to the inferences that can be drawn from this study, I was able to identify habitat variables that further narrow the definition of Timberline Sparrow habitat presented by Walker (2000). The PCA identified broad relationships of habitat characteristics that are associated with Timberline Sparrows (fig 3). These are high tree and shrub density, large patches of trees, taller trees, and a high proportion of fir. Spruce and herbs were not used. If you superimpose quadrants on the scatterplot of the variables on the 1^st and 2^nd principal components, Timberline Sparrows occupied all sites (with the possible exception of ac) that fall in the upper-left quadrant (tall dense trees, fir, dense shrubs, and larger cluster size). No site in the lower-right quadrant was used. Inferences about the relative significance of composition, structure and distribution of vegetation can be made through the results of the logistic regression models.

Vegetation composition was the most important predictor of Timberline Sparrows at my sites. FIR was used in both multivariate models, was highly significant in the univariate model, and Timberline Sparrows were never found at a site where less than 12% of the ground cover was fir. The two firs present on the sites, Douglas and subalpine, while taxonomically different enough to be placed in separate genera, are similar enough in structure that they appear to serve the same biological purpose for the birds. Indeed, in the absence of cones, which are relatively rare in krummholz, I had difficulty distinguishing between the two taxa. PIEN and HERB also were used in the most parsimonious multivariate model. However, HERB has a moderate negative correlation with FIR (r = -0.558, P = 0.009), possibly accounting for some of its apparent importance. There were two surprises in regards to composition. First, contrary to earlier observations (Cox et al. 2000, Walker 2000), pine was not negatively associated with sparrow presence, and the amount of pine on the used and unused sites was not statistically different. Whitebark pine, the dominant high-elevation pine in the area, is a superior pioneer, often serves as a nurse plant for subalpine fir (Callaway 1998) and thus, is a common component of many fir krummholz stands in the region (Arno and Weaver 1989). Second, spruce was negatively associated with presence; this was unexpected as spruce is used by Timberline Sparrows in some parts of their range (Swarth 1926, B. McGillivray, in Doyle 1997). PERSH was not a particularly strong predictor but there was a higher shrub component at the used sites.

Structural variables appear to be less significant to the birds, although they do have some predictive power. Tree height was a predictor of sparrow presence in this dataset, but as some bias might have been introduced in site selection, caution should be used in drawing conclusions about its importance to the birds. It can be concluded that the upper limit of tree height acceptable to Timberline Sparrows is greater than the 2 m suggested by Walker (2000). Almost all the used sites had one or more plots with trees > 2 m tall, and trees at jc and ol averaged greater than 2 m and in some instances were as tall as 4 m. TRAREA and SHAREA give a measure of size of the patch of trees or shrub. Although TRAREA was not significant in the univariate logistic regression, it did have some predictive power in the PCA and multivariate logistic regression, although it is difficult to say if the important feature to the birds is large groups of trees or a correlated higher tree density. The largest and smallest values for TRAREA occurred at used sites, indicating that as measured, that variable alone cannot adequately explain Timberline Sparrow site use. SHDST and TRDST were a measure of the distribution of vegetation, and were the most heavily weighted variables in the 1^st principal component although they were insignificant in the univariate regressions.

High elevation environments are harsh; the wind blows almost constantly at these sites and it snowed 12” at jc on June 1^st. I suggest that the Timberline Sparrow selects habitats so as to avoid the most extreme conditions that are found at some treeline sites. More than many communities, the patterns of vegetation at treeline are strongly influenced by microsite climate and topography (Arno and Weaver 1989, Peet 2000). On exposed ridges, wind can scour the ground of snow, and thus unprotected from desiccation, trees are unable to establish or persist (Peet 2000). In sheltered areas, particularly on north facing slopes, the persistence of snowpack into or through the growing season can also prevent tree establishment (Peet 2000). Between these two extremes, trees can grow in a variety of forms, from low-growing cushions only a few inches tall to fully erect (Arno and Weaver 1989, Peet 2000). Often they are found in clumps, a result of vegetative reproduction through layering or the tendency for establishment to occur in the shelter of a pioneer tree or a rock (Callaway 1998, Peet 2000). Somewhat correlated with wind and snowpack, although also influenced by slope and aspect, moisture availability strongly influences the species composition on a site. Spruce is associated with the wettest sites, subalpine fir can tolerate a range of conditions but is sometimes replaced by Douglas fir on drier sites, with whitebark and limber pine are able to tolerate the most xeric conditions (Arno and Weaver 1989).

The relatively tall krummholz and even upright trees in which I found Timberline Sparrows, combined with their absence in lower krummholz, indicates that they are not present at the windiest and most exposed sites. Large patches of trees and shrubs offer more foraging opportunities without changing patches (Rotenberry and Weins 1998), while patches in close proximity mean less exposure time to fly between patches. Fir krummholz is extremely dense; often it is virtually impenetrable. It is likely that fir provides superior protection from the wind and late season snow. The absence of Timberline Sparrows on north-facing slopes may also be indicative of selection for selection of relatively moderate environments.

Although predictions of regional climate change remain somewhat uncertain (Schneider 1993), models generally predict warmer winter temperatures and higher snowfall at high elevation, continental sites (e.g., Dickinson 1986). There are indications that these changes are, in fact, both real and currently occurring in the Rocky Mountains (Inouye et al. 2000). Temperature and snowfall changes will have significant effects on the vegetation in areas where Timberline Sparrows breed. The complex mosaic of topography and microclimate, results in different conditions, and different communities, at even a fine scale, it is difficult to predict exactly how changes will affect these birds. On the Olympic peninsula, subalpine fir was able to invade alpine meadows during periods of wetter than average weather (Woodward et al. 1995). If increased snowpack is not accompanied by warmer than usual summers, some areas that are currently too dry for seedling establishment may develop stands of krummholz fir, increasing available habitat for Timberline Sparrows. However, these sites may be too exposed for the birds. Conversely, there is evidence that some krummholz are beginning to develop a more upright posture (reviewed in Lloyd and Graumlich 1997). Timberline Sparrows could likely continue to use this taller fir for some time, but if the stands eventually grow to full-sized timber, they will become unusable. The interplay of white pine blister rust, climate, and the requirements of subalpine fir for a nurse-plant may affect the distribution of habitat across the landscape as well, possibly in ways we cannot foresee.

Acknowledgements

This work was funded in part by an IBS-CORE Undergraduate Research Fellowship to Suzanne Cox through a grant from the Howard Hughes Medical Institute to the University of Montana. Funding was also provided by a grant from the Five Valleys Audubon Society and a Watkins Scholarship. I would like to thank my advisor, Dr. R.L. Hutto for help in all areas. Dr. D. Patterson and Steve Hoekman helped with the statistical analysis and study design. Melissa Hart of the Wildlife Spatial Analysis Lab at the University of Montana provided the slope, aspect and elevation values based on my GPS points. Paul Griffin and Brett Walker provided moral support and general advice. Salem kept me company.

LITERATURE CITED

Arno, S.F. and Hammerly, R.P. 1990. Timberline: mountain and arctic forest frontiers. – The

Mountaineers, Seattle, WA.

Arno, S.F. and T. Weaver. 1989. Whitebark pine community types and their patterns on the

landscape. Proceedings – symposium on whitebark pine ecosystems: Ecology and

management of a high mountain resource. Bozeman, MT.

Callaway, R.M. 1998. Competition and facilitation on elevation gradient in subalpine forests of

the northern Rocky Mountains, USA. Oikos 82: 561-573.

Cox, S.A., B.L. Walker, and M. Hart. 2000. Using GIS to predict the distribution of Timberline Sparrows in northwestern Montana. Proceedings of the national conference on

undergraduate research.

Dickinson, R. 1986. The climate system and modelling of future climate change. Pages 207-270

in B. Bolin, B. Doos, J. Jager, and W.A. Warrick, editors. The greenhouse effect, climate

change, and ecosystems. Wiley, Chichester, UK.

Doyle, T.J. 1997. The Timberline Sparrow, Spizella (breweri) taverneri, in Alaska, with notes

on breeding habitat and vocalizations. Western Birds 28(1): 1-12.

Gates, D.M. 1990. Climate change and forests. Tree physiology 7:1-5.

Inouye, D.W., B. Barr, K.B. Armitage, and B.D. Inouye. 2000. Climate change is affecting

altitudinal migrants and hibernating species. Proceedings of the National Academy of

Science 97(4): 1630-1633.

Lloyd, A.H. and L.J. Graumlich. 1997. Holocene dynamics of treeline forests in the Sierra

Nevada. Ecology 78(4): 1999-1210.

Luikart, G. and F.W.Allendorf. 1996. Mitochondrial-DNA variation and genetic-population

structure in Rocky Mountain bighorn sheep (Ovis canadensis canadensis). Journal of

Mammology 77(1): 109-123.

McTaggart-Cowan, I. 1946. Notes on the distribution of Spizella breweri taverneri. Condor

48:93-94.

Peet, R.K. 2000. Forests and meadows of the Rocky Mountains. Pages 75-121 in M.G. Barbour

and D. Billings, editors. North American Terrestrial Vegetation 2^nd ed. Cambridge

University Press: Cambridge, UK.

Peteet, D. 2000. Sensitivity and rapidity of vegetational response to abrupt climate change.

Proceedings of the National Academy of Science 97(4): 1359-1361.

Rotenberry, J.T. and J.A Wiens. 1998. Foraging patch selection by shrubsteppe sparrows. Ecology 79: 1160-1173.

Rotenberry, J.T., M.A. Patten, and K.L. Preston. 1999. Brewer's Sparrow (Spizella breweri). In The Birds of North America, No. 390 (A. Pool and F. Gill, eds.). The Birds of North America, Inc., Philadelphia, PA.

Schneider, S.H. 1993. Scenarios for global warming. Pages 9-23 in Biotic Interactions and Global Change. P.M. Kareiva, J.G. Kingsolver, and R.B. Huey, eds. Sinauer Associates Inc.: Sunderland, Massachusetts.

Szeicz, L.M. and G.M. MacDonald. 1995. Recent white spruce dynamics at the sub-arctic alpine treeline of north-western Canada. Journal of Ecology 83: 873-885.

Swarth, H.S. 1926. Birds and Mammals of the Atlin region of southeast Alaska. University of California Publications in Zoology 30: 130-131.

Walker, B. 2000. The distribution, abundance, breeding status and identification of the Timberline Sparrows, Spizella (breweri) taverneri, in Glacier National Park, 1998-1999. Unpublished report to the National Park Service.

Woodward, A., E.G. Schreiner, and D.G. Silsbee. 2000. Climate, geography, and tree establishment in subalpine meadows of the Olympic Mountains, Washington, U.S.A. Arctic and alpine research 27(3): 217-225.

Table 1. Site information. Site name is given with site code in parentheses, date visited indicates day of vegetation sampling, status (A=Timberline Sparrows absent, P= Timberline Sparrows present), and means for the site for elevation, aspect, slope and vegetation variables.

Site	Date Visited	Status	Elevation (m)	Aspect ˚fromN	%Slope	TREE_HEI (m)	PIEN	HERB	FIR	PERSH	TRAREA (m^2)	TRDST (m)	SHDST (m)
Apikuni Creek (AC)	7/1	A	1838	219	37	1.7	.01	.30	.16	15.6	92.8	6.9	5.9
Apikuni Creek (AC)	6/30-7/1	P	1992	179	49	1.7	.00	.44	.12	9.4	48.9	10.0	5.3
Forty Mile Creek (FC)	6/5-6/6	A	1842	48	27	1.4	.01	.46	.08	12.1	60.2	10.9	11.9
Forty Mile Creek^a (FC)	6/5-6/6	P	1828	56	20	1.5	.01	.51	.12	18.2	224.3	11.6	11.4
Gable Mountain (GM)	7/3	A	2092	286	16	1.2	.03	.52	.09	17.6	61.3	7.9	7.6
Gable Mountain (GM)	7/2	P	2095	279	19	1.3	.01	.32	.23	11.8	258.0	7.2	9.5
Clary Coulee (CC)	6/4, (6/17^b)	A	2075	146	32	1.1	.00	.42	.09	15.3	24.3	8.9	9.2
Jones Creek (JC)	6/3, (6/1-6/2, 6/18^b)	P	1876	220	17	2.2	.00	.44	.20	17.9	64.2	6.9	2.7
Mount Baldy (MB)	7/8-7/9	A	2063	164	22	1.3	.03	.59	.15	2.5	124.4	7.5	20.0
Mount Baldy (MB)	7/9	P	1970	118	44	1.3	.03	.36	.24	21.0	122.9	3.9	3.6
Mount Wright^a (MW)	6/21-22	A	2298	120	50	1.7	.03	.42	.16	12.0	50.9	8.2	6.7
Mount Wright (MW)	(6/19^b), 6/21	P	2307	53	52	2.0	.00	.36	.16	5.9	306.6	9.1	8.4
Preston Park (PP)	6/26	A	2252	176	27	1.3	.00	.30	.15	4.2	83.2	12.5	21.6
Otokomi Lake (OL)	6/29	P	2156	134	63	2.0	.00	.22	.23	11.7	21.1	6.7	6.1
Mount Patrick Gass (PG)	6/12, (7/18^b)	A	2258	253	61	1.1	.02	.43	.02	6.7	7.9	7.7	8.3
Mount Patrick Gass^a (PG)	6/11	P	2198	244	51	1.3	.00	.42	.18	8.1	202.4	6.9	6.1
Sunrift Gorge (SG)	6/27	A	1745	240	46	1.3	.03	.67	.06	15.4	82.9	10.3	5.7
Sunrift Gorge (SG)	6/25	P	1922	235	56	1.6	.00	.49	.12	29.3	155.2	9.3	6.3

Table 1(cont).

Site	Date Visited	Status	Elevation (m)	Aspect ˚fromN	%Slope	TREE_HEI (m)	PIEN	HERB	FIR	PERSH	TRAREA (m^2)	TRDST (m)	SHDST (m)
Scenic Point (SP)	6/8	A	2182	116	22	1.4	.00	.63	.09	3.1	256.3	14.3	21.7
Scenic Point (SP)	6/7	P	2144	97	16	1.5	.00	.50	.21	5.8	467.7	8.1	16.4
Divide Mountain (DM)	7/14	P	2228	60	72	1.1	.01	.36	.19	19.7	77.9	5.9	4.4

^a New site in 2000.

^b Initial attempts to sample Jones Creek failed because of a snowstorm. Both Jones Creek and Clary Coulee were visited on additional days as part of another study. Mt.Wright was located two days prior to sampling. Two birds were banded on Mount Patrick Gass on the second visit.

Table 2. Description of elevation, slope, and aspect distributions for used and unused sites.

		Minimum	Maximum	Mean	Standard Deviation
Presence sites
	Elevation (m)	1829	2307	2065	156
	Aspect (˚ from N)	54	279	153	83
	% Slope	16	72	42	20
Absence sites
	Elevation (m)	1745	2298	2065	196
	Aspect (˚ from N)	49	287	177	73
	% Slope	16	62	35	14

Table 3. Identification of all variables measured and their means and standard deviations at used and unused sites. Variable function refers to the vegetation attribute that I was attempting to quantify. Test-statistics and P-value for independent-samples t-tests are shown.

	Variable function	Variable	PRES_ABS	Mean	Standard Deviation	t-statistic^a	Significance (2-tailed)
Estimated tree height on plot	Structure	TREE_HEI	A	1.340	.207	-1.965	.064
			P	1.586	.344
% alder (Alnus)	Composition	ALDER	A	.000	.001	-1.046	.309
			P	.011	.032
% serviceberry (Amelanchier)	Composition	AMAL	A	.004	.012	0.445	.661
			P	.003	.005
% sage (Artemisia)	Composition	ARTR	A	.002	.005	1.052	.306
			P	.000	.000
% Juniper (Juniperis)	Composition	JUNIPER	A	.038	.047	-0.920	.369
			P	.056	.042
% Twinflower (Lonicera)	Composition	LONIC	A	.001	.002	-1.170	.257
			P	.003	.007
% mountain boxwood (Paxistima)	Composition	PAMY	A	.000	.000	-0.951	.353
			P	.003	.009
% shrubby cinquefoil	Composition	PEFL	A	.035	.028	0.292	.773
(Pentaphylloides)			P	.032	.019
% ribes	Composition	RIBES	A	.003	.004	-0.520	.609
(ribes)			P	.004	.006
% willow	Composition	SALIX	A	.014	.027	-0.336	.741
(Salix)			P	.019	.048
% Canada buffaloberry	Composition	SHCA	A	.011	.012	0.829	.418
(Sheperdia)			P	.007	.010
% huckleberry	Composition	VAME	A	.002	.005	-1.091	.289
(Vaccinium)			P	.007	.014
% subalpine fir	Composition	ABLA	A	.090	.063	-2.022	.057
(Abies lasiocarpa)			P	.152	.077
% Douglas fir	Composition	PSME	A	.014	.028	-0.699	.493
(Psuedotsuga menziesii)			P	.029	.060
% whitebark pine	Composition	PIAL	A	.013	.018	0.095	.925
(Pinus albacaulis)			P	.012	.012
% spruce	Composition	PIEN	A	.015	.013	2.324	.031
(Picea englemannii)			P	.005	.008
% limber pine	Composition	PIFL	A	.009	.027	0.637	.532
(Pinus flexilis)			P	.003	.010
% lodgepole pine	Composition	PICO	A	.001	.002	1.052	.306
(Pinus contorta)			P	.000	.000

Table 3 (cont):

	Variable function	Variable	PRES_ABS	Mean	Standard Deviation	t-statistic^a	Significance (2-tailed)
% aspen	Composition	POTR	A	.003	.008	.367	.718
(Populus tremuloides)			P	.002	.004
% herbaceous cover	Composition	HERB	A	.475	.129	1.492	.152
			P	.403	.087
%bare ground	Composition	BARE	A	.102	.049	0.548	.590
			P	.090	.049
% rock or scree	Composition	ROCK	A	.124	.093	0.023	.982
			P	.123	.103
% downed woody debris, dead shrubs, etc	Composition	DEBRIS	A	.026	.021	-0.721	.480
			P	.032	.016
% ground covered by snow	Composition	SNOW	A	.016	.037	1.383	.183
			P	.000	.002
% all pines^b	Composition	PINE	A	.022	.028	0.713	.485
			P	.016	.014
% all firs^b	Composition	FIR	A	.104	.049	-3.756	.001
			P	.181	.045
estimated % shrub cover	Composition	PERSH	A	10.443	5.782	-1.366	.188
			P	14.426	7.385
estimated % tree cover	Composition	PERTR	A	15.172	5.321	-2.674	.015
			P	22.193	6.569
average height of nearest shrub in each quadrant	Structure	SHHT	A	.439	.076	-0.703	.491
			P	.464	.087
average area of nearest tree or cluster of trees in each quadrant	Structure	TRAREA	A	84.412	69.081	-1.972	.063
			P	177.202	133.191
average area of nearest shrub or cluster of shrubs in each quadrant	Structure	SHAREA	A	4.937	3.286	-0.615	.546
			P	5.892	3.776
average height of nearest tree in each quadrant	Structure	TRHT	A	1.087	.148	-1.992	.061
			P	1.345	.384
average distance to nearest tree in each quadrant	Structure	TRDST	A	9.501	2.441	1.729	.100
			P	7.780	2.122
average distance to nearest shrub in each quadrant	Structure	SHDST	A	11.858	6.622	1.943	.067
			P	4.423	3.941

^aDegrees of freedom for all tests were 19.

^bThese variables were created by combining other variables. They were not directly measured in the field.

Table 4. Results of logistic regression an individual variables. ** denotes variable kept for further analysis.

Variable entered	Variable function	Significance of β	Exp(β)	95% C.I. for Exp(β)
				Lower	Upper
ELEVATIO		.994	1.000	.995	1.005
SLOPE		.318	1.027	.975	1.082
ASPECT		.456	.996	.984	1.007
**TREE_HEI	structure	.088	30.967	.603	1591.431
JUNIPER		.353	19777.311	.000	2.3 E+13
PEFL		.760	.002	.000	1.2 E+14
RIBES		.596	9.4 E+22	.000	6.53 E+107
SALIX		.728	66.577	.000	1.3 E+12
SHCA		.397	.000	.000	2.8 E+20
PIAL		.920	.051	.000	8.9 E+23
**PIEN	composition	.051	.000	.000	1.613
**HERB	composition	.159	.001	.000	2.852
BARE		.570	.005	.000	513858.1
ROCK		.981	.894	.000	8345.626
DEBRIS		.459	1.2 E 08	.000	3.2 E+29
PINE		.470	.000	.000	6.3 E+11
**FIR	composition	.017	1.4 E+15	514.694	3.9 E+27
**PERSH	composition	.188	1.104	.953	1.280
PERTR		.032	1.215	1.017	1.451
SHHT		.474	69.522	.001	7756398
SHAREA		.526	1.088	.839	1.410
**TRAREA	structure	.090	1.010	.998	1.022
**TRDST	dispersion	.118	.691	.435	1.098
**SHDST	dispersion	.096	.839	.682	1.032

Figure 1. Diagram of sampling at a hypothetical site. b represents the location of Timberline Sparrows, X the points sampled as used, and Y the points samples as unused. Distances between any two X’s or any two Y’s = 50-m and no X is > 50-m from a b. No Y is < 150-m from an X.

N

70.0 km

Figure 2. Sites that were sampled where Timberline Sparrows were present are shown in yellow. Site identification code is shown to the right of each site. Unused study sites are not shown because in most cases, they cannot be differentiated from their used counterpart at this scale.

Figure 3. A depiction of the sites as located in relationship to the 1^st and 2^nd principal components. Component 1 is a gradient from taller trees and fir to more herbs, and trees and shrubs more scattered. Component 2 is a gradient from spruce and shrubs to larger patches of trees, taller trees and more fir. One site, dm, was not used in this analysis. Its inclusion does not alter the results perceptibly.

Back to Abstract