---

---

Introduction

Forest habitats in North Dakota occupy about 2% of the State's area (Stewart 1975). The Pembina Hills in Cavalier and Pembina counties represent a unique forest community because it is one of only two major deciduous forest ecosystems in the State. The heavily wooded Turtle Mountains in Bottineau and Rollette counties account for about half of the remaining forest area in North Dakota. Contributing to the uniqueness of the Pembina Hills is the fact that the Pembina River basin is a meeting area for three major North American ecosystems: Aspen Parkland, Tall Grass Prairie, and Eastern Deciduous Forest (U.S. Army Corp of Engineers 1976).

The breeding avifauna of the Pembina Hills is a unique aspect of North Dakota's resources. Stewart (1975) listed 14 species as primary indicators of this region, and 33 species as secondary indicators. Among these, ruffed grouse (scientific names of bird species are given in the Annotated Species List), broad-winged hawk, ruby-throated hummingbird, yellow-bellied sapsucker, yellow-throated vireo, Philadelphia vireo, northern waterthrush, mourning warbler, and scarlet tanager are restricted primarily to the eastern third of the State, and attain their greatest breeding densities in the Pembina Hills or in the Turtle Mountains farther West.

Fig. 1. Geographic location of the study area within the watershed of the Pembina River.

The objectives of the present study were to determine densities and estimate the population size of breeding bird species, and describe the vegetative characteristics of the major forest and shrub habitats of the Pembina River gorge. Information is needed to assess the potential impact to the area of the Pembilier Dam, proposed by the U.S. Army Corps of Engineers to control floods downstream on the Pembina River. The resulting reservoir would permanently inundate at least 3.2 km² of woodland. Overall, about 12.1 km² of land would be affected. The reservoir would permanently inundate 15.3 km of free-flowing river, and 33.8 km of river would be inundated when under floodwater storage conditions (U.S. Army Corps of Engineers 1976).

Study Area

Physiography

The Pembina River gorge lies entirely within the Drift Plain physiographic region of North Dakota (Arndt 1975). Land area included in the Pembina River drainage totals about 10,032 km² in southern Manitoba and northeastern North Dakota. The geographic location of the study area is shown in Fig. 1. Predominant bedrock of the region includes the Cretaceous age Carlisle and Niobrara formations which are made up primarily of shale overlain by deposits of glacial till occurring in varying depths. The Pembina Escarpment which forms the boundary of the Drift Plain and Red River Valley is the dominant physiographic feature of the region. Land use of the study area was primarily agricultural; farming activities were restricted to the floodplain.

The climate is continental, with hot, dry summers and cold winters. Mean annual precipitation is about 50 cm (Simpson 1929). Mean June and July temperatures are 17.5 and 20.6° C (U.S. National Climatic Center 1981). In 1981, the year this study was conducted, precipitation for June was 5.2 cm above normal, and in July 3.7 cm below normal. Precipitation fell on 19 days in June and 10 days in July. Temperatures during June and July 1981 were 2.4 and 1.5° C below normal.

Habitats

Our study area extended upstream from the proposed dam site on the Cavalier and Pembina county lines to the Canadian border (Fig. 2). Five wooded habitats were predominant in the study area: bur oak (Quercus macrocarpa) forest, quaking aspen (Populus tremuloides) forest, lowland forest, willow (Salix sp.) shrub, and serviceberry (Amelanchier alnifolia) thickets. The bur oak type was the most extensive.

Fig. 2. Map of the Pembina Hills Study Area. The stippled area represents the full pool contour of the proposed impoundment.

Bur Oak Forest

Bur oak forest made up about 85.5% of the study area. It occurred on practically all topographic features of the study area except those directly adjacent to the Pembina River and its major tributaries. Predominant trees included bur oak and green ash (Fraxinus pennsylvanica). Subdominant tree species were quaking aspen, cottonwood (Populus deltoides), and American elm (Ulmus americana). Important components of the shrub layer included beaked hazelnut (Corylus cornuta), pin cherry (Prunus pennsylvanica), and serviceberry.

Quaking Aspen Forest

Quaking aspen forest occupied about 7.4% of the study area. Aspen occurred in "clones" of nearly homogeneous stands and in mixtures with bur oak. Predominant trees also included box elder (Acer negundo), white birch (Betula papyrifera), and American elm. Important shrub layer species were wolfberry (Symphoricarpos occidentalis), green ash saplings, beaked hazelnut, red-osier dogwood (Cornus stolonifera), and poison ivy (Toxicondendron rydbergii).

Lowland Forest

Lowland forest was restricted to lands adjacent to the Pembina River. This type was probably more extensive in the recent past, but has been reduced considerably by clearing for agricultural purposes along the valley floor where only small isolated stands remain. Currently, lowland forest makes up about 1% of the study area. Predominant tree species included American elm, cottonwood, box elder, black ash (Fraxinus nigra), and peach-leaf willow (S. amygdoloides). In drier portions of the forest, bur oak and hackberry (Celtis occidentalis) were occasionally encountered. Important in the shrub layer were red-osier dogwood, chokecherry (Prunus virginiana), and highbush cranberry (Viburnum opulus). Lowland forest sites typically had well-defined canopy and understory layers.

Willow Shrub

Willow shrub made up about 5% of the study area and occurred in areas directly adjacent to the Pembina and Little South Pembina rivers. It was probably an early successional stage of regenerating lowland forest. The vegetation was dominated almost exclusively by peach-leaf willow. Red-osier dogwood and serviceberry were sub-dominant components of the community. Because of the early age of this habitat, no well-defined shrub layer or canopy had developed.

Serviceberry Thicket

Serviceberry thicket was rather limited, occupying about 1% of the area primarily along the drainage of the Little North Pembina River. The area occupied apparently had been recently burned, and serviceberry was the early successional stage that became dominant. Serviceberry was the prevalent shrub species; red-osier dogwood and highbush cranberry occurred in scattered locations. As with the willow shrub community, no well-defined shrub or canopy layer had developed because of the young age of the habitat.

Methods

Census Plot Selection

Eleven census plots were established in the study area and an additional plot was located in the Deltaic Sand area south of Walhalla, Pembina County. Wherever possible, we attempted to locate plots in relatively homogeneous stands of habitat to allow for accurate estimates of bird population sizes by habitat type. Thus, in several instances, plot size was smaller than normally recommended. Area of each major census plot was determined from aerial photographs (scale 1:15,840).

Census Techniques

Breeding bird populations were censused during 7 June to 2 July 1981 by the spot-mapping method (Williams 1936). Each of 11 census plots was visited 6 times during the count period. Count times were usually from local sunrise to no later than 1100. Several early evening counts were made to census thrush (Turdidae) populations. The starting and ending point in each plot was changed with each count to ensure uniform coverage during all periods of the morning or evening.

In addition to census plots, periodic investigations of major habitats throughout the study area were made to record the presence of additional species not found on census plots. These data were used to compile a more complete species list for the study area. The results of these investigations are provided in the Annotated Species List.

Observations of bird use of the study area were concentrated primarily during the breeding season. Several other trips were made to gather data on species using the area during migration. Data from the migration periods are incomplete and should not be considered as estimates of population density.

Data Analysis

Populations of breeding birds were estimated from the number of indicated pairs encountered on census plots. Densities reported for each species are expressed as pairs per square kilometer. Population estimates for breeding birds follow the methods described by Faanes (1982), including the calculation of highest probability density (HPD) intervals (Johnson 1977). Bayesian confidence intervals were used because the number of breeding pairs is known to be non-negative, and because HPD intervals are more precise. The species diversity index, H' (Tramer 1969), was calculated for each habitat type.

Breeding bird densities were calculated by determining the number of territories on each plot. After completing the censuses, composite maps delineating territorial boundaries were compiled for each species on each census plot. For individuals recorded 3-4 times, or those having only a portion of their territory on the plot, a density of 0.5 pair was given. Species recorded 2 times on a plot were assigned a density value of 0.3 pair per plot. Species recorded once were typically considered to be visitors, i.e., probably breeding in the area but not on the census plot, and were assigned a density of 0.1 pair per plot. These density figures, especially for diurnal raptors, appeared to be close to the number of presumed breeding pairs encountered throughout the study area.

Vegetation

We quantified vegetation on each census plot using the methods of James and Shugart (1970). Habitat variables evaluated on each census plot are summarized in Table 1. These data were used to determine important components of the habitat occupied by each species in the study area. Vegetation sampling was conducted on each bird census plot during 21-22 July.

Results and Discussion

Vegetation Analysis

Quaking Aspen Forest

Six tree species were recorded in this habitat. Mean values among characteristic tree species indicated that quaking aspen was highest in both relative dominance (57.2%) and relative density (54%). Dead quaking aspen trees, important for natural nest cavities and perch sites, made up 18.7% of dominance and 16% of density. Green ash was third in density (11%) and American elm was third in dominance (9%).

This habitat supported the highest number of trees per hectare among all types (Table 1), especially "pole"-stage trees (7.6-22.8 cm in diameter). This relates both to the growth form of quaking aspen and the relative age of the forest. Most quaking aspen in the study area was approaching old age. Canopy cover of 47.5% and the relatively high ground cover (77%) were related to the density of dead quaking aspen in the forest. Subsequent increases in penetration of sunlight to the forest floor resulted in high groundcover values.

Shrub density in quaking aspen was second lowest among all forest types and resulted in substantial areas of growth among herbaceous plants in the ground level. Most of the shrubs were serviceberry, highbush cranberry, and green ash. The distribution of shrubs in the forest was related to the presence of openings in the canopy caused by dead trees. Percent ground cover was third highest in aspen forest. The low number of shrubs was one indication that the forest had recently reached maturity, and suggested that the forests on the study area were approaching the climax stage.

The herbaceous layer was composed primarily of poison ivy, various forbs, and Pennsylvania sedge (Carex pennsylvanica). This provided a double layering effect in the ground layer with an upper "canopy" of poison ivy about 0.5 m high, and a lower layer composed of forbs and grasses. Thus, a mosaic of several levels occurred within the ground and shrub layers.

Lowland Forest

Five tree species were recorded in the lowland forest plots. American elm was highest in relative dominance, but box elder was highest in relative density. Thus, although there were more box elder trees present (density), the size of the American elm (dominance) trees exerted greater influence on the volume of woody vegetation present in the forest. Relative density of American elm was only 16.7%, suggesting that few were in the forest, but those present were large.

Black ash was second highest in relative density and third in relative dominance. This species is rare in North Dakota; its distribution is restricted to the extreme northeastern corner of the State. Cottonwood made up 13.2% of relative dominance and 14.2% of density. Peach-leaf willow was a relatively unimportant component of this community, making up 1.3% of dominance and 3.0% of density.

Lowland forest supported the second highest number of trees per hectare. The distribution of trees within various diameter classes in the lowland forest showed the greatest amount of variation of all habitat types; trees ranged from 7.6 to 61 cm in diameter. All trees greater than 22.9 cm in diameter were American elm. High basal area and dominance by American elm is a regular characteristic of lowland forests east of the Great Plains.

Canopy cover was greatest in lowland forest, and percent ground cover and shrub density were the lowest among all habitats. The high percentage of canopy cover is characteristic of mature lowland deciduous forests. The low values for shrub density and percent ground cover in lowland forest are related to reduced sunlight reaching lower layers of the forest. The result is a well-defined canopy layer and a well-defined ground layer, with a poorly developed shrub layer.

The ground layer was dominated by an extensive growth of wood nettle (Laportea canadensis). Individual wood nettle plants were widely spaced, and leaves were large and broad in response to reduced sunlight. Few herbaceous plants grew below the layer of nettle leaves, which resulted in many areas of bare soil.

Mean vegetation height was highest in lowland forest. This was a reflection of the growth form of typical lowland forest tree species and was related to the generally mesic to wet-mesic character of the community.

Bur Oak Forest

The bur oak forest represented the most xeric habitat in the study area. Distribution of this type was primarily on drier south-facing slopes, and, in many instances, was associated with scattered openings supporting native prairie grasses.

The mean number of trees per hectare in bur oak forest ranked third among all habitat types (Table 1). Considerable variation existed within the stands studied. This variation was probably caused by past fire activity which resulted in the vegetation of one plot supporting only 79 trees per hectare.

Distribution of trees within each diameter size class reflected the low number of trees in this habitat and the relative age of the forest. Over 60% of the trees were less than 15 cm in diameter. The mean of 7 trees/ha in the 22.9- to 38.1-cm class reflected older bur oak trees.

A dramatic increase in the shrub density was related primarily to the relatively young age of this forest type. The mean shrub density in bur oak forest was over 3 times greater than in quaking aspen forest. The shrub layer was dominated primarily by bur oak saplings ranging from 1 to 2.5 m tall. Serviceberry was of minor importance. Regeneration of green ash was not apparent.

Canopy cover was 27%, which was about 64% lower than the mature lowland forest. Percent ground cover was second highest among all habitats and reflected the open nature of this forest type. Mean height of canopy was the lowest of all habitats studied. Range of canopy height was 3.0-8.3 m, another indication of the young age of the forest and probably poorer growth on drier soils.

Willow Shrub

The willow shrub community was composed almost entirely of peach-leaf willow. The mean number of trees per hectare was lowest among all habitats supporting trees (Table 1). Basal area and trees by diameter class reflected the low frequency of trees in this habitat. Willow shrub occurred exclusively in a narrow band adjacent to the Pembina River. Mean height of vegetation was 2.7 m. Height of peach-leaf willows was uniform throughout the study area, which suggested a relatively even-aged community.

Shrub density was second highest among all habitats. Most shrubs were in the 2.5- to 5.0-cm size class. Percent canopy cover was 5.5% and ground cover was 61.5%. Ground layer vegetation consisted primarily of various grasses and sedges; there were few forbs present. In other habitats studied, there was a direct relation between canopy and ground cover. However, in the willow shrub community the relation was reversed. Although there was little canopy cover, ground layer vegetation was poorly developed. This was probably related to the sandy soils prevalent throughout much of the floodplain.

Serviceberry Thicket

This distinct community occurred within the floodplain of the Pembina and Little North Pembina rivers, but on better-drained soils that appeared to contain higher amounts of silt and loam. No trees greater than 3 m high were identified in the area studied (Table 1). Percent ground cover was the highest of all habitats studied. Ground-layer vegetation consisted of several grasses and small raspberry bushes (Rubus sp.). Shrub density was the highest of all study sites and the mean vegetation height was lowest. Shrub density consisted primarily of serviceberry, although red-osier dogwood occurred in scattered patches.

Avian Use

In the study area 120 bird species were recorded in 1981. The most recent compilation of North Dakota birds indicates that 353 valid species have been recorded in the State (Faanes and Stewart 1982). Thus, 34% of the total avifauna of the State was recorded in the Pembina River gorge during April-September 1981. The breeding avifauna consisted of at least 79 species for which evidence of nesting was obtained.

Stewart (1975) listed 196 species that nest, have nested, or probably nest in North Dakota. The 79 breeding species recorded in the Pembina River gorge represented 40.3% of the known breeding avifauna of the State. One major factor limiting the number of breeding birds in the Pembina River gorge is the absence of prairie wetlands, which are common across most of the State. Stewart (1975) listed about 152 bird species that nest regularly in the eastern half of North Dakota (east of 100° W longitude). Of these, 44 (or 29%) are associated with prairie wetlands. Removing these from the total, there are 108 breeding species that are not associated with wetlands in eastern North Dakota. The Pembina River gorge supported 73.1% of this avifauna of eastern North Dakota in 1981.

Breeding Bird Populations

Fifty-seven breeding bird species were recorded on the census plots. Records were obtained for an additional 22 species that probably nested in the study area. These species were recorded during supplemental surveys throughout the study area. Among the 57 species recorded in census plots, data for 7 were not used in the population estimates or in calculation of various habitat variables. These species (redtailed hawk, broad-winged hawk, American woodcock, belted kingfisher, golden-winged warbler, orange-crowned warbler, and Canada warbler) occurred locally in the study area and populations were represented by 1-15 pairs. However, because expansion of the population data would result in grossly overestimated figures, these species were not included.

The breeding avifauna was dominated by the order Passeriformes, but included nine non-passerine species represented by 4,619 pairs, or about 6.1% of the total. Orders other than Passeriformes were Columbiformes (2,422 breeding pairs), Apodiformes (1,730), Piciformes (254), Galliformes (126), Cuculiformes (71), Accipteriformes (8), and Meleagridiformes (8).

The breeding avifauna included 21 families. The Parulidae (wood warbler) were most abundant with seven species and an estimated population of over 28,000 breeding pairs (Table 2). The sparrow family (Fringillidae) was second in abundance with seven species and an estimated population of over 17,500 pairs. Five families representing 23 species made up about 80% of the population.

Based on census plot data, the total breeding population in the study area was estimated at 75,972 breeding pairs (HPD interval, 58,166-79,899 breeding pairs). Twenty species made up about 90% of the population; five species made up over 51% of the total (Table 3).

The yellow warbler was the most abundant breeding species, making up 19.4% of the total breeding population. This species was recorded in all habitat types investigated, reaching highest density in the willow shrub community. The American redstart was second in total abundance. The estimated breeding population of 7,973 pairs was 10.5% of the total avifauna breeding population. Breeding American redstarts occurred only in the more mature forest habitats. Largest breeding densities were in bur oak forest. The clay-colored sparrow was third in total abundance. The estimated breeding population of 7,226 pairs represented 9.5% of the total. Largest breeding densities were in the serviceberry community and in young bur oak forest.

Included in the total breeding avifauna were species not recorded on census plots. These species included waterfowl, shorebirds, swallows, and raptors, which normally have either large, wide-ranging territories or nest in colonial or semi-colonial associations. Although not of numerical significance, these species were unique components of the avifauna. The total breeding population of these species, based on intensive observations throughout the area, was estimated at 290 pairs. This represented less than 0.4% of the projected population of breeding birds.

Habitat Distribution of Breeding Birds

Lowland Forest

The highest calculated species diversity index and second highest mean density of breeding birds per square kilometer occurred in lowland forest (Table 4). Thirty-two breeding bird species were recorded on the lowland forest census plots and 12 species reached their highest mean density per square kilometer in this habitat, including American robin, blackbilled magpie, great crested flycatcher, eastern kingbird, eastern phoebe, hairy woodpecker, least flycatcher, mourning dove, red-eyed vireo, rose-breasted grosbeak, ruffed grouse, and warbling vireo. The black-billed magpie was recorded only in this habitat.

Among breeding birds, the yellow warbler was the most numerous species in lowland forest, making up 15.4% of total abundance. The veery (7.7%) and least flycatcher (7.2%) were second and third in abundance. Only six species accounted for about 50% of total abundance, and 14 species made up 80% of the population. The warbler family represented by four species was most numerous, making up 18.5% of total abundance. The flycatchers represented by five species accounted for 17.9% of abundance.

Case (1964) reported a similar density from Illinois floodplain forests, where six species (indigo bunting, American redstart, red-eyed vireo, cardinal (Cardinalis cardinalis), great crested flycatcher, and eastern wood pewee) accounted for 50% of total abundance. Fourteen of the 41 species recorded in Case's study area are typical of the southeastern United States. Stamp (1978) reported a mean density of about 1,689 pairs/km² from riparian cottonwood forests in south-central Arizona. Finches and doves collectively made up over 35% of total abundance on Stamp's study area.

The vertical distribution of breeding birds in lowland forest was remarkably uniform. Among 30 species that could be classified to vertical stratification, 12 were characteristic of the shrub layer, 10 were characteristic of the mid-level, and 8 were primarily canopy-nesting species. This degree of stratification was one factor responsible for the high species diversity exhibited in lowland forest.

Quaking Aspen Forest

The estimated density of breeding birds in quaking aspen forest was the lowest of all habitats studied (Table 4). Species diversity ranked second and species richness ranked first. Thirty-eight breeding bird species were recorded on quaking aspen census plots. Ten species (wild turkey, blackbilled cuckoo, downy woodpecker, yellow-bellied sapsucker, eastern wood pewee, black-capped chickadee, white-breasted nuthatch, Philadelphia vireo, black-and-white warbler, and northern oriole) reached their highest mean breeding density in this habitat.

Least flycatcher was the most numerous breeding bird species in quaking aspen (about 12% of total abundance). Eight species (least flycatcher, veery, yellow warbler, mourning dove, red-eyed vireo, black-and-white warbler, American redstart, and American goldfinch) made up over 50% of total abundance. About 85% of the population was made up by 22 species. The warbler family, represented by six species, was most numerous (21.6% of the population). The flycatchers represented by four species made up about 29%, thrushes with two species accounted for 9.3%, and vireos with three species made up 9.1%.

Quaking aspen forest supported seven species that exhibit a limited breeding range in North Dakota. Most unusual were Philadelphia vireo and mourning warbler. Both species are restricted in the State almost exclusively to the Pembina Hills region or the Turtle Mountains. Five species (wild turkey, black-billed cuckoo, eastern wood pewee, Philadelphia vireo, and northern oriole) occurred only in quaking aspen forest in the study area.

Breeding bird densities in mature and pole-stage aspen forests are variable across broad geographic areas (Back 1979). Erskine (1969) reported 326 pairs/km² from a mature New Brunswick aspen forest. The ovenbird was the most abundant species, making up nearly one-third of the breeding population. Sixteen species were found on Erskine's study area including seven that did not occur in the Pembina Hills. Erskine (1972) reported 520 pairs/km² from a mature aspen forest in west-central Manitoba. Fifteen species were recorded including four not found on Pembina Hills aspen forest plots.

Densities ranged from 225 pairs/km² in Saskatchewan to 852 pairs/km² in North Dakota among five pole-stage aspen forests analyzed (Back 1979). In both pole-stage and mature aspen forests, the ovenbird was the most abundant breeding bird species. The veery and red-eyed vireo were second and third in abundance in pole-stage forests. The red-eyed vireo and least flycatcher were second and third in abundance in mature aspen forest. Breckenridge (1956) pointed out that the degree of openness in the canopy was an important component of least flycatcher habitat. Canopy cover in quaking aspen forests of the Pembina Hills was 47.5%, thus providing a degree of openness almost identical to the conditions described by Breckenridge. Erskine (1972) reported that least flycatchers on his Manitoba study area were restricted to areas of mature quaking aspen.

An important factor influencing the high species richness in quaking aspen was the well-developed growth of saplings in the shrub layer. Erskine (1972) attributed the exceptionally high density of American redstarts on his study area to dense growth in the shrub layer. Nearly 40% of the breeding birds recorded in quaking aspen forest on our study area were associated with the shrub layer.

Dead quaking aspen accounted for about 19% of forest dominance in the Pembina Hills. Dead trees are a source of natural cavities and are frequently mentioned for their importance in "snag" management. Dead trees also provide frequent openings in the canopy that are regularly used by eastern wood pewees for song perches. About 18% of the breeding avifauna of quaking aspen forest were species that typically nest in cavities. The two most numerous species were black-capped chickadee and yellow-bellied sapsucker.

Bur Oak Forest

The second highest species richness among habitats was recorded in bur oak forest. This habitat supported the lowest species diversity and one of the lowest total breeding densities (Table 4). Thirty-four breeding bird species were recorded on bur oak census plots, and 10 species reached their highest mean density per square kilometer there: American redstart, ruby-throated hummingbird, blue jay, indigo bunting, scarlet tanager, mourning warbler, vesper sparrow, yellow-throated vireo, common crow, and house wren. Among these species, indigo bunting, vesper sparrow, and yellow-throated vireo were restricted to bur oak forest.

Among breeding birds, the yellow warbler was the most numerous species in bur oak forest, making up 18.7% of total abundance. American redstart was second in abundance (12.5%) and clay-colored sparrow third (10.5%). Five species made up 50% of total abundance and 16 species made up 80% of the population. The warbler family, represented by five species, made up 39.3% of the population in bur oak. Sparrows (three species) made up 13.6% of the population, finches made up 9.1%, and vireos accounted for 7. 1% of the population.

Probst (1979) observed that species richness in oak forests ranged from 9 to 29 species, with a mean of 21.2. Mean breeding density ranged from 160.5 to 963.3 pairs/km², with a mean of 548.3. Probst also noted that mesic stands can have higher densities than xeric stands. Anderson (1970) reported a mean density of 1,358.5 pairs/km² in Oregon white oak (Quercus garryana) forests. Anderson suggested that the stage of succession of Oregon white oak forests was an important factor in supporting the exceptionally high density of breeding birds he observed.

Anderson (1970) also argued that western oak forests supported lower numbers of permanent residents among breeding birds. MacArthur (1959) reported that about 90% of eastern oak forest breeding bird species were migratory. About 82% of the breeding bird species occupying bur oak forests in the Pembina Hills were migratory.

Density of breeding birds in bur oak forests was related to several vegetative factors including successional stage, canopy cover, and density of the shrub layer. Eighteen species on our study plots were indicative of early successional stage vegetation. Also, three species not included in the population analysis for bur oak forest (American woodcock, golden-winged warbler, and orange-crowned warbler) were recorded only in early successional-stage bur oak forest. Seven of the 10 species reaching maximum density in bur oak were characteristic of early successional-stage vegetation.

The diversity of bur oak forest breeding birds was divided equally between ground-foraging seedeaters and canopy or sub-canopy flycatchers (nine species each). Also, an equal number of foliage gleaners and ground-layer insect eaters (five species each) were observed. Bond (1957) reported that foliage gleaners and ground-feeding insectivores were more common in mesic forests and plant eaters were more common in xeric forests.

Probst (1979) found that the highest number of birds in oak forests were ground foragers. Predominant among these were the thrushes, rufous-sided towhee, and ovenbird. Ground cover was second highest in bur oak forest and serviceberry was an important shrub layer species. Although quantitative data are lacking, sizable populations of flying insects were observed in the shrub layer of bur oak forest. During late June, an abundance of ripe seeds was present, which was heavily exploited by most ground- and shrub-layer foraging species.

Willow Shrub Community

The highest mean density of breeding birds and the third highest species diversity were recorded in the willow shrub community (Table 4). Nineteen breeding bird species were recorded on willow shrub census plots. Twelve species reached their highest mean density per square kilometer in this habitat: alder flycatcher, bank swallow, barn swallow, brown-headed cowbird, cedar waxwing, common yellowthroat, gray catbird, northern waterthrush, red-winged blackbird, song sparrow, veery, and yellow warbler. No species was restricted entirely to this habitat.

Among breeding birds, the yellow warbler was the most numerous species in willow shrub, making up 25.7% of total abundance. The alder flycatcher (12.9%) and song sparrow (10.6%) were second and third in abundance. These three species collectively accounted for about 50% of the total breeding population in the willow shrub community. Eight species accounted for about 80% of the population. The warbler family, represented by three species, made up 38.5% of the population. Sparrows and flycatchers, each represented by two species, made up 13.6 and 13.2% of the population.

The breeding bird density recorded in the willow shrub community was nearly twice as large as any other habitat type in the study area. The general pattern of succession in avian communities suggests that diversity and populations increase from bare ground and grasslands through early seral stages and non-mature forests to maximum diversity in mature mixed forests. Diversity and species abundance decline when forests approach climax stage. Along this gradient, the willow shrub community would occupy one of the lower levels of succession.

Because the willow shrub community is poorly studied, few comparative data are available. Baumgartner and Slack (1977) studied a structurally similar shrub community in New York. On their plot (9.3 ha), 15 species were recorded and the mean density was 892 pairs/km². A second New York plot of similar size examined by Lucid et al. (1977) supported 16 species and 505 pairs/km². On both of these New York study areas, the yellow warbler was overwhelmingly the most numerous breeding bird species. On Baumgartner and Slack's area, the yellow warbler made up 31.3% of the population and on the plot studied by Lucid et al., about 20% of the population. Six species were common to both our North Dakota study area and the New York plots. The willow flycatcher (Empidonax trailii) replaced the alder flycatcher in New York, although abundance on one plot was practically identical to the mean values reported here.

The yellow warbler population occupying willow shrub in the Pembina Hills is probably indicative of maximum habitat exploitation by a species. Morse (1966) reported that yellow warblers on his study area favored wet areas supporting alders (Alnus sp.) and willow, 1 to 4 m high. Kendeigh (1941) reported exceptionally high yellow warbler densities in wolfberry thickets associated with prairie grasslands. An important component of these habitats was the presence of numerous elevated song perches. As Morse (1966) reported, usually the only other Parulid associated with the yellow warbler was the common yellowthroat. The willow shrub community in the Pembina Hills supported three Parulids: yellow warbler, common yellowthroat, and northern waterthrush. The yellowthroat and waterthrush forage at ground level.

Willow shrub in the Pembina Hills was a prominent edge habitat. The importance of ecotonal areas to breeding bird species abundance has been reported extensively. On our study area, willow shrub typically occurred at the edge of mature lowland forests, in wet areas of overlap between quaking aspen and lowland forest, and in areas bordering the Pembina River. Because most willow shrub communities existed as patches of similar vegetation adjacent to more advanced seral stages, many breeding bird species occupied this habitat as a part of their overall breeding territory. This overlap in habitat use, especially evident in the song sparrow, gray catbird, common yellowthroat, and veery, greatly increased total avian use, and the perceived number of breeding territories was correspondingly higher.

Serviceberry Thicket

The lowest number of breeding bird species was recorded in the serviceberry thicket (Table 4). Mean breeding density per square kilometer and species diversity were second lowest among all habitats studied. Fourteen breeding bird species were recorded in serviceberry thickets. Four species reached their highest mean density per square kilometer: clay-colored sparrow, American goldfinch, solitary vireo, and chestnut-sided warbler. The vireo and warbler were recorded only in this habitat.

Among breeding birds, the yellow warbler was the most numerous species, making up 28.6% of total abundance. Clay-colored sparrow was second and American goldfinch was third. These three species collectively accounted for nearly 62% of the total population. Three species of the warbler family made up about 39% of the population in serviceberry thickets, two species of sparrows made up 19.8%, and the one finch accounted for over 14%.

The serviceberry thicket is a community occurring in limited extent at the forest-prairie ecotone. Because of its limited extent, virtually no previous field work has been conducted in this community. The vegetative structure characteristics are similar to other early successional stage habitats, so generalized statements and comparisons can be made.

The breeding birds of this community were dominated by ground-foraging species including gray catbird, common yellowthroat, cedar waxwing, veery, song sparrow, and brown-headed cowbird. Ground-layer flycatchers were represented by yellow warbler, least flycatcher, solitary vireo, and chestnut-sided warbler. The lack of trees in this community precluded use by higher-level flycatching species, although eastern kingbirds were regularly observed hunting over adjacent areas of similar habitat.

Serviceberry thicket appears to be an intermediate, early seral stage in the succession of bur oak forest. Although serviceberry and red-osier dogwood were the predominant shrub species present, chokecherry and sapling bur oak were occasionally encountered. Examination of plant data for the bur oak forest revealed that serviceberry and red-osier dogwood were the most numerous species in the shrub layer. Thus, through normal succession, this community can be expected to develop bur oak forest as the next seral stage.

Breeding bird composition of serviceberry thickets was very similar to bur oak forest; 12 species were common to each habitat. The rufous-sided towhee was conspicuously absent from serviceberry thickets, although it was one of the more common species in bur oak. Shrub density was one-third greater in serviceberry thickets than in bur oak, resulting in a very dense vegetation layer close to the ground. Ground cover was 17% more open in bur oak, which probably allowed more foraging space for towhees to use.

Habitat characteristics of yellow warbler territories, as discussed earlier, would suggest that although elevated song perches were generally lacking, serviceberry thickets provided most of the important components of yellow warbler breeding habitat. A primary characteristic of clay-colored sparrow breeding habitat is the clumped arrangement of wolfberry patches (Stewart 1975; Kantrud 1981; Faanes 1982). Serviceberry thickets are structurally similar, at least outwardly, to wolfberry clumps. Thus, the heavy use of serviceberry by clay-colored sparrows was not unexpected. Our observations are the first, however, of heavy use of Saskatoon serviceberry by clay-colored sparrows (Kantrud 1981).

Conclusion

The habitats of the Pembina River valley of North Dakota are at the periphery of their distributions. Populations of many species with eastern or western, northern or southern affinities meet there. This was evident both with plant and avian species. Ground-layer vegetation of some quaking aspen forests on north-facing slopes supported wild ginger (Asarum canadense), which is indicative of moist southeastern forests, growing in profusion with Canada mayflower (Maianthemum canadense), a species typical of moist northern mixed coniferous-deciduous forest. Similar associations were evident with bird species. Clay-colored sparrows and black-billed magpies, both primarily western species, occurred on the same lowland forest census plot with eastern species such as mourning warbler and rubythroated hummingbird. Quaking aspen forest supported the Philadelphia vireo, which is characteristic of the Canadian Aspen Parkland, along with clay-colored sparrow and scarlet tanager.

An outstanding feature of the Pembina River gorge forests was their well-developed understory and numerous brushy edges that flourished in the absence of grazing by domestic livestock. Across much of North Dakota, livestock grazing has impacted riparian vegetation. Most of the Little Missouri River forest community outside Roosevelt National Park is heavily grazed. Much of the riparian forest habitat along the remaining unflooded portion of the Missouri River is heavily grazed and regeneration of saplings occurs only in isolated stands. In the extensive forests of the Turtle Mountains, only public lands are not being grazed.

The breeding warblers of the Pembina Hills were among the most unique aspects of the avifauna. Stewart (1975) listed 14 warbler species that occurred in North Dakota during the breeding season. Among these species, 10 were present on our study area. We also found one breeding pair of golden-winged warblers on territory. A nest found 30 June was the first recorded for North Dakota. Thus, 11 warbler species were present during the breeding season. Of the four warblers listed by Stewart (1975) that were not found on our study area during the breeding season, the Tennessee warbler and cerulean warbler (Dendroica cerulea) have been recorded a total of only 8 times during the breeding season in North Dakota; yellow-rumped warblers occur only in pine forests of extreme western North Dakota, and yellow-breasted chat (Icteria virens) occupies dry thickets, generally west of the Missouri River.

A recent examination of North Dakota forest resources (Jakes and Hackett 1981) revealed that the area of commercial forests in the State has declined by over 22,000 ha since 1954. Land clearing was identified as the primary cause for this decline, but forest land lost to reservoir construction was also considered important. The Pembina Hills region was identified by Jakes and Hackett (1981) as one of only three North Dakota areas supporting commercial forests. Forest vegetation and forest soils were also identified as contributing to the State's water quality, and to reduced runoff through infiltration.

Forest resources, especially in North Dakota, are also important for aesthetic reasons. Kantrud (1973) identified two Cavalier County sites as significant natural areas. One of these, the Little North Pembina Gorge, occurred in the study area. Kantrud described the values of that site as "scenic grandeur amid bur oak-American elm-green ash- basswood forest." Important animal species included orange-crowned warbler, rufous-sided towhee, whitethroated sparrow, lynx (Lynx canadensis), and moose (Alces alces). Total area of the site described by Kantrud was 3,887 ha. Among forest habitats in the study area, the presence of black ash was considered unique and unusual. Although black ash is more widespread in the eastern United States, in North Dakota its distribution is restricted to the Pembina Hills region.

Density and diversity of breeding birds was related to the interspersion of forest habitat types. The patterns of distribution shown by several species were in response to habitat interspersion. Black-capped chickadees that nested in bur oak and quaking aspen forest were regularly observed foraging in Saskatoon serviceberry thickets. Eastern kingbirds nested at the edge of bur oak forests and fed over various shrub communities and cropland.

In New Jersey hardwoods, Forman et al. (1976) found that increased forest size was responsible for increased numbers of breeding bird species. These authors postulated that the species increases were through the addition of numerous minimum territory size requirements. Whitcomb et al. (1977) reported similar results from Maryland; they attributed high breeding bird densities partly to the interrelationship of forest fragments sufficient to support large numbers of species including wide-ranging groups such as raptors. Large forest tracts are also necessary to attract and maintain rare or uncommon forest species.

The importance of riparian vegetation as habitat for birds has been well established (Anderson and Ohmart 1979; Tubbs 1980). The yearly loss of riparian vegetation across the United States has been estimated at over 1,000 km² (McCormick, unpubl. rep.). Throughout much of the Great Plains region, riparian habitats are being altered or destroyed to accommodate the increasing demands of society. Tubbs (1980) identified reduced water flows, channelization, and damming of streams for flood control as three important impacts to this community. Stauffer and Best (1980) identified four levels of tolerance to habitat alterations by breeding birds in riparian ecosystems. Among these, bird species with low tolerance to alterations occurred in fewer habitats, and those species were affected most adversely by loss of preferred habitat.

Jakes and Hackett (1981) identified about 5,200 km² of forest vegetation in North Dakota, exclusive of planted shelterbelts and windbreaks. Through 1981, only 424 km² (8.2%) of forest habitat in North Dakota was protected. To maintain maximum species diversity among North Dakota forest birds, large tracts such as the Pembina River gorge deserve protection. Whitcomb (1977) summarized the situation by stating that thousands of contiguous hectares need to be protected to preserve forest-interior bird species. Preservation of large contiguous forest tracts will also preserve forest-edge species by protecting and maintaining maximum interspersion of habitat diversity.

---

Acknowledgments

This research was funded through a cooperative agreement among the U.S. Army Corps of Engineers; the U.S. Fish and Wildlife Service, Division of Ecological Services, Bismarck, North Dakota; and the Northern Prairie Wildlife Research Center. This report benefited from critical reviews by R. J. Greenwood, D. H. Johnson, D. Simpson, S. Young, and M. S. Zschomler. C. S. Shaiffer prepared the figures.

---

References

---

---

NPWRC Home | Site Map | About Us | Staff | Search | Contact | Web Help | Copyright